The I. scapularis genome (Gulia-Nuss et al., 2016) was searched with the specific names of genes encoding for enzymes involved in the major carbohydrate metabolic pathways, glycolysis, gluconeogenesis, pentose phosphate, tricarboxylic acid cycle (TCA), glyceroneogenesis, and mitochondrial oxidative phosphorylation and β-oxidation. When records were not obtained using specific enzyme names, then the I. scapularis genome was searched with the Blastp tool from the Basic Local Alignment Search Tool (BLAST) using the human ortholog as “query” (Altschul et al., 1990; Madden et al., 1996). The sequences with the lowest E-value were selected. The conserved domains of identified protein sequences were classified using the protein families database Pfam (Finn et al., 2014). The I. scapularis orthologs found in the genome were double-checked by searching the Homo sapiens genome database using as queries the tick homologs identified in the previous step.

The quantitative transcriptomics and proteomics data for uninfected and A. phagocytophilum-infected I. scapularis nymphs, female midguts and salivary glands, and ISE6 cultured cells were obtained from previously published results (Ayllón et al., 2015; Villar et al., 2015) and deposited at the Dryad repository database, NCBI's Gene Expression Omnibus database and ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD002181 and doi: 10.6019/PXD002181. For transcriptomics and proteomics analysis in I. scapularis nymphs, female midguts and salivary glands, the procedures were described in Ayllón et al. (2015). Briefly, nymphs and adult female I. scapularis were infected with A. phagocytophilum by feeding on a sheep inoculated intravenously with ~1 × 10⁷ A. phagocytophilum (NY18 isolate)-infected HL-60 cells (90–100% infected cells). In this model, over 85% of ticks become infected with A. phagocytophilum in nymphs, midguts and salivary glands. Ticks (200 nymphs and 100 female adults) were removed from the sheep 7 days after infestation, held in the humidity chamber for 4 days and dissected for DNA, RNA, and protein extraction from whole internal tissues (nymphs) or midguts and salivary glands (adult females). Adult midguts and salivary glands were washed in PBS after collection to remove hemolymphs-related cells. Uninfected ticks were prepared in a similar way but feeding on an uninfected sheep. Two independent samples were collected and processed for each tick developmental stage and tissue. After RNA sequencing on an Illumina Hiseq 2000, TopHat was used to align the reads to the I. scapularis (assembly JCVI_ISG_i3_1.0; http://www.ncbi.nlm.nih.gov/nuccore/NZ_ABJB000000000) reference genome. Raw counts per gene were estimated by the Python script HTSeq count [http://www-huber.embl.de/users/anders/HTSeq/] using the reference genome. The raw counts per gene were used by DEGseq to estimate differential expression at P < 0.05. For peptide identification by liquid chromatography-tandem mass spectrometry (LC-MS/MS) using iTRAQ labeled peptides, all spectra were analyzed with Proteome Discoverer (version 1.4.0.29, Thermo Fisher Scientific) using a Uniprot database (http://www.uniprot.org) containing all sequences from Ixodida, Anaplasmataceae and Ruminantia. Peptide identification was validated using the probability ratio method and false discovery rate (FDR) was calculated using inverted databases and the refined method with an additional filtering for precursor mass tolerance of 12 ppm. Only peptides with a confidence of at least 95% were used to quantify the relative abundance of each peptide. Outliers at the scan and peptide levels and significant protein-abundance changes were detected from the z-values (the standardized variable used by the model that expresses the quantitative values in units of standard deviation) by using a FDR threshold of 5%. Results were the mean of two replicates. For transcriptomics and proteomics analysis in tick cells, the I. scapularis embryo-derived tick cell line ISE6, provided by Ulrike Munderloh, University of Minnesota, USA, was cultured in L-15B300 medium (Munderloh et al., 1999), except that the osmotic pressure was lowered by the addition of one-fourth sterile water by volume. The ISE6 cells were first inoculated with A. phagocytophilum (human NY18 isolate; Asanovich et al., 1997)-infected HL-60 cells and maintained until infection was established and routinely passaged. Uninfected and infected cultures (N = 3 independent cultures with ~10⁷ cells each) were sampled at 7 days post-infection (dpi; percent infected cells 71–77%; Ave ± SD, 74 ± 3). Transcriptomics data was obtained as described above for nymphs and adult female ticks. Two biological replicates were used for each of uninfected and infected tick cells and genes differentially expressed in response to A. phagocytophilum infection were selected with P ≤ 0.05. The proteomics analysis followed the same pipeline describe above in ticks and the MS/MS raw files generated with Xcalibur (version 2.1, Thermo Fisher Scientific) were searched against a compiled database containing all sequences from Ixodida and Anaplasmataceae (http://www.uniprot.org). Three biological replicates were used for each of uninfected and infected tick cells. For the quantitative analysis of tick proteins, after discarding Anaplasma proteins in infected cells, the total number of peptide-spectrum matches (PSMs) for each tick protein were normalized against the total number of PSMs in tick cells and compared between control and infected cells by Chi2-test (P ≤ 0.05). Although the percent of infected ticks and cultured cells was determined as described above, the bacterial load on these samples was not considered in the analysis.

The identified genes in the carbohydrate metabolic pathways were searched against the transcriptomics and proteomics data to characterize their mRNA and protein levels in response to A. phagocytophilum infection.

The tertiary structures of the partial protein sequences of I. scapularis hypoxia-inducible factor 1 alpha HIF-1α (XP_002414889) and HIF-1β (XP_002416629) proteins, which are the DNA binding N-terminus domains, were modeled using the Swiss-Model server (Biasini et al., 2014). The tertiary models were optimized using the Protein Preparation Wizard (Li et al., 2007) in the Schrödinger's Maestro software package. The Protein Preparation Wizard clusters at the highest degree of hydrogen bonding in equilibrium. Monte Carlo orientations are performed (100,000) for each cluster. The optimized structure is based on electrostatic and geometric scoring functions. Any remaining steric clashes were eliminated by minimization of the entire system with the default settings in the Schrodinger's Maestro package. The ternary structure that includes the hypoxia response element (HRE) was constructed via superimposition by using the Swiss-Model server used the mouse HIF-1α, HIF-1β, and the HRE (Dalei et al., 2015) as a template for both I. scapularis hypoxia-inducible protein sequences. As a final optimization step, the ternary structure (HIF-1α/HIF-1β/HRE) was processed using the normal mode ready-made script in the Metropolis Monte Carlo-based Protein Energy Landscape Exploration (PELE) server (Borrelli et al., 2005). The PELE software implements an anisotropic network model (Atilgan et al., 2001) for perturbations of the alpha-carbon backbone causing structural conformational changes. The PELE server can be accessed at https://pele.bsc.es/.

Female ticks fed on A. phagocytophilum-infected and uninfected sheep and fixed with 4% paraformaldehyde in 0.2 M sodium cacodylate buffer were embedded in paraffin and used to prepare sections on glass slides as previously described (Ayllón et al., 2015). The paraffin was removed from the sections through two washes in xylene and the sections were hydrated by successive 5 min washes with a graded series of 100, 96, and 65% ethanol and finally with distilled water. Next, the slides were treated with Proteinase K (Dako, Barcelona, Spain) for 7 min, washed with 0,1% PBS-Tween 20 (Sigma-Aldrich, St. Louis, MI, USA) and blocked with 2% bovine serum albumin (BSA; Sigma-Aldrich) in PBS-Tween 20 during 1 h at room temperature. The slides were then incubated overnight at 4°C with mouse anti-Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) monoclonal antibodies (ab50567; Abcam, Cambridge, UK) diluted 1:100 in 2% BSA/PBS-Tween 20. Preimmune serum was used as control. After 3 washes with PBS-Tween 20, the slides were incubated for 1 h with rabbit anti-mouse IgG conjugated with FITC (Sigma-Aldrich) diluted 1:160 in 2% BSA/PBS-Tween 20. Finally, after two washes with PBS the slides were mounted on ProLong Diamond Antifade Mountant with DAPI reagent (Thermo Scientific, Madrid, Spain). The sections were examined using a Zeiss LSM 800 with Airyscan (Carl Zeiss, Oberkochen, Germany).

The I. scapularis embryo-derived tick cell line ISE6 were cultured as described above, infected with A. phagocytophilum (human NY18 isolate; Asanovich et al., 1997) or mock-infected and maintained according to Munderloh et al. (1999).

A. phagocytophilum-infected ISE6 cells were left untreated or treated for 24 or 48 h with 5 mM 2-Deoxy-D-Glucose (ab142242, Abcam) to inhibit glycolysis (Wang et al., 2011), 100 μM LY294002 (ab120243, Abcam) to inhibit the phosphatidylinositol 3-kinase (PI3K; Sultana et al., 2010), 100 nM Chetomin (ab144222, Abcam) to inhibit the activity of HIF-1α (Misra et al., 2012) or 0.5 mM Deferoxamine mesylate (Sigma-Aldrich) to activate HIF-1α (Choi et al., 2004). After treatment, cells were harvested for the preparation of whole cell lysates to determine HIF-1α activity and for DNA extraction. Tick cell lysates were prepared by adding to cell pellets the RIPA lysis buffer (Thermo Scientific) supplemented with a protease inhibitor cocktail (cOmplete Mini, EDTA-free, Roche, Sigma-Aldrich). HIF-1α activity was determined in the cell lysate supernatants with the HIF-1 alpha Transcription Factor Assay Kit (ab133104, Abcam) following manufacturer's recommendations. A. phagocytophilum DNA levels were characterized by msp4 real-time PCR normalizing against tick 16S rDNA as described previously (Ayllón et al., 2013). Optical density values (O.D. 450 nm) for HIF-1α activity and normalized Ct values for A. phagocytophilum DNA levels were compared between treated and untreated control cells by Student's t-test with unequal variance (P = 0.05; N = 4 biological replicates).

Seven major pathways involved in carbohydrate metabolism were selected for characterization (Table 1). A total of 79 genes coding for the proteins involved in glycolysis, gluconeogenesis, pentose phosphate pathway (PPP), glyceroneogenesis, TCA, mitochondrial oxidative phosphorylation (OXPHOS) and β-oxidation were identified in the I. scapularis genome (Table 1). Based on these results, a model for glucose metabolism in ticks was proposed (Figure 1). At least in humans, pyruvate carboxylase (PC) catalyzes the irreversible carboxylation of pyruvate to form oxaloacetate, which is then transformed in phosphoenolpyruvate by the cytoplasmic enzyme phosphoenolpyruvate carboxykinase (PEPCK-C; Berg et al., 2002). However, the PC orthologue was not identified in the I. scapularis genome.

The carbohydrate metabolic response to A. phagocytophilum infection was then characterized using the quantitative transcriptomics and proteomics data generated from uninfected and A. phagocytophilum-infected I. scapularis ticks and ISE6 cultured cells (Ayllón et al., 2015; Villar et al., 2015). Most of the identified carbohydrate metabolism genes were differentially regulated in response to A. phagocytophilum infection in at least one of the analyzed tick tissues (Figure 2). Twenty-eight (35%), 23 (29%), 62 (78%), and 64 (81%) carbohydrate metabolism components were identified in both transcriptome and proteome of ISE6 cells, nymphs, adult female midguts, and salivary glands, respectively (Figure 2). Of these genes, 59 (75%), 14 (18%), 60 (76%), and 41 (52%) were up-regulated, while 20 (25%), 65 (82%), 16 (20%), and 35 (44%) were down-regulated in response to infection in ISE6 cells, nymphs, adult female midguts, and salivary glands, respectively (Figure 2). Many of the carbohydrate metabolism proteins were not identified by mass spectrometry (Figure 2), but the results showed similar differential regulation at the mRNA and protein levels for 50% of the identified proteins.

Glucose transporters, and mainly facilitative glucose transporters (GLUT) are an important component of the glycolytic pathway because they transport glucose from the extracellular space to the cellular cytoplasm, which makes the glucose accessible to hexokinase (HXK; Augustin, 2010; Li et al., 2015). We found 11 putative glucose transporters in the genome of I. scapularis, 2 members (sglt1 and sglt2) of the sodium-glucose linked transporter family (SGLT), and 9 members (2 isoforms of glut1, glut3, 4 isoforms of glut8, glut10, and glut12) of the GLUT family. In response to A. phagocytophilum infection, glut1B, glut3, glut8C, glut10, glut12, and sglt1 genes were up-regulated in tick midguts, and the GLUT1A and SGLT2 proteins were over-represented in infected midguts when compared to uninfected controls (Figure 3).

As in previous experiments (Villar et al., 2015), these results supported a role for carbohydrate metabolism during A. phagocytophilum infection in I. scapularis, and suggested tissue-specific differences in response to infection.

Based on these results, the putative glucose metabolic pathways affected by A. phagocytophilum infection in different I. scapularis tissues and ISE6 cells were proposed (Figures 4, 5). The results showed that the genes involved in glycolysis were all up-regulated in tick midguts after A. phagocytophilum infection, except for phosphofructokinase (pfk) and hxk genes that did not change in response to infection (Figure 2). The up-regulation of glycolytic genes correlated with protein over-representation in tick midguts, except for the cofactor-independent phosphoglycerate mutase (ipgm) and pyruvate kinase (pk) that despite gene up-regulation in infected midguts, protein was under-represented in response to infection (Figure 2). However, since iPGM catalyzes a reversible reaction, it is not a site of major regulatory mechanisms for the glycolytic pathway. In contrast, HXK and PFK proteins, which catalyze irreversible steps of glucose glycolysis and therefore constitute major regulatory steps of this pathway, were over-represented and did not change, respectively in response to infection of tick midguts (Figure 4). The expression of the p53 target TP53-inducible glycolysis and apoptosis regulator (tigar, ISCW020485), which was recently shown to inhibit glycolysis (Bensaad et al., 2006), was down-regulated in tick nymphs, midguts, salivary glands, and ISE6 cells. The over-representation of all glycolytic enzymes correlated with up-regulation of their respective genes, suggesting transcriptional regulation. These data suggested that A. phagocytophilum infection activates glucose uptake and degradation in I. scapularis midguts. In agreement with an increase in glucose catabolism, the levels of glucose were significantly reduced (0.51 ± 0.06 vs. 0.31 ± 0.05, p < 0.005) in infected ISE6 cells when compared to uninfected controls (Villar et al., 2015).

PK catalyzes the final irreversible step of glycolysis, the transfer of a phosphate group from phosphoenolpyruvate to ADP, yielding one molecule of pyruvate and one molecule of adenosine 5′-triphosphate (ATP; Li et al., 2015). This gene was up-regulated, but protein was under-represented in response to infection in tick midguts (Figures 2, 4). Due to the regulatory role of PK, low levels of this enzyme may hamper the flux of pyruvate toward the mitochondrial matrix, decreasing the activity of TCA cycle. PK generates pyruvate, which is transported to the mitochondrial matrix by mitochondrial pyruvate carriers 1 and 2 (MPC1 and MPC2). The homologs for mpc1 and mpc2 genes were identified in I. scapularis and were up-regulated in tick midguts, but down-regulated in salivary glands in response to A. phagocytophilum infection (Figure 3). In the mitochondrial matrix, pyruvate is decarboxylated by pyruvate dehydrogenase E1 (PDE1; Berg et al., 2002). The decarboxylation of pyruvate is the first step of a series of enzymatic reactions that transform pyruvate into acetyl coenzyme A (acetyl-Coa), which is oxidized to CO₂ and water in the TCA cycle (Berg et al., 2002). PDE1 is an important regulator of the TCA cycle (Berg et al., 2002). Pyruvate dehydrogenase kinase 1 (PDK1) inhibits, while pyruvate dehydrogenase phosphatase catalytic subunit 1 (PDPC1) activates PDE1 (Berg et al., 2002). We found that in the midguts of infected ticks, PDE1 was under-represented and the genes pdk1 and pdpc1 were up-regulated and down-regulated, respectively in response to infection (Figure 2). This result strongly suggested that A. phagocytophilum infection inhibited the TCA cycle in I. scapularis midguts. In agreement with this suggestion, several TCA cycle enzymes such as citrate synthase (CS), aconitase (ACON), isocitrate dehydrogenase [subunits α (IDH3A) and γ (IDH3G1)], 2-oxoglutarate dehydrogenase E1 (OXOE1), succinate dehydrogenase (flavoprotein subunit (SDHA), and iron-sulfur subunit (SDHB) were under-represented in midguts from infected ticks when compared to uninfected controls (Figures 2, 4).

The NADH generated by the TCA cycle is fed into the OXPHOS pathway to produce ATP (Li et al., 2015). The mRNA and protein levels of components of the OXPHOS complex I–IV present in I. scapularis were examined. Higher mRNA and protein levels were found for most of the respiratory components in infected tick midguts when compared to uninfected controls (Figure 2). This result suggested that ticks might have a regulatory mechanism to maintain the levels of ATP in response to the inhibition of the TCA cycle.

Finally, we found that although most genes were up-regulated, all proteins of gluconeogenesis (synthesis of glucose from non-carbohydrate precursors) were under-represented in tick midguts, except for fructose-1,6-bisphosphatase (FBP; Figure 2). FBP catalyzes an irreversible reaction in the gluconeogenesis, the conversion of fructose 1,6-bisphosphate into fructose 6-phosphate, and therefore it is a regulatory step of this anabolic pathway. However, the enzyme glucose 6-phosphatase that catalyzes the last step of gluconeogenesis, the transformation of glucose 6-phosphate into free glucose was under-represented. These results suggested that midgut cells in A. phagocytophilum-infected ticks tend to keep glucose 6-phosphate within the cells for other metabolic processes (e.g., glycogen synthesis).

HIF-1 is a heterodimeric transcription factor consisting of a constitutively expressed β-subunit (HIF-1β) and an oxygen-regulated α-subunit (HIF-1α; Pagé et al., 2002; Déry et al., 2005; Ziello et al., 2007; Badeaux and Shi, 2013). The gene encoding for HIF-1α (ISCW023657), which is a major transcriptional activator of glycolytic genes (Hu et al., 2006), was significantly up-regulated in tick midguts, but not in nymphs, salivary glands or ISE6 cells in response to A. phagocytophilum infection (Ayllón et al., 2015; Villar et al., 2015). The expression of the gene encoding for the HIF-1β (ISCW023999), which is a molecular partner of HIF-1α, was up-regulated in midguts from infected ticks, but was down-regulated in nymphs and did not change in salivary glands and ISE6 cells in response to infection (Ayllón et al., 2015; Villar et al., 2015).

Both I. scapularis HIF-1α and HIF-1β (a.k.a, aryl hydrocarbon receptor nuclear translocator or ARNT) possess the conserved basic helix-loop-helix (bHLH) domain and two PAS domains (Figures 6A,B). The bHLH domain of HIF-1α and HIF-1β were highly conserved with identical interacting residues as those previously identified (Dalei et al., 2015). The PAS-B domain of HIF-1α was disordered due to its low conservation and the presence of small insertions and deletions when compared to the mouse HIF-1α (Dalei et al., 2015). The PAS-B of HIF-1α interacts with both the PAS domain of HIF-1β, thereby making it difficult to determine interacting residues. One key feature of HIF-1α is the presence of two residues lysine (Lys) and glycine (Gly), which interact with the nucleotides of hypoxia response element (HRE; Dalei et al., 2015). The I. scapularis HIF-1α had the conserved Lys, but showed an asparagine (Asn) to Gly substitution (Figure 6).

It has been demonstrated that a number of non-hypoxic stimuli are highly capable of inducing the expression of hif-1 (Pagé et al., 2002; Déry et al., 2005). The PI3K pathway and its downstream effectors, mTOR and p70S6 kinase (p70S6K), may induce a hypoxia-independent increase in HIF-1α levels (Déry et al., 2005). In addition, diacylglycerol (DAG)-sensitive protein kinases C (PKC) were shown to up-regulate hif-1α gene expression in a hypoxia-independent manner (Pagé et al., 2002; Déry et al., 2005). Infection with A. phagocytophilum activates PI3K in I. scapularis (Sultana et al., 2010). Therefore, we characterized the mRNA and protein levels of tick PI3K components in response to A. phagocytophilum infection.

Most PI3K-mTOR pathway components found in other organisms (Pilot-Storck et al., 2010) were identified in the I. scapularis genome (Table 2). The tick PI3K components were differentially regulated in response to A. phagocytophilum infection (Figure 7). The p70S6K and PI3K components (p85α and p110α) were up-regulated in tick midguts and salivary glands in response to infection. Additionally, p70S6K and p85α proteins were over-represented in midguts from infected ticks when compared to uninfected controls (Figure 7). Oxygen-independent activation of hif-1α may involve PI3K components, p70S6K and mTOR (Déry et al., 2005). However, we did not find an mTOR ortholog in the I. scapularis genome. DAG-sensitive PKCs were also suggested to play a role in oxygen-independent activation of hif-1α (Pagé et al., 2002). Protein PKCα was found to be over-represented in tick midguts and salivary glands, while the PKCε gene was up-regulated in tick midguts in response to infection.

Functional studies were focused on glycolysis by targeting the process at different levels (Figures 8A,B). A. phagocytophilum-infected ISE6 cells were left untreated or treated for 24 or 48 h with 2-Deoxy-D-Glucose to inhibit glycolysis (Wang et al., 2011), LY294002 to inhibit the PI3K (Sultana et al., 2010), Chetomin to inhibit the activity of HIF-1α (Misra et al., 2012) or Deferoxamine mesylate to activate HIF-1α (Figures 8A,B). Both HIF-1α (Ayllón et al., 2015; Villar et al., 2015), which is a major transcriptional activator of glycolytic genes (Hu et al., 2006), and most PI3K components (Figure 7) were affected in response to A. phagocytophilum infection. The results showed that LY294002 was the only compound active in ISE6 tick cells (Figure 8A). The lack of effect of chetomin and Deferoxamine mesylate on tick HIF-1α activity may be due to structural differences between tick and mammalian HIF-1α. The treatment of tick cells with LY294002 for 48 h resulted in lower HIF-1α activity when compared to untreated control cells (Figure 8A). Treatment with LY294002 has been shown in I. scapularis and other organisms to inhibit the PI3K pathway (Sultana et al., 2010), which is involved in the induction of HIF-1α to activate the glycolysis (Déry et al., 2005). Therefore, the results obtained in tick cells supported a role for PI3K in HIF-1α induction in tick cells. The decrease in A. phagocytophilum infection after LY294002 treatment for 24 and 48 h (Figure 8B) provided additional support to these results. As expected from the lack of effect on HIF-1α activity after incubation of tick cells with the other compounds, treatment did not affect A. phagocytophilum infection in these cells. Additionally, the immunofluorescence assay of GAPDH, which was over-represented in I. scapularis midguts and salivary glands in response to A. phagocytophilum infection (Figures 4, 5), corroborated the results of the proteomics analysis in tick salivary glands (Figure 8C).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.