A proteomics approach reveals molecular manipulators of distinct cellular processes in the salivary glands of Glossina m. morsitans in response to Trypanosoma b. brucei infections

Glossina m. morsitans is the primary vector of the Trypanosoma brucei group, one of the causative agents of African trypanosomoses. The parasites undergo metacyclogenesis, i.e. transformation into the mammalian-infective metacyclic trypomastigote (MT) parasites, in the salivary glands (SGs) of the tsetse vector. Since the MT-parasites are largely uncultivable in vitro, information on the molecular processes that facilitate metacyclogenesis is scanty. To bridge this knowledge gap, we employed tandem mass spectrometry to investigate protein expression modulations in parasitized (T. b. brucei-infected) and unparasitized SGs of G. m. morsitans. We annotated the identified proteins into gene ontologies and mapped the up- and downregulated proteins within protein-protein interaction (PPI) networks. We identified 361 host proteins, of which 76.6 % (n = 276) and 22.3 % (n = 81) were up- and downregulated, respectively, in parasitized SGs compared to unparasitized SGs. Whilst 32 proteins were significantly upregulated (> 10-fold), only salivary secreted adenosine was significantly downregulated. Amongst the significantly upregulated proteins, there were proteins associated with blood feeding, immunity, cellular proliferation, homeostasis, cytoskeletal traffic and regulation of protein turnover. The significantly upregulated proteins formed major hubs in the PPI network including key regulators of the Ras/MAPK and Ca2+/cAMP signaling pathways, ubiquitin-proteasome system and mitochondrial respiratory chain. Moreover, we identified 158 trypanosome-specific proteins, notable of which were proteins in the families of the GPI-anchored surface glycoproteins, kinetoplastid calpains, peroxiredoxins, retrotransposon host spot multigene and molecular chaperones. Whilst immune-related trypanosome proteins were over-represented, membrane transporters and proteins involved in translation repression (e.g. ribosomal proteins) were under-represented, potentially reminiscent of the growth-arrested MT-parasites. Our data implicate the significantly upregulated proteins as manipulators of diverse cellular processes in response to T. b. brucei infection, potentially to prepare the MT-parasites for invasion and evasion of the mammalian host immune defences. We discuss potential strategies to exploit our findings in enhancement of trypanosome refractoriness or reduce the vector competence of the tsetse vector.


(Continued from previous page)
Conclusions: Our data implicate the significantly upregulated proteins as manipulators of diverse cellular processes in response to T. b. brucei infection, potentially to prepare the MT-parasites for invasion and evasion of the mammalian host immune defences. We discuss potential strategies to exploit our findings in enhancement of trypanosome refractoriness or reduce the vector competence of the tsetse vector.
Keywords: LC-MS/MS, Protein-protein interaction, Metacyclic trypomastigotes, Metacyclogenesis, Vector competence, Trypanosome refractoriness Background Trypanosoma brucei group causes African trypanosomoses, a group of neglected zoonotic tropical diseases endemic in 37 sub-Saharan African countries [1], against which there are no effective vaccines or drugs [2,3]. Amongst the medically and agriculturally important tsetse fly species, Glossina morsitans morsitans is highly significant in the savannah woodlands and is the primary vector of the species of T. brucei group [4].
In order to complete their complex developmental cycles in the tsetse vector, trypanosomes face two critical replicative stage barriers: colonization of the midguts and establishment in the salivary glands (SGs). The pathway followed by the parasites between these two independent multi-replicative stages is time-dependent, irreversibly transient and eventually influences metacyclogenesis in the SGs (differentiation into the mammalian-infective metacyclic trypomastigote (MT) parasites) [5][6][7][8][9]. Critical replicative events occur~3 days post-infective blood meal when only a small proportion (≤ 10 %) of the parasites is able to pass through the midgut barrier [8]. Upon establishment of infection, parasite numbers per fly gut remain remarkably constant [6,8]. However, since the SGderived MT-parasites are uncultivable in vitro, molecular mechanisms that promote metacyclogenesis remain to be investigated. Whilst the SGs determine success of metacyclogenesis, there is limited knowledge on how trypanosomes adapt to and evade the host defence responses in the SGs [10]. It is however generally known that trypanosomes modulate SGs microenvironment [11], and that factors such as parasite genotypes, midgut antioxidant status, lectins and environmental stimuli influence parasite maturation [12,13]. Most trypanosome research has focussed on the bloodstream and/or in vitro cultured procyclic forms, thus creating a knowledge gap with regard to metacyclogenesis. The availability of complete genomes of G. morsitans [14] and T. brucei [15] makes it possible to identify proteins involved in the mechanisms that facilitate metacyclogenesis. Such proteins are ideal candidates to develop improved strategies for tsetse and trypanosomosis control, especially via the sterile insect technique (SIT) programs [16], which have so far been employed in areas without active disease transmission [2].
Vector competence is one of the key pillars in the SIT programs, which involves mass release of sterile males into wild populations of the target species. More importantly, the sexually sterilized males are still competent trypanosome vectors, thus increasing risks of disease transmission when millions of sterile males are released into trypanosome-infested areas. Attempts have been previously made to make the sterile males vector incompetent via drug-supplemented blood meals, an approach now known to be inefficient [17]. Thus, alternative and/ or complementary approaches are necessary, especially with the risk of trypanosomes developing resistance to the trypanocidal drug supplements. In this regard, metacyclogenesis and transmission of the mammalian-infective MT-parasites potentially represent vulnerable and attractive intervention points to enhance the natural trypanosome refractoriness or reduce the vectorial competence of the sterile males used in the SIT campaigns.
We hypothesized that T. b. brucei manipulates expression of proteins involved in pathways that specifically prepare the MT-parasites for successful transmission to and infection of susceptible mammalian hosts. To test this hypothesis, we employed tandem mass spectrometry (LC-MS/MS) to determine T. b. brucei-induced protein expression modulations in parasitized SGs of G. m. morsitans compared to unparasitized SGs. We also aimed at highlighting major metacyclic T. b. brucei-specific proteins, which are potentially critical for parasite survival in the SGs and transmission to susceptible mammalian hosts. We discuss our findings from the perspective of potential approaches to enhance trypanosome refractoriness in tsetse as an antivector strategy against African trypanosomosis.

Tsetse flies and parasites infections
Male G. m. morsitans were obtained from the Institute of Tropical Medicine (Antwerp, Belgium) and infected with a highly transmissible T. b. brucei strain (EATRO 1125 AnTaR 1) [11]. Male flies were used because they mature significantly more midgut infections than females [5]. For the infections, teneral flies (24-48 h post-adult eclosion) received their first blood meals supplemented with~12 μg of parasites/fly [11]. Fully-engorged flies were selected, reared for 28 days post-infection (dpi) in controlled insectaria (65 % relative humidity; 26°C) and subsequently fed on clean (trypanosome-free) defibrinated horse blood (in vitro; 2-3 times/week) [18]. Control flies were prepared from males of the same batch, age and feeding history as the parasitized flies. The control flies were maintained on clean blood meals and handled as described for their parasitized counterparts. All assays consisted of biological triplicates (20 flies per group) in small holding cells (3.5 cm in diameter × 6 cm height).

Parasite scoring and SG dissections
Forty-eight h after the last blood meal, the flies were scored for the presence of the MT-parasites in the SGs using phase-contrast microscopy (×400) as previously described [11]. Briefly, the SGs were considered parasitized if stuffed with trypanosomes (i.e. the fly's spit full of the parasites). Fly's spit (from the control group) completely devoid of trypanosomes were considered unparasitized. For mass spectrometry, 10 intact pairs of SGs were selected from each of the replicated fly groups and immediately preserved at -20°C in 150 μl sterile saline supplemented with complete protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany).

SG extracts preparation and SDS-PAGE
SGs were individually homogenized using a glass/Teflon homogenizer and ultra-sonicated (Sonifier cell disruptor, Branson Instruments Co., Stanford, Connecticut, USA). Homogenates were freeze-thawed and clarified three times by centrifugation (7500× g; 10 min; 4°C) to completely remove cell debris. Supernatants were pooled and proteins quantified using the standard BCA method (Bio-Rad, Hercules, California, USA). Equal protein quantities (600 ng) were resolved in 12 % SDS-PAGE gels and stained (CBB stain; NuPAGE Novex; Invitrogen Life Technologies, Carlsbad, California, USA). The middle sections of entire gel lanes were longitudinally excised (from top to bottom), divided into five equal fractions (covering the entire gel lanes) and cut into small pieces (~1 mm 3 ) as previously described [19].

LC-MS/MS measurements and protein identification
Tryptic peptides for subsequent LC-MS/MS measurements were prepared as previously described [19]. Briefly, the gel pieces were washed with 50 mM ammonium bicarbonate (ABC) buffer and ABC buffer/50 % acetonitrile (ACN) and proteins reduced and alkylated using dithiothreitol and iodoacetamide. Gels were washed with ABC/ ABC-ACN buffer, followed by in-gel trypsin digestions and LC-MS/MS measurements [20]. Proteins were identified by searching the MS/MS spectra (using MaxQuant/Andromeda [21,22]) against G. m. morsitans and T. brucei databases (downloaded from UniProt), a decoy database (constructed by reversing all the protein sequences) and a database of common contaminants (available from Max-Quant). MaxQuant search parameters included fixed carbamidomethylation (C), oxidation (M), acetylation and deamidation (N and Q). Two peptides (at least one unique and unmodified) matching the same protein were required for protein identification at a maximum false discovery rate (FDR) of ≤ 0.01. The unique identifiers of the proteins downloaded from the UniProt databases were used to identify and classify the LC-MS/MS peptides as specific to the host (G. m. morsitans) and the parasite (T. b. brucei). LC-MS/MS peptide hits to the decoy database and hits with modified peptides only were deleted from the final list of protein groups.

Protein quantification and normalization
Unique and 'razor' (non-unique) peptides were used for peptide assignments and protein quantification [21,23]. For each of the above-described five gel fractions, peptides were matched across different MS/MS runs based on mass and retention time ('match between runs' of 2 min). Label-free quantification (LFQ) was enabled. To minimize technical variations and to easily compare abundances of the same proteins (parasitized vs unparasitized), Log 10 normalized LFQ was used across the biological triplicates. To compare levels of different proteins from the same samples (parasitized and unparasitized), Log 10 iBAQ (intensity-based absolute quantitation) was used [24]. To determine up-and downregulated proteins (parasitized vs unparasitized SGs), t-tests were performed on Log 10 LFQ using the MaxQuant's Perseus module. Proteins were considered to be up-or downregulated when their Log 10 iBAQ ratios (parasitized vs unparasitized) were larger or smaller than zero, respectively, and significantly upregulated when the FDR was ≤ 0.05.

Gene ontology (GO) annotations and PPI network analyses of upregulated SG proteins
Blast2GO v. 3.2 [25] was used to classify the identified proteins along three biological aspects, i.e. biological process (BP), molecular function (MF) and cellular component (CC) Gene Ontology (GO) terms. It should be noted that the GO terms are descriptions of the different protein functional classes and how they relate to each other. To provide a broader overview of the ontologies, the GO classes were grouped into GO-slim terms using CateGOrizer [26]. To correctly place the significantly upregulated proteins within signaling pathways and networks, computational predictions of protein-protein interactions (PPIs) were inferred using interolog mapping [27]. For this, human orthologs to the identified G. m. morsitans proteins were obtained from Ensembl [28] and used to generate an exhaustive list of possible interacting protein combinations using custom in-house Python scripts. Putative interactions for these combinations were then determined using FpClass [29] (probability ≥ 0.2) and mapped back onto the SG protein datasets obtained from the above-mentioned LC-MS/MS measurements. The resulting significant interactome was rendered in Cytoscape v3.30 [30].

Results
Parasitized SG proteome of G. m. morsitans Analysis of the LC-MS/MS data resulted in 5469 and 4366 total and unique peptides, respectively. These peptides mapped to 874 protein groups. Removal of the common contaminants, applying extra filter steps such as removing single peptide hits (thereby decreasing FDR to below 0.01) and hits to the decoy database resulted in 523 non-redundant (nr) protein groups (Additional file 1: Table S1). Of these, 363 protein groups had specific peptide hits to the tsetse vector (G. m. morsitans) proteins (Additional file 1: Table S2) and 158 protein groups had specific peptide hits to the parasite (T. b. brucei) proteins (Additional file 1: Table S3). We also obtained peptide hits to four proteins specific to the bacterial endosymbionts Wigglesworthia glossinidia (n = 3) and Sodalis glossinidius (n = 1) (Additional file 1: Table S4). These symbiont proteins have been reported in the genome of G. m. morsitans [14]. It should be noted that there were no host and/or parasite and/or symbiont proteins that grouped together in one protein group. Figure 1 shows the abundance distribution of the proteins identified in parasitized SGs compared to unparasitized SGs. Table 1 presents the topmost abundant 25 proteins, amongst which were the following blood feeding-associated proteins: tsetse salivary gland proteins-1 & 2 (Tsal1/2), tsetse salivary growth factors-1 & 2 (TSGF-1/2), salivary antigen-5-protein (TAg5), 5′-nucleotidase-related saliva protein (5′Nuc), adenine deaminase (ADA) and 5′-nucleotidase-related SG protein-3 (Sgp3). Also abundant were proteins related to cellular proliferation/differentiation (adenosine deaminase-related growth factors, muscle LIM protein at 84B), cytoskeletal proteins of the actin and myosin families, lectins, molecular chaperones (e.g. 14-3-3 protein zeta family chaperones) and enzymes (e.g. trehalose-6-phosphate synthase, isovaleryl-CoA dehydrogenase, and aspartate aminotransferase).
Amongst the least abundant proteins that we identified included dihydrolipoamide transacylase α-ketoacid dehydrogenase (DBT) complex proteins (for amino acid  Except salivary secreted adenosine (significantly downregulated), all the other proteins shown in this table were insignificantly upregulated (n = 10) or downregulated (n = 15) in parasitized SGs compared to the unparasitized GGs. Upregulated and downregulated proteins are indicated by upward (↑) and downward (↓) arrows, respectively. The unique peptides for each of the proteins listed in the table are shown in square brackets in column 5 metabolism), GTP-binding proteins, Ras-related protein and clathrin adaptor complex proteins (for endocytotic trafficking), 26S proteasome regulatory complex proteins (for protein turnover), bifunctional ATP/sulfurylaseadenosine 5′-phosphosulfate (APS) kinase (for uptake/ assimilation of inorganic sulphate) and transloconassociated (TRAP) complex proteins (for endoplasmic reticulum (ER)-targeting of nascent polypeptides) (see Additional file 1: Table S2). The parasitisation of G. m. morsitans SGs appears not to have affected the expression of at least four host proteins, including bifunctional methylene-tetrahydrofolate dehydrogenase (MTHF-folD) (for thymidylate/methionine/purine synthesis) and 25-/28-kDa glutathione Stransferase 1 (GST1) proteins (for detoxification and lipophilic compound transport) (Additional file 1: Table S8).
Gene ontology (GO) and PPI network analyses of parasitized SGs during T. b. brucei infection All of the significantly upregulated proteins were associated with at least one GO term for a total of 9323 term occurrences. Using the CateGOrizer, the GO terms were trimmed down and grouped into 127 Go-Slim terms associated with biological processes (BP; 55.1 %), molecular functions (MF; 25.9 %) and cellular components (CC; 19 %) ontologies (Fig. 2). Proteins involved in metabolism and cell-or organelle reorganization were present amongst the key BP ontologies (Additional file 1: Table  S9), while the majority of MF ontologies represented proteins associated with catalytic, binding and hydrolase activities (Additional file 1: Table S10). Topmost in the CC ontology were proteins localized in the cellular, intracellular and cytoplasmic compartments of the SGs (Additional file 1: Table S11).
The predictions of protein-protein interaction network of the upregulated host proteins resulted in 10,861 putative interacting protein pairs, of which 225 pairs had significant interacting probabilities (Additional file 1: Table S12). A single high-scoring network was obtained consisting of 88 nodes, a clustering coefficient of 0.107 and an average number of neighbours of 5.0 (Fig. 3). Eighteen of the significantly upregulated proteins formed nine of the main PPI network hubs (Fig. 3). Seven of these proteins formed network nodes with the most edges including SUMO, TcP-1s, mitochondrial NADHubiquinone oxidoreductase (nuo-24), V-ATPase-D, Drk and GS.
Here, it is important to note that FAZ and C-terminal motor kinesin are important in the adjustments of the flagellar positions/sizes [32], depending on the parasite life-cycle stages. Twenty-five of the parasite protein groups were of unknown functions (denoted as 'uncharacterized proteins' in Additional file 1: Table S3). Top-BLASTp analyses (bitscore >100; E-value ≤ 1.0E-6) of these proteins against the nr-NCBI protein database yielded eight hits to proteins with known functions in flagellate protozoans ( Table 5). Six of these were hits to proteins reported in the recently sequenced genome of the African crocodilian trypanosome, T. grayi (vectored by G. palpalis), which is more closely related to T. cruzi than T. brucei [33]. Other homologies included proteins associated with antigenic variations (Pro-Glu/polymorphic GC-rich repeat (PE-PGRS) protein [34] and acyl-CoA-binding protein [35]), parasite proliferation (auxin-induced in root cultures 9 (AIR9)-like protein [36]) and the chemosensational intraflagellar transport, osmotic avoidance abnormal protei3 (OSM3)-like kinesin [37].
Discussion A robust immune system in tsetse midguts makes the flies naturally Trypanosoma-refractory [38]. Following an infected blood meal, the absolute parasite numbers drastically drop at the midgut barrier (days 1-3), then proliferates (day 4) and stably colonizes the midguts where the established population reaches approximately 5 × 10 5 trypanosomes [8,39]. Trypanosomes then migrate to the ectoperitrophic space (day 5) [39], congregate within the proventriculus (days 6-8) and subsequently colonize and complete metacyclogenesis in the SGs (days 12-18) [8]. The MT-parasites then detach from the SG epithelium into the lumen and are uniquely adapted to infect and survive in susceptible mammalian host [32]. The entire process takes approximately 3-5 weeks [40], implying that at the time of the SG dissections (i.e. 28 dpi), MT-parasites were continuously produced.

Trypanosoma b. brucei suppresses expression of abundant host proteins to promote MT-parasite transmission
Previous transcriptomic analyses on G. m. morsitans SGs reported seven protein families: (i) Tsal1/2; (ii) 5′ Nuc/apyrase; (iii) ADA; (iv) TAg5 family-related proteins; (v) Glycine-glutamate (GE)-rich proteins; (vi) C-type lectins; and (vii) housekeeping genes [13,41]. We identified these protein families with similar abundance profiles as reported across different tsetse species [11,[41][42][43][44][45][46]. Among these, the expression of tag5, tsal1/2, tsgf1/2, 5′-nuc and spg3 genes were previously reported to be decreased in parasitized flies [11]; these previous results agree with our results (see Table 1 and Additional file 1: Table S7). We however did not detect the anticoagulant tsetse thrombin inhibitor (TTI), echoing reports by   Capello et al. [47] and Telleria et al. [13], who noted moderate down-regulation of TTI in parasitized SGs. It is notable that the differentially modulated proteins are involved in key processes such as blood feeding which, despite their high abundance, were downregulated in the parasitized SGs compared to unparasitized SGs (Table 1). Reduction in the expression of these proteins could reduce fly feeding performance, which in turn increases tsetse biting frequencies to achieve full engorgement [11]. One could conclude that an increase in biting frequencies promotes the competence of the tsetse vector in trypanosome transmission.

Trypanosoma b. brucei induces upregulation of host proteins essential for parasite differentiation and survival
Cellular proliferation and homeostasis are important conditions for parasite survival. We identified several proteins, whose upregulation may be advantageous to T. b. brucei (see Table 2). For instance, CaMK is one of the most important regulators of stage-specific morphological differentiation of Trypanosoma, Leishmania and Plasmodium parasites [48]. Upregulation of Serp-2 is also notable owing to its central roles in ER stress, an important cue for Plasmodium parasites to switch to transmissible sexual stages in mosquitoes [49]. Further, one of the parasitic survival tactics is to alter intra-cellular locations, binding partners and/or functions of specific proteins and antagonizing other protein modifications. One of the quickest ways to achieve these changes is via sumoylation [50,51], a posttranslational modification mediated by protein SUMO. Notable also is the upregulation of proteins in the family of the ubiquitous and highly conserved V-ATPases; at least eight different V-ATPase subunits were variously upregulated in parasitized SGs (see Additional file 1: Table S5). Mosquito cells over-expressing V-ATPases are reported to be preferentially invaded by Plasmodium parasites [52]; perhaps trypanosomes has a similar invasion preference for V-ATPase expression in SGs. Homeostasis-associated proteins such as ArgK are also important for parasite survival; dsRNA-mediated silencing of ArgK significantly reduced Plasmodium parasite loads in Anopheles gambiae [53]. In our study, ArgK was not only the most abundant protein (Table 1), but also amongst the upregulated proteins (see Additional file 1: Table S5).

Glossina m. morsitans overexpresses immunity-related proteins in responses to SG parasitisation
Overcoming host immune responses is one of the most challenging parts of trypanosome life-cycle because immunity counter-defence provides a nutritious and equable environment. Lehane et al. [54] reported that out of the 68 putative immunity-related genes, 15 genes were actually overexpressed in the midguts of T. b. brucei-infected G. m. morsitans. Midgut expression of 12 of the 15 genes in T. b. gambiense-infected G. p. gambiensis was recently quantified by Hamidou et al. [55], who noted a time-dependent and variable high expression of five out of the 12 genes. Eight of these previously reported immune-related proteins were upregulated in parasitized SGs (see Table 3), a result in agreement with another transcriptome-based study on parasitized SGs of G. m. morsitans [13]. The only exemption was that only Serp-2 and CatB were significantly upregulated in parasitized SGs; CatB was not reported by Hamidou et al., [55]. These differences potentially indicate tissue-specific differential expression patterns of these proteins, for instance in the midgut versus the SGs. In addition to the above-mentioned previously reported tsetse immunity genes, we also identified other proteins implicated in the immunity of other insects (see Table 3). These included Rac/Rab GTPases (for immune surveillance against pathogens and parasites in Drosophila [56]), ubiquitin protein ligase (for Drosophila humoral systemic immune response [57]), a homolog to the Pro-Glu/polymorphic GC-rich repeat (PE-PGRS) protein (for antigenic variations in intracellular parasites [34]), Tsp42Ed and integrins (for activation of immune signaling Plasmodiuminfected mosquitoes [53,58]) and Drk (implicated in Drosophila hemocyte immunity [59]). Finally, loosely associated to insect immunity is amino acid metabolism; some of the upregulated proteins we identified are related to amino acid metabolism (see Table 1). However, a link between upregulation of proteins involved in amino acid metabolism and parasite infection in the SGs is unclear. This notwithstanding, it is rather obvious that following a blood meal, increase in amino acid metabolism is important for insect immune responses (via activation of proteolytic immunity cascades such as prophenoloxidase) and detoxification (removal of amino acid metabolites). Although likely to occur in the hemolymph, these processes are reportedly present in the SGs of locusts, leafhoppers and aphids [60]. Similar scenario cannot be totally ruled out in the case of tsetse immune responses to trypanosome infections in the SGs.

Trypanosoma b. brucei controls the efficiency of protein turnover in the SGs of T. b. brucei
It is notable that > 20 % of the downregulated hostspecific proteins in parasitized SGs were RPs. This modulation of RP expression is in agreement with recent transcriptome-based studies in the differential expression of the RP genes in T. b. gambiense-infected G. p. gambiensis midguts [61]. Downregulation of RPs implicates protein translational regulation to prepare mammalianinfective MT-parasites, or reflects the fly's adaptive response to SG parasitisation. Notably, the downregulation of RPs appears to be accompanied by upregulation of at  least 17 proteins related to the UPS pathway, suggesting that metacyclogenesis requires protein translation and/or degradation efficiency. Moreover, the importance of protein turnover/folding/modifications during parasite infections is underscored by differential modulations of the UDE homolog [62], TcP-1 components [63] and various molecular chaperones identified in the current study.

Trypanosoma b. brucei induces upregulation of key regulators (host proteins) of diverse pathways
The identification of the significantly upregulated SUMO as one of the top interacting proteins in the predicted PPI network (Fig. 3) is interesting given its central roles in the regulation of various cellular processes including transcription, replication, chromosome segregation and DNA repair. SUMO is required for activation of the Ras/MAPK pathway in Drosophila S2 cells [64]; Drk is one of the essential components of the pathway. Additionally, the 14-3-3 protein family, which constitute a multitude of functionally diverse signalling proteins related to behaviour, is also a SUMO substrate. Sumoylation of 14-3-3 proteins results in modulation of diverse cellular processes, including cell cycle regulation, metabolism control, apoptosis and gene transcription control [65]. Furthermore, sumoylation and ubiquitin-proteasome systems are known to selectively modify the functions, sub-cellular locations and half-life of proteins in a very specific manner to maintain cellular homeostasis [66]. According to our PPI predictions (see Fig. 3 and Additional file 1: Table S12) sumoylation targets include several proteins of the TcP-1 complex,  members of the 70-kDa heat shock protein family (Hsp70-4 and Hsp70-5), translation initiation factors and several enzymes (see Fig. 3).
On the other hand, members of the NADH-ubiquitin oxidoreductases (in our case the nuo-24 protein; significantly upregulated; Table 2) are important for energy metabolism and nucleotide synthesis; these proteins are also the initial enzymes in mitochondrial transport chain (mtECT). In fact, mtECT-related proteins were reported to be upregulated during the invasion of mosquito SGs by Plasmodium parasites [67]. Following a blood meal, GS (a key enzyme in glutamine biosynthesis) is upregulated in mosquito midguts [68], which agrees with the results shown in Additional file 1: Table S5. Glutamine is eventually used for production of chitin, which is also produced in other tissues including the SGs. Perhaps the upregulation of GS is related to chitin biosynthesis in Glossina, requiring modulation during parasite infection as can be observed in our PPI predictions, i.e. Serp-4 and ubiquitinrelated proteins were first neighbours of GS (see Fig. 3). Finally, the central hub occupied by the V-ATPases in the predicted PPI network is noteworthy because these electrolyte/solute trans-epithelia transporters are critical in saliva secretion in other insects such as the blowflies [69]. Trypanosoma. b. brucei may therefore modulate the expression of V-ATPases to facilitate saliva-mediated transmission to the mammalian host, which possibly explains the upregulation of various V-ATPases (Additional file 1: Table S5). Taken together, the central hubs occupied by the significantly upregulated proteins in the predicted PPI network potentially indicate manipulations of diverse cellular processes in response to trypanosome infection. These processes require strict control of the participating proteins and enzymes.
Trypanosoma. b. brucei MT-parasites express specific proteins in preparation to invade mammalian host During metacyclogenesis, the MT-parasites re-acquire the VSG/ISG coat and enter the SG lumens in preparation for invading susceptible mammalian hosts [70]. Expressions of VSG and the VSG-shielded ISGs are activated in the SGs to mediate the parasite's evasion of and/or escape the immune surveillance of the host, and successful establishment of mammalian infections [71,72]. Trypanosomes have other VSG-shielded proteins or proteins that are 'hidden' within the flagellar pocket, where they mediate numerous host-parasite interactions. For instance BARPs are predominantly used by the parasites to attach to the SG plasma membranes and promote the release of VSGs [32,73]. In fact, mutations of BARPs are reported to block cytokinesis in T. b. brucei [74]. Further, expression site-associated gene (ESAG) proteins are widely distributed amongst different Trypanosoma species albeit with low expression [72,75], probably explaining the detection of only few ESAG variants in our study (see Table 4 and Additional file 1: Table S3). Notably, to enhance adaptive immunity, T. brucei switches the expression of VSGs with compatible ESAG variants [76]. Moreover, the identification of members of the diverse and rapidly evolving RHS family in our study is noteworthy (see Table 4). Since RHS proteins are localized in polymorphic subtelomic regions [77], they are likely to confer the MTparasites with selective advantages (via antigenic variations). Additionally, detoxification is also important in parasite metabolism, which underscores the importance of the abundance of TXNs, members of the peroxiredoxin AhpCtype family (Table 4). Abundant in all T. brucei life stages, TXNS are crucial in nucleotide synthesis and adaptive defence (hyperoxide detoxification) [78,79].
The interchange between the functionally and morphologically distinct forms of trypanosomatids (from epimastigotes to trypomastigotes) is accompanied by efficient regulation of gene expression, which is largely posttranslational [85]. Among other strategies, this regulation can be through the control of the expression of translation factors; translation of RPs is reduced during differentiation of trypanosomatids into the MT-parasites [86]. The possibility of this form of translational repression partially explains our detection of only few RPs (see Table 4), potentially because the MT-parasites are essentially growtharrested. Our results echo studies on the on differential expression of RPs in the closely related Chagas disease causative agent, T. cruzi [86].
During active replication, trypanosomes heavily rely on membrane transporters for nutrient and metabolite salvage from the host. The main transporters include purine/ pyrimidine transporters (trypanosomes are incapable of de novo purine synthesis) [87], hexose/sugar transporters (to meet parasite's nutritional requirements) [88], amino acid transporters (for incorporation into proteins by the parasite) [89,90], aquaporins (for osmotic protection and metabolism) [91,92] and fatty acids transporters (for preservation of organelle vitality and expansion of vacuole sizes to the accommodate growing parasite progenies) [93]. Trypanosoma brucei contains a repertoire of 305 transporters clustered in 23 families [94]. We identified only three transporters, all with low abundances (see Table 4). Potentially, the growth-arrested MT-parasites may not require expression of many transporters; expression of the transporter genes is probably activated in the next life-cycle stage, i.e. proliferating bloodstream forms in the mammalian host.

Potential approaches to enhance trypanosome refractoriness in tsetse
Factors such as the lack of effective vaccines, resistance to pesticides/drugs and inefficient public health infrastructures make African trypanosomosis increasingly important in sub-Saharan Africa. The fact that sterile males are still competent trypanosome vectors reduces the efficacy of tsetse and trypanosomosis control via the SIT programs. One of the strategies to solve this problem is to enhance the natural trypanosome refractoriness or reduce the competence of the tsetse vector. This can be achieved by identifying candidate genes and creating transgenic sterile males that are incapable of transmitting trypanosomes to mammalian hosts. Ideally, the transgenic sterile males released into wild tsetse populations could eventually block trypanosome transmission. This would be important especially in the event that the areas into which these males are released have active trypanosome circulation.
Although a proteomic approach is far better (in protein identification and observed quantitative expression changes) than a transcriptomic approach, the former approach is often not precise enough to directly make valid biological inferences, especially due to the posttranslational modifications that play important roles in protein functions. Nevertheless, our proteomics and pathway analyses are important milestones in the identification and characterization of SG proteins that potentially contribute to the trypanosome infection status (refractoriness) in the tsetse vector. Having confirmed the protein abundances and expression patterns (parasitized vs unparasitized SGs), we are now designing bioassays to functionally appraise the biological consequences (by immunoblotting, qPCR, RNAi gene silencing) of the protein level changes of some of the proteins identified in our proteomics data sets. After validation, the next objective is to use modern tools such as the piggyBac transposon or CRISPR/Cas9 systems [95], or paratransgenesis [96] to enhance trypanosome refractoriness of the sterile males that are used in SIT programs. Previous studies have provided proof of principle that these methods are feasible in creation of parasite-refractory insect vectors. For instance, using the piggyBac system, Ito et al. [97] created Plasmodium-refractory anopheline mosquitoes by overexpressing a blood-inducible anti-plasmodium gene (salivary gland/midgut-binding peptide 1; SM1) in a tissuespecific manner. Recently, Gantz et al. [98] used the CRISPR/Cas9 system to create Plasmodium-resistant mosquitoes by expression of blood-inducible and tissuespecific single chain variable fragment antibodies (scFvs) against Plasmodium proteins. In the case of tsetse flies, De Vooght et al. [96] demonstrated that transformed Sodalis can secrete significant amounts of functional Nanobodies against Trypanosoma VSG epitopes. The concept of paratransgenesis has also been demonstrated in the control of Leishmania transmission by sand flies [99]. Taken together, our data take us a step closer towards improved anti-vector methods against tsetse and African trypanosomiasis. We have made significant progress towards this direction in that the Joint FAO/IAEA Division of Nuclear Techniques in Food and Agriculture has brought together research experts in different disciplines in a collaborative coordinated research project to explore ways to enhance trypanosome refractoriness in tsetse [2].

Conclusions
Our data suggest that T. b. brucei modifies SG protein composition and functions (suppression of abundant SG proteins) and induces SG cellular proliferation (upregulation of immunity, stress, homeostasis and translatomerelated proteins). Further, the repertoire of T. b. bruceispecific proteins largely consisted of proteins reminiscent of non-replicative MT-parasites (suppression of RPs and transporters), and proteins critical for preparing the trypomastigotes for invasion and evasion of mammalian host immune responses (over-representation of immunity, signal transduction and virulence-related proteins). In response to T. b. brucei infection of G. m. morsitans SGs, divergent cellular processes appear to be manipulated via modulations of proteins involved in various pathways, which is accompanied by modulations of proteins involved in the control of protein turnover. Similar to the tsetse midgut barrier, SG micro-environment is a critical bottleneck with key determinants to T. b. brucei life-cycle transitions. Our proteomic data provide evidence that these genes are not only transcribed (as evidenced by previous transcriptome-based studies), but are also translated into (potentially functional) proteins. These proteins, especially the immunity-related proteins, present an attractive platform to enhance trypanosome refractoriness as an antivector strategy to control tsetse and African trypanosomosis. However, as we have discussed above, we are aware of the need to experimentally validate our proteomics data, a process which is now on-going.