Shotgun and TMT-Labeled Proteomic Analysis of the Ovarian Proteins of an Insect Vector, Aedes aegypti (Diptera: Culicidae)

Abstract Aedes aegypti [Linnaeus in Hasselquist; yellow fever mosquito] transmits several viruses that infect millions of people each year, including Zika, dengue, yellow fever, chikungunya, and West Nile. Pathogen transmission occurs during blood feeding. Only the females blood feed as they require a bloodmeal for oogenesis; in the bloodmeal, holo-transferrin and hemoglobin provide the females with a high iron load. We are interested in the effects of the bloodmeal on the expression of iron-associated proteins in oogenesis. Previous data showed that following digestion of a bloodmeal, ovarian iron concentrations doubles by 72 hr. We have used shotgun proteomics to identify proteins expressed in Ae. aegypti ovaries at two oogenesis developmental stages following blood feeding, and tandem mass tag-labeling proteomics to quantify proteins expressed at one stage following feeding of a controlled iron diet. Our findings provide the first report of mosquito ovarian protein expression in early and late oogenesis. We identify proteins differentially expressed in the two oogenesis development stages. We establish that metal-associated proteins play an important role in Ae. aegypti oogenesis and we identify new candidate proteins that might be involved in mosquito iron metabolism. Finally, this work identified a unique second ferritin light chain subunit, the first reported in any species. The shotgun proteomic data are available via ProteomeXchange with identifier PXD005893, while the tandem mass tag-labeled proteomic data are available with identifier PXD028242.

phase when oocytes become competent to incorporate vitellogenin, and little follicle growth occurs. At the start of blood feeding, the initiation phase begins and lasts 3-10 hr, during which time follicle growth recommences and new microvilli and coated pits are observed on the surface of the oocyte (Clements 2000). Following the follicular growth re-initiation, the follicular epithelium separates allowing hemolymph contact with the oocyte, and incorporation of vitellogenin into yolk bodies. In the trophic phase, vitellogenin uptake increases linearly until 24 hr post bloodmeal (PBM); the basal lamina surrounding the follicular epithelium acts as a coarse filter permeable to lipids, polysaccharides, and proteins of up to 500 kDa or <11 nm, for incorporation into the oocyte. Passage of nutrients from the hemolymph into the oocyte concludes by 48 hr PBM, when deposition of an intact endochorionic layer between the oocyte and follicular epithelium ends the trophic phase. The oocyte grows to its final size and assumes its final form in the post-trophic phase with deposition of the exochorion by the follicular epithelium in preparation for fertilization and oviposition that occurs between 72 and 96 hr PBM.
Proteomics has proven valuable for evaluating the expression of large numbers of proteins from yeast (~4,000) to humans (>10,000) (Richards et al. 2015). Given the iron load of the bloodmeal, the transport of meal iron to the ovaries, and the requirement for iron during animal development (Gonzalez-Morales et al. 2015, Rivera-Perez et al. 2017, we employed two proteomics techniques, shotgun and tandem mass tag-labeling (TMT-labeling), to identify proteins present in this tissue following a bloodmeal. To our knowledge, this is the first report to identify proteins expressed at two different stages of oogenesis in Aedes ovaries PBM utilizing high throughput, high-resolution mass spectrometry. Shotgun proteomics identified several putative proteins involved in mosquito iron metabolism including a second ferritin light chain subunit (LCH2), the first identified in any species. We also employed TMT-labeling proteomics to quantify the expression of a subset of ovarian proteins in response to an iron meal.

Mosquito Rearing
Mosquitoes were raised as noted elsewhere (Geiser et al. 2017); mated female mosquitoes were fed on Porcine blood eight to nine days post-eclosion. Ae. aegypti were a gift from Dr. Michael Riehle (Department of Entomology, University of Arizona).

Tissue Collection
In Ae. aegypti, blood feeding is required for oogenesis. We reasoned that we would identify the greatest number of proteins by analyzing samples during an early stage of oogenesis (the trophic phase, 24 hr PBM), and a very late stage (72 hr PBM) because, as noted, these times signify important gates in the development process (Anderson and Spielman 1971). Further, Akbari et al. (2013) reported that in a pair-wise comparison of the transcriptome at various times during egg development, every development stage was highly correlated with the adjacent stages with the exception of 36-48 and 60-72 hr, suggesting that these times represented developmental physiological transitions. To allow for biological variation, we blood fed animals on six different dates. For each date, we collected ovaries from 15 cold anesthetized animals at each developmental stage PBM (24, 72 hr, Supplement 1 [online only]); ovary pairs were dissected into disruption buffer (10 mM Tris-HCl, pH 7.9; 1.5 mM MgCl 2 ; 0.5 mM DTT added fresh) with 2x Protease Inhibitor cocktail (Catalogue # 539131; EMD Millipore, Billerica, MA) added fresh. As ovaries were pooled from each time point, samples represented 6 different laboratory-raised Ae. aegypti mosquito populations and a total of 90 ovary pairs.
Peptide identification was performed using SeQuence IDentfication (SQID) (Li et al. 2011) against the Ae. aegypti database downloaded from the National Center for Biotechnology Information (NCBI; Genome ID: 44, downloaded on 23 June 2012) for the combined results for 8 gel pieces per time point. The search was performed with the forward database appended with a reverse database; the false discovery rate (FDR) was determined as FDR = 2 * ReverseID/(ForwardID + ReverseID). The Ae. aegypti forward and reverse database was used to establish criteria for protein identification. A maximum of 2 missed cleavages were allowed and methionine oxidation (M+16) and carbamidomethylation (C+57) were searched as variable modifications. The result spectra list for each time point was generated in Scaffold (Version 3.1.2; Proteome Software, Inc., Portland, OR) and ranked using a SQID score column to calculate peptide FDR. A 5% peptide FDR, a minimum of two unique peptides per protein, and a 1% protein FDR were used as the threshold for protein identification. Experimental results represent two time points each analyzed in duplicate. Running duplicates minimized the chances that a low abundance protein was unidentified at a time point due to sensitivity rather than a true difference between time points.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Deutsch et al. 2020) via the PRIDE (Perez-Riverol et al. 2019) partner repository with the dataset identifier PXD005893 and 10.6019/PXD005893. Proteomic data include spectra, number of unique peptides identified, total number of peptides identified, and percent coverage for each protein identified in the following tables and analyses can be acquired from the original Scaffold file downloaded from the PRIDE Archive (https:// www.ebi.ac.uk/pride/archive/) and viewed in the most recent version of Scaffold via free download from the Scaffold website (https:// www.proteomesoftware.com/products/scaffold-5).

Criteria for Identifying Proteins at Each Development Stage
A candidate protein was validated as being only detected at either early or late development by meeting the following three criteria for duplicate analyses: (1) we identified a minimum of two unique peptides per protein; (2) all peptides had a greater than 95% identification confidence; and (3) peptides were only identified at one time point.

Mosquito Rearing, Tissue Collection, and Sample Preparation
Since we anticipated identifying the greatest number of proteins from ovaries during the trophic phase of oogenesis (24 hr PBM), we chose to quantitate differences in ovary protein expression based on iron content of the diet at this developmental stage by labeling sample peptides using a tandem mass tag (TMT) Isobaric and Isotopic Mass Tagging kit. Mosquitoes were maintained as described above. Mated, female mosquitoes were fed one of two isoproteinic diets or a Porcine bloodmeal (Porcine BM, 603 ng Fe/μl) maintained at 27°C in glass feeders for 2 hr. The isoproteinic diets were: Kogan's (1990) artificial bloodmeal (ABM) with hemoglobin ((+)Fe): 0.8% (w/v) Porcine Hemoglobin (Sigma, St. Louis, MO); 1.5% (w/v) Porcine IgG (Sigma); 10% (w/v) Porcine Albumin (Sigma); and 5 mM ATP (Sigma) in feeding buffer (~56 ng Fe/μl) and Kogan's (1990) ABM without hemoglobin ((-)Fe): 10.7% (w/v) Porcine Albumin; 1.6% (w/v) Porcine IgG; and 5 mM ATP in feeding buffer (~24 ng Fe/μl). Tissues were collected as described above for early stage of oogenesis (24 hr PBM). Ovary proteins were extracted, and protein concentrations were determined as described above. Samples were flash frozen in liquid nitrogen and stored at −80°C until TMT-labeling.

Ovary Protein TMT-Labeling
Differences in ovary protein expression can be quantified by combining TMT-labeled peptides from ovaries of animals fed different diets and analyzing the peptides simultaneously by LC-MS/MS. Ovary samples at the trophic phase of oogenesis (24 hr PBM) for the (+)Fe diet, (-)Fe diet, and the Porcine BM were collected and processed as previously described and TMT-labeled following the manufacturer instructions for the TMT6plex kit (Catalogue # 90061; Thermo Fisher Scientific). Briefly, to prepare samples for TMTlabeling, 6 μl of each sample ((+)Fe: 33.9 μg; (-)Fe: 34.4 μg; Porcine BM: 41.4 μg) was diluted in sterile ddH 2 O up to 50 μl, then 45 μl 100 mM triethyl ammonium bicarbonate (TEAB) was added and mixed with each sample. Since these samples were complex protein mixtures, 5 μl 2% SDS was mixed in to bring the final volume to 100 μl. Tris(2-carboxyethyl)phosphine [TCEP (5 μl 200 mM)] was added to each sample, mixed, and incubated at 55°C for 1 hr in the dark. Further, 5 μl 375 mM iodoacetamide (IAA) was mixed into each sample and incubated at room temperature (RT) for 30 min in the dark. Finally, 1 ml acetone, prechilled at −20°C, was mixed with each sample and allowed to precipitate overnight at −20°C and then centrifuged 10 min at 8,000 × g at 4°C. Supernatants were decanted, and the precipitated pellets were dried at RT for 10 min. Sample pellets were suspended in 50 μl 100 mM TEAB, and trypsin digested in the solution overnight (1 µg trypsin (Promega):15 µg sample protein; digestion conditions determined during protocol optimization) at 37°C, following the protocol (https://proteomics.arizona.edu/sites/ proteomics.arizona.edu/files/Solution_trypsin_digestion_2.pdf) provided by the Analytical & Biological Mass Spectrometry Facility (MassSpec@email.arizona.edu). TMT6plex labels were dissolved in 41 μl acetonitrile for 5 min, RT, with occasional vortexing, and 6 μg (based on original sample protein concentration) trypsin-digested sample was added to a unique TMT6plex label, and incubated for 1 hr, RT. Hydroxylamine (8 μl, 5%) mixed with each sample and incubated for 15 min, RT was used to quench the labeling reaction. Samples were dried down to pellets using a SpeedVac vacuum concentrator with medium heat. In preparation for LC-MS/MS, sample pellets were dissolved in 6 μl 0.1% trifluoroacetic acid (TFA) pH 3.0, yielding sample concentrations of 1 μg/μl (based on original sample protein concentration). Next, 2 μl of each TMT-labeled diet sample (Table 1) was combined into one tube for a total volume of 6 μl (Tube 1), which was then further divided into 2 tubes of 3 μl each (Tubes A and B). Each tube had a total protein concentration of 3 μg (based on original sample protein concentration). Samples were stored at −80°C until LC-MS/MS analysis.

TMT-Labeled Ovary Proteome Analysis
Prior to analysis, each sample, Tube A or Tube B, was thawed and desalted using the protocol described: (http://proteomics.arizona.edu/ sites/proteomics.arizona.edu/files/SPE_C18_Clean_up_ZipTip.pdf). Each TMT-labeled sample, Tube A or Tube B, was analyzed in triplicate by LC-MS/MS using equipment and methods described by Yuan et al. (2015); peptides on the analytical column were fractionated by holding at 5% solvent B (acetonitrile, 0.1% formic acid) for 10 min, followed by a solvent B gradient of 5-20% over 65 min, then by a solvent B gradient of 20-35 % over 45 min, then a 35-95% gradient of solvent B (0.1 min), ending with a solvent B (95%) hold for 5 min. Running multiple analyses of each tube minimizes the chances that a low abundance protein goes unidentified. Data dependent scanning was done as described in Jiang et al. (2014) with the following changes: the Orbitrap analyzer scanned m/z 350-1600 followed by collision-induced dissociation (CID) MS/MS of the ten most intense ions in the linear ion trap analyzer, and the dynamic exclusion was set on an exclusion list for 30 s after a single MS/MS.
The raw file of the MS/MS spectra output and the spectra were searched using Proteome Discoverer version 2.4.0.305 software (Thermo) for quantitative analysis against the Ae. aegypti database downloaded from UniProt (Aedesaegypti_UniprotKB_2020_7159. fasta). For protein identification, the following options were used: MS/MS spectra matches reflected fully tryptic peptides with up to 2 missed cleavage sites, variable modifications considered during the search included methionine oxidation (15.995 Da), TMT6plex on the N termini (+229) and cysteine carbamidomethylation. Proteins were identified and validated according to Yuan et al. (2015) and the results displayed with Scaffold Q+S v 4.11.0 (Proteome Software, Inc.); TMT 6 plex (N-term) and TMT 6 plex (K). Variable modifications included Met residue oxidation (M) and filtering at an FDR ≤ 0.01; for quantification, the median of only unique peptides of the protein are used to calculate the protein ratios, and for experimental bias, all peptide ratios were normalized by the median protein ratio, and the median protein ratio should be 1 after normalization.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Deutsch et al. 2020) via the PRIDE (Perez-Riverol et al. 2019) partner repository with the dataset identifier PXD028242 and 10.6019/PXD028242. Proteomic data include spectra, number of unique peptides identified, total number of peptides identified, and percent coverage for each protein identified in the following tables and analyses can be acquired from the original Scaffold and Proteome Discoverer files downloaded from the PRIDE Archive (https://www.ebi.ac.uk/pride/archive/). The Scaffold files can be viewed in the most recent version of Scaffold Q+S via free download from the Scaffold website (https://www.proteomesoftware. com/products/scaffold-quant). The Proteome Discoverer files can be opened in the most recent version of Proteome Discoverer via free trial download from the ThermoFisher Scientific website (https:// www.thermofisher.com/us/en/home/industrial/mass-spectrometry/ liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/ multi-omics-data-analysis/proteome-discoverer-software.html).

Criteria for Identifying Proteins Responsive to Dietary Iron
A candidate protein was validated as being regulated by iron in the diet during the trophic phase of oogenesis by meeting the following criteria for multiple analyses: (1) a minimum of two unique peptides per protein were identified; (2) all peptides had a greater than 95% identification confidence; (3) protein was identified in both the Proteome Discoverer and Scaffold outputs from the same analysis; (4) protein was identified in two separate analyses; (5) the difference between increased or decreased expression of a protein due dietary iron was significant (p < 0.05) using the Holm-Sidak method for multiple unpaired t-tests in Prism 6, Version 6.07 (Graph Pad Software, Inc., San Diego, CA); and (6) the direction of protein expression, either upregulated or downregulated, was consistent among analyses.

Bioinformatic Analysis of the Ovary Proteome and TMT-Labeled Ovary Proteome
Scaffold or Proteome Discoverer identified protein GenBank GI numbers were analyzed for their correct unique identifiers using Batch Entrez (Coordinators 2014). Those identifiers that were found to be an old version were updated with a new GI number, while those identifiers that were found to be eliminated were further analyzed by BLASTP (Coordinators 2014) to obtain the correct GI Number. UniProtKB (UniProt 2015) analysis was performed using the corrected GI numbers; entry information for each identified protein was obtained (UniProtKB ID: Status, Protein names, Gene names, and Length; Gene Ontology (GO): Biological Process, Molecular Function, and Cellular Component; GO Identifiers: Developmental Stage, Induction, Tissue Specificity, Pathway, Subcellular Location, Protein Families, Metal Binding, PubMed ID, Cross-reference (STRING), and Cross-reference [InterPro]). Duplicate protein entries (the same protein identified with separate GI numbers) are acknowledged in supplementary tables in italics. The Database for Annotation, Visualization, and Integrated Discovery (DAVID version 6.7 [Huang da et al. 2009a,b]) was used to analyze the Gene names obtained from the UniProtKB analysis; DAVID identifiers were recorded and the proteomic data set was annotated and grouped by GO terms for biological process, molecular function, and cellular component. Proteins that were not in the DAVID database in all pertinent tables and supplementary data are identified in bold-faced font. Proteins found in the DAVID database were further analyzed using functional annotation clustering with a minimum p < 0.05 to identify enrichment of GO categories against the Ae. aegypti genomic background.

Characteristics of the Aedes PBM Ovary Proteome
We used shotgun proteomics to identify proteins present in two developmental stages; if a protein that met the criteria noted was present in both replicates, it was identified at that developmental stage. Data from Scaffold analyses identified 1,525 proteins of which 29 were duplicates providing a database of 1,496 unique proteins. The Ae. aegypti ovary proteome database with ontology analysis using general GO terms from the NCBI, UniProtKB, and DAVID databases, is available in Supplement 3 (online only). Of the 1,496 proteins, 1,199 were assigned an ontology category (Fig. 1A, Table  2). Many proteins identified to more than one functional term as shown in the Venn diagram (Fig. 1A). General GO categories and specific GO terms used in our analysis only indicate likely gene functions in processes for the GO term name, and do not provide definitive function without further empirical research. The numbers and percentages of proteins that are sorted to various pathways or processes are shown in Table 3. We found 242 proteins (16%) involved with ATP and GTP processes indicating that development provokes a high demand for energy. Of the 185 ATP-associated proteins, 12% are part of the proton-or ion-transporting mechanism (22/185) and 84% (156/185) are involved in ATP-binding. Among ATP-binding proteins, 16% are involved in phosphatase/kinase activities (25/156), 6% in stress response (9/156), and 6% are involved in DNA replication (10/156). There are 60 GTP-associated proteins and 93% (56/60) are involved in GTP-binding. Of the GTP-binding proteins, 42% are active in signal transduction (24/56), 9% in catabolism (5/56), and 13% with cytoskeletal motor activity (7/56). About 7% of the ATP-or GTP-binding proteins play a role in catabolic process and 6% in cytoskeleton motor activity (12/211). In addition to the ATP-associated proteins involved in DNA replication, as we might expect there is strong response for proteins involved in cellular multiplication and transcription, 5% as nucleic acid binding proteins (72/1496), 4% as DNA processing proteins (61/1496), and 2% as proteins of DNA transcription (2%, 23/1496) that includes those engaged in DNA binding (35%, 8/23) and polymerase activity (35%, 8/23).
The ovary PBM proteome contains 10% (146/1496) transport proteins that include proteins involved in both vesicle (20/146) and intracellular transport (40/146). Among the vesicular transporters, the clathrin and COP vesicular transport constitutes the largest aggregate (10/20). The COP-II vesicle pathway is important for iron metabolism as our previous work indicates that ferritin is secreted from mosquito cells using this pathway (Geiser et al. 2009).
Blood feeding induces oxidation reactions in the mosquito (Lima et al. 2012, Bottino-Rojas et al. 2015. To handle the redox toxicity of a bloodmeal, Ae. aegypti dedicate ~2% (35/1496) of the identified proteins for oxidation-reduction processes with the largest component as oxidoreductases (49%,17/35). This also could reflect the engagement of these proteins in the movement of electrons and electron capture.
Female mosquitoes seek a specific host for blood feeding. About 3% (39/1496) of the ovary and egg proteome is invested in odorant binding proteins known for their involvement in host preference detection (International Glossina Genome 2014). Akbari, et al. (2013) found the ovarian transcriptome enriched in odorant transcripts during oogenesis. Odorant proteins have been observed in Drosophila male reproductive organs and could be transferred to females during mating. Odorant proteins were also found in Aedes eggshell and Anopheles eggs (Yamamoto and Takemori 2010, Sirot et al. 2011, Mastrobuoni et al. 2013, Marinotti et al. 2014). Akbari, et al., (2013) found that gene expression was upregulated 5-fold in the ovary PBM and that the transcriptome signature differed during oogenesis. Thus, as expected, some ovary proteins in the current study are differentially expressed at each developmental stage:108 proteins that were detected only at the early developmental stage, while 76 proteins were expressed only at the later stage ( Fig. 1B, Tables 4 and 5, respectively. Although many of these proteins are found in organisms from other phylogenetic lineages (conserved hypothetical proteins), they remain functionally uncharacterized in Ae. aegypti (Galperin and Koonin 2004), while others are predicted from an open reading frame (hypothetical proteins) with no experimental evidence of translation (Desler et al. 2009). These data offer initial proof that the hypothetical proteins are expressed. Supplement 4 (online only) lists early (24 hr) and late (72 hr) stage-specific ovary proteins obtained from DAVID functional annotation cluster analysis that provides gene enrichment information against the background of the whole Ae. aegypti genome. According to Huang et al. (2009a, b), the DAVID gene enrichment analysis highlights the most over-represented (enriched) biological annotations out of thousands of linked terms and contents found in a genomic background. While the DAVID results demonstrate the enrichment of some proteins compared to the whole Aedes database, it does not account for all the proteins identified in our analysis. Thus, we provide this analysis as supplementary data, and we report protein numbers/category rather than DAVID enrichment.

Ovary Metal-Associated Proteins Expressed PBM
The second most abundant category of PBM ovary proteins identified by GO analysis is the metal-associated proteins. These proteins represent 11% (171/1496) of the female ovary proteome, 22% have iron (38/171) or zinc (38/171) as a co-factor ( Fig. 2A), and 21% (36/171) are proteins associated with calcium. Of calcium-associated proteins, 17% (6/36) are integral membrane proteins. The database of the Ae. aegypti ovary metal-associated proteins identified from our ovarian database is available in Supplement 5 (online only). Of the 171 unique proteins in this database, 17 were identified only at 24 hr PBM ( We also identified proteins associated with iron/heme metabolism (Table 6) and analyzed the iron/heme gene enrichment against the whole Ae. aegypti genome (Supplement 6 [online only]). Of the iron-binding proteins, 26% are engaged with iron-sulfur clusters (11/42). Iron sulfur clusters are required for proteins involved in numerous processes, including the electron transport chain as well as defense/detoxification (e.g., NADH-ubiquinone oxidoreductase (AaeL_AAEL005508, AaeL_AAEL012552), ABC transporter (AaeL_AAEL010059), ubiquinol-cytochrome c reductase ironsulfur subunit (AaeL_AAEL003675), Maio 2017, Braymer andLill 2017). We also found several homologs of proteins that participate in mammalian iron metabolism. We identified ferrochelatase (FECH/protoheme ferrolyase; AaeL_AAEL005415), the enzyme that catalyzes the insertion of Fe 2+ in the terminal step in heme biosynthesis converting protoporphyrin IX into heme (Hamza and Dailey 2012); mitochondrial aconitase (mAco/ACO2: AaeL_AAEL003734), an iron-sulfur cluster enzyme involved in the regulation of cellular energy metabolism (Lushchak et al. 2014); succinate dehydrogenase (SDH: AaeL_AAEL010330), an enzyme involved in the Krebs cycle that is regulated by iron in Drosophila (Melefors 1996, Kim et al. 2012; and cytoplasmic aconitase (cAco/ ACO1, AaeL_AAEL008216). In mammals, under adequate iron conditions, an iron sulfur cluster present in cytoplasmic aconitase gives the protein enzymatic capacity to convert citrate to isocitrate. However, in iron deficient conditions, the cluster is lost from the protein and cytoplasmic aconitase becomes the iron regulatory protein 1 (IRP1). IRP1 regulates translation of several proteins involved in iron homeostasis and was previously characterized in mosquitoes (Zhang et al. 2002, Ghosh et al. 2015, Holmes-Hampton et al. 2018. We also found a multicopper oxidase identified by GO analysis, Laccase-like multicopper oxidase 1 (AaeL_AAEL007415). In mammals, multicopper oxidases involved in iron metabolism allow ferrous to ferric conversion (Vashchenko and MacGillivray 2013). More detailed information on multicopper oxidases in insects can be found in the review by Dittmer and Kanost (2010). As expected, we identified transferrin (TF: AaeL_AAEL015458) and the ferritin subunits. Although mosquito TF binds a single iron atom, it does not serve as a major transport protein in mosquitoes (Geiser and Winzerling 2012). Available information indicates it is more likely involved in mosquito immunity (Yoshiga et al. 1997, Harizanova et al. 2005) and development (Puri et al. 2008). Ferritins can store up 4,500 atoms of iron (Theil 2012, Plays et al. 2021. Ferritin nanocages self-assemble from mixtures of 24 or 12 catalytically active (heavy chain [H/FTH1]) and inactive, stabilizing (light chain [L/FTL1]) subunits (Theil 2013). In mosquitoes, ferritin functions as both a major iron storage protein and an iron transporter in hemolymph from the gut to the tissues (Zhou et al. 2007, Geiser et al. 2009).
We anticipated identifying the HCH (AaeL_AAEL007385) and LCH1 (AaeL_AAEL007383) ferritin subunits; surprisingly, we also found a second ferritin light chain homologue (LCH2, AaeL_ AAEL002158) expressed at both development stages PBM. We empirically evaluated this subunit and found that LCH2 differs from all other ferritin subunits in that it is unresponsive to iron exposure. We reported high levels of LCH2 message and LCH2 protein in ovaries of Ae. aegypti (Geiser et al. 2017). Despite numerous studies on mosquito ferritin, the LCH2 had not been previously identified, perhaps, because available antiserums failed to react with LCH2 and the mass of LCH2 is similar to that of the other ferritin subunits.
Some mosquito iron/heme-associated proteins are differentially expressed PBM by developmental stage (Fig. 3); 2 proteins were detected only at the early stage (Table 6, light grey) and 5 proteins were detected only at the later stage (Table 6, dark grey). Procollagenlysine,2-oxoglutarate 5-dioxygenase (AaeL_AAEL008099), an orthologue to Drosophila Procollagen lysyl hydroxylase (Plod) that was recently shown to contribute to egg elongation in flies (Jha et al. 2015), was found only at 24 hr PBM, as was one of the four cytochrome P450s, AaeL_AAEL004054. A different cytochrome P450, AaeL_AAEL002046, was detected only at 72 hr PBM. In Ae. aegypti, cytochrome P450s are a family of 160 proteins that act on paired donors with incorporation or reduction of molecular oxygen (Stevenson et al. 2012) participating in heme and iron binding, oxidoreductase activity, and monooxygenase activity. AaeL_AAEL004054 gene expression is upregulated by nitroquine treatment, an antimalarial drug, in An. stephensi [Liston (Diptera: Culicidae); Indo-Pakistan malaria mosquito] (Zhang et al. 2014), while elevated levels of AaeL_AAEL002046 gene expression has been shown in pyrethroid resistant Ae. aegypti populations (Stevenson et al. 2012). Two peroxinectins also were detected only at 72 hr PBM (AaeL_AAEL004401 and AaeL_AAEL003612). These are orthologous to the Drosophila peroxinectin-like gene (Dpxt) that is expressed at high levels in late oogenesis, and may be involved in microbicidal and apoptotic cell phagocytosis as well as the cell adhesion processes (Vazquez et al. 2002). Two conserved hypothetical proteins, AaeL_AAEL000496 and AaeL_AAEL000507, also are present only at 72 hr PBM, and are predicted to have peroxidase activity, bind heme, and respond to oxidative stress.
The other conserved hypothetical proteins present at both time points PBM are predicted to be involved in oxidation-reduction (AaeL_AAEL001185) and in iron-sulfur cluster binding (AaeL_ AAEL004234). While the only hypothetical protein in this list is  a duplicate entry of the peroxinectin gene, AaeL_AAEL004390, that binds heme, participates in the response to oxidative stress and in eggshell chorion assembly, and has peroxidase activity (Vazquez et al. 2002, Schmidt et al. 2010, Sivakamavalli et al. 2016. Notably, during our analysis, two proteins were identified from species other than Ae. aegypti: UniProtKB ID: Q5QIR4 and UniProtKB ID: D3JAI6. Both are from the Wolbachia endosymbionts, which is a maternally inherited intracellular bacterial symbiont that has been shown to regulate iron metabolism and fecundity in insects (Kremer et al. 2009, Gill et al. 2014).

Comparison of Our Shotgun Ovarian Proteome Database with Other Relevant Published Databases
Ovarian proteins, including ferritin, could come from hemolymph, tissue synthesis, or both. Paskewitz and Shi (2005) reported ferritin heavy and light chain1 subunits in a proteomic study of hemolymph proteins from An. gambiae [Giles (Diptera: Culicidae); African malaria mosquito]. The presence of ferritin in hemolymph and transcripts for all three subunits in ovarian tissues of Ae. aegypti has been documented previously (Geiser et al. 2017). When we compared our protein database to that provided by Paskewitz and Shi using the protein name as an identifier, we also found putative    Due to limited resources, we were able to analyze only two development stages and selected those that we thought would be most likely to express the greatest numbers of different proteins. We did not evaluate proteins expressed by ovaries before blood feeding. However, Dias-Lopes et al. (2016) identified proteins in the reproductive tract of sugar-fed An. aquasalis [Curry (Diptera: Culicidae); South and Central American malaria mosquito]. We evaluated our database for the presence of the proteins in the Dias-Lopes et al. database using the protein name/description as provided in their reported data as an identifier (Online Resource 1 [Dias-Lopes et al. 2016]). We included their 'identified' and 'putative' proteins, however, excluded their 'uncharacterized' proteins and proteins identified with only one peptide. Of more than 600 proteins, we found 403 (67%) with a potential match in our database, 250 identified proteins and 153 putative proteins (Supplement 7 [online only]). Thus, it is likely that these proteins are expressed in ovaries before blood feeding, as well as during oogenesis. Indeed, Dias-Lopes et al. (2016) reported identifying nine proteins involved in oogenesis expressed in the sugar-fed female including ferritin HCH, ferritin LCH1, and transferrin. The remaining proteins in the Dias-Lopes et al. database could not be matched by protein name or description to proteins in our database. A further analysis showed that, of the iron-associated proteins we identified using shotgun proteomics, only ferritin HCH, ferritin LCH1, and transferrin were present in the Dias-Lopes et al. database. This would suggest shotgun proteomics successfully detected several iron-associated proteins with increased expression during development, and that these proteins when further evaluated could help to complete the picture of iron metabolism in these animals. Sreenivasamurthy et al. (2018) also reported a comparison of proteins expressed by several tissues including the ovary in sugar-fed An. stephensi, the reader is referred to this source for further information on proteins expressed before blood feeding. Marinotti, et al. (2014) used shotgun proteomics to evaluate the eggshell proteome from Ae. aegypti ovaries. They obtained the eggs by dissection at 72 hr PBM and extracted only the eggshell proteins. Of the 127 proteins they reported, we found 51% matching proteins in our database. We might expect a lowered matching with the eggshell proteome, as we did not obtain the eggshell proteins by a separate dissection and extraction. Akbari et al. (2014) reported the transcriptome for Ae. aegypti ovaries for several developmental time points (nonblood fed,12,24,36,48,60,and 72 hr PBM). We compared our protein database to their gene database using the short gene name as an identifier. Using annotated genes only, we found matching transcripts for 99.4% of the proteins in our database (Supplement 8 [online only]). Of the proteins we identified only at the early development stage, 60.7% of Fig. 3. Ae. aegypti Ovary Iron/Heme-Associated Proteins Expressed at Different Developmental Stages. Scaffold identified proteins were categorized using GO terms obtained as described in Fig. 1 to find iron/hemeassociated proteins. Proteins detected in early development (24 hr) and late development (72 hr) PBM were identified using the stringency criteria described in the Materials and Methods. Of 38 iron/heme-associated proteins identified using GO terms, 2 proteins were detected only at 24 hr PBM, while 5 proteins were detected only at 72 hr PBM.

Fig. 2. A. Types of Metal-Associated Proteins Identified in
Ae. aegypti Developing Ovaries. Proteins identified using Scaffold were categorized using GO terms obtained as described in Fig. 1 to find metal-associated proteins. Metal-associated protein categories were expressed as a percentage of the total number of metal-associated proteins identified from both time points PBM. B. Metal-Associated Proteins Expressed at Different Developmental Stages. Metal-associated proteins detected in early development (24 hr) and late development (72 hr) PBM were identified using the stringency criteria described in the Materials and Methods. Of 171 metal-associated proteins identified using GO terms, 17 proteins were detected only at 24 hr PBM, while 7 proteins were detected only at 72 hr PBM.  We also identified transcripts for 100% of the proteins identified as iron-associated in our ovarian proteome database. Although it is beyond the scope of this project, it will be interesting to see if the proteins we identified have transcripts and/or genes with known iron or metal responsive control elements.

TMT-Labeling Proteomics Analysis of Proteins Expressed on Different Diets
At the time we obtained our data, methodology for quantitating shotgun proteomic data were not available. To obtain some quantitative data on ovarian protein expression in Ae. aegypti, we evaluated proteins expressed in response to an iron-controlled meal using proteomics with TMT-labeling. Due to limited resources, we selected to analyze proteins expressed only during the early developmental stage (24 hr PBM). The two runs (each in triplicate) identified 134 proteins (Tube A) and 96 proteins (Tube B) for a total of 230 proteins (Supplement 9 [online only]). Of these 230 proteins, 88 proteins were identical between runs and 65 of the identical proteins showed significantly different expression on the varied diets. A comparison of the TMT-labeled protein database with our shotgun ovarian database showed of the 65 differentially expressed proteins 47 were present. Why our subset of TMT-labeled proteins did not exactly match to the shotgun proteomic database we do not know. It is likely that proteins expressed on the artificial diets differed from those expressed following blood feeding as of the 18 proteins that did not match, 15 showed greater expression on the artificial diets than for animals fed on blood. Alternatively, the great differences in how the samples were prepared for analysis could account for a substantial amount of variation. For the shotgun proteomics experiments all samples were separated on gels, digested in-gel, and the individual gel slices analyzed, and then the data combined for each sample. Whereas the TMT-label experiments were from artificial bloodmeal and Porcine BM samples, trypsin digested in solution, isobaric labeled for each individual diet, and then combined for the analysis. Twenty-nine of the proteins with significantly different expression (45%, Table 7, (-)Fe>(+)Fe) showed greater expression on the (-)Fe diet than on the (+)Fe diet, and of these, 23 (79%) showed greatest expression on (-)Fe diet relative to either the (+)Fe diet or the blood meal ((-)Fe>Porcine BM). Nineteen of these 29 proteins (65%, Table 7, grey shaded) are involved in ribosomal structure or function, including AAEL001420-PA (putative kinase involved in RNA translation), AAEL010097-PA (role in mRNA localization during oogenesis [Akbari et al. 2014]), Elongation factor 1-alpha (transfer of aminoacyl tRNAs to the ribosome [Sasikumar et al. 2012]) and RNA helicase. Among the remaining proteins, 14-3-3 protein is from a family of regulatory proteins found in most species that are essential components of phosphorylation-mediated signaling (Aitken 2006, Tinti et al. 2012, while reduced heat shock cognate 70 has been shown to permit cell proliferation (Miller and Fort 2018). AAEL011197-PA and AAEL005097-PA are two predicted members of the actin family.
Only four proteins (Table 7, boldfaced, (+)Fe>(-)Fe) were identified with a significantly increased expression on the (+)Fe diet relative to Fe(-) diet. These included: vitellogenin-A1, a protein synthesized Difference between increased or decreased expression of a protein due dietary iron was significant (p<0.05) using the Holm-Sidak method for multiple unpaired t-tests in Prism 6 as described in the Materials and Methods. UniProtKB ID = UniProt (http://www.uniprot.org/); Boldfaced = Increased expression on the (+)Fe diet; Grey shaded = Involved in ribosomal structure or function; Underlined = Significantly different (SD) expression for both runs performed in triplicate; Italics = SD expression for one run performed in triplicate; (+)Fe = ABM with Iron; (-)Fe = ABM without Iron; Porcine BM = Porcine bloodmeal. in fat body and imported into the ovaries during egg formation, as well as AAEL017516-PB and 60S ribosomal protein L7, both constituents of the ribosome.
A comparison of the expression of proteins from (+)Fe fed animals with those of Porcine blood-fed animals identified 38 (Table 7, boldfaced, (+)Fe>Porcine BM) with a greater expression on the (+)Fe diet. These included 25 proteins involved in ribosomal structure or function, as well as, Tubulin alpha chain, Enolasephosphatase E1, AAEL005170-PA (putative cytochrome c oxidase subunit iv, VectorBase), AAEL005515-PF (heterogeneous nuclear ribonucleoprotein), and AAEL011584-PA (putative 60 kDa heat shock protein).
A comparison of proteins from animals fed (-)Fe with Porcine blood-fed animals yielded 48 proteins that showed greater expression on the (-)Fe diet than for Porcine BM (Table 7, (-)Fe>Porcine BM). Thirty-three of these proteins (69%, grey shaded) are involved in ribosomal function and structure-including AAEL004500-PA, a putative translation elongation factor. Of the remaining proteins, AAEL001872-PA is a putative voltage-dependent anion channel (VDAC) mitochondrial porin. Mitochondrial porins function in the outer membrane of the mitochondria to efficiently import carrier precursors (Ellenrieder et al. 2019). An interaction of a VDAC with tubulin has been shown to control mitochondrial metabolism in mammalian cells (Fang and Maldonado 2018). AAEL009642-PA (Cathepsin b), one of only five proteins to demonstrate a significant decrease in the expression on the (-)Fe diet compared to the Porcine BM, was identified in the recent paper by Martins et al. (2021) examining proteins of the head and salivary glands of Ae. aegypti.

Conclusions
We used proteomic analysis to identify proteins in the developing ovaries of Ae. aegypti to provide further insight into mosquito oogenesis and on the expression of iron-associated proteins. As far as we can determine ours is the first report of proteins expressed in this tissue following a bloodmeal and the first to report protein expression at two developmental stages of oogenesis. Our database matches well with the transcript database of Akbari et al. (2014), confirms the expression of the proteins for numerous transcripts they identified and can provide some insight into the timing of translation for proteins expressed in specific developmental stages. Our database also can serve as a platform for others engaged in seeking candidates for interference in the development process of these animals. It is clear from the numbers of hypothetical proteins found by our work that the function of many proteins involved in oogenesis and egg development remains to be elucidated.
We show that metal-associated proteins play an important role in Ae. aegypti oogenesis. In fertilized chicken eggs, it was demonstrated that the yolk is the major source of the minerals, such as Mn, P, Fe, Ca, Cu, and Zn that are essential for early, as well as later, embryonic development (Yair and Uni 2011). Significantly, iron-associated proteins make up 22% of the identified metal-associated proteins in Ae. aegypti ovaries. Iron is required for developing mammals (Gambling et al. 2011, Lipinski et al. 2013, and several minerals, including iron, require protein association for storage, transport, and utilization. Little is known about iron movement in Ae. aegypti cells. In other species, iron movement inside cells requires chaperone proteins and membrane iron transporters. It remains to be determined if the putative ABC transporter and cytochrome B5 we identified might serve in these processes. We identified several proteins engaged in the binding of or the formation of iron sulfur clusters implying these processes are necessary for development in mosquitoes similar to that reported for Drosophila (Marelja et al. 2018). We also identified ferrochelatase, and several proteins known to bind heme in other species. The importance of heme was demonstrated in in Schistosoma mansoni [Larue (Diplostomida: Schistosomatidae); human blood fluke] where inhibition of heme uptake significantly reduced egg production and retarded development (Toh et al. 2015). The importance of iron for oogenesis also was demonstrated in Ixodes ricinus [Linnaeus (Ixodida: Ixodidae); castor bean tick], another blood feeding arthropod (Hajdusek et al. 2009), where RNA silencing of ferritin reduced hatch rate. Iron/heme proteins also are essential for the metabolic response to oxidative stress, iron homeostasis, heme biosynthesis, eggshell chorion assembly, and energy production.
Importantly, through this work we identified the LCH2, and subsequently obtained empirical data supporting the expression of LCH2 in ovaries and its lack of control by iron (Geiser et al. 2017). Why this species expresses a second LCH subunit is not known. Mammalian ferritin is composed of 24 subunits of two types, heavy chain or light chain, and protein isoforms reflect the numbers of each type of subunit present in the molecule (Liu and Theil 2005). So, for example, the storage isoform consists of more light chain subunits than heavy chain subunits. X-ray crystal structure of insect ferritin shows 12 heavy chain subunits and 12 light chain subunits (Hamburger et al. 2005). Thus, we suggest that insects might substitute the different light chain subunits (LCH1 and LCH2) to form ferritin isoforms in these animals. If this is the case, it seems probable that other LCH subunits remain to be found as cDNAs and multiple genes for LCH ferritin subunits in mosquitoes have been reported (Dunkov and Georgieva 2006).
To obtain quantitative data, we employed TMT-labeling of proteins expressed by animals fed an iron-controlled diets. The differentially expressed proteins we identified using this technique were obtained with high confidence in the expression differences according to the criteria defined in the Materials and Methods. Differential expression suggests that iron could be involved in the regulation of expression of some of the proteins we identified either by repressing expression for proteins with the greatest expression on the (-)Fe diet relative to the (+)Fe diet, or by enhancing expression of proteins that showed the greatest expression on the (+)Fe diet relative to that of the (-)Fe diet or the bloodmeal.
Why several proteins involved in ribosomal structure, function, and translation processes show greater expression for animals fed the artificial diets than for animals fed the Porcine BM, we do not know. Given that the ribosomal apparatus is present following blood feeding as shown by the shotgun proteomics analysis and that vitellogenin (a protein imported from fat body into the ovaries) shows the greatest levels in the Porcine blood-fed animals, we speculate that the processes of oogenesis that involve protein synthesis by ovaries vary with time during egg development and in blood-fed animals are reduced by 24 hr relative to earlier time intervals. This suggests that the rate of development for animals fed artificial diets lags that of blood-fed animals. The bloodmeal contains many constituents including numerous nutrients and signaling factors not found in the artificial meals. Possibly some of these factors increase the rate of ovarian and egg development in blood-fed animals. If this is the case, then rate of development would be an important factor to consider when comparing results from animals fed artificial diets with those fed on blood.
In summary, we used two very different approaches to identify potential proteins involved in oogenesis with differential expression associated with iron metabolism. The data collected from the shotgun proteomics analysis identified 1496 proteins including 38 with a known or a potential relationship to iron metabolism. The fitness of this approach has been demonstrated by empirical evaluation of LCH2. The second study showed isobaric labeling could detect differential expression of proteins for animals fed two types of artificial diets varying in one nutrient and identified some candidates that showed expression control by iron. The further usefulness of this technique remains to be seen as identified proteins are explored for their roles in iron metabolism.
Since we conducted these analyses, others have shown that isobaric labeling can be coupled with shotgun proteomics to produce reliable large-scale quantification. This is important when seeking proteins involved in mineral metabolism as many could be less abundant proteins and would be missed using only isobaric labeling methods without shotgun proteomics. In support of this notion, Martins et al. (2021) recently combined the techniques of shotgun proteomics with isobaric protein labeling and reported results for the proteome from heads and salivary glands of Ae. aegypti infected with the Wolbachia wMel strain and ZIKV. They recovered more than 4,000 proteins. Their work as well as that of others (De Mandal et al. 2020) supports the power and the recommendation of combining these two methods in future studies with similar goals.

Supplementary Material
Supplementary data are available at Journal of Insect Science.