An updated antennal lobe atlas for the yellow fever mosquito Aedes aegypti

The yellow fever mosquito Aedes aegypti is a prolific vector of arboviral and filarial diseases that largely relies on its sense of smell to find humans. To facilitate in-depth analysis of the neural circuitry underlying Ae. aegypti olfactory-driven behaviors, we generated an updated in vitro atlas for the antennal lobe olfactory brain region of this disease vector using two independent neuronal staining methods. We performed morphological reconstructions with replicate fixed, dissected and stained brain samples from adult male and female Ae. aegypti of the LVPib12 genome reference strain and determined that the antennal lobe in both sexes is comprised of approximately 80 discrete glomeruli. Guided by landmark features in the antennal lobe, we found 63 of these glomeruli are stereotypically located in spatially invariant positions within these in vitro preparations. A posteriorly positioned, mediodorsal glomerulus denoted MD1 was identified as the largest spatially invariant glomerulus in the antennal lobe. Spatial organization of glomeruli in a recently field-derived strain of Ae. aegypti from Puerto Rico was conserved, despite differences in antennal lobe shape relative to the inbred LVPib12 strain. This model in vitro atlas will serve as a useful community resource to improve antennal lobe annotation and anatomically map projection patterns of neurons expressing target genes in this olfactory center. It will also facilitate the development of chemotopic maps of odor representation in the mosquito antennal lobe to decode the molecular and cellular basis of Ae. aegypti attraction to human scent and other chemosensory cues.

Introduction Aedes aegypti is a prolific vector of yellow fever, dengue, chikungunya, Zika and filariasis in tropical and subtropical regions around the world. This highly anthropophilic mosquito species uses its sense of smell to track volatile chemicals present in human skin odor and breath and orientate towards us. For example, carbon dioxide [1] and other volatile constituents of human body odor, including lactic acid [2], trigger behavioral activation and steer directional flight of this mosquito species towards humans to acquire a blood meal [3]. Olfaction also influences other critical aspects of Ae. aegypti biology including oviposition [4], mating [5] and nectar-seeking [6]. Molecular and cellular substrates underlying the Ae. aegypti sense of smell therefore represent key targets for the control of mosquito-borne diseases transmitted by this cosmopolitan and urbanized disease vector.
In the Ae. aegypti adult, olfactory sensory neurons (OSNs) expressing chemoreceptors that detect odorants are primarily localized to peripheral olfactory organs [1,[7][8][9][10] including the antennae, maxillary palps and proboscis. Dendritic sensory nerve endings of OSNs are found in various morphological classes of specialized porous sensilla distributed along the length of these organs. Each Ae. aegypti sensillum typically houses 2-3 OSNs [7][8][9]. The axonal processes of these neurons project centrally to the mosquito brain, to the first center of olfactory information processing called the antennal lobes. The antennal lobes are made up of clusters or foci of synaptic connectivity called glomeruli, where olfactory information is transmitted from OSNs to second order neurons [11][12][13].
In the vinegar fly Drosophila melanogaster, an individual glomerulus is the point at which the axonal projections of OSNs typically expressing a unique complement of ligand-binding chemoreceptors converge [13,14] and synapse with the dendritic arbors of projection neurons (PNs) and local interneurons (LNs) [15]. Each glomerulus is therefore tuned to a unique subset of chemicals. Anterograde dye-filling of peripheral OSNs that respond functionally to the same subset of odors and subsequent tracing of their glomerular projections indicates this pattern is likely to be conserved in mosquitoes [12,16,17].
An initial morphological study of the Ae. aegypti antennal lobe by Bausenwein and Nick [18] denoted the presence of approximately 35 antennal lobe glomeruli in a single Ae. aegypti female examined. However, it was noted that due to the staining methods employed in this study that subdivision of antennal lobe glomerular neuropil in some areas was unclear. Subsequently, Ignell and coauthors [11] estimated that the antennal lobes of this species were comprised of between 49-50 glomeruli using improved immunohistochemical techniques.
A significant volume of the Ae. aegypti antennal lobes was previously ascribed a non-olfactory mechanosensory function by Ignell [11]. This large core of neuropil on the anterior surface of the antennal lobe was annotated as the Johnston's Organ Centre (JOC) based on silver nitrate Golgi staining. This analysis revealed the putative sensory innervation of this antennal lobe region by afferent fibers from the Johnston's Organ (JO), an observation mirrored shortly thereafter by the same authors in Anopheles gambiae [19]. However, a recent reassessment of antennal lobe morphology in Anopheles coluzzii failed to observe the JOC in the antennal lobes of this species, instead identifying discrete glomeruli in this spatial region [20].
To assist basic and applied studies of Ae. aegypti olfaction, we have used improved neuropil labeling and confocal imaging techniques to re-examine the neuroanatomy of the Ae. aegypti antennal lobe. We present three-dimensional (3D) models based on manual segmentation and reconstructions of antennal lobe glomeruli from replicate male and female brains from the LVPib12 Ae. aegypti genome reference strain. To facilitate spatial orientation and stereotypical identification of glomeruli within the antennal lobe, we also present detailed two-dimensional (2D) maps wherein individual glomeruli are named based on their spatial arrangement in this brain region using nc82-and phalloidin-based neuronal staining methods. Finally, we examine antennal lobe neuroanatomy and organization in a recently field-collected strain of Ae. aegypti from the municipality of Patillas, Puerto Rico. Altogether, this work describes a standardized framework for glomerular nomenclature in the Ae. aegypti antennal lobe. This atlas will serve as a useful reference for neurogenetic and functional imaging studies of the antennal lobe in this anthropophilic disease vector.

Mosquito maintenance and rearing
Aedes aegypti LVPib12 [21] and Patillas [24] strains were maintained at 27˚C, 80% relative humidity with a 12 hr light:dark photoperiod. Immature stages were reared using a standardized protocol optimized for yielding synchronous pupation. Briefly, on day 0 (hatching), 500mL of a 1% w/v autoclaved broth solution of finely milled TetraMin (Tetra, 16110) in dH 2 O was placed in a clean polycarbonate pan (Cambro, 24CW135) with covering lid (Cambro, 20CWCH135). Oviposition papers containing desiccated eggs were then completely submerged in the broth solution inside each allocated pan to facilitate larval hatching. On day 1, 500 ml of additional dH 2 O and 1g of finely milled Tetramin was added to the pan to bring the total volume to 1L. On day 2, oviposition papers were removed, and hatched larvae thinned to a set density of 200 larvae in 2L of fresh dH 2 O per pan and fed one coarsely dissociated 1g Tetramin tablet. On day 3, larvae in each pan were fed with an additional 1g Tetramin tablet. Thereafter, larvae in each pan were fed two 1g Tetramin tablets per day until pupation which fully occurred on day 6. After emergence, adult mosquitoes were provided constant access to a 10% w/v sucrose solution.

Ethics statement
For routine colony maintenance, female Ae. aegypti were blood fed using anesthetized mice (Johns Hopkins Animal Care and Use Committee protocol # MO18H391).

Neuronal staining
Neuropil in the central brain of mated, 5-10 day old, male and nulliparous female Ae. aegypti were initially stained with the primary monoclonal antibody nc82 (DSHB, nc82-s, AB_2314866) that labels the pre-synaptic active zone protein Bruchpilot. To dissect brains, mosquito heads were severed from the thorax and tissue fixed for 3 hr at 4˚C in Milonig's buffer: 0.1M PBS pH 7.2 with 4% v/v paraformaldehyde (PFA) as a fixative. Next, adult brains were dissected out into 0.1 M PBS using fine forceps (Dumont No 5, 100 nm tips) to carefully remove the head capsule as well as the pigmented ommatidia over the optic lobes and any floating air sacs connected to the brain. Dissected brains were washed 3 times for 20 min each at room temperature in 0.25% PBST consisting of 0.25% v/v Triton-X-100 detergent as a permeabilizing agent prepared in 0.1 M PBS. The brains were then treated in a blocking solution consisting of 2% w/v normal goat serum (NGS) and 4% v/v Triton-X in 0.1 M PBS and incubated on a nutator overnight at 4˚C. This step was followed by incubation in a primary antibody solution for 3 days at 4˚C consisting of mouse nc82 (1:50 v/v in 0.1 M PBS with 0.25% Triton-X and 2% NGS). Next, the brains were washed three times at room temperature in 0.25% PBST and incubated for 3 days at 4˚C in a secondary antibody solution consisting of goat anti-mouse Cy3 (Jackson ImmunoResearch 115-165-062, 1:200 v/v in 0.1 M PBS with 0.25% Triton-X and 2% NGS). Following three final washes at room temperature for 20 min each in 0.25% PBST, the brains were mounted within the space created by two coverslip (Number 2-170 μm) bridges on glass slides in 20μl of Slow-Fade Gold Antifade Mountant (Invitrogen, S36936), to preserve 3D structure.
As an alternative method to immunohistochemistry, adult brains were stained with fluorophore conjugated-toxin phalloidin to mark cytoskeletal F-actin filaments present ubiquitously across neuropilar structures including OSN axonal processes. Adult heads were first fixed in Milonig's buffer for 3 hr at 4˚C and brain tissue dissected in 0.1 M PBS as described above. After washing 3 times in 0.25% PBST, the dissected brains were incubated in 0.25% PBST with 1/80 (v/v) Alexa Fluor 488 Phalloidin (Invitrogen, A12379) and 2μg/ml DAPI nuclear counterstain (Invitrogen, D1306), for 3 days at 4˚C. Brains were then mounted on glass slides as described and imaged.

Image acquisition settings
Images of nc82 and phalloidin stained brains were taken on a single-point laser scanning, Carl-Zeiss LSM 780 confocal microscope. To capture images of the entire adult brain for both methods, a 10X objective lens (0.3 NA, Plan-Apochromat) was used. To yield images to perform 3D reconstructions of the antennal lobes, images were taken with a 20X objective lens (0.8 NA, Plan-Apochromat). Images were processed using Imaris software (Oxford Instruments).
For nc82-stained brains, a solid-state laser line at 561 nm (1% laser power) was used to excite the Cy3 signal and the gain of the GaAsP detector was set to 600. 100 z-slices, with a zstep size of 1 μm and a 1024 X 1024 pixel size were acquired. For 3D reconstructions, 60 zslices with a z step size of 1 μm were acquired. The power of the 561 nm laser was adjusted to 2% with detector gain at 620. These settings ensured the acquisition of high intensity images, allowing for the accurate de-lineation of glomerular boundaries while scanning through the entire volumes of both antennal lobes, including a few slices above the surface of the antennal lobe, and a few slices below the deep-seated MD1 glomerulus, so as to closely replicate threedimensionality of the AL structure in the reconstructed model. Images were imported in the � . lsm-Zeiss, Zen format into the Amira software program. We generated inverted images of the z-slices, in Image j, first using the gray lookup table (LUT) to get a black and white image, followed by the invert LUT. No brightness and contrast adjustments were made.
For phalloidin stained brains, a 488 nm laser line (2% laser power) was used to excite the Alexa-488 phalloidin fluorophore, with the detector gain at 500. The DAPI stain was excited with a 405 nm diode laser (2% laser power and detector gain at 400). 60 z-slices with a z step size of 1 μm and a 1024 X 1024 pixel size were acquired. Phalloidin and DAPI stained confocal slices were inverted using Adobe Photoshop, without altering brightness or contrast.

Antennal lobe reconstructions
3D reconstructions of the antennal lobe were performed using Amira software (FEI Houston Inc). Confocal files were opened in the '.lsm' format and viewed using the Image Segmentation Editor tool. For each z-slice typically ranging from 0 μm to -60 μm through the anterior-posterior axis of the antennal lobe, olfactory glomeruli were identified by shape and position. We defined the start of the antennal lobe (denoted 0μm) as the first z-slice where the surface of the first landmark and most anterior glomerulus AD1 is located. Next, an image segmentation step was carried out in which the voxels occupied by individual glomeruli were first highlighted using the Amira paintbrush tool and then assigned a unique label and color. After segmentation of the voxels corresponding to all antennal lobe glomeruli, the image volume was surface rendered to create a 3D model to allow visualization of antennal lobe morphology along x-y-z axes.
Volumes of individual glomeruli were obtained (in μm 3 ) using the Material Statistics tool in Amira. We used the ortho slice tool to generate 2D cartoons of all the confocal slices and saved the images in 'tif' format. Each antennal lobe glomerulus was named based on its spatial position relative to other glomeruli. For each strain and staining method used, we classified glomeruli as spatially 'invariant' or 'variant' based on their frequency of identification across combined male and female, left and right antennal lobe datasets. A threshold frequency of �80% was designated for the classification of spatially invariant glomeruli.

Statistics
Based on our 3D atlas of the antennal lobe of female and male LVPib12 Aedes aegypti, we calculated the volume of each annotated glomerulus and the total volume of the antennal lobe (constituting the cumulative volume of all stained glomeruli in a given antennal lobe) from each brain sample. To test whether mean volumes of invariant glomeruli differed significantly within each lobe, we performed a One-Way ANOVA and Tukey's HSD post hoc test to correct for multiple comparisons. To evaluate size differences of homologous invariant glomeruli between the left and right antennal lobe and between sexes, the non-parametric Mann-Whitney U-test was performed. Statistical analysis was performed using GraphPad Prism Software version 8.1.1 (GraphPad Software, Inc.). P< 0.05 were considered significant.
The numerical data used in all figures and tables are included in S1 Data.

Three-dimensional reconstruction of the antennal lobes of male and female Aedes aegypti mosquitoes
To define boundaries of the antennal lobe of the LVPib12 genome reference strain of Ae. aegypti [21], we initially stained whole dissected and fixed mosquito brains using the monoclonal antibody nc82 as a synaptic marker and acquired 1 μm confocal sections of the central brain starting from the surface of the antennal lobes through the central complex. Similar to other dipteran insects such as Drosophila melanogaster [22], the deutocerebrum [11] in Ae. aegypti is composed of two paired neuropils: the antennal lobe olfactory center and the periesophageal neuropils which include the antennal mechanosensory and motor center (AMMC). The left antennal lobe (LAL) and the right antennal lobe (RAL) flank the esophageal foramen and are positioned dorsal to the AMMC (Fig 1A and 1B). The antennal lobes are innervated by the antennal (flagellar) [8] nerves that collectively include the axons of OSNs originating from the third antennal segment [11,16]. In addition, maxillary afferents from the distal segment of the maxillary palp terminate in the antennal lobes [16], supporting its role as the first center in the central mosquito brain to receive signals from the peripheral olfactory organs.
Confocal sections of the LVPib12 antennal lobe revealed that its constituent glomeruli varied both in terms of morphology and volume, and are arranged at different depths relative to i)

PLOS NEGLECTED TROPICAL DISEASES
Updated antennal lobe atlas for Aedes aegypti the anterior to posterior axis of the mosquito body and ii) along the dorsal to ventral neuraxis [22]. To re-ascertain glomerular counts relative to previously published studies, we initially reconstructed the left and right antennal lobes from 5 male and 5 female LVPib12 brains that had been stained with nc82 antibodies.
We used the software program Amira to perform segmentation of the antennal lobe by manually painting over all the voxels corresponding to neuropil representing individual glomeruli, which were identified based on their size, shape, spatial position and depth. We found that each antennal lobe in all LVPib12 brains examined was comprised entirely of discrete demarcated glomeruli and we were unable to identify the wedge shaped, JOC non-glomerular core which was described previously [11]. The number of glomeruli were found to vary slightly across our reconstructions from 80 ± 2 in the female antennal lobe to 77 ± 2 glomeruli in the male antennal lobe (mean ± s.d.) (Fig 1G and 1H).

Spatial arrangement of glomeruli in the Aedes aegypti antennal lobe
We further sought to understand the pattern of glomerular arrangement within the LVPib12 antennal lobe. To facilitate intuitive annotation of glomeruli within the Ae. aegypti antennal lobe based on their spatial position, we decided to apply standard conventions for naming insect brain axes [22] to our updated atlas. We first defined the anterior-posterior (A-P) as the longitudinal body axis which runs from the mosquito head to the thorax. Secondly, we defined the dorsal-ventral neuraxis (D-V) as the brain axis positioned perpendicular to the A-P axis, with the dorsal brain surface found at the top of the mosquito head. Finally, we defined the medial-lateral axis (M-L), with medial features positioned closer to the D-V midline of the brain relative to those that are lateral. Confocal sections of the brain were acquired in the A-P orientation.
Based on spatial position along these standard A-P, D-V and M-L axes, we initially aimed to categorize glomeruli in the male and female antennal lobe into spatial groups, with reference to a previous antennal lobe model that annotated 13 spatial groups of glomeruli in the Ae. aegypti antennal lobe using a non-conventional coordinate system [11]. However, as we identified approximately 30 additional glomeruli in our reconstructions relative to this extant map, we first re-annotated spatial descriptors for each group in our atlas to accommodate these extra glomeruli and better reflect their spatial positions along the standard coordinate axes [22] of the antennal lobe (Table 1). These revised groupings organize the Ae. aegypti antennal lobe into 13 new glomerular clusters that aid in identifying and naming all glomeruli in each lobe. Each of these groups was assigned a specific color in our reconstructed 3D antennal lobe models (Fig 1C-1F).
Viewing the antennal lobe along the A-P axis, on the anterior slices of the antennal lobe we annotated the Antero-Dorsal (AD) group surrounded by the Antero-Lateral (AL) group, laterally; the Antero-Medial (AM) and Antero-Central (AC) groups, medially; and the Ventro-Central (VC) group, ventrally. The groups moving towards the posterior sections of the antennal lobe include the Postero-Medial (PM) group, Postero-Central (PC) and the deeper Medio-Dorsal (MD) groups, along the medial axis; Postero-Lateral (PL) group located on the lateral side; Centro-Dorsal (CD) group located centrally; and Postero-Ventral (PV) group located ventrally. In the most posterior slices, the dorsal group is named Dorsal (D); and the ventrally located group is named Ventral (V).

A two-dimensional neuroanatomical map of the Aedes aegypti antennal lobe
Based on each reconstructed three-dimensional LVPib12 antennal lobe model, we generated two-dimensional maps, showing all manually segmented glomeruli in each confocal slice as observed along the anterior to posterior axis of in vitro antennal lobe imaging. As a visual guide, we first selected eight glomeruli with salient features such as a large volume, shape and consistent position across all reconstructions to serve as landmarks for identifying the positions of other glomerular groups. These eight landmark glomeruli are shown in the 3D models ( Fig 1C-1F) and include AD1 and VC6 in anterior antennal lobe slices and D5, PL9, CD4, V1, V8 and MD1 in the more posterior antennal lobe. We did not select landmark glomeruli from the PC, PM, AM and AC groups, as the boundaries of these peripheral glomeruli cannot always be reliably demarcated by the in vitro staining methods used here.
To facilitate stereotypical identification of glomeruli based on their relative position to the eight landmark glomeruli, we generated a reference key detailing the typical spatial arrangement of glomeruli in the Ae. aegypti antennal lobe (Table 2). In each reconstructed antennal  Tables 3 and 4 and S1 Table. https://doi.org/10.1371/journal.pntd.0008729.t001 lobe, we assigned a name to each glomerulus based on its spatial group and depth. Glomeruli within the spatial groups are typically arrayed in (i) rows and are numbered based on depth along the medial lateral axis, (ii) in a vertical line (column) or (iii) in a circular manner and numbered along the dorsal-ventral axis. Glomeruli within some spatial groups may also appear to be randomly arrayed. Useful reference arrays of glomeruli within and between spatial groups that assist with identification and naming of glomeruli based on their spatial position relative to landmark glomeruli are detailed in Table 2 and overlaid upon an example antennal lobe reconstruction in S1A-S1J Nomenclature for the vast majority of spatial groups and glomeruli reported here have been systematically revised with respect to previous studies. For an explanation of these changes relative to Ignell et al. (2005) see Tables 1, 3 and 4 and S1 Table. 2 Description of the geometric arrangement of glomeruli within each spatial group. Glomeruli are typically arrayed in columns † , rows ‡ , circles˚, or randomly arrayed and numbered based on depth § . Each glomerulus may occupy a medial M , lateral L , dorsal D or ventral V position relative to the preceding glomerulus in its array. Landmark glomeruli are underlined. Commonly observed variant glomeruli are indicated by an asterisk � . See also S1 Fig. 3 The depth of glomerular arrays in dissected brains is subject to variation based on the extent to which the antennal lobe is distorted by removal of the esophagus, fixation and/or coverslip placement. 4  <80% across reconstructions, according to these criteria. The number of spatially invariant and variant glomeruli found in each of the 13 spatial groups are described as follows; The anterior surface of the antennal lobe is comprised of five groups of glomeruli: 1. Antero-Dorsal group. In anterior slices of the antennal lobe, AD1 is a landmark glomerulus, distinguishable by its large size and dorsal position. It is located towards the center of the antennal lobe. AD1 is the first antennal lobe glomerulus visualized when stepping down a z-series along the A-P axis, and the typical depth of this glomerulus from its anterior surface (denoted 0 μm) to posterior surface is 15 μm. AM1-3, AC1, VC1, AL1-3 glomeruli routinely cluster around the AD group. AD2 and AD3 are found ventral and towards the medial and lateral side of AD1, respectively. AD2 and AD3 glomeruli are typically found in the space between the AD1 and the VC group and are characteristic satellite glomeruli flanking the AD1 landmark. . We note that in our reconstructions, this region consists entirely of discrete glomeruli that we have reclassified as the VC and PV spatial groups. The VC group is comprised of an anterior surface layer consisting of the VC1, VC2 and VC3 glomeruli arranged in a row ordered along the medial to lateral axis, or alternatively in a triangle with VC1 positioned dorsal to VC2 and VC3; and the VC4, VC5 and VC6 glomeruli in a more posterior bottom row ordered along the medial to lateral axis. The most posterior layer of this group comprises VC7, VC8 and VC9 that are arranged centrally in this region of the antennal lobe (Table 2). VC1 is a large glomerulus positioned ventral to the landmark glomerulus AD1. VC6 is a large landmark glomerulus located on the lateral side of the antennal lobe and ventral to VC3. Two variant glomeruli (VC10-VC11) in a ventral to dorsal column are sometimes observed more posterior to the VC7-9 cluster. See S1B Fig and S2E Fig.   3. Antero-Lateral group: The AL group is observed in anterior slices within the dorsal and lateral region of the antennal lobe. The AL1 glomerulus is located laterally, relative to the landmark AD1. AL1, AL2, AL3 glomeruli are positioned in a D-V column on the lateral side of the antennal lobe in order of increasing depth along the A-P axis. For example, AL1 is found dorsal and anterior to AL2. In some reconstructions, three variant glomeruli were sometimes found ventral and posterior to AL3 and were named AL4-AL6. See S1C

Antero-Medial group:
The AM group is located medial to the AD group. It is comprised of the AM1, AM2 and AM3 glomeruli arranged in a column along the D-V axis. AM1 is a large glomerulus located above AD1 towards the medial axis of the antennal lobe. A variant glomerulus AM4 was observed ventral to AM3 in some reconstructions. See S1D Fig and  S2C and S2D Fig.   5. Antero-Central group: The AC group is located medial to the AM group bisecting the equator of the antennal lobe in a D-V column flanking the medial side. In more posterior slices, the boundaries of some the AC glomeruli were difficult to demarcate as this spatial group is sandwiched between the AM and the more posterior PM spatial groups. AC1 and 3 were spatially invariant glomeruli, whereas variant glomeruli AC2, AC4 and AC5 were observed in some reconstructions. See S1E Fig and S2C and S2D Fig. In mid-sections of the antennal lobe, we describe the following groups: 6. Postero-Medial group: This group is located posterior to the AM glomeruli. It is comprised of glomeruli along a D-V column on the medial side of the antennal lobe. The boundaries of PM1, PM2, and PM3 are noted for their small size, and position between the anterior AC and posterior D groups. PM4 is positioned more ventrally and is larger in volume relative to PM1-PM3. In more posterior slices, the ventral PM5 and PM6 are classified as variant glomeruli. See S1F Fig and S2D and S2J Fig). 9. Ventral group: The V group is found posterior and dorsal to the PV group, in the deepest layers of the antennal lobe. The landmark glomerulus V1 is found ventral and posterior to PV1. V1 flanks the medial margin of the antennal lobe. In the medial to lateral direction the other glomeruli of this group are arranged in three layers. The first layer consists of V1, V2, V3, and V4 aligned in a medial-lateral row. The second more posterior layer consists of V5, V6, V7 and V8 are arranged in an approximate lateral-medial row. V8, which is posterior and medial relative to V7, is a landmark glomerulus that is easy to identify because of its lateral adjacent position to MD1. V8 is found posterior to V1. In some reconstructions we identified more posterior variant V glomeruli (V9-V11). See S1I and S1J Fig and S2G, S2H and S2J Fig.   10. Centro-Dorsal group: The CD group is the central group in dorsal regions of the posterior-most slices of the antennal lobe. CD1, CD2, CD3, CD4, CD5, and CD6 are arranged almost in a circle. CD1 and CD2 lie centrally in a dorsal position relative to other CD glomeruli, yet ventral to the D group. CD3 and CD4 are more medial in position and ventral to CD1 and CD2. CD5 and the more lateral CD6 are aligned towards the lateral side of the antennal lobe. CD4 is a landmark, conspicuous because it is the most ventral CD glomerulus and due to its large, irregular shape. This glomerulus is located dorsal to the V group.
Overall, the CD glomeruli can be distinguished from the other posterior antennal lobe groups by their smaller volume. In some reconstructions, more than 6 CD glomeruli were observed. These variant glomeruli were found in more posterior slices and named CD7-CD12. See S1J Fig and S2H Fig.  11. Postero-Lateral group: The PL group of glomeruli are located along the lateral axis and form a distinct cluster on the dorso-lateral corner as confocal slices approach the posterior surface of the antennal lobe. This is especially apparent in the male antennal lobe. The PL group lies more posterior than the AL group in both sexes. PL1 is the most ventral glomerulus in this group. Thereafter PL glomeruli are named with increasing numbers based on dorsal position relative to PL1 and depth. PL9 is a landmark that is characterized by its large size, visible through the most posterior sections of the antennal lobe. PL9 is positioned laterally diagonal and opposite to the MD1 glomerulus. PL1-9 are spatially invariant glomeruli. Four residual variant glomeruli are sometimes observed in more posterior sections (PL10-PL13). See S1J and S1K Fig and S2J Fig.   12. Representative confocal stacks and color-coded 2D maps detailing the typical spatial arrangement of invariant and variant glomeruli in the female left and right antennal lobes as shown by nc82 staining, are illustrated in Fig 2 and S3 Fig, respectively. Examples for the left and right antennal lobes of adult males are illustrated in Fig 3 and S4 Fig, respectively. The gross 3D shape of reconstructed nc82 stained antennal lobes varied noticeably between individual mosquitoes (see S5 Fig). However, the eight characteristic landmark glomeruli could reliably be identified in all reconstructed models we generated, facilitating assignment of names to all remaining glomeruli based on their relative spatial position.

Changes in glomerular nomenclature relative to previous in vitro models of the Aedes aegypti antennal lobe
We used our reference key (Table 2) to apply our updated system for glomerular nomenclature to the extant in vitro models of the Ae. aegypti antennal lobe reported by Ignell et al. (2005). These authors previously annotated 50 and 49 glomeruli in female and male antennal lobe of the Rockefeller Ae. aegypti strain, respectively.   In total, we identified 40 glomeruli in their synapsin:phalloidin-based model of the female antennal lobe [11] and 39 glomeruli in their nc82-based model of the male antennal lobe [11] that appear to occupy orthologous positions to spatially invariant glomeruli annotated in this study. Using our updated system for glomerular nomenclature, only 7 of these glomeruli retain their previously annotated name in both sexes across both studies (AL1-3, AM1-3 and MD1), with the names for all other glomeruli changing (Table 3).

PLOS NEGLECTED TROPICAL DISEASES
Of those renamed as reported in Table 3, 23 glomeruli in their female model and 24 glomeruli in their male model were reclassified into new glomerular spatial groups with different nomenclature and revised organization. The remaining 10 renamed glomeruli in their female model and 8 renamed glomeruli in their male model were assigned new numerical descriptors to reflect their arrayed position within their extant spatial group-predominantly belonging to the PL, CD and MD groups in both sexes. Across all spatially invariant glomeruli we identified within their models, only 14 (35-36%) share the same descriptor annotated by Ignell et al. (2005) in both sexes (Table 3), consistent with considerable variation in the way glomeruli were ascribed names by these authors in their reported models with reference to relative spatial position.
In addition to the complement of orthologous glomeruli described above, 12 new spatially invariant glomeruli in females and 13 new spatially invariant glomeruli in males were identified in our updated atlas that likely result from segmentation of the Johnston's Organ Center (JOC) from Ignell et al. (2005) (Table 3). We further annotated 11 additional new spatially invariant glomeruli in the updated PL, D, CD, V and MD groups of females, and 11 new spatially invariant glomeruli in the updated PC, PL, CD, V and MD groups of males in our atlas that were not evident in the reconstructed maps of Ignell et al. (2005) (Table 3).
In the antennal lobe models reported by Ignell et al. (2005), we were unable to assign updated names to PC1 and PL6 for the female (Table 4), and additionally AD4 for both sexes (Table 4 and S1 Table). These glomeruli appear positioned at the boundary of two different spatial groups, therefore making definitive assignment to either group difficult. AD4 for example, is positioned at the interface of updated PC and PM groups; PC1 could be assigned either to the updated PC or CD groups, and similarly, PL6 could be considered as an updated V or PL group glomerulus.

Volumetric analysis of the Aedes aegypti antennal lobe
With our updated system for glomerular nomenclature established, we then performed volumetric analyses on the left and right antennal lobes from all nc82-stained LVPib12 brains that we reconstructed (n = 5 brains per sex, 20 antennal lobes total in this dataset). The mean total antennal lobe volume is 123,319 μm 3 in females, and 85,370 μm 3 in males (Mann-Whitney, P<0.0001). The total antennal lobe volume of females is therefore approximately 1.4 times larger than that of male mosquitoes.
For both sexes, there were no significant differences in total antennal lobe volume between the left and right sides of the brain (S2 Table). Within antennal lobes derived from females (Fig 4) and males (Fig 5), the mean volumes of glomeruli varied significantly (one-way ANOVA, P<0.0001 for each lobe, S6 Fig, S7 Fig, S8 Fig, S9 Fig). The largest glomerulus in both the female and male antennal lobe is MD1, with a mean volume of 5,222 μm 3 and 4,090 μm 3 , which accounts for 4.26% and 4.81% of their total antennal lobe volumes, respectively.
To identify candidate sexually dimorphic glomeruli, we then calculated relative volumes for all 63 invariant glomeruli that we annotated in females and males (S2 Table). To do this, we determined the percentage volume that each invariant glomerulus sampled occupies within its concordant antennal lobe, and then compared grouped relative glomerular volumes by sex using the non-parametric Mann-Whitney U-Test. Using this approach, we identified two glomeruli with significantly larger relative volumes in females: AD1 (P = 0.005) and VC1 (P = 0.036). Of these, the landmark glomerulus AD1 has a mean volume of 4,708 μm 3 in females and 2,456 μm 3 in males which accounts for 3.82% and 2.89% of their total antennal lobe volumes, respectively. In addition, four glomeruli were significantly larger in males: AM3 (P = 0.004), CD3 (P = 0.019), CD4 (P = 0.015) and MD3 (P = 0.016). AM3, which is positioned medially on the anterior surface of the antennal lobe, has a mean volume of 1,315 μm 3 in females and 1,431 μm 3 in males which accounts for 1.06% and 1.67% of their total antennal lobe volumes, respectively.

Validating the revised neuroanatomical model of the Aedes aegypti antennal lobe
To verify glomerular counts and the spatial organization of the antennal lobe model reconstructed from nc82-staining, we utilized a complementary and independent neuronal labeling method to visualize glomeruli. Since the antennal lobes across diverse insect species are known to contain a rich aggregation of F-actin filaments in OSNs [23], the fluorophore-conjugated toxin phalloidin was used as a cytoskeletal marker of OSN axonal processes to visualize antennal lobe morphology. Using this method, we could observe strong staining of all the major neuropil in the central Ae. aegypti brain (Fig 6A and 6B). We then performed reconstructions on the left and right antennal lobe from 3 male and 3 female LVPib12 mosquitoes.
We counted 79 ± 1 glomeruli in females and 77 ± 3 glomeruli in males (mean ± s.d.) (Fig 6G  and 6H). Relative to pre-synaptic nc82-staining, the diffuse labeling pattern of phalloidin from overlapping OSN axonal processes traversing the antennal lobe made delineation of glomeruli using this method challenging at times, particularly in the anterior and central antennal lobe. As a result, the confocal images had to be magnified 4-5 times in order to perform segmentation. Furthermore, at lower magnification we observed in many reconstructions that PV and VC glomeruli appeared to be stained as a single large cluster of neuropil, rather than as individual glomerular units that could be clearly demarcated with antibody labeling. This cluster has a similar morphology to the non-glomerular JOC that was described in an earlier Ae. aegypti antennal lobe model generated using phalloidin staining [11].
We selected representative reconstructions of the male and female antennal lobes in which we were able to easily demarcate all landmark glomeruli, as described for the nc82-stained model (Fig 6C-6F). We then created 2D anatomical maps of female left and right antennal lobes (S10 Fig and S11 Fig) to comprehensively name all glomeruli in this dataset. Consistent with the nc82 model, we found that the antennal lobe was divisible into 13 distinct spatial groups following our reference key. Applying our thresholding criteria to categorize spatially invariant and variant glomeruli within this data set, we identified the exact complement of spatially invariant glomeruli (n = 63) seen with nc82-stained brains. Remaining glomeruli found at a frequency of <80% in the antennal lobe of both sexes were categorized as variant glomeruli, consistent with morphological distortions being inherent to both nc82 and phalloidinbased neuronal staining methods. Of note, the PL group of glomeruli was positioned more dorsally in reconstructions of the male left antennal lobe (Fig 6F).
To further assist with user identification of glomeruli in the LVPib12 strain using nc82 and phalloidin-based neuronal staining methods, we have generated a series of digital models that allow users to interactively view the spatial arrangement of glomeruli in left antennal lobe of a female brain sample stained with each method from different perspectives and depths (S1 Movie, S2 Movie, S3 Movie, S4 Movie).

Conservation of antennal lobe organization across geographically divergent strains of Aedes aegypti
Finally, we compared the antennal lobe neuroanatomy of the inbred LVPib12 strain [21] originally of African origin, with a recently established strain of Ae. aegypti derived from Patillas, Puerto Rico (Fig 7A) that is routinely supplemented with field-caught mosquitoes [24]. We applied nc82-staining to dissected and fixed Patillas brain samples and reconstructed the left and right antennal lobes of three male and three female Ae. aegypti mosquitoes in total. Overall, the gross shape of antennal lobes from the LVPib12 and Patillas strains differed subtly. In particular in the Patillas strain, the female left and right antennal lobes were rounded, while the PV group from the male left and right antennal lobes appeared to taper out of the central region. Using our LVPib12 antennal lobe models as a guide, we generated 3D models of the Patillas male and female antennal lobe, in which we classified and color-coded glomeruli into the 13 glomerular spatial groups (Fig 7B-7E).
We then segmented 3D models from a representative Patillas male and female brain sample to yield 2D maps to name individual glomeruli. Using our reference key (Table 2), we named all antennal lobe glomeruli in these brain samples, and then applied our thresholding criteria to identify spatially invariant and variant glomeruli within the data set. We identified all 63 spatially invariant glomeruli found in the LVPib12 strain and denoted the presence of residual variant glomeruli consistent with our previous models. In total, we counted 81 ± 4 glomeruli in the female antennal lobes (Fig 7F) and 77 ± 2 in the male antennal lobes (mean ± s.d.) ( Fig  7G). We conclude that despite gross morphological differences in antennal lobe shape, the spatial organization of antennal lobe glomeruli in these geographically diverse populations of Ae. aegypti appears conserved.

Discussion
Aedes aegypti is an emerging model system for mosquito olfaction [3,[25][26][27][28][29][30][31]. An accurate anatomical map of the antennal lobe is crucial towards studies aimed at understanding how odors are detected and represented by the olfactory system of this disease vector. In this study, we generated an updated atlas for the Ae. aegypti antennal lobe providing systematic nomenclature and a standardized framework for identifying glomeruli within this primary olfactory processing center of the mosquito brain.
We determined the Ae. aegypti antennal lobe is composed of approximately 80 morphologically discrete glomeruli in both sexes. We surmise that the advanced confocal imaging techniques employed in our study may have facilitated clearer delineation of glomerular boundaries, thus revealing additional glomeruli relative to previous studies that found 35 or 49-50 glomeruli in the Ae. aegypti antennal lobe, respectively [11,18]. In comparison with the Drosophila melanogaster antennal lobe which has 56 glomeruli [32], the Aedes aegypti antennal lobe has markedly increased subdivision making assignment of names to individual glomeruli not a trivial matter.
To facilitate stereotypical identification of glomeruli, we first subdivided the antennal lobe into 13 spatial groups with revised nomenclature, merging several extant groups from a previous antennal lobe model [11], and annotating others with new descriptors to better reflect their true spatial position and accommodate the expanded complement of glomeruli we found (Table 1). Subsequently, we identified eight landmark glomeruli based on their characteristic shape, volume and position across these groups, with all other glomeruli annotated by their spatial position relative to these landmarks. Guided by replicate morphological reconstructions, we annotated 63 spatially invariant glomeruli that can be reliably identified in the same geometric arrangement within dissected and stained antennal lobe preparations. Despite intra-and inter-strain variability in the overall shape of antennal lobes in this study, the spatial organization of invariant glomeruli from LVPib12 and Patillas strains was shared and discernable. These results suggest that antennal lobe organization is likely conserved across geographically divergent strains of Ae. aegypti. Our revised atlas may therefore serve as a useful reference for studying antennal lobe organization in other Ae. aegypti strains supplemental to those reported here.
We determined the largest invariant glomerulus in Ae. aegypti was MD1. Anterograde tracing experiments in this species previously revealed that MD1 likely receives projections from maxillary palp-derived OSNs [12] which are responsive to the volatile gas carbon dioxide (CO 2 ). The increased volume of this glomerulus in both sexes may be reflective of the critical role that CO 2 plays in Ae. aegypti host-seeking behavior [3,30,33]. We additionally identified six candidate glomeruli including AD1 and AM1 with sex-specific differences in relative volume. In the future, identifying what odorants these glomeruli detect may help to further clarify their role in sexually dimorphic olfactory behaviors in this mosquito species.
We noted considerable variation in the spatial position of approximately 20% of glomeruli annotated within our in vitro antennal lobe reconstructions. These variant glomeruli tended to have smaller volumes and were clustered around boundaries between spatial groups, making them difficult to accurately name. Typically, variant glomeruli were positioned in more posterior layers of the antennal lobe and were frequently noted in PM, PL, CD and V groups. Such spatial variability in glomerular position is likely due to technical artifacts associated with dissection, fixation, protocol duration and coverslip placement for imaging that commonly distort antennal lobe morphology [34]. The Ae. aegypti antennal lobes are also positioned in close proximity to the anterior esophagus, a thick structure lined with cuticle, epithelial cells and muscle [35]. The esophagus runs through the esophageal foramen and it was removed during our dissections to prevent it from occluding the antennal lobes, possibly causing distortion or disassociating glomeruli, and altering the shape of each antennal lobe.
As recently described for a revised antennal lobe model for Anopheles coluzzii [20], we did not observe any large clusters of neuropil orthologous to the Johnston's Organ Center (JOC) [11] in any of our reconstructions from the LVPib12 or Patillas Ae. aegypti strains. Instead, we identified morphologically discrete glomeruli in the anterior region of the antennal lobe where this structure was previously annotated. The JOC was assigned based on evidence from silver nitrate Golgi staining experiments [11] that suggested putative neuronal connectivity between the antennal lobe and the sound-receptive Johnston's Organ (JO). However, other anterograde tracing studies in Ae. aegypti and D. melanogaster indicate afferent JO neurons project to the antennal motor and mechanosensory center (AMMC) of the deutocerebrum, rather than the antennal lobe [36,37]. Further in-depth neuroanatomical studies are clearly warranted to definitively validate any afferent or efferent neuronal connectivity between the JO and antennal lobe in Ae. aegypti and other mosquito species. It will also be important in the future to investigate whether any differences in antennal lobe morphology are apparent between the Rockefeller strain of Ae. aegypti from which the JOC was originally annotated [11] and the LVPib12 and Patillas strains investigated here.
The Ae. aegypti genome encodes large chemoreceptor gene families including Odorant Receptors, Ionotropic Receptors and Gustatory Receptors [28,38]. To date, chemoreceptors have only been localized to OSNs on the Ae. aegypti antennae, maxillary palps and proboscis using bulk transcriptomics and immunohistochemical approaches [25,28,39]. We propose this model atlas will be useful to map projection patterns of neurons expressing chemoreceptors and other genes involved in olfactory signaling to their associated glomeruli in the Ae. aegypti antennal lobe using genetic labeling approaches [20,34,40,41]. Such studies will be critical to validate our antennal lobe model that is currently based on morphology alone, toward understanding the molecular and cellular organization of the Ae. aegypti olfactory system. Finally, this updated atlas lays the foundation for improved spatial registration during functional imaging within the Ae. aegypti antennal lobe with genetically encoded calcium indicators and dyes [30,42,43]. In particular, studies with activity-dependent ratiometric indicators such as CaMPARI [44,45], that provide a permanent readout of neural activity and are also compatible with post-hoc neuronal staining to aid in spatial registration of such signals, will benefit greatly from using the in vitro antennal lobe atlas reported here as a reference. Such analyses may help to reliably identify glomeruli activated or inhibited by diverse odorants that are behaviorally relevant to Ae. aegypti such as human odors, plant odors, pheromones and insect repellents. This information may further accelerate reverse engineering of synthetic odorant blends mimicking patterns of neurophysiological activity elicited by these natural olfactory stimuli, to yield novel applied strategies to modulate Ae. aegypti olfactory behaviors.
Supporting information S1 Data. Excel spreadsheet containing, in separate sheets, the underlying numerical data for Figure panels 1g, 1h, 4a, 4b, 5a, 5b, 6g, 6h, 7f, 7g and S6 Fig, S7 Fig, S8 Fig, S9 Fig and  S2 Table. (XLSX) S1  Glomeruli within intra-group arrays belonging to the thirteen color-coded spatial groups are highlighted from various perspectives of the antennal lobe. Scale bars = 10 μm. All remaining glomeruli were made transparent to highlight these example arrays and their typical spatial position. Textual annotations that denote the typical spatial arrangement of glomeruli within intra-group glomerular arrays are shown below each perspective of the model lobe. Underlined glomeruli in arrays (AD1, VC6, D5, V1, V8, MD1, CD4 and PL9) are landmarks that facilitate identification of other glomeruli based on their relative spatial position. Glomeruli are typically arrayed in columns † , rows ‡ , circles˚, or randomly arrayed and numbered based on depth § . Each glomerulus may occupy a medial M , lateral L , dorsal D or ventral V position relative to the preceding glomerulus in its array. Landmark glomeruli are underlined. Commonly observed variant glomeruli are indicated by an asterisk � . A full series of intra-and inter-group glomerular arrays are detailed in Table 2. Textual annotations that denote the typical spatial arrangement of glomeruli within an intergroup glomerular array are shown below each perspective of the model lobe. Underlined glomeruli in arrays (AD1, VC6, D5, V1, V8, MD1, CD4 and PL9) are landmarks that facilitate identification of other glomeruli based on their relative spatial position. Each glomerulus may occupy a medial M , lateral L , dorsal D or ventral V position relative to the preceding glomerulus in its array. A full series of intra-and inter-group glomerular arrays are detailed in Table 2