Increased mitochondrial biogenesis and ROS production accompany prolonged CD4⁺ T cell activation

Abstract

Activation of CD4⁺ T cells to proliferate, drives cells towards aerobic glycolysis for energy production while using mitochondria primarily for macromolecular synthesis. In addition, the mitochondria of activated T cells increase production of reactive oxygen species (ROS) providing an important second messenger for intracellular signaling pathways. To better understand the critical changes in mitochondria that accompany prolonged T cell activation, we carried out an extensive analysis of mitochondrial remodeling using a combination of conventional strategies and a novel high-resolution imaging method. We show that for four days following activation, mouse CD4⁺ T cells sustained their commitment to glycolysis facilitated by increased glucose uptake through increased expression of GLUT transporters. Despite their limited contribution to energy production, mitochondria were active and showed increased ROS production. Moreover, prolonged activation of CD4⁺ T cells led to increases in mitochondrial content and volume, in the number of mitochondria per cell and in mitochondrial biogenesis. Thus, during prolonged activation, CD4⁺ T cells continue to obtain energy predominantly from glycolysis but also undergo extensive mitochondrial remodeling resulting in increased mitochondrial activity.

INTRODUCTION

Activation in response to foreign molecules results in a cascade of changes in immune cells. These changes include not only rapid proliferation and differentiation but also the extensive cellular remodeling that accompanies them. The cellular changes that immune cells go through in order to meet the needs of activated cells have gained considerable interest in recent years (1, 2). Most of the advances in the area focus on how immune activation is coupled to changes in cellular metabolism and how these metabolic changes are maintained. Although cellular activation and differentiation is a process that demands high energy production, the way cells meet this demand is not uniform. For instance, recent studies showed that both B and T lymphocytes rely on aerobic respiration and fatty acid oxidation for their quiescent state energy needs, and upon activation, they shift towards glucose as the main energy source (3–5). However, in contrast to T cells which, upon activation, remodel their energy production machinery predominantly towards glycolysis with limited increase in oxidative phosphorylation (OXPHOS) (6), activated B cells show a more proportional increase in both glycolysis and OXPHOS and they maintain part of their oxidative phosphorylation capacity by diverting a portion of glucose towards oxidation in mitochondria through increased pyruvate dehydrogenase activity (7). So, as compared to activated T cells, mitochondria in activated B cells contribute more to overall energy production (3).

The process through which activated T cells use their mitochondria for generation of macromolecular intermediates, such as lipid biosynthesis from citrate and nucleic acids through 1-carbon metabolism, rather than energy production resembles a similar choice that exists in rapidly proliferating tumor cells (8–11). This phenomenon, termed the Warburg Effect, represents a shift in rapidly proliferating cells towards glycolysis and lactate production even in the presence of oxygen (12). Despite the inefficiency of glycolysis as a source of energy compared to OXPHOS, this commitment allows for the utilization of mitochondrial TCA cycle for macromolecular synthesis in proliferating T cells (9, 10, 13–15). Furthermore, recent studies showed that mitochondrial ROS production increases upon T cell activation and this acts as a second signal in regulating multiple downstream elements (15, 16).

However, despite the advances in our understanding of T cell immunometabolism, key questions remain unanswered. Because most studies focus on early time points after T cell activation, we do not clearly know whether the shift towards glycolysis is sustained after prolonged activation. In addition, we do not know what type of structural mitochondrial remodeling, if any, accompanies sustained T cell activation.

Here, we addressed these key questions by applying a range of metabolic and cellular analyses to naïve and CD4⁺ T cells activated both in vitro and in vivo in a comparative fashion. Our data showed a continued dominance of glycolysis over OXPHOS as the source of energy in activated T cells even four days after activation. This was accompanied by a gradual increase in the expression of GLUT transporters which allowed for increased glucose uptake. Using a novel mitochondrial imaging strategy optimized for lymphocytes, we showed that mitochondrial remodeling goes in favor of increasing mitochondrial size, volume and number in activated cells which highlights the importance of mitochondria in rapidly proliferating activated T cells despite their limited role in energy production.

MATERIALS and METHODS

Animals, cells and reagents

8–12 weeks old C57BL/6 female mice purchased from the Jackson laboratories (Bar Harbor, ME USA) were used for isolation of lymphocyte subsets. CD45.1⁺ CD45.2⁺ mice were generated by breeding C57BL/6 mice with B6.SJL-Ptprca Pepcb/BoyJ mice (purchased from Jackson Laboratories). OT-II TCR transgenic mice were obtained from Taconic Farms (Hudson, NY, USA). Mice were maintained at NIAID animal facilities according to Animal Care and Use Committee Standards.

For lymphocyte isolation mice were euthanized by CO₂ asphyxiation followed by cervical dislocation and spleens were harvested. Naïve CD4⁺ T cells were isolated using an isolation strategy described elsewhere (17). Cell purities were checked by staining with phenotyping antibodies and found to be over 95% as measured by flow cytometry. For the isolation of dendritic cells, spleens were removed and flushed by complete RPMI containing Liberase Blendzyme II and 2 µg/ml DNase, both purchased from Roche (Indianapolis, IN, USA). Spleens were then fragmented and incubated at 37˚C for 30 min. After incubation, RBCs were lysed with ACK-lysing buffer. DCs were isolated using CD11c Microbeads (Miltenyi Biotec; Auburn, CA, USA) and autoMACS (Miltenyi Biotec) according to manufacturer’s protocol. For experiments involving cell culture, cells were maintained in complete media (RPMI 1640 media containing 50 U/ml penicillin, 50 µM 2-mercaptoethanol, 50 µM streptomycin, 2 mM L-glutamine, 0.1 mM non-essential amino acids, 10% Fetal calf serum, 1 mM sodium pyruvate and10 mM HEPES.)

To test the involvement of Nitric Oxide (NO) and mTOR in the regulation of metabolic remodeling, NO scavenger 2-(4-Carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide potassium (Carboxy-PTIO) and mTOR inhibitor Rapamycin (both purchased from Sigma Aldrich; St. Louis, MO, USA) were added to cultures at 500 µM and 200 nM final concentrations respectively.

In vitro CD4⁺ T cell activation

For nonpolarizing CD4⁺ T cell (Th₀) activation, a protocol previously described elsewhere was used (18). Briefly, aseptically purified naïve CD4⁺ T cells were resuspended in complete media supplemented with 100 U/ml IL-2 (Peprotech; Rocky Hill, NJ, USA) and 10 µg/ml anti-TGF-β (Bioxcell; West Lebanon, NH, USA) and plated on sterile 24 well tissue culture plates (Corning/Life Sciences; Tewksbury, MA, USA) previously coated with 4 µg/ml anti-mouse CD3ε (Clone 145–2C11, Biolegend; San Diego, CA, USA) and 4 µg/ml anti-mouse CD28 (Clone 37.12, Biolegend). Cells were incubated for up to 96 h in a 5% CO₂ humidified tissue culture incubator at 37°C.

Adoptive transfer and in vivo CD4⁺ T cell activation

For adoptive transfer experiments, naïve CD4⁺ T cells purified from spleens of CD45.1⁺CD45.2⁺ WT mice and CD45.2⁺ OT-II transgenic mice were mixed at a 1:1 ratio, stained with e450 cell proliferation dye (ThermoFisher) according to the manufacturer’s recommendations, and resuspended in PBS. 200 μl PBS containing 2 × 10⁶ e450-stained CD4⁺ T cells were transferred into a CD45.2 WT mouse i.v. via tail vein injection. 24 h post T cell transfer, mice were injected i.v. with 100 μl PBS containing 7.5 X10⁵ dendritic cells (DCs) previously loaded with either OVA_(323–339) or LCMV GP_(61–80) peptides. Four days post DC transfer, mice were euthanized and spleens were harvested.

Preparation of samples of light microscopy

18 mm No.1.5 circular glass cover slips (Cat No:64–0714; Warner Instruments; Hamden, CT, USA) were placed inside wells of 12 well tissue culture plates (Corning) and 600 µl 0.01% Poly-L-lysine solution (Sigma Aldrich; St. Louis, MO, USA) was carefully applied on top of the coverslip avoiding spillover to the underside. After 5 minutes of incubation at RT, liquid was aspired and coverslip was dried for 2 h to O/N in a ventilating tissue culture hood.

Cells harvested upon activation or purification were stained with RPMI without phenol red supplemented with 2% FCS, 1% HEPES and 1/250 dilution of LIVE/DEAD™ Fixable Green Dead Cell Stain (Thermofisher; Waltham, MA, USA) for 20 min at 4°C protected from light. Cells were then, washed and resuspended with pre-warmed RPMI without phenol red supplemented with 2%FCS, 1%HEPES and 150 nM Mitotracker Red CMXROS (Thermofisher). Cells were incubated at 37 °C incubator for 30 min. followed by two washes in a protein free buffer. (RPMI without phenol red, PBS or HBSS) After the final wash cells were resuspended in protein free buffer at approximately 3 × 10⁶ cells/ ml and 500 µl of the cell suspension was transferred onto the Poly-L-lysine coated coverslip. (NB: While these numbers yielded optimal coating density for lymphocytes tested here, adjustments might be required if dramatic alterations in size of cells are anticipated due to specific experimental conditions). After 10 minutes of incubation at RT, suspension was aspirated and washed using FACS buffer (HBSS containing 2% FCS and sodium azide). This was followed by a 10 min incubation with anti CD16/32 antibody (Biolegend). Cells were then washed with FACS buffer, fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences; San Jose, CA, USA). Rabbit anti mouse TOM20 (2 µg/ml) (Abcam) followed by ATTO-647N conjugated anti rabbit secondary (2 µg/ml) (Sigma-Aldrich) was used to stain TOM20 intracellularly. Fixation, permeabilization and intracellular staining steps were carried out by following the BD Cytofix/Cytoperm protocol. DAPI staining of nuclei was carried out next by incubating the cells with PBS supplemented with 300 nM DAPI for 5 min at RT. Coverslips were washed once and mounted on slides by using approximately 10–15 µl mounting media prepared by mixing dissolved Mowiol 4–88 (Polysciences Inc; Warrington, PA) with %0.1 aqueous solution of p-phenylenediamine at 9:1 ratio according to manufacturer’s guidelines. Slides were left at RT in dark for O/N to ensure proper hardening before imaging. For long term storage samples were kept at −20°C.

Light microscopy and image analysis

Images were collected on a Leica TCS SP8 STED 3X system equipped with white light and UV excitation lasers, a pulsed 775 nm depletion laser, a HC PL APO 100x/1.40 oil STED White objective, and gated HyD detectors. To exclude dead cells from images, an initial single slide image was taken from the area of interest using all four lasers and upon confirmation of cell viability, 488 nm laser used to detect the Live/DEAD stain was switched off to decrease sampling time and photobleaching.

Image analysis was carried out using Imaris software version 9.1 (Bitplane; Zurich, Switzerland). Channel arithmetic extension in Imaris was used to sum the TOM20 and Mitotracker fluorescence intensities together and resulting channel was selected to create 3D surface rendering of the mitochondrion via surface creation and watershed splitting algorithm. The seed points were determined with a region growing estimated diameter of 0.5 µm. Other parameters used in the 3D surface creation such as surface grain size and diameter of the largest sphere were adjusted individually according to the fluorescence intensities.

Transmission Electron Microscopy

For Transmission Electron Microscope imaging, activated and naïve CD4⁺ T cells were harvested and fixed by directly mixing the cell culture suspensions with excess amounts of the modified Karnovsky fixative (0.1M sodium phosphate buffer containing 4% paraformaldehyde and 2.5% glutheraldehyde) (Electron Microscopy Sciences, Hatfield, PA). Samples were processed following a previously published microwave irradiation strategy (19) with the following modifications: Centrifugations were carried out at 800G, cells were infiltrated in Araldite resin (SPI, Inc., West Chester, PA), sections were cut at 200 nm thickness and no additional staining was performed on sections. Ultrascan 4000 camera (Gatan, Inc., Pleasanton, CA) was used for image collection.

Extracellular Flux Assay

Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) measurements were performed using Seahorse XF96 analyzer (Agilent Technologies, Santa Clara, CA) following a previously published protocol (20). For each experiment, activated and naïve CD4⁺ T cells were FACS sorted prior to assay to exclude non-viable cells. Chemicals used for glycolysis and mitochondrial stress tests were administered at the following final concentrations: Oligomycin (1 µM); 2,4 Dinitrophenol (2,4 DNP) (0.1 mM), Antimycin A (1 µM), Rotenone (1 µM), glucose (10 mM), 2- Deoxy-D-glucose (2-DG) (50 mM)

Flow cytometry

For measurement of mitochondrial membrane potential, a final concentration of 5 µM FCCP (depolarizes the membrane to show the background staining) or 6 µM oligomycin (hyperpolarizes the membrane to show maximum achievable potential) or plain media (to show current potential) were added directly to the cell cultures. After 10 min incubation at 37°C, staining mixture containing Tetramethylrhodamine, Methyl Ester, Perchlorate (TMRM) (ThermoFisher) (30 nM final concentration) and SYTOX Blue dead cell stain (Thermofisher) (1 µM final concentration) were added to the cells. Cultures were incubated for an additional 30 min and then analyzed in flow cytometry. [100 X (MFI _{(TMRM alone)}-MFI _(TMRM+FCCP) ) / ( MFI _{(TMRM + Oligomycin)} - MFI _(TMRM+FCCP) )] formula was used to standardize the measured TMRM MFI levels in the percent of the maximum potential format.

For flow cytometric detection of reactive oxygen species production, 5 μM MitoSOX Red (Thermofisher) or 500 nM CellROX (Thermofisher) were used together with 1 μM SYTOX Blue Dead Cell Stain.

For intracellular staining of GLUT transporters and mitochondrial markers, harvested cells were resuspended in FACS buffer supplemented with 1/250 dilution of Live/Dead Near IR dead cell stain (Thermofisher) and any relevant surface staining antibodies. Upon 30 min incubation at 4 °C, cells were washed, fixed and permeabilized by either BD CytoFix/Cytoperm kit (for GLUT stains) or Biolegend Foxp3 stain kit (for mitochondrial stains). Fluorochrome conjugated monoclonal antibodies against GLUT-1, GLUT-3, VDAC1, TOM 20, COXIV used in these experiments were purchased from Abcam (Cambridge, MA, USA).

For identification of the origin of adoptively transferred cells, antibodies against mouse CD4, CD45.1 and CD45.2 were used. Activation states of cells were detected by antibodies against CD44. mTOR complex 1 (mTORC1) activity was evaluated by measuring surface expressions of mTOR dependent markers CD71 and CD98 (21–23) using fluorescently labelled monoclonal antibodies. These antibodies were purchased from Biolegend.

Real time glucose uptake of the cells were monitored by adding 2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose (2-NBDG) (Thermofisher) to cultured cells. For this purpose, activated and naïve T cells were stained with a viability marker and then incubated with 2-NBDG for various durations. Cells remained in 37 °C culture until being harvested at various time points and then rapidly analyzed in flow cytometer.

Data acquisition was done using BD LSR II or BD X20 cytometers and data were analyzed using FlowJo software (version 10) (FlowJo LLC; Ashland, OR, USA).

Western Blot

For mitochondrial content detection by western blot, freshly isolated naïve CD4⁺ T cells and in vitro activated CD4⁺ T cells were stained with a viability dye and live cells were FACS sorted into protein free buffer. Equal numbers of viable cells from both conditions were lysed using RIPA buffer. Lysates were run on SDS-PAGE and transferred to nitrocellulose membranes. The following antibodies were used: Histon 3 (9715S, Cell Signaling Technology), HSP60 (4870S, Cell Signaling Technology), SIRT3 (5490S, Cell Signaling Technology), and TOM20 (sc-11415, Santa Cruz Biotechnology). Images were captured using the Odyssey system (Li-Cor).

qPCR

Ratio of mitochondrial to genomic DNA was used to assess the mitochondrial biogenesis potential of T cells. For this purpose, viable cells were FACS sorted and DNA was isolated using DNeasy Blood and Tissue Kit (Qiagen; Hilden, Germany) and qPCR was carried out with different dilutions of DNA using iQ™ SYBR® Green Supermix (Bio-Rad; Hercules, CA, USA). To represent genomic DNA Mouse 18S ribosomal DNA was amplified using (5’ TAGAGGGACAAGTGGCGTTC 3’ and 5’ CGCTGAGCCAGTCAGTGT 3’). To represent mitochondrial DNA mouse cytochrome oxidase subunit I was amplified using (5’ GCCCCAGATATAGCATTCCC 3’ and 5’ GTTCATCCTGTTCCTGCTCC 3’) and 12S ribosomal DNA was amplified using (5’ ACCGCGGTCATACGATTAAC 3’ and 5’ CCCAGTTTGGGTCTTAGCTG 3’). The transcriptional activity of various genes encoding antioxidant enzymes were measured using RNA isolated from freshly isolated naïve CD4⁺ T cells and viable FACS sorted four day activated CD4⁺ T cells following a protocol and primer sets published elsewhere (7).

Statistical Significance Analysis

Methods used to analyze statistical significance between experimental groups and the P value ranges are described in respective figure legends. Statistical analysis was calculated using Graph Pad Prism Version 7.

RESULTS

Prolonged activation of CD4⁺T cells results in cellular changes in order to facilitate glycolysis

Naïve CD4⁺ T cells harvested from mouse spleens were activated in vitro using plate bound antibodies specific for CD3 and CD28 in media containing soluble IL-2 and anti-TGF-beta antibody for four days in order to promote nonpolarized activation and proliferation. As expected, T cells proliferated and upregulated the activation marker CD44 as assessed by flow cytometry (Supplementary Fig 1A). We initially compared freshly isolated naïve and activated CD4⁺ T cells in Seahorse analyzer using a glucose stress test as described previously (20). For this assay, both cell types were immobilized in chambers and starved of glucose for 30 min. Following three baseline recordings of ECAR which correlates to lactate production by glycolysis, glucose was added to the wells to stimulate glycolysis. Next Antimycin/Rotenone (A/R) was added to block complex I and III of the respiratory chain which is expected to increase glycolytic capacity to its maximum (24). Finally, 2DG was added to the wells to stop glycolysis in order to demonstrate the direct link between the ECAR and the glycolytic performance. We observed that activated T cells rapidly responded to glucose by increasing their ECAR values, that was further enhanced by adding A/R. In contrast, naïve T cells did not show any significant response to stimulation showing the relative insignificance of glycolysis in resting T cell energy production (Fig.1A,B). These observations were in line with the findings of a previous study which showed gradual increases in ECAR values and lactate production for both human CD4⁺ and CD8⁺ T cells during the course of a three day in vitro activation (25).

Figure 1.

Activated CD4⁺ T cells remodel their glucose uptake machinery in order to meet increased energy demand. For all experiments purified naïve and 96 h ex vivo activated CD4⁺ T cells were used. A-B) Cells were FACS sorted for viability and 5×10⁵ viable cells were immobilized onto each well of a 96 well Seahorse analyzer where they were starved of glucose for 30 min and basal ECAR levels were measured. Glucose, A/R and 2DG were added consequently after this period and changes in ECAR values were recorded. A) Arrow heads indicate the time points treatments were added. Symbols and error bars refer to mean and SD of triplicates. B) ECAR values pooled from four independent glycolysis stress tests (Glycolysis= ECAR_post-glucose-ECAR_basal, glycolytic capacity=ECAR_post-A/R-ECAR_post-2DG). Symbols demonstrate the means of individual experiments, lines mark the mean of the pooled data. C-F) The expression levels of GLUT-1 (C,D) and GLUT-3 (E,F) as measured by flow cytometry. Representative histograms (C,E) and bar graphs (D,F) demonstrating the MFI values are shown. Bars and error bars represent mean and standard deviation of triplicates. G) Cells were stained with Live-DEAD and incubated in the presence of 10 µM 2-NBDG for up to 90 min. At each time point aliquots were harvested and run on flow cytometer. MFI values were then normalized by subtracting the background MFI. Symbols and error bars represent mean and standard deviation of quadruplicates respectively. H) Expressions of Slc2a1 and Slc2a3 genes encoding GLUT-1 and GLUT-3 respectively are quantified using RNA isolated from naïve and activated T cells. Bars and error bars represent mean and SEM of five independent experiments respectively. Data in (A,C-G) represent four independent experiments. Statistical significance was measured with Welch’s t-test (B,D,F,H) or Two-way ANOVA with Sidak’s multiple comparisons analysis (G). (0.01<P≤0.05=*, 0.001<P≤0.01=**, P ≤0.0001= ****)

Next, we ruled out the possibility that confounding effects resulting from glucose starvation might be accounting for these findings, by repeating the experiment in the absence of glucose deprivation which showed similar results (Supplementary Fig. 1B). Consistent with these observations, as compared to naïve T cells, activated T cells increased GLUT-1 and GLUT-3 expressions (Fig. 1C-F) that corresponded to their increased ability to take up glucose as measured by 2-NBDG (Fig.1G). Activated T cells were actively transcribing Slc2a1 and Slc2a3, the genes encoding GLUT1 and GLUT3 respectively. These genes were up to 40 folds more transcriptionally active in activated T cells compared to naïve (Fig. 1H). Thus, upon prolonged activation, T cells sustained increased glycolytic activity; and the high expression levels of GLUT1 and 3 facilitated glucose uptake to fuel the glycolytic activity.

T cells preserve their mitochondrial reserves upon prolonged activation

We performed a mitochondrial stress test to determine the contribution of mitochondria to cellular energy production in activated and naïve CD4⁺ T cells using Seahorse extracellular flux analyzer. Cells were immobilized and their basal oxygen consumption was recorded. This was followed by oligomycin treatment to inhibit ATP production that leads to a decrease in oxygen consumption. 2,4 DNP was added to uncouple oxidative phosphorylation from electron transport, which increases oxygen consumption to its maximum possible level. Finally, the electron transport chain through complex I and III was blocked using A/R which dropped the oxygen consumption to the basal level confirming the link between the oxygen consumption measurements and mitochondrial activity. We showed that, compared to naïve T cells, activated T cells used their mitochondria more actively for energy production as shown by significantly higher oxygen consumption levels (Fig 2A,B). However, the OCR/ECAR ratio was significantly lower in activated cells indicating that the prolonged activation induced increase in glycolysis surpasses the increase in mitochondrial oxidative phosphorylation (Fig 2C).

Figure 2.

T cells continuously activated in vitro, also increase their mitochondrial respiration despite greatly accelerated glycolysis. For all experiments purified naïve and 96 h ex vivo activated CD4⁺ T cells were used A-C) Cells were FACS sorted for viability and 5×10⁵ viable cells were immobilized onto each well of a 96 well Seahorse analyzer. Upon basal OCR measurements cells were sequentially treated with oligomycin, 2,4 DNP and A/R on time points indicated with arrow heads. A) Graph represents the changes in OCR values. Symbols and error bars refer to mean and SD of each recording. B) OCR values pooled from four independent mitochondrial stress tests (Basal: OCR_initial - OCR_post-A/R, maximal respiration=OCR_post-DNP - OCR_post-A/R, ATP-coupled respiration= OCR_Basal - OCR _{post-Oligomycin}). Symbols demonstrate means of individual experiments, lines mark the mean of the pooled data. C) Ratio of the basal OCR to basal ECAR values, pooled from the mitochondrial stress tests above. D-F) Mitochondrial membrane potentials were measured using TMRM either alone or in combination with oligomycin or FCCP. Representative histograms (D), MFI bar graphs (E) and percent of the maximum potential graphs (F) are shown. Data represent four independent experiments each carried out with triplicates. Statistical significance was calculated using Welch’s t-test. (P>0.05=ns; 0.01<P≤0.05=*, 0.0001<P≤0.001=***)

Next, we compared the mitochondrial performance between naïve and activated CD4⁺ T cells by measuring their relative mitochondrial membrane potential using TMRM either alone or in combination with oligomycin or FCCP. We showed that activated T cells have higher total membrane potential (Fig. 2D,E) but use a similar percent of their maximum potential compared to naïve T cells (Fig. 2F). These findings suggest that the increase in TMRM staining may either represent a continuation of mitochondrial membrane hyperpolarization that was shown to occur shortly after activation (26, 27) or merely a consequence of a greater total mitochondrial content in activated T cells. Since both activated and resting T cells use less than 30% of their mitochondrial membrane potential, both can be considered efficient in terms of mitochondrial ATP use with adequate reserves.

Activated T cells maintain their mitochondrial activity despite prolonged ROS production

It was shown previously that T cell activation leads to rapid increases in mitochondrial ROS production which acts as a second messenger to induce downstream cellular changes (16). We asked whether this early increase in ROS production is sustained upon prolonged activation. We observed higher ROS production in activated T cells as measured by both CellROX (Fig.3A), a marker for total intracellular ROS production, and MitoSOX (Fig.3B), a specific dye indicating mitochondrial ROS production. The increases in intracellular ROS often triggers transcriptional activation of anti-oxidant enzymes such as SOD1 and SOD2 to prevent pathologic consequences of oxidative stress such as induction of mitochondrial swelling by increasing mitochondrial membrane permeability (28). However, we observed no change in transcriptional activity of the genes encoding anti-oxidant enzymes in activated T cells (Fig. 3C) suggesting either ROS levels were insufficient to induce transcription or there are other mechanisms that prevent upregulation of antioxidant genes in order to preserve beneficial increase in ROS levels. Consistent with a beneficial role of increased ROS production we observed no increase in percent of maximum mitochondrial membrane potential (Fig. 2) which would rise during mitochondrial dysfunction. Moreover, electron micrograms of both naïve and activated T cells showed healthy cristae structures and absence of swollen or vesicular-swollen mitochondrial matrix areas similar to those exemplified in (7, 29, 30), all of which rule out any ROS-induced gross pathology in mitochondrial structures (Fig.3D). Therefore, sustained increase in ROS can be considered as a regulatory component of activation rather than a signature of pathology.

Figure 3.

Increased ROS production following T cell stimulation persists during prolonged activation without causing oxidative stress induced mitochondrial dysfunction. For all experiments purified naïve and 96 h ex vivo activated CD4⁺ T cells were used. A,B) Cells were stained with SYTOX Blue viability dye together with CellROX (A) or MitoSOX (B). Representative histogram overlays (top panels) and graphs of MFI values (bottom panels) are shown. Bars and error bars represent mean and standard deviation respectively. Data represents three independent experiments each carried out with quadruplicates. C) Transcription of genes coding for anti-oxidant enzymes glutathione reductase (Gsr), glutathione peroxidase 1 (Gpx1) superoxide dismutase (Sod1 and Sod2) and catalase (Cat) as measured by qPCR are shown for both activated and naïve T cells. Bars and error bars refer to mean of individual samples obtained from five independent experiments and standard error of the mean respectively. D) TEM electron micrograms of naïve and activated T cells. Images represent sections obtained from 10 individual cells. N refers to nucleus (Scale bars = 400 nm). Statistical significance was calculated using Welch’s t-test. (P>0.05=ns, 0.001<P≤0.01=**, 0.0001<P≤0.001=***)

Prolonged T cell activation leads to increases in mitochondrial content

We explored whether during prolonged activation, T cells increase their mitochondrial content. To do so, we measured the expression levels of various proteins specifically localized in mitochondria and are widely used to assess mitochondrial content (7, 31–33). T cells activated for four days in vitro as compared to naïve T cells showed increased levels of mitofilin, cytochrome c oxidase subunit IV (COXIV) as measured by flow cytometry; sirtuin 3 (Sirt3), heat shock protein 60 (Hsp60) as measured by western blot; voltage dependent anion channel 1 (VDAC1) and translocase of the outer membrane 20 (TOM20) as measured by both (Fig. 4A-C). Upregulation was observed for all these markers demonstrating an increase in the total mitochondrial content in T cells upon prolonged activation consistent with the increased TMRM staining shown in (Fig. 2D,E). To assess the mitochondrial biogenic potential of the cells we compared the ratio of COI or 12S ribosomal DNA, genes encoded in the mitochondrial genome, to 18S ribosomal DNA, a gene encoded in the nuclear genome. This analysis revealed that four days post stimulation in vitro, T cells retain an increased mitochondrial DNA mass which further supports increased mitochondrial biogenesis capacity (Fig 4D).

Figure 4.

Prolonged T cell activation leads to an increase in total mitochondrial content. For all experiments purified naïve and 96 h ex vivo activated CD4⁺ T cells were used A,B) Cells were stained with a viability dye together with various mitochondria specific markers and analyzed in flow cytometry. Representative histogram overlays (A) and MFI graphs (B) are shown. Bars and error bars represent mean of triplicates and standard deviation respectively. C) Live cells were FACS sorted and equal numbers of naïve and activated cells were lysed, separated in protein gel and immunoblotted for mitochondrial markers as well as H3 for loading control. Representative western blot image is shown. Data represent two independent experiments. D) Total DNA was isolated from naïve and FACS sorted live activated CD4⁺ T cells. Graphs show mitochondrial DNA copy numbers relative to genomic DNA that were measured using qPCR. Bars and error bars refer to mean of individual samples obtained from five independent experiments and standard error of the mean respectively. Statistical significance was calculated using Welch’s t-test. (*=0.01<P≤0.05; ***=0.0001<P≤0.001; P ≤0.0001= ****)

Metabolic remodeling following T cell activation is linked to NO mediated increases in mTOR activity.

Our data, so far, showed that prolonged T cell activation leads to a remodeling that involves both glucose metabolism and mitochondria. In order to confirm that these changes directly stem from TCR stimulus, we repeated our four day culture experiments using different combinations of activation conditions. This comparative analysis revealed that the observed increases in GLUT and mitochondrial markers can be recapitulated almost perfectly by plate bound anti CD3 stimulation alone and that addition of IL2 in culture or costimulation with CD28 have no effects in the absence of TCR stimulation (Suplementary Fig 2A,B).

Next, to have a more in depth understanding of the dynamics of metabolic and cellular remodeling resulting from prolonged T cell activation, we measured the changes in the expression of CD44, GLUT1 and mitochondrial markers on daily basis during the course of four day CD4⁺ T cell activation in vitro (Fig. 5A,B). This analysis showed gradual increases in the expression levels of the activation marker CD44 and glucose transporter GLUT1. On the other hand, markers of mitochondrial content increased within the first 24 h and plateaued after that showing that the need for increased glucose uptake is corelated with the duration of activation while majority of mitochondrial remodeling occurs within the first day of activation. This data also complemented the sustained high transcriptional activity we showed in Slc2a genes in Fig. 1H.

Figure 5.

Metabolic remodeling due to activation can be prevented by the inhibition of NO or mTOR pathways. Purified naïve CD4⁺ T cells were activated up to 96 h ex vivo. A-B) Time dependent changes in activation status (CD44), GLUT1 expression and mitochondrial content (TOM20 and COXIV). Representative histogram overlays (A) and MFI graphs (B) are shown. Symbols and error bars represent mean of triplicates and standard deviation respectively. Time points were compared against 0 h. C-E) Naïve CD4⁺ T cells were activated for 24 h ex vivo in the presence of Rapamycin or Carboxy-PTIO where indicated. Representative histogram overlays and MFI graphs for Nitric Oxide levels (C), mTOR dependent surface markers (D) and metabolic markers (E) are shown. Bars and error bars represent mean of triplicates and standard deviation respectively. Data are representative of two independent experiments. Freshly isolated naïve CD4⁺ T cells were used as control. Statistical significance was measured using one-way ANOVA with Dunnett’s (B) or Tukey’s (C-E) multiple comparisons analysis (P>0.05=ns, *=0.01<P≤0.05, 0.001<P≤0.01=**, ***=0.0001<P≤0.001, P ≤0.0001= ****).

Earlier studies on human samples showed a direct link between T cell stimulation induced NO production and the increase in mitochondrial calcium levels and mitochondrial membrane hyperpolarization (34, 35). NO, through induction of mTORC1 activity, was shown to be responsible for prevention of mitophagy by Rab4A-mediated depletion of Drp1(36–38). Furthermore, a recent clinical trial in SLE patients showed that mTOR blockade leads to a reduction in mitochondrial mass (39). In order to address whether a similar link between T cell activation, NO production and mTOR activity might be responsible for the changes we observed in the expression of GLUT transporters and/or mitochondrial markers, we activated CD4⁺ T cells in vitro in the presence or absence of NO scavenger (Carboxy-PTIO) or mTOR inhibitor (Rapamycin). We first showed that T cell activation induced NO can be efficiently scavenged in the presence of Carboxy-PTIO (Fig. 4C) and that the blockage of NO results in a level of reduction in mTORC1 dependent cell surface markers CD71 and CD98 similar to those observed in the presence of Rapamycin (Fig. 4D). Upon confirming the link between T cell activation mediated NO production and the induction of mTOR activity, we showed that both rapamycin and Carboxy-PTIO prevented the upregulation of GLUT1, VDAC1 and TOM20 in activated T cells. Carboxy-PTIO mediated inhibition of the increase in mitochondrial content and glucose transporters were more pronounced compared to the level of inhibition, observed in the presence of rapamycin. This suggests that NO may be utilizing both mTOR dependent and independent pathways during T cell activation induced metabolic remodeling.

Changes observed in in vitro activated T cells can be recapitulated in vivo

Thus far, we have shown that prolonged in vitro activation of isolated T cells results in increases in glucose transporters, total mitochondrial content and mitochondrial ROS production (Fig 1–4). To test whether these changes also occur upon in vivo activation of the cells, we carried out an adoptive transfer experiment (Fig. 6A). For this purpose, naïve TCR transgenic CD4⁺ T cells isolated from CD45.1⁻ CD45.2⁺ OT-II mice were mixed with naïve polyclonal CD4⁺ T cells isolated from WT CD45.1⁺ CD45.2⁺ congenic mice at 1:1 ratio. Cells were labeled with e-450 cell proliferation dye and adoptively transferred to WT CD45.2⁺ animal. 24 h post-transfer, recipient mice were injected with dendritic cells (DCs) pulsed with either OVA_(323–339) or control peptide, namely LCMV GP_(61–80). Four days post-stimulation, spleens were harvested and analyzed in flow cytometer using a gating strategy outlined in (Fig. 6B). As expected, OT-II transgenic T cells that were activated by OVA_(323–339) pulsed DCs in vivo had gone through a few rounds of proliferation and induced the expression of activation marker CD44 (Fig. 6C). In vivo activation of OT-II T cells also increased the fluorescent intensities of GLUT-1, COXIV, TOM20 and MitoSOX as compared to control groups indicating that changes accompanying in vitro activation can be recapitulated in vivo (Fig 6D-G)

Figure 6.

In vivo activation of CD4⁺ T cells leads to changes that are similar to those observed in vitro. A) Schematic outline of the adoptive transfer experiment. B) Gating strategy used to discriminate adoptively transferred OT-II and polyclonal T cells in recipient mouse spleens. C) Histogram overlays showing the levels of CD44 expression (left) and the extent of proliferation (right) in adoptively transferred polyclonal and OT-II T cells four days post stimulation with DCs pulsed with control GP peptide (top) or OVA peptide (bottom). D-G) T cells were activated in vivo for four days as illustrated in (A). Histogram overlays (left) and MFI graphs (right) show the levels of GLUT1 (D), TOM20 (E), COXIV (F) and MitoSOX (G). Each circle is an individual mouse. Data represent two independent experiments. Statistical significance was calculated using Two-way ANOVA with Sidak’s multiple comparisons test. (P>0.05=ns; P ≤0.0001= ****)

Multicolor imaging strategy offers a comprehensive analysis of activation-induced mitochondrial changes in T cells.

Having observed that both mitochondrial content and biogenesis increase during prolonged activation, we wanted to visualize the morphology of the resulting mitochondria. Visualization of mitochondria in live T cells is challenging given the small cytoplasmic volume and the relative fragility of T cells ex vivo.

To visualize T cell mitochondria, we developed a strategy using a combination of a Mitotracker dye (MitoTracker Red CMXRos) that accumulates in the mitochondrial matrix (40) and a fluorescently labelled antibody specific for a subunit of TOM20 (41). The nucleus was stained with DAPI and dead cells were marked by a viability dye (Fig 7A). We used high resolution STED microscopy to distinguish individual mitochondria inside the small cytoplasmic volume of T cells. To reduce excessive photobleaching due to high laser power with STED imaging, we used photostable ATTO-647 fluorochrome to label TOM20-specific antibody. This dual mitochondrial staining approach both eliminated staining artifacts and also provided a more precise image of mitochondrial morphology. We then applied computer algorithms to sum the fluorescence intensities of TOM20 and Mitotracker stains and to create artificial surfaces based on the combined fluorescence intensity to predict the gross mitochondrial morphology and measure total mitochondrial volume. The surfaces, consisting of both matrix and membrane compartments of the mitochondria, were also split via watershed splitting algorithm to estimate the size and number of individual mitochondria in each cell (Fig 7B) (Supplementary Video 1).

Figure 7.

High-resolution multicolor imaging of lymphocytes reveals mitochondrial changes in CD4⁺ T cells induced upon prolonged activation. Cells activated for four days in vitro or freshly isolated from spleens of C57BL/6 mice were stained with Live/Dead and MitoTracker CMXROS red, then mounted on Poly-L lysine coated coverslips. Cells were then fixed, permeabilized and stained for TOM20 and DAPI as described in methods. A) Microscope image demonstrating viable (white arrow heads) and non-viable (yellow arrow heads) cells in the same area of interest (Scale bar 4 µm). B) Outline of the sequential image processing strategy used for distinguishing mitochondrial boundaries in lymphocytes. Naïve CD4⁺ T cells, stained as outlined above, were imaged for TOM20, Mitotracker, and DAPI in STED system Images left to right show microscope images of TOM20, Mitotracker, their merged view with DAPI, combined channel showing sum of both mitochondrial markers in one channel, computer generated surface of the combined channel and individual mitochondria estimates based on the watershed splitting algorithm applied to the combined surface (Scale bars: 1 µm). C-F) Naïve and activated T cells were imaged and analyzed in groups of five cells per area of interest as described above. Representative STED microscope images (C) and data derived from image analysis showing average total mitochondrial volumes per cell (D), average number of mitochondria per cell (E) and volume distribution of individual mitochondria (F) are given. Lines represent mean values, symbols represent averages of five cells (D, E) or individual mitochondrion (F). Data represent two independent experiments each with at least 25 cells analyzed per condition. Statistical significance was calculated using Welch’s t-test. (P ≤0.0001= ****)

We used this imaging strategy to compare freshly isolated naïve CD4⁺ T cells with T cells that were activated for four days in culture. The images showed fewer, sparsely distributed mitochondria in naïve T cells as compared to activated T cells. (Fig 7C). Image analysis showed that the average total mitochondrial volume per cell is almost three folds higher in activated cells (Fig. 7D). Furthermore, compared to naïve cells, activated T cells had at least a two-fold increase in the estimated number of mitochondria (Fig 7E) and average volume of individual mitochondria in these cells were larger (Fig. 7F). These results not only complement our observations on total mitochondrial content (Fig 4–6) but also shows that the increased total mitochondrial content in activated T cells is a direct consequence of both individual mitochondrial enlargement and increased mitochondria numbers. Altogether, our data suggest that activation-induced mitochondrial biogenesis serves the purpose of increasing mitochondrial content in multiple ways.

DISCUSSION

In this study, we have addressed key questions related to metabolic and cellular changes that accompany prolonged T cell activation and proliferation. Our data showed that the early increases in glycolysis and OXPHOS in activated T cells are maintained during prolonged activation. We have shown that after four days of both in vitro and in vivo activation, rapidly proliferating CD4⁺ T cells have already increased their GLUT transporters and are more able to import glucose into the cell and yet are still transcriptionally active to induce further GLUT expression.

Our findings mainly focused on the changes in mitochondrial dynamics. This was an unexplored area due to the lack of high-resolution imaging strategies optimized for lymphocytes. These cells are not only difficult to handle and culture ex vivo (20), but also their massive nucleus to cytoplasm ratio poses a great difficulty for any researcher aiming to visualize cytoplasmic compartments. We addressed these shortcomings by introducing viability staining to exclude non-viable cells as well as targeting the mitochondria with two independent markers for added specificity. Taking advantage of this novel multicolor mitochondria staining protocol in a high-resolution STED microscope setting and processing the raw data further by using an optimized computer algorithm, we have been able to visualize individual mitochondria and estimate their boundaries even at close proximity areas.

These analyses showed that upon continuous activation, CD4⁺ T cells upregulate their total mitochondrial volume and number as well as the size of individual mitochondria. Our comparative analysis on the ratio of mitochondrial DNA to genomic DNA revealed that activated T cells have increased mitochondrial biogenesis. In addition to the structural mitochondrial remodeling, we showed that increases in ROS production is sustained through long-term activation which is likely maintained at least in part by lack of increase in anti-oxidant enzyme synthesis. Nevertheless, despite sustained increases in cellular ROS levels, our data demonstrated that the functional and structural integrity of mitochondria is preserved which is in line with the previous observations suggesting a physiological role for the ROS production.

For many cell types, the function of mitochondria goes far beyond its role in energy production. Recent studies showed that there is a clear correlation between the intracellular organization of mitochondria and the stress response of cells (42, 43). Mitochondria are also known to actively participate in programmed cell death (44). Furthermore, recently, we have identified a novel contribution of mitochondria in modulating activation induced cell death in B lymphocytes. Our assays showed that antigen stimulation of B cells requires a spatiotemporally distinct second stimulus in the form of either cognate B-T interaction or TLR activation. Lack of this second signal triggered a cascade of events including mitochondrial swelling, inefficient mitochondrial respiration and increased production of reactive oxygen species (ROS) all of which eventually led to the death of the cell (7). All of these examples show the importance of accurately assessing mitochondrial changes.

Altogether our findings provide the most comprehensive mitochondria-centric approach to decipher changes that accompany T cell activation and highlight the functional importance of mitochondria in activated T cells despite its limited role in energy production. The data presented here not only increase our understanding of mitochondrial remodeling following prolonged T cell activation but also offer novel methodological approaches that can be applied to decipher changes in the mitochondrial dynamics in other cell types under numerous experimental settings.