The Mitochondrial Isoform of FASTK Modulates Nonopsonic Phagocytosis of Bacteria by Macrophages via Regulation of Respiratory Complex I

Fas-activated serine/threonine kinase (FASTK) is the founding member of a novel family of mitochondrial proteins. This family is composed of six members, FASTK and its homologs FASTKD1–5. FASTK was initially proposed to act as a protein kinase (1), but critical active site residues are not conserved within the family, and this activity has been questioned since (2). All members have been found only in vertebrates (2) and were identified as RNA-binding proteins on the basis of mRNA-bound proteome studies (3, 4). Through these interactions, members of this family regulate various aspects of mitochondrial RNA metabolism, including mitochondrial RNA granules, RNA processing, degradation, and translation (5–12). FASTK can localize to different cellular compartments due to the presence of an alternative translational initiation site. About 50% of the endogenous FASTK protein resides in mitochondria (mitoFASTK, ∼50 kDa species), whereas the rest distributes between the nucleus and the cytosol (cytoFASTK, ∼60 kDa species). FASTKD1–5 are exclusively mitochondrial (2).

In mitochondria, mitoFASTK interacts with ND6 mRNA, which encodes an essential subunit of mitochondrial respiratory complex I (CI, NADH:ubiquinone oxidoreductase). Through this interaction, FASTK protects ND6 mRNA from degradation by the degradosome (12). Accordingly, depletion of FASTK results in the loss of ND6 mRNA and a defect in CI activity (12). In the cytoplasm, cytoFASTK localizes to stress granules and processing bodies and counteracts TIA-1–mediated inhibition of mRNA translation (13, 14). In the nucleus, FASTK regulates the alternative splicing of exons flanked by weak splice site recognition sequences. Accordingly, FASTK promotes the inclusion of exon IIIb of the fibroblast growth factor receptor 2 (FGFR2) premRNA and exon 6 of Fas premRNA (15, 16).

Still very little is known about the role of FASTK in the pathogenesis of human disease. FASTK was found to be overexpressed in PBMCs derived from patients with rheumatoid arthritis, other autoimmune diseases, and asthma (17, 18). These data drew our attention and led us to hypothesize that FASTK might play a role in the pathogenesis of the diseases in which its transcript is upregulated. In our attempt to identify the true function of FASTK in immune system and immune-related diseases, we generated FASTK knockout (FASTK^─/─) mice. These mice lack the two isoforms of FASTK, mitoFASTK and cytoFASTK, because we deleted the entire gene (19). FASTK^─/─ mice were born at the expected Mendelian frequency, survived to maturity, and appeared healthy (19). Our first experiments consisted in inducing experimental allergic asthma and inflammatory arthritis in FASTK^─/─ mice. First, we exposed mice with house dust mite for 3 wk to identify the role of FASTK in the pathogenesis of asthma. The neutrophil recruitment into the lung and alveolar spaces was markedly reduced in FASTK^─/─ mice compared with wild type (WT) controls. Similarly, LPS-induced alveolar neutrophil recruitment was markedly reduced in FASTK^─/─ mice (19). This was accompanied by reduced concentrations of various cytokines and chemoattractants in bronchoalveolar lavage fluids (19). Recently, we have also identified FASTK as a driver of joint inflammation in the K/BxN model. In this model, arthritis has been shown to be crucially dependent on innate immune cells, and FASTK^─/─ mice developed less severe arthritis compared with WT mice (20). The precise molecular and cellular mechanisms leading to the anti-inflammatory phenotype of FASTK^─/─ remain to be determined but could reflect additional targets for directed therapy against inflammatory diseases.

One of our current lines of research aims to delineate the role of FASTK and mitochondria in the functions of macrophages in the innate immune response. In this study, we reveal a novel role for FASTK in the regulation of phagocytosis. Macrophages from FASTK^─/─ mice exhibited marked increase in nonopsonic phagocytosis of bacteria. The analysis of the molecular mechanisms underlying this phenotype reveals that mitochondria, namely CI activity, are key in the control of phagocytosis.

WT and FASTK^─/─ mice were derived from the founder stock previously described (19). All procedures and experiments were carried out according to institutional guidelines for the Animal Care and Use Committee at University of Valladolid (Spain). Elicited macrophages were obtained by i.p. injection of 1.5 ml of 3% sterile Brewer thioglycolate medium 5 d prior to harvesting. The murine macrophage RAW 264.7 cell line was a gift from Dr. A. Alonso (Instituto de Biología y Genética Molecular, Valladolid). Cell viability was determined by trypan blue exclusion assay or by staining with crystal violet solution (0.1% in PBS) after cell fixation in 4% paraformaldehyde. Apoptosis was examined using TUNEL assay (Roche). Nuclei were stained with 50 ng/ml Hoechst 33258 (Sigma-Aldrich). Digital images were captured using a Leica fluorescence microscope (Leica DMI3000 B).

RAW 264.7 cells were transduced with lentivirus carrying short hairpin RNA (shRNA) against FASTK, NDUFS3, and NDUFS4. The target sequences within the coding regions (coding sequences [CDS]) of the genes of interest were as follows: mouse FASTK (RefSeq accession number NM_023229, https://www.ncbi.nlm.nih.gov/nuccore/NM_023229) 5′-GTCAGCTCATCATCCGAAA-3′, mouse NDUFS3 (RefSeq accession number NM_026688, https://www.ncbi.nlm.nih.gov/nucleotide/NM_026688) 5′-GAGAGTATGTGGCTGAAAT-3′ and mouse NDUFS4 (RefSeq accession number NM_010887.2, https://www.ncbi.nlm.nih.gov/nuccore/NM_010887.2) 5′-GCTATGATGTGGAAGAGAA-3′. The target sequence within the 5′ untranslated region (5′UTR) of FASTK (RefSeq accession number NM_023229. https://www.ncbi.nlm.nih.gov/nuccore/NM_023229) was as follows: 5′-GCTGGACTGCGATTGGCGTCT-3′. Inserts were prepared by annealing shRNA oligonucleotides containing two complementary target sequences linked by a short loop and then were cloned into the BamHI/EcoRI sites of the shRNA lentiviral vector pGreenPuro, all according to vector manufacturer’s instructions (System Biosciences). The resultant plasmids were screened for insert by PCR and then sequenced with the following primers (all according to vector manufacturer's instructions): forward 5′-AATGTCTTTGGATTTGGGAATCTTAT-3′ and reverse 5′-TGGTCTAACCAGAGAGACCCAGTA-3′. Lentiviral stocks were produced by transient cotransfection into the human 293FT cell line (Thermo Fisher Scientific) with the appropriate lentiviral expression plasmid and lentiviral helper plasmids (psPAX2 packaging vector and pMD2.G envelope-encoding vector) using TransIT-X2 (Mirus Bio). psPAX2 (Addgene plasmid no. 12260) and pMD2.G (Addgene plasmid no. 12259) were gifts from D. Trono. RAW 264.7 cells were then transduced and selected with 7 μg/ml puromycin for at least 2 wk before assaying. Gene silencing efficiency was assessed by quantitative RT-PCR using SYBR Green (Thermo Fisher Scientific), and it was of 74% for shNDUFS3 CDS, 67% for shNDUFS4 CDS, 85% for shFASTK CDS, and 78% for shFASTK 5′ UTR. β-actin was used as a housekeeping gene. The primers used for quantitative RT-PCR were as follows: mouse FASTK, 5′-CTGGAAGCCATTGCTCATTTC-3′ (forward) and 5′-CTCTCAAGGCAGGGCATAAA-3′ (reverse); mouse NDUFS3, 5′-GATGAGCTGACACCCATTGA-3′ (forward) and 5′-ATGTCCCTCGAAGCCATAATC-3′ (reverse); mouse NDUFS4, 5′-TCTGACCTTCAGTGCCAAAG-3′ (forward) and 5′-GTAGCCAGCTCCAACCTATTT-3′ (reverse); mouse β-actin, 5′-GACATGGAGAAGATCTGGCA-3′ (forward) and 5′-GGTCTCAAACATGATCTGGGT-3′ (reverse). RAW 264.7 cells stably expressing shRNA against the 5′UTR of mouse FASTK were used for rescue experiments. These cells were transfected with constructs encoding for red fluorescent protein (RFP) alone (Clontech), C-terminal RFP-tagged WT human FASTK, or human FASTK deletion mutant starting at the internal methionine at position 35 (Δ1–34). RFP-tagged FASTK constructs were a generous gift of Dr. N. Kedersha and Dr. P. Anderson (Harvard University). Transfected cells were selected with 1 mg/ml G418 (Sigma-Aldrich), and the expression of RFP-tagged constructs was assessed using fluorescence microscopy and Western blotting.

Escherichia coli strain DH5-α was from Invitrogen (Thermo Fisher Scientific). Staphylococcus aureus Cowan 1 strain was from American Type Culture Collection (ATCC 12598). Recombinant E. coli (DH5-α) bacteria expressing GFP were obtained by introducing a pUC-based plasmid carrying a GFP gene (ampicillin-resistant) into chemically competent bacteria. GFP-expressing S. aureus were a gift from Dr. A. Abadía Molina (University of Granada, Spain). When necessary, bacteria were labeled with fluorochromes FITC (Sigma-Aldrich) or tetramethylrhodamine-5- (and-6)-isothiocyanate (Thermo Fisher Scientific). Briefly, heat-killed bacteria (80°C, 15 min) were incubated with a 0.1 mg/ml solution of fluorochrome at 37°C for 30 min and washed three times with PBS prior to use. Zymosan particles were from Sigma-Aldrich and were labeled as described for bacteria. Polystyrene latex beads were 2.0 μm in diameter and yellow-green fluorescent (Molecular Probes, Thermo Fisher Scientific). For the preparation of serum-opsonized particles, fluorochrome-labeled, heat-killed bacteria, zymosan, and latex beads were incubated in 50% mouse serum in PBS at 37°C for 30 min and then washed to remove unbound serum. As to Ig-opsonized particles, the following reagents were purchased from Molecular Probes (Thermo Fisher Scientific): zymosan A bioparticles and E. coli (K-12 strain) bioparticles, both Texas Red conjugates, S. aureus (Wood strain without protein A) bioparticles fluorescein conjugate, and purified rabbit polyclonal IgG Abs specific to each of the three bioparticles. IgG-opsonized polystyrene latex beads were prepared by incubation with 3 mg/ml rabbit IgG (Sigma-Aldrich) in PBS for 30 min at 37°C followed by extensive washing with PBS.

Macrophages (5 × 10⁵) were seeded on 12-mm coverslips in 24-well plates and were grown overnight. To remove all traces of seric proteins, the cells were washed twice in serum/antibiotics-free DMEM medium supplemented with GlutaMAX and were preincubated at 37°C for 30 min. All further incubations were performed in the absence of serum. Nonopsonized fluorochrome-labeled bacteria, zymosan, or latex beads were added to macrophages at a 100:1 ratio for 30 min at 37°C. Where indicated, Ig- or serum-opsonized particles were used. After incubation periods, macrophages were washed three times with cold PBS, and extracellular FITC-labeled bacteria were quenched with a 60-s wash in trypan blue (0.2 mg/ml). Macrophages were fixed with 4% paraformaldehyde. Phagocytosed fluorochrome-labeled bacteria were counted under a confocal fluorescence microscope (the characteristics of the equipment are detailed in this 2Materials and Methods section under 8Evaluation of phagolysosomal fusion). Phagocytic index (PI) was calculated as the percentage of positive macrophages × the mean number of either bacteria or FITC-labeled latex beads (Sigma-Aldrich) of positive macrophage. The fluorescence of the cells was also determined by flow cytometry using a Beckman Coulter Gallios (Beckman Coulter). In this case, the cells were plated in 12-well plates (1 × 10⁶ macrophages per well). The phagocytosis test was done in a similar manner to that described for 24-well plates.

Bactericidal activity was measured using a gentamicin protection assay. Briefly, elicited peritoneal macrophages and RAW 264.7 cells were plated into 24-well plates at a density of 5 × 10⁵ cells per well (three wells per experimental condition) and allowed to adhere overnight. To remove all traces of seric proteins, the cells were washed twice in serum/antibiotics-free DMEM supplemented with GlutaMAX and were preincubated at 37°C for 30 min. All further incubations were performed in the absence of serum. Nonopsonized bacteria were added to cells at a multiplicity of infection of 50. After 15 min of incubation at 37°C and 5% CO₂, the nonadherent extracellular bacteria were removed by washing with PBS, and the adherent extracellular bacteria were subsequently killed by the addition of serum-free DMEM containing gentamicin (100 μg/ml). At time 0′ (which corresponds to 15 min after addition of gentamicin) and subsequent time points, cells were washed three times with PBS and lysed with 1 ml of sterile distilled water. Serial dilutions of cellular lysates were plated onto Luria–Bertani agar plates, and the number of CFUs was determined after 24-h growth at 37°C. The percentage of killed bacteria was calculated as follows: % killing = 100 − [(no. of CFU at time X/no. of CFU at time 0′) × 100].

The oxidative burst in macrophages was triggered by exposure to bacteria (ratio of bacteria to cells was 100:1) and detected using a luminol-based bioassay. Chemiluminescence was detected in 323-μl samples that contained 2 × 10⁵ macrophages, 200 μM luminol (Thermo Fisher Scientific), and 16 U of HRP (Sigma-Aldrich) in a microplate luminometer (Appliskan; Thermo Fisher Scientific).

We monitored the kinetics of phagolysosomal fusion under experimental conditions similar to those of the gentamicin protection assay described above. Elicited peritoneal macrophages (5 × 10⁵) were seeded on 12-mm coverslips in 24-well plates and were grown overnight. Cells were switched to serum/antibiotics-free conditions and pulsed for 15 min with nonopsonized GFP-expressing E. coli (at a multiplicity of infection of 50). Cells were chased for different times in the presence of gentamicin (at 100 μg/ml) and were loaded with Lysotracker Red DND-99 (500 nM) for the last 5 min of each chase period. Time 0′ corresponds to 15 min after addition of gentamicin. Before fixation, coverslips were washed three times in PBS and then fixed in 4% paraformaldehyde for 15 min at room temperature. Nuclei were stained with 5 μ/ml Hoechst 33342, and after washing with PBS, the coverslips were mounted and viewed. Confocal images were obtained with a 63× oil immersion objective (HCX PL Apo CS NA = 1.4; Leica) attached to a confocal DMI 6000B microscope with a TCS SP5 confocal system (Leica) equipped with acousto-optical beam splitter and acousto-optic tunable filter systems. We obtained confocal sections under constant conditions to minimize image acquisition variation. Images were stored as 1024 × 1024 pixels and 8-bit TIFF files. The colocalization of E. coli (green) with lysotracker (red) was enumerated. At least 20 fields were examined per slide, and representative pictures were taken.

Abs were purchased from the following sources: FITC-conjugated anti-CD80 (16-10A1), FITC-conjugated anti-CD86 (GL1), FITC-conjugated anti-CD40 (3/23), FITC-conjugated anti-CD38 (90/CD38), FITC-conjugated anti-CD14 (rmC5-3), FITC-conjugated anti-CD206 (19.2), and APC-conjugated CD11b (M1/70) were purchased from BD Biosciences. PE-conjugated F4/80 (Cl:A3-1) was purchased from AbD Serotec. Anti-TLR 2 (6C2) and anti-TLR4 (MTS510) were purchased from eBioscience (Thermo Fisher Scientific), and PE-conjugated anti-rat Ig was obtained from BD Biosciences. When indicated, cells were labeled with 2.5 μM MitoSOX Red indicator (Molecular Probes, Thermo Fisher Scientific). In all cases, the fluorescence of the cells was determined by flow cytometry using a Beckman Coulter Gallios (Beckman Coulter). Data were analyzed using either FlowJo software (Tree Star) or Kaluza software (Beckman Coulter). Fluorescence intensity was plotted on a log scale. The number of cells acquired for each sample depended on the subpopulation studied, and at least 10,000 cells were analyzed.

Protein was extracted by RIPA buffer (20 mM tris, pH 7.4; 1% NP-40; 1% sodium deoxycholate; 0.1% SDS; 150 mM NaCl; 1 mM EDTA) containing a mixture of protease inhibitors from Sigma-Aldrich (catalog no. P8340) and 50 mM sodium fluoride. Twenty-microgram protein samples separated by SDS-PAGE and transferred to PVDF membranes were probed with Abs against AMP-activated protein kinase α (AMPKα, catalog no. 2532; Cell Signaling Technology), phospho-AMPKα Thr172 (catalog no. 2531; Cell Signaling Technology), RFP (5F8; Chromotek), and actin (AC-40; Sigma-Aldrich) following the manufacturer’s instructions. Band intensities were determined by densitometric analysis using ImageJ software.

Respiratory activity of complexes I, II, III, and IV were performed in a Shimadzu UV-1800 Spectrophotometer as previously described (21) with slight modifications.

All analyses were performed using Prism software (GraphPad). Data are expressed as mean ± SEM and were analyzed by using either one-way ANOVA with Bonferroni correction, the unpaired Student t test, or Mann–Whitney U test as appropriate. In all graphs, *p < 0.05, **p < 0.01, ***p < 0.001.

To further explore the role of FASTK in innate immune system, we assessed whether the absence of this protein impacted one of the most important functions of the macrophages, which is their ability to phagocytose and kill bacteria. We took advantage of the FASTK^─/─ mice we had previously generated (19). FASTK^─/─ mice do not show defective development of the immune system (19). In this study, we used thioglycolate-elicited macrophages, and we again excluded the possibility of developmental defects because WT and FASTK^─/─ macrophages had no difference in their expression of cell surface markers (Supplemental Fig. 1A).

Phagocytosis was performed using both nonopsonized and opsonized phagocytic particles, and the incubation period was 30 min in all the experiments. As shown in Fig. 1A, the PI for uptake of nonopsonized E. coli was significantly increased in FASTK^─/─ macrophages compared with WT controls. Similarly, although to a lesser extent, the PI for nonopsonized S. aureus was markedly increased. Almost identical results were obtained when the incubation period was only 15 min (Supplemental Fig. 1B). No differences were observed at 4°C (Supplemental Fig. 1C), indicating that internalization, rather than binding, was affected. Phagocytosis of FASTK^─/─ macrophages was not globally increased, as the phagocytosis of nonopsonized zymosan and latex beads was similar to that of WT (Fig. 1A top panel). We next explored the phagocytosis of Ig- and serum-opsonized bacteria, zymosan, and latex beads (Fig. 1A bottom panel and Supplemental Fig. 1D). Opsonic phagocytosis was unaltered in FASTK^─/─ macrophages. Altogether, these data indicate that FASTK selectively modulates the nonopsonic phagocytosis of bacteria.

We next examined the bactericidal activity of FASTK^─/─ macrophages using the in vitro gentamicin protection assay in serum-free conditions to eliminate the contributions of Fc and/or complement receptors (Fig. 1B). Nonopsonized bacteria were added to cells for 15 min. The nonadherent extracellular bacteria were removed by washing, and the adherent extracellular bacteria subsequently killed with gentamicin (100 μg/ml) for 15 min. At the indicated time points, intracellular CFUs were enumerated to monitor bacterial burden (Fig. 1B, left panel). As we indicated in the 2Materials and Methods section, time 0′ corresponds to 15 min after addition of gentamicin and 30 min after the beginning of the exposure to bacteria. As shown in the left panel of Fig. 1B, the number of surviving E. coli and S. aureus was significantly higher in FASTK^─/─ macrophages than in WT controls at all time points. This result was expected, as the number of bacteria recovered from the gentamicin protection assay at early points correlates well with the phagocytic capacity. Based on the numbers of surviving bacteria, we next calculated the percentage of killed intracellular bacteria as an approach to quantify the intracellular bactericidal activity (Fig. 1B, right panel). The graphs show that the bactericidal activity is unaltered in FASTK^─/─ macrophages at all time points. At the end of the time course, both types of macrophages had eliminated ∼60% of the ingested E. coli and ∼50% of the ingested S. aureus. Production of reactive oxygen species (ROS) by the NADPH oxidase is essential for efficient bacterial killing by macrophages and neutrophils. As shown in Supplemental Fig. 2A, FASTK^─/─ macrophages’ ability to produce ROS in response to bacteria was unaltered as measured in a luminol-based assay that correlated well with the intact killing activity in the absence of FASTK. Phagocytosed bacteria are contained within phagosomes that mature into phagolysosomes. LysoTracker selectively labels late endosomes and lysosomes and colocalizes with LAMP (22, 23). We monitored the maturation of phagosomes containing GFP-expressing E. coli by their ability to colocalize with Lysotracker Red over time under confocal fluorescence microscopy (Supplemental Fig. 2B). Pulse-chase incubation times corresponded to those of the gentamicin protection assay described above. The percentages of colocalization between these two colored markers were similar in WT and FASTK^─/─ macrophages at all time points. Colocalization increased gradually over time and reached ∼50% at time 120′ in both types of cells. Unaltered kinetics of the maturation of phagosomes containing E. coli in FASTK^─/─ macrophages were consistent with the above explained bacterial killing assay results.

Altogether, these results demonstrate that FASTK modulates the phagocytic capacity of macrophages toward E. coli and S. aureus but not their bactericidal activity.

FASTK can be found in the nucleus, processing bodies, and stress granules in the cytoplasm, and mitochondrial RNA granules. The mitoFASTK isoform has been the best characterized. It is synthesized from an alternative downstream translation initiation site and lacks the first 34 aa present at the N-terminus, exposing a mitochondrial targeting signal. mitoFASTK is essential for mitochondrial ND6 mRNA biogenesis. Thus, in the absence of mitoFASTK, the ND6 mRNA is undetectable, resulting in a ∼60% decrease in CI activity in the skeletal and cardiac muscles of FASTK^─/─ mice (12). To explore the molecular mechanisms involved in the phagocytic phenotype, we first studied whether the activity of CI was also affected in FASTK^─/─ peritoneal macrophages. As shown in Fig. 2, we observed a significant decrease of CI activity by up to 45% in FASTK^─/─ macrophages, whereas the activity of other complexes was not significantly altered. The decrease in CI has also been linked to other processes that resemble phagocytosis: efferocytosis by microglia and autophagy in HeLa cells (24, 25). We thus hypothesized that CI-decreased activity was responsible for the bacterial phagocytic phenotype in FASTK^─/─ macrophages.

To determine whether decreased CI activity could underlie increased phagocytosis of bacteria in FASTK^─/─ macrophages, we tested the role of CI in macrophage phagocytosis by specifically inhibiting CI activity using either rotenone or shRNA against of NDUFS3 and NDUFS4. Rotenone has been reported to be a specific and potent inhibitor of CI. A rotenone dose-response curve on WT peritoneal macrophages demonstrated that 30 nM rotenone produced a similar inhibition of CI activity (∼50% decrease) to that observed in FASTK^─/─ macrophages (Supplemental Fig. 3A). For this reason, we used 30 nM rotenone in this study. Similarly, RAW 264.7 cells also showed 50% inhibition of CI activity when treated with 30 nM rotenone (data not shown).

WT peritoneal macrophages were incubated with 30 nM rotenone for 24 h before assessment of bacterial phagocytosis. Rotenone at 30 nM did not result in any toxicity as assessed by the trypan blue exclusion or TUNEL assays (Supplemental Fig. 3B). Interestingly, treatment with rotenone resulted in an increase in the phagocytosis of nonopsonized bacteria after 30 min incubation comparable to the increase observed in FASTK^─/─ macrophages (Fig. 3A). As in FASTK^─/─ macrophages, the PI of nonopsonized zymosan and latex beads remained unchanged by rotenone treatment (Fig. 3A). Nearly identical results were obtained when the incubation period was only 15 min (Supplemental Fig. 3C). Importantly, rotenone pretreatment did not alter the killing activity of WT macrophages at 120′/T2 time point (Fig. 3B) nor the ability to generate ROS (Supplemental Fig. 2A). The increased intracellular CFU numbers in macrophages pretreated with rotenone correlates well with increased bacterial uptake.

To further confirm the role of CI in the regulation of bacterial phagocytosis, we generated two macrophage models of CI deficiency by silencing two of its subunits: NDUFS3 and NDUFS4 (Fig. 4A). As a catalytic subunit, NDUFS3 has been shown to play a vital role in the proper assembly of CI (26). NDUFS4 is not directly involved in electron transport but plays a role in the stability of the entire complex (27). We stably depleted NDUFS3 and NDUFS4 in the murine macrophage RAW 264.7 cell line using lentiviral-based shRNA constructs. RAW 264.7 cells expressing shRNA against FASTK CDS or pretreated with rotenone (30 nM) were used as positive controls. As shown in Fig. 4B, NDUFS3 and NDUFS4 depletion resulted in an increase in the PIs of nonopsonized bacteria PI, whereas PI of nonopsonized zymosan or latex beads remained unchanged. Finally, we explored the bactericidal activity in these stable cell lines. Depletion of NDUFS3 and NDUFS4 did not alter the killing activity of RAW 264.7 cells. It is noteworthy that parental RAW 264.7 cells exhibited increased bactericidal activity against S. aureus as compared with WT peritoneal macrophages. At 120′/T2 time point, the percentage of killed S. aureus was ∼70% in parental RAW 264.7 cells and ∼50% in WT peritoneal macrophages. In addition, we observed that silencing of NDUFS3, NDUFS4, and FASTK as well as pretreatment with rotenone caused a small, nonsignificant increase in the bactericidal activity of RAW 264.7 cells against S. aureus.

In conclusion, the inhibition of CI activity using either rotenone or shRNA against NDUFS3 and NDUFS4 induced similar effects on macrophages as those induced by the deletion or depletion of FASTK. These observations suggest that mitoFASTK is responsible for the phenotype and functions as a negative regulator of phagocytosis by targeting CI.

To test that the mitochondrial isoform of FASTK is indeed responsible for the phagocytic phenotype, we performed rescue experiments. First, we stably depleted FASTK in RAW 264.7 cells with lentiviral-based shRNA construct targeting the 5′UTR of FASTK. We used an shRNA that targets the 5′UTR because the expression can be easily rescued by vectors expressing the coding region of FASTK. FASTK-depleted cells were then stably transfected with WT FASTK, which contains the two translation initiation sites and encodes both mitoFASTK and cytoFASTK and a mutant FASTK that starts at the internal methionine at position 35 (Δ1–34) and only encodes mitoFASTK (Fig. 5A). Both constructs carried RFP at their C terminus. As control, we also generated FASTK-depleted RAW 264.7 cells expressing RFP alone. Expression was evaluated by fluorescence microscopy (data not shown) and Western blot with anti-RFP Ab (Fig. 5B). As shown in Fig. 5C, the re-expression of FASTKΔ1–34 alone was able to rescue CI activity and the phagocytic phenotype of FASTK-depleted cells. Again, phagocytosis experiments were performed with nonopsonized E. coli and S. aureus. These results demonstrate that mitoFASTK is a negative regulator of nonopsonic phagocytosis of bacteria by macrophages via regulation of respiratory CI.

CI dysfunction triggers transient and sustained changes in metabolism and is accompanied by the phosphorylation of AMPKα, a crucial cellular energy sensor. Glucose supply and AMPKα subunit have been proven to be crucial for cell survival during CI dysfunction (28). AMPKα has also been reported to enhance the phagocytic ability of macrophages and neutrophils (29). We thus hypothesized that a decrease in CI activity augments the phagocytosis of bacteria by macrophages via activation of AMPK.

We first examined the phosphorylation status of AMPKα (threonine 172) and the total AMPK levels in WT and FASTK^─/─ peritoneal macrophages unexposed or exposed to bacteria E. coli and S. aureus. Interestingly, the basal phosphorylation level of AMPKα was increased ∼2-fold in FASTK^─/─ macrophages over WT cells (Fig. 6A, 6B). Exposure to bacteria led to a significant increase in AMPKα phosphorylation in WT macrophages. We also observed an increase in phosphorylated AMPKα in FASTK^─/─ macrophages upon exposure to bacteria, but in this case, the increase over already increased baseline levels was not found to be significant. AICAR was used as positive control for AMPK phosphorylation. Total AMPK levels remained unchanged (Fig. 6A).

To further explore the role of AMPK activation in FASTK-mediated modulation of phagocytosis, we examined the effects of Compound C AMPK inhibitor on the phagocytic phenotype of FASTK^─/─ macrophages. Cells were pretreated with 10 μM of Compound C prior to phagocytosis assays. As shown in Fig. 6C, treatment with Compound C reverted the phagocytic phenotype induced by the absence of FASTK in murine primary macrophages. Very similar results were obtained using FASTK-silenced RAW 264.7 cells (Supplemental Fig. 4A). These results demonstrate that AMPK is a key player in the modulatory role of FASTK in phagocytosis

The present work reveals a novel role for mitoFASTK in the regulation of nonopsonic phagocytosis of bacteria by macrophages. The molecular mechanisms responsible for this phenotype are the decrease in CI activity and AMPK activation.

Although opsonic phagocytosis is more effective, nonopsonic phagocytosis is a crucial mechanism for host control during the initial stages of infection and in anatomic spaces where opsonins may be limiting, such as the alveolar space (30). Different pattern recognition receptors (PRRs) have been described to be involved in the binding and capture of nonopsonized microorganisms, which include transmembrane proteins such as the TLRs and C-type lectin receptors as well as cytoplasmic proteins such as the retinoic acid–inducible gene-I–like receptors and NOD-like receptors (31). PRRs recognize microbial components; for example, C-type lectin receptors Dectin-1 and Dectin-2 are responsible for sensing β-glucans from fungi, and NOD1 and NOD2 recognize bacterial peptidoglycan-derived structures (31). The TLR family is one of the best-characterized PRR families and are unusual in that some can recognize several structurally unrelated ligands (32). Although they recognize microorganisms, TLRs do not function as phagocytic receptors; however, they can cooperate with other nonopsonic receptors to stimulate phagocytosis (33). In this regard, it is noteworthy that TLR2- and TLR4-signaling have been reported to be required for the maturation of phagosomes containing bacteria (34). Thus, PRRs and the cross-talk between them confer a certain degree of specificity to innate immunity that could underlie the specific increase in bacterial phagocytosis in the absence of FASTK and/or decrease in CI activity.

Accumulating evidence has shown that sensing of Gram-negative bacteria and the exposure to TLR4 ligands decreases OXPHOS and increases glycolysis (35–37). A recent report by Garaude et al. (38) shows that recognition of viable Gram-negative bacteria in fact decreases the abundance of assembled CI, which results in decreased CI activity and subsequent cellular metabolic adaptations. This metabolic rewiring has also been described upon the activation of toll receptors TLR2 and TLR9 in dendritic cells (39). Still little is known about the impact of this metabolic reprogramming on immune function. Moreover, the importance of certain pre-existing mitochondrial dysfunctions in immune function has also been largely unexplored. In this article, several lines of evidence implicate CI in the modulation of bacterial phagocytosis by macrophages: 1) FASTK^─/─ macrophages show decreased CI activity and increased nonopsonic bacterial phagocytosis, and the re-expression of mitoFASTK alone is able to rescue the phenotype; 2) treatment with rotenone, a specific inhibitor of CI, and shRNA-mediated silencing of NDUFS3 and NDUFS4 also lead to increased phagocytosis of nonopsonized bacteria by macrophages. NDUFS3 is an integral subunit of the Q module of the mitochondrial respiratory CI (26), whereas NDUFS4 is an accessory subunit required for stability of CI (27, 40). Mutations in NDUSF3 and NDUFS4 genes result in Leigh syndrome, a progressive neurodegenerative disorder (41–44).

CI (NADH:ubiquinone oxidoreductase) represents the entry point for most electrons into the respiratory chain. It is a large, multimeric enzyme complex composed of 44 different subunits, the assembly of which is influenced by the oxidative environment. As mentioned above, the recognition of bacteria transiently decreased the assembly and activity of CI (38). In this study, we observed that a decrease in CI, in turn, increases the phagocytic activity against bacteria in macrophages. We speculate that CI acts as sensor and effector in the regulation of bacterial phagocytosis by macrophages and that this is the result of evolutionary adaptation of macrophages to combat bacteria. The role of FASTK and CI in regulating nonopsonic phagocytosis of bacteria by macrophages represents a novel and exciting finding that deserves further investigation. It is important to mention that CI deficiency does not lead to a global increase in phagocytosis. This finding suggests that stimuli associated with bacteria contribute to the role of CI in phagocytosis. Specific bacterial stimuli may underlie the finding that the increase of bacterial phagocytosis associated with CI deficiency is greater for Gram-negative bacteria than for Gram-positive bacteria. Finally, it is also important to highlight that the decrease in CI has also been linked to other processes that resemble phagocytosis: efferocytosis and autophagy. In fact, the decrease in CI activity using metformin and rotenone is associated with increased phagocytosis of apoptotic cells and inert particles by microglia (24). Similarly, silencing of CI subunit GRIM-19 in HeLa cells induces autophagy (25).

In this article, we further investigated the molecular mechanisms of action of FASTK protein in regulating the phagocytic capacity of macrophages for nonopsonized bacteria. AMPK is a heterotrimeric kinase consisting of an α catalytic subunit and two regulatory subunits, β and γ (45). AMPK is activated by phosphorylation of threonine 172 within the catalytic subunit α (45). AMPK is a sensor of energy status that maintains cellular energy homeostasis. In this context, CI dysfunction triggers metabolic response and adaptation that depends on AMPK activation (28). Accordingly, FASTK^─/─ macrophages show increased basal levels of AMPKα phosphorylation. Exposure to bacteria led to increase in AMPKα phosphorylation, which only was significant for WT macrophages. Bacteria do not increase significantly AMPKα phosphorylation in FASTK^─/─ macrophages probably due to the fact that AMPKα is almost fully phosphorylated under basal conditions. CI dysfunction has also been linked to increased mitochondrial ROS (mitoROS) levels (46). We used MitoSOX to evaluate mitoROS generation in WT and FASTK^─/─ macrophages. As shown in Supplemental Fig. 4B, exposure to bacteria led to an increase in mitoROS generation that was higher in FASTK^─/─ macrophages compared with WT macrophages. Whether mitoROS plays a role in AMPK activation or in phagocytosis events remains unclear. However, some studies suggest that local production of mitoROS concomitant with TLR-mediated detection could contribute to macrophage bactericidal activity (47). AMPK has been reported to enhance the phagocytic ability of macrophages and neutrophils (29). AICAR treatment increased phagocytosis of bacteria, whereas inclusion of Compound C in the macrophage cultures prior to AICAR exposure diminished uptake of bacteria (29). Similarly, FASTK^─/─ macrophages reverted to WT phagocytic phenotype by supplementation of Compound C. AMPK is involved in the microtubule and actin network formation during phagocytosis (29). We show in Supplemental Fig. 4C that FASTK^─/─ macrophages showed increased rise in F-actin content following phagocytosis of E. coli. Further studies are needed to unveil the detailed molecular mechanisms by which FASTK-mediated activation of AMPK controls phagocytosis. Our work represents the first study, to our knowledge, to document the role of FASTK and respiratory CI in modulating the nonopsonic phagocytosis of bacteria by macrophages. Despite having extensively characterized the phenotype in vitro, it is still necessary to prove its translatability to in vivo models to verify the relevance of this finding in the context of the infection. Even taking into account this limitation, we expect our discovery will open up whole new avenues of investigation that may lead eventually to novel treatments for increasing uptake and clearance of bacteria in infections prior to the initiation of an inflammatory response or in infections in anatomic spaces with limiting amount of opsonins.

We thank Alicia Ortega for technical support.

The authors have no financial conflicts of interest.