Targeting T cell Bioenergetics by Modulating P-glycoprotein Selectively Depletes Alloreactive T cells to Prevent GVHD

Abstract

T lymphocytes play a central role in many human immunologic disorders including autoimmune and alloimmune diseases. In hematopoietic stem cell transplant, acute graft-versus-host-disease (GVHD) is caused by an attack on the recipient’s tissues from donor allogeneic T cells. Selectively depleting GVHD-causing cells prior to transplant may prevent GVHD. In this report we have evaluated 24 chalcogenorhodamine photosensitizers for their ability to selectively deplete reactive T lymphocytes, and identified the photosensitizer 2-Se-Cl that accumulates in stimulated T cells in proportion to oxidative phosphorylation (OXPHOS). The photosensitizer is also a potent stimulator of P-glycoprotein (P-pg). Enhanced P-gp activity promotes the efficient removal of photosensitizer not sequestered in mitochondria, and protects resting lymphocytes essential for antipathogen and antitumor responses. To evaluate the selective depletion of alloimmune responses, donor C57BL/6 splenocytes were cocultured for 5 days with irradiated Balb/c splenocytes, and then photodepleted (PD). PD-treated splenocytes were then infused into lethally irradiated BALB/c (same-party) or C3H/HeJ (third-party) mice. Same-party mice that received PD-treated splenocytes at the time of transplant lived 100 days without evidence of GVHD. In contrast, all mice that received untreated primed splenocytes and third-party mice that received PD-treated splenocytes died of lethal GVHD. To evaluate the preservation of antiviral immune responses, acute lymphocytic choriomeningitis virus (LCMV) infection was employed. After PD, expansion of antigen-specific naïve CD8+ T cells and viral clearance remained fully intact. The high selectivity of this novel photosensitizer may have broad applications and provide alternative treatment options for patients with T lymphocyte mediated diseases.

Introduction

T lymphocytes are central to the development of adaptive immune responses, but may also become pathologic and mediate many human immunologic disorders including both autoimmune and alloimmune diseases. In hematopoietic stem cell transplant (HSCT) acute graft-versus-host-disease (GVHD) is associated with significant morbidity and mortality, and is caused by an attack on the recipient’s tissues from donor allogeneic T cells (1). Multiple organs are targeted including the skin, liver, lungs and gut (2). Depletion of T lymphocytes by two to three logs from the HSCT graft prior to transplant effectively reduces the incidence of acute GVHD (3). However, this approach has been associated with graft failure, and an increased risk of disease recurrence (4, 5). The goal of selective depletion is to prevent acute GVHD by removing only the GVHD-causing T cells from the graft prior to transplant. Pre-clinical experiments demonstrate that when GVHD-causing cells are selectively eliminated, healthy lymphocytes remain that may mediate anti-leukemia, antiviral, and antifungal immune responses (6, 7). This technique requires the co-culturing of leukemia-free, patient-derived antigen presenting cells with donor lymphocytes. Alloactivated donor lymphocytes can then be selectively targeted for removal. Recently, two methods have been employed to selectively remove alloreactive T cells: 1) the use of monoclonal antibodies against activation markers such as CD25, or FasL-mediated induction of apoptosis, and 2) the use of the photosensitizer 4,5-dibromorhodamine methyl ester (TH9402) to target P-glycoprotein differences of activated cells (8-10). Although these techniques effectively decreased the incidence of severe acute GVHD, insufficient depletion of alloreactive cells and non-specific depletion of cells important for regulatory, antiviral, and antifungal immunity occurred, resulting in persistent, chronic GVHD and recurrent infections (11, 12). Consequently, further efforts are required to improve selective depletion by building on the successes and overcoming the limitations of these prior techniques.

A challenge in developing a new selective depletion technique is identifying a target unique to activated cells. We hypothesize that the increased oxidative phosphorylation (OXPHOS) of activated cells may be used to identify and remove alloreactive, GVHD-causing cells prior to HSCT. In general, cells generate ATP by aerobic glycolysis and OXPHOS. In 1924 Otto Warburg observed that cancer cells have a unique bioenergetic profile with an increase in aerobic glycolysis over OXPHOS compared to cells in normal tissues, which is often referred to as the “Warburg Effect” (13). Although aerobic glycolysis is less efficient yielding only 2 ATP compared to the possible 36 ATP generated by OXPHOS, increased aerobic glycolysis may provide the macromolecules and reducing equivalents required to support proliferation (13). More recently, this bioenergetic configuration has been identified in pathogenic T cells, and may represent metabolic adaptations to chronic stimulation (14, 15). Additionally, memory T cells have recently been shown to utilize both glycolysis and OXPHOS to a greater extent than naïve T cells to support the rapid and prolonged proliferation required for secondary immune responses (16). The rapid recall response of memory T cells is the result of increased cellular mitochondria content and the associated bioenergetic advantage. The greater mitochondrial mass in memory cells facilitates a rapid induction of OXPHOS to produce substantial ATP upon activation. ATP production promotes conversion of glucose into glucose-6-phosphate by mitochondria-associated, ATP-dependent hexokinases, which is required for the first step of glycolysis (17). As a result, the rapid induction of OXPHOS directly engages glycolysis in memory T cells, creating the bioenergetic configuration seen in malignant cells and pathogenic T cells. This observation suggests that the Warburg Effect is not unique to pathogenic cells, but represent a bioenergetic reconfiguration that may occur in all cells to support rapid proliferation (15, 16).

None of the photosensitive agents in use today have demonstrated selectivity for activated cells without significant toxicity occurring in resting cells. We have previously demonstrated that prolonged intracellular resident times associated with the photosensitive agent TH9402 resulted in non-selective depletion of susceptible lymphocyte subsets (11). TH9402 is highly dependent on P-glycoprotein (P-gp) for cell extrusion, and cells that express low P-gp activity are susceptible to increased intracellular accumulation. Consequently, lymphocyte subsets with low P-gp activity, such as CD4⁺ and memory T cells, are disproportionately depleted when using this photosensitizer. In the clinical setting of HSCT, the use of TH9402 has resulted in the non-selective depletion of lymphocytes important for normal immune responses, and poor patient outcomes (12).

The ability of chalcogenorhodamines photosensitizers to both modulate P-gp activity and inhibit OXPHOS provides the basis for our new approach to selective depletion. In this study, we evaluated whether a novel photosensitizer with these dual properties would improve selectivity for activated T lymphocytes, and protect resting cells important for normal immune responses. For this purpose, we created an in vitro model with a bioenergetic profile similar to pathogenic T cells. We then designed a small library of 24 photosensitive chalcogenorhodamines (figure S1A), and measured their effects on the bioenergetics of resting T cells to determine potential for toxicity. We identified a photosensitizer (2-Se-Cl) from this library that potently stimulates P-gp to protect resting cells and selectively inhibits OXPHOS of activated cells, and then confirmed selectivity by evaluating 2-Se-Cl for the ability to selectively deplete an immune response and prevent GVHD.

Materials and Methods

Nomenclature

Compound 2-Se-Cl is shown in figure 1, and the 24 photosensitizers investigated in this report are shown in supplemental figure S1A. Compounds are identified by degree of P-gp ATPase stimulation, and are referred to as “strong” or “weak” P-gp stimulators in the remainder of the manuscript.

Figure 1. The chalcogenorhodamine photosensitizer 2-Se-Cl.

The photosensitizer 2-Se-Cl accumulates in stimulated T cells in proportion to the degree of OXPHOS, and also potently stimulates P-glycoprotein (P-pg) to protect resting lymphocytes essential for antipathogen and antitumor responses.

Cell isolation and stimulation

To examine the effects of our photosensitizers on human T cells, peripheral blood mononuclear cells (PBMCs) were separated using Ficoll-Hypaque density gradient centrifugation (Organon Teknika, Durham, NC), and rested in RPMI supplemented with 10% heat-inactivated fetal calf serum. To produce stimulated T cells, human PBMCs were cultured in RPMI 1640 (LifeTechnologies, Gaithersburg, MD) with 50 ng/mL Staphylococcus enterotoxin B (SEB, Toxin Technology, Inc., Sarasota, FL) for 72 hours.

Photodepletion (PD)

For all PD experiments, cellss were suspended in a photosensitizer-rich media of 5.0 × 10⁻⁸ M for 20 minutes followed by 30 minutes in a photosensitizer-free media. Cells were then exposed to 5 J/cm² of light (600nM, 65-Watt equivalent LED) and 180 rotations per minute. Afterward, cells were washed twice to facilitate photosensitizer removal in RPMI 1640 supplemented with 10% heat-inactivated fetal calf serum (FCS).

Animals

All studies were approved by the Animal Care and Use Committees of Wake Forest University. BALB/c, C57BL/6, and C3H/HeJ mice (The Jackson Laboratories. Bar Harbor, ME) were female and 8 to 12 weeks of age at the time of transplant. P14 TCR transgenic (H-2^b and Thy1.2⁺) mice were purchased from National Cancer Institute (Frederick, MD). Thy1.1⁺ P14 transgenic mice were generated by crossing P14 mice onto a B6.PL-Thy1a/Cy (H-2^b and Thy1.1⁺) background as described previously (18). All mice were housed in a specific pathogen-free facility for the duration of the study.

BMT

To evaluate the selective depletion of alloimmune responses, a well-established complete MHC antigen-mismatched murine model of HSCT was employed (19). To prepare the PD – treated primed splenocytes, donor C57BL/6 splenocytes were cocultured for 5 days with irradiated (20 Gy) Balb/c splenocytes, and then photodepleted. On the day of HSCT, 10 × 10⁶ donor C57BL/6 T-cell–depleted bone marrow cells (< 0.1% mature CD3⁺ T cells) accompanied by 5 × 10⁶ PD – treated (treatment group) or untreated (control group) primed splenocytes were infused via tail vein injection into irradiated same-party BALB/c (9 Gy) or third-party C3H/HeJ (9.5 Gy) recipients. All mice were monitored for signs of GVHD according to an established mouse GVHD grading system, and mice with severe GVHD (overall score >5) were euthanized (20, 21).

P14 cell isolation, transfer, and viral infection

Splenocytes were isolated from naive P14 Thy1.1⁺ mice, photodepleted, and then rested overnight. Cells were then enumerated by staining with α-CD8α and D^bGP33–41, followed by i.v. transfer of 10⁵ naive GP33-specific CD8⁺ T cells into naïve C57BL/6 Thy1.2⁺ hosts. Mice were then infected i.p. with 2 × 10⁵ pfu of LCMV-Armstrong . LCMV was prepared and quantitated as described previously (18). After 8 days animals were sacrificed, and FACS analysis performed on splenocytes for enumeration and function. Viral levels in spleens were determined by plaque assays as previously described (22).

Reagents for flow cytometry

The following monoclonal antibodies were used and purchased from eBioscience: mouse anti-human α-CD3-Cyanin-7-allophycocyanin (APC-Cy7; clone OKT3); α-CD4-Pacific Blue (PB; clone RPA-T4); α-CD8-fluorescein isothiocyanate (FITC; clone SK1); α-CD25-Cyanin-7-phycoerytherin (PE-Cy7; clone BC96); and rat anti-mouse α-CD19-PE (clone 1D3); α-CD3-FITC (clone 17AD); α-CD90.1- Peridinin-chlorophyll protein-Cy5.5 (PerCP-Cy5.5) or –FITC (clone HIS51); α-CD8-PB or –PE or –V500 (clone 53-6.7); α-CD27-PE-Cy7 (clone LG.7F9); α-CD44-FITC or –PerCP-Cy5.5 (clone IM7); α-CD62L-APC-Cy7 (clone MEL-14); α-CD127-FITC (clone A7R34); α-KLRG1 (clone 2F1); INF-γ-FITC (clone XMG1.2); MIP-1α-PE (clone DNT3CC); TNF-α-PE-Cy7 (cloneMP6-XT22); and IL-2-APC (clone JES6-5H4) . The following monoclonal antibodies used to identify naïve and memory T cell subsets were purchased from BD Pharmingen: mouse anti-human α-CD27-PE (clone M-T271) and α-CD45RO-FITC (clone UCHL1). Apoptosis was assessed by FACS analysis of annexin V and 7AAD-stained cells (BD Pharmingen) as previously described (23). MitoTracker Red, MitoTracker Green, and LysoTracker Green were used and purchased from LifeTechnologies (Gaithersburg, MD).

Cell proliferation assay

Measurement of in vitro proliferation was performed by the method of CFSE dilution. Human PBMCs or mouse splenocytes were stained with 0.5 μM CFSE (Invitrogen) and incubated at 37°C for 7 minutes. CFSE was quenched by adding 1mL normal AB serum (NABS, Gemini Bio-Products) for 2 minutes, and the cells were washed twice in RPMI 1640 containing 10% NABS. 1 - 5 × 10⁵ cells were then stimulated in a 48 well plate for 5 – 6 days, followed by staining with mAb to CD3, and then acquired on a BD FACS Canto II flow cytometer. Data were analyzed using FlowJo, and the division, proliferative indices, and the percent divided parameters were calculated using the Proliferation Platform in the FlowJo software package.

Photosensitizer retention and extrusion experiments

For experiments comparing the retention of the different photosensitizers, SEB stimulated human PBMCs were stained with mAb to CD3 and CD25. Cells were then washed and suspended at a concentration of 2 × 10⁶ cells/mL in phosphate buffered saline (PBS) containing 2.5 × 10⁻⁸ M of the photosensitizer for 20 minutes. Cells were than washed and suspended in photosensitizer-free media for 30 minutes (RPMI supplemented with 10% heat-inactivated FCS). To measure extrusion kinetics, HUT-78 (ATCC, Manassas, VA) were suspended at a concentration of 2 × 10⁶ cells/mL in PBS containing 2.5 × 10⁻⁷ M of the photosensitizer for 20 minutes. HUT-78 cells were than washed and suspended in photosensitizer-free media for 18 hours. Data were acquired on a BD FACS Canto II flow cytometer, and analyzed using FlowJo software.

Measuring T cell bioenergetics

For experiments on resting T cells, CD3⁺-selected cells (Miltenyi Biotech, Bergisch Gladbach, Germany) were immediately purified from human PBMCs. For experiments on SEB stimulated T cells, CD25⁺-selected T cells were isolated from culture and compared with the CD25⁻ fraction. Isolated T cells were then exposed to photosensitizer alone for 20 minutes (coloration only), or photodepleted as described above. Cells were then washed twice in a protein rich media (10% FCS) to absorb and remove all photosensitizer, and then kept in the dark (37°C, 5% CO2). Real-time analysis of the extracellular oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of isolated T cells was measured with the XF24 Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA) according to the manufacturer’s protocol, and performed 1 hour after coloration or PD. In brief, cells were plated in XF-24 culture plates with 1.5 × 10⁶ T cells/well. At the indicated time points, cells were washed and analyzed in XF Running Buffer per the manufacturer’s instructions. Bioenergetic profiles of the T cells were measured in a basal state, and after the addition of 1M oligomycin (to block ATP synthesis), 1.5 M fluoro-carbonyl cyanide phenylhydrazone (FCCP; to uncouple ATP synthesis from the electron transport chain), and 100 nM rotenone (to block complex I of the electron transport chain) (all Sigma-Aldrich).

Statistical analysis

Unless otherwise stated, all experiments were performed in triplicate. Data are represented as mean ± SE. Graphs were generated using GraphPad Prism (GraphPad software Inc., USA). For comparison of bioenergetics, the area under the curve (AUC) of rates 1-3 for basal OCR and ECAR was calculated and adjusted for baseline (after oligomycin injection), and a significant difference between groups was determined by one-way ANOVA and Tukey’s HSD method for post-hoc analysis. Survival data were analyzed by log rank test. All t-tests were 2 sided, and p-values up to .05 were considered significant.

Results

Bioenergetics of superantigen stimulated T cells resembles the Warburg Effect

Alloreactive T cells that mediate GVHD increase both aerobic glycolysis and OXPHOS, which is also the bioenergetic phenotype of pathogenic T cells seen in other disease settings (15, 24, 25). Consequently, we hypothesized that this bioenergetic configuration results from a chronic or robust stimulation in context of disease, and may therefore be induced in vitro. For this purpose, we used staphylococcal superantigens (SAgs) to create a model of robust T cell stimulation. To test our hypothesis, human PBMCs were cultured with 50 ng/mL of SEB for 72 hours, at which time the bioenergetic profiles were measured by real-time analysis. For this analysis, we used the extracellular oxygen consumption rate (OCR) as an indicator of OXPHOS, and the extracellular acidification rate (ECAR) as an indicator of aerobic glycolysis. After 3 days of stimulation, 80% of T cells expressed CD25 (figure 2A), representing an activated population. The OCR of CD25⁺ T cells was 3 to 4-fold higher than in the CD25⁻ T cell population, indicating increased OXPHOS in response to SEB stimulation (figure 2B). However, no difference in spare respiratory capacity (SRC) was noted between activated and resting T cells as indicated by similar % OCR in response to the uncoupler carbonyl cyanide-4-(trifluoromethoxy)phenylhydralazone (FCCP) compared to baseline (measured after oligomycin injection, figure 2C). The SRC is the extra capacity available in cells to produce energy in response to increased stress or work, and may reflect increased mitochondrial mass (16). To evaluate for differences in cellular mitochondrial content, we quantified mitochondrial mass with MitoTracker Green (MTG). Fluorescence intensity of MTG was 3 to 4-fold greater in CD25⁺ T cells compared to CD25⁻ T cells (figure 2D). These results indicate that superantigen stimulation with SEB rapidly expands T cells with greater mitochondrial content compared to resting cells, and that the increased mitochondrial mass of SEB-stimulated T cells enables increased basal OCR. In the setting of similar SRC between the populations, the increased mitochondrial content and baseline OCR of activated T cells indicates a greater degree of mitochondrial inner membrane (IM) uncoupling and IM instability at baseline. Consistent with the recent observations that the increased OXPHOS of proliferating cells will directly engage aerobic glycolysis, we observed a parallel increase in OXPHOS and ECAR after SEB stimulation (figures 2E and 2F). As observed in the Warburg phenomenon, aerobic glycolysis increased to a greater extent than OXPHOS, as evidence by a lower OXPHOS/ECAR ratio in CD25⁺ T cells (figure 2G). These results demonstrate that superantigen stimulation with SEB produces a bioenergetic profile shared with highly proliferative malignant cells and chronically activated T cells in the setting of disease, and that the relative uncoupled IM of cells with this bioenergetic configuration may be susceptible to targeted therapy (15, 24-26).

Figure 2. Bioenergetic profile of superantigen stimulated T cells resembles the Warburg Effect.

PBMCs were stimulated with SEB for 72 hours. (A) Activated (CD3⁺CD25⁺) and resting (CD3⁺CD25⁻) T cells were then identified by FACS analysis, as shown for one representative experiment. Activated and resting T cells were then isolated, and bioenergetics were measured by real-time extracellular flux analysis. The (B) oxygen consumption rate (OCR) and (C) spare respiratory capacity (SRC) were measured in a basal state, and after the addition of oligomycin (to block ATP synthesis), FCCP (to uncouple ATP synthesis from the electron transport chain), and rotenone (to block complex I of the electron transport chain), as shown for one representative experiment. (D) The mean fluorescent intensity (MFI) of MitoTracker Green, (E) the basal OCR, (F) the basal extracellular acidification rate (ECAR), and (G) the OCR/ECAR ratio are shown for both activated and resting cells. Mean ± SE are plotted from 5 independent experiments. * p < 0.05, ** p < 0.01.

Modulation of P-gp improves the selectivity of photosensitizers for activated T cells

We have previously demonstrated that lymphocytes with low P-gp activity may be disproportionately depleted by phototherapy (11). P-gp (also known as MDR1 or ABCB1) is a member of the ATP-binding cassette (ABC) superfamily and was the first efflux protein identified. Recently, simple substitutions in a series of chalcogenorosamine/rhodamine structures have been shown to create molecules that possess a high affinity for P-gp and are either highly stimulating or inhibiting for ATPase activity (27, 28). Based on this observation, we developed a small library of 24 photosensitive agents that modulate P-gp. We hypothesized that P-gp stimulation would decrease photosensitizer accumulation in resting cells, and improve selectivity for stimulated T cells. To measure differences in cell extrusion kinetics between photosensitizers, the T cell line HUT-78 was used for uniform P-gp expression (29). Of the photosensitizers analyzed in this report, the weak P-gp stimulating photosensitizers (thioamide scaffolds 1, 3, 5, 7, and the collected julolidine rhodamines 5, 6, 7, 8 in table SI and figure S1) were associated with prolonged intracellular retention compared to their paired P-gp stimulating analogues (amide scaffolds 2, 4, 6, 8, and the collected half – julolidyl rhodamines 1, 2, 3, and 4). These results demonstrate that small structural changes in our photosensitizers dictate P-gp ATPase activity to determine cell extrusion kinetics (table S1).

We next evaluated whether SEB-stimulated T cells will preferentially retain photosensitizers with the potential to inhibit OXPHOS. The ATP production of OXPHOS is driven by the electrochemical gradient across the mitochondrial IM. Early after T cell activation, the mitochondrial IM becomes hyperpolarized and negatively charged (30). In this activated state mitochondria attract positively charged molecules (31). With an appropriately designed cationic photosensitizer, cells that possess greater mitochondrial mass and activity, such as effector and alloreactive cells, experience increased intracellular accumulation. Upon exposure to light, a well-designed photosensitive agent will then produce reactive oxygen species that uncouple the mitochondrial IM and selectively impede OXPHOS. Based on this reasoning, we also designed the 24 photosensitizers investigated in this report with the potential for intra mitochondrial accumulation. To evaluate for selective uptake in activated T cells, SEB stimulated cells were washed and suspended at a concentration of 2 × 10⁶ cells/mL in 2.5 × 10⁻⁷ M of photosensitizer for 20 minutes, followed by suspension in photosensitizer-free media for 30 minutes. Fluorescence intensity of the photosensitizers was 5 to 7-fold higher in CD25⁺ T cells compared to CD25- T cells (Figure 3A), and was in proportion to the extent of P-gp modulation. Strong P-gp stimulators were associated with a significantly higher retention differential between stimulated and resting T cells (figure 3B). These results demonstrate that the photosensitizers investigated in this report are preferentially retained in stimulated T cells, and that P-gp stimulation further increases this retention differential, and therefore may improve the selectivity of phototherapy.

Figure 3. Photosensitizers are preferentially retained in stimulated T cells.

Stimulated T cells by were identified by FACS analysis for CD3⁺ and CD25⁺ coexpression. Resting T cells were identified as CD3⁺ without expression of CD25. A) A representative histogram of photosensitizer fluorescence is shown. B) Bar graphs represent the ratio of the mean fluorescent intensity (MFI) of CD25⁺/CD25⁻ T cells for the strong and weak P-gp stimulating photosensitizers. Mean ± SE are plotted from 3 independent experiments. ** p < 0.01.

Photosensitizers with rapid extrusion kinetics minimally impede the bioenergetics of resting T cells

None of the photosensitive agents in use today have demonstrated selectivity for activated cells without significant toxicity occurring in resting cells. We hypothesized that if longer intracellular resident times were associated with greater toxicity, then the photosensitizers associated with rapid extrusion kinetics would have lower potential for toxicity. To test our hypothesis, we next compared the influence of the 24 photosensitizers on the bioenergetics of resting T cells. For these experiments, ex vivo selected T cells were exposed to 5 × 10⁻⁷ M of photosensitizer for 20 minutes in the absence of light, followed by the real-time measurement of OCR and ECAR. The weak P-gp stimulators, which are associated with slower cell extrusion kinetics, affected T cell bioenergetics to a greater extent than the strong P-gp stimulating photosensitizers (figures 4A and 4B). These results demonstrate photosensitizers that are rapidly extruded from resting cells are associated with a low potential for bioenergetic impedance and dark toxicity. To better understand the toxicity profile of our photosensitizers, we then evaluated the effects of increasing concentrations. For this purpose, we used the photosensitizer 2-Se-Cl and found no change in the bioenergetics of resting T cells at 5 × 10⁻⁷ M concentration. However, an initial increase in OCR at 6 × 10⁻⁷ M was followed by a linear decline, where OXPHOS was inhibited in proportion to 2-Se-Cl and accompanied by an increase in aerobic glycolysis (figures 4C and 4D). The decrease in OXPHOS with an increase in aerobic glycolysis was evidenced by a decrease in the OCR/ECAR ratio (figure 4E). These results demonstrate that OXPHOS inhibition is associated with a compensatory increase in aerobic glycolysis in resting T cells.

Figure 4. Stimulation of P-gp protects cells from toxicity.

For photosensitizer alone experiments, isolated T cells were exposed to either 5 × 10⁻⁷ M of photosensitizer, or PBS (control) for 20 minutes. For photodepletion experiments, isolated T cells were exposed to 5 × 10⁻⁸ M of photosensitizer or PBS (control) for 20 minutes, washed, and then exposed to 5J/cm² light. (A) Bar graphs represent mean oxygen consumption rate (OCR) of 24 photosensitizers as a percent of control in resting T cells for the strong and weak P-gp stimulators. The (B) OCR, (C) extracellular consumption rate (ECAR), and (D) the OCR/ECAR ratio are shown for increasing concentration of 2-Se-Cl for one representative experiment. (E) The effects of PD on the bioenergetics of resting T cells using 2-Se-Cl and 2-S-Cl is compared to control in a basal state, and after the addition of oligomycin (to block ATP synthesis), FCCP (to uncouple ATP synthesis from the electron transport chain), and rotenone (to block complex I of the electron transport chain) for one representative experiment. (F) Bar graphs demonstrate the effects of PD on the OCR and survival of cells compared to control for photosensitizers 2-S-Cl, 2-Se-Cl, 4-S-Cl, and 4-Se-Cl. Three donors were used in 4 independent experiments. Mean ± SE are plotted. * p < 0.05 compared to control.

Based on the rapid extrusion kinetics and low potential for toxicity in resting cells, we selected four strong P-gp stimulating photosensitizers for further analysis (2-S-Cl, 2-Se-Cl, 4-S-Cl, and 4-Se-Cl), and evaluated the effects of PD on the bioenergetics of resting T cells. For all PD experiments, immunomagnetically-selected CD3⁺ cells were suspended in a photosensitizer-rich media of 5.0 × 10⁻⁸ M for 20 minutes followed by 30 minutes in a photosensitizer-free media. Cells were then exposed to 5 J/cm² of light followed by real-time measurement of OCR and ECAR. The SRC was decreased in resting cells after PD with all 4 photosensitizers (figure 4F). However, of the four photosensitizers, only the two selenorhodamine analogues (2-Se-Cl and 4-Se-Cl) did not significantly impede the basal OCR (figures 4F and 4G). We next evaluated the effects of PD on cell survival. For these experiments, FACS analysis was performed 18 hours after PD. Cell survival was identified by failure to bind Annexin 5 and 7AAD, and percent survival was calculated as the difference in the absolute number of cells between PD and control (non-PD samples) samples. Significant cell death occurred with use of 2-S-Cl and 4-S-Cl. In contrast, minimal cell death was observed when either 2-Se-Cl or 4-Se-Cl were used for PD, without selectivity for CD4⁺, CD8⁺, Naïve, or Memory cells subsets in resting T lymphocytes (figure 4G). These results demonstrate that photosensitizers that stimulate P-gp ATPase are rapidly extruded from cells, and protect resting cells from both dark and phototoxicity. When using these photosensitizers, only suppression of basal OCR and not SRC affected survival. These findings indicate that basal OXPHOS is sufficient to support cell homeostasis and survival with a reduced SRC. After PD, the suppression of SRC reflects mitochondrial respiratory inhibition due to uncoupling and loss of the electron gradient across the mitochondrial IM. However, in the case of PD with 2-Se-Cl and 4-Se-Cl, the degree of mitochondrial IM uncoupling was insufficient to induce apoptosis.

PD with 2-Se-Cl selectively affects the bioenergetics and survival of activated T cells

We next evaluated the selectivity of PD for activated T cells. Although both the photosensitizers 2-Se-Cl and 4-Se-Cl demonstrated a high degree of selective accumulation in activated T cells and a low potential for toxicity in resting T cells, we focused on 2-Se-Cl. To confirm intracellular localization of 2-Se-Cl, FACS analysis using ImageStream Technology (Amnis) was performed (figure 5A). Colocalization of MitoTracker Green (MTG) with 2-Se-Cl in the mitochondria of HUT-78 T cells (figure 5B, similarity score = 2.56 ± 0.30) was clearly evident. A statistical analysis of the similarity of localization of MTG and MitoTracker Red (MTR) gave a mean bright detail similarity score of 3.06 ± 0.38 for 2420 cells, indicating a high degree of co-localization of these two agents. In contrast, a low similarity score between LysoTracker Green (LYS) and MTR of 0.55 ± 0.21 for 2590 cells was observed, demonstrating differences in the accumulation of the reporter fluorescent dyes in mitochondria and lysosomes. The high similarity score of 2-Se-Cl and MTG (figure 5C) demonstrates the specific localization of 2-Se-Cl in mitochondria. These results confirm the mitochondrial specificity of 2-Se-Cl, and the absence of any significant localization of this photosensitizer in the lysosomes of T cells.

Figure 5. Photosensitizer 2-Se-Cl localizes in mitochondria.

Malignant T cells (HUT-78) were stained with MitoTracker Green (MTG), MitoTracker Red (MTR), LysoTracker Green (LYS) and 2-Se-Cl for comparison by FACS analysis. A) A bright field image (BF) demonstrating MTG fluorescence, 2-Se-Cl fluorescence (Dye), and a merged image of MTG/2-Se-Cl fluorescence (MTG/Dye) is shown in one representative sample. B) Histogram demonstrates the pixel-by-pixel statistical analysis of each cell analyzed (n = 2205), in which the y-axis is the number of cells and x-axis is the similarity coefficient between MTG and 2-Se-Cl. C) Bar graphs represent the average similarity coefficient comparing the similarity score for colocalization of 2-Se-Cl and MTG with the score for colocalization of LYS and MTG, and with the score for colocalization of MTR and MTG. Mean ± SE are plotted.

Next we evaluated the differential effects of PD on bioenergetics of activated and resting T cells. For these experiments, immunomagnetically-selected CD25⁺ and CD25^? T cells were isolated (> 95% purity) after SEB stimulation. PD was then performed, and bioenergetics were measured within 1 hour. We hypothesized that the greater intra mitochondrial accumulation of 2-Se-Cl in activated T cells would selectively uncouple the mitochondrial IM to impede basal ATP production required for cell survival. The percent of basal OCR devoted to ATP production was determined by comparing basal OCR to baseline OCR (after oligomycin injection). PD with 2-Se-Cl significantly impeded OXPHOS associated ATP production in activated T cells, but not of resting T cells from the same culture (figure 6A), and while not affecting aerobic glycolysis of either population (figure 6B). These results indicate that the increased mitochondrial metabolism drives the potential for greater photosensitizer accumulation. Upon exposure to light, the higher concentration of 2-Se-Cl selectively disrupted OXPHOS in activated T cells.

Figure 6. Photodepletion with 2-Se-Cl selectively affects the bioenergetics and survival of activated T cells.

PBMCs were stimulated with 50ng/mL staphylococcal enterotoxin B (SEB) for 72 hours and then photodepleted (PD) with 5 × 10⁻⁸ M of 2-Se-Cl and 5 J/cm² light. (A) The bar graphs represent the average area under the curve (AUC) summations for basal OCR/baseline OCR and (B) the ECAR measurements for resting and activated T cells of PD and non-PD (control) samples. Nine donors were used in 3 independent experiments. (C) Cell survival was measured 18 hours after light exposure and enumerated by FACS analysis by exclusion of Annexin V and 7AAD. D) Percent survival compared to control was determined in 3 independent experiments. Mean ± SE are plotted. ** p < 0.01.

To determine whether the selective impedance of basal ATP production affected cell survival, we next performed FACS analysis 18 hours after PD (figure 6C). Cell survival was identified by failure to bind Annexin 5 and 7AAD, and percent survival was calculated as the difference in the absolute number of cells between PD and control (non-PD samples) samples. Greater than 90% of activated T cells were eliminated from culture with minimal to no cell death occurring in the resting T cell population (figure 6D). Both the CD4⁺ and CD8⁺ cell compartments were depleted equally. These results demonstrate that PD with 2-Se-Cl selectively disrupts OXPHOS in activated T cells to induce cell death, while resting T cells remain intact.

PD with 2-Se-Cl selectively depletes immune responses

To directly test the hypothesis that PD with 2-Se-Cl will selectively eliminate an immune response, PBMCs were stimulated with 50ng/mL staphylococcal enterotoxin B (SEB) for 72 hours, and then photodepleted using 2-Se-Cl as described above. Cells were then rested overnight, stained with CFSE, and rechallenged with SEB or toxic shock syndrome toxin 1 (TSST-1) in culture for 6 days. After PD, no proliferation occurred in response to SEB (figure 7A right upper panel). In contrast, when challenged with TSST-1, a superantigen that stimulates a different range of the T cell receptor (TCR) repertoire compared to SEB, a robust response was observed (figure 7A right lower panel). Both SEB and TSST-1 bind to specific TCR sequences, which represent about 20% of the TCR repertoire. The loss of SEB-specific T cells enriched the TSST-1-specific T cells in the remaining PBMCs, and accounts for the increase percentage of dividing cells and the higher division index (the average # of cell divisions for all cells) in response to TSST-1 (figures 7B and C). Although the average proliferation index (the average # of divisions for proliferating cells) for T cells responding to TSST-1 after PD was lower than that of the control (3.9 vs 2.6), a robust response to TSST-1 was maintained, demonstrating that the resting T cells remain intact and functional after PD (figure 7C).

Figure 7. Photodepletion with 2-Se-Cl selectively depletes immune responses.

(A) PBMCs were stimulated with 50ng/mL staphylococcal enterotoxin B (SEB) for 72 hours, and then photodepleted (PD) with 5 × 10⁻⁸ M of 2-Se-Cl and 5 J/cm² light. Cells were then rested overnight, stained with CFSE, and rechallenged with SEB or toxic shock syndrome toxin 1 (TSST-1) in culture for 6 days. Histograms of CFSE fluorescence for stimulated (dashed lines) and non-stimulated (solid lines) T cells are shown for one representative sample. (B) Bar graph represents the percent of the total cells proliferating in response to SEB or TSST-1 for PD and control (non-PD) samples. C) Bar graph represents the division index (average # of cell divisions for all cells) and proliferation index (the average # of divisions for proliferating cells) for TSST-1 stimulated cells. (D) Responder splenocytes from C57BL/6 mice (H-2^b) were cultured with irradiated (2000 cGy) stimulator Balb/c (H-2^d) splenocytes for 5 days, and then photodepleted. Cells were then rested overnight, stained with CFSE, and then rechallenged with same-party Balb/c or 3rd -party C3H.HeJ (H-2^k) splenocytes for 5 days. Histograms of CFSE fluorescence for stimulated (dashed lines) and non-stimulated (solid lines) T cells are shown for one representative sample. (E) Bar graph represents the percent of the total cells proliferating in response to same-party Balb/c or 3^rd-party C3H.HeJ for PD and control samples. F) Bar graph represents the division index and proliferation index for 3^rd-party C3H.HeJ stimulated cells. Three donors were used in 3 independent experiments. Mean ± SE are plotted. ** p < 0.01.

Next, selective depletion of an allogeneic immune response was evaluated. For these experiments, a MHC-mismatched murine model was employed. Responder splenocytes from C57BL/6 mice (H-2^b) were cultured with irradiated (2000 cGy) stimulator Balb/c (H-2^d) splenocytes for 5 days, and then photodepleted. Cells were then rested overnight, stained with CFSE, and then rechallenged with same-party Balb/c or 3^rd -party C3H.HeJ (H-2^k) splenocytes for 5 days. Similar to the results of our superantigen model using human cells, no proliferation occurred after PD in response to same-party cells (figure 7D right upper panel, and 7E). In contrast, after PD a robust proliferative response was observed against 3^rd-party cells (figure 7A right lower panel, and 7F). These studies demonstrate that our novel photosensitizer 2-Se-Cl, designed to stimulate P-gp, will selectively accumulate in activated T cells to inhibit OXPHOS, and as a result, will selectively deplete an alloimmune response while leaving intact resting cells with a normal response potential.

PD with 2-Se-Cl selectively prevents GVHD

To determine whether selective depletion of antigen-specific, alloreactive T cells by PD with 2-Se-Cl will prevent GVHD in vivo, a complete MHC-mismatched murine model of rapidly lethal acute GVHD was used. To prepare the PD – treated primed splenocytes, donor C57BL/6 (H-2^b) splenocytes were cocultured for 5 days with irradiated (20 Gy) Balb/c (H-2^d) splenocytes, and then photodepleted. On the day of HSCT, C57BL/6 PD-treated splenocytes were infused into lethally irradiated BALB/c (same-party) or C3H/HeJ (H-2^k) (third-party) mice. Same party mice that received PD-treated splenocytes at the time of transplant lived 100 days without evidence of GVHD (figure 8C), and with less than 10% weight loss occurring throughout the 100 day post-transplant period (figure 8A), which was equivalent to the mean weight loss noted in the same-party control groups (TCD BM only). In contrast, all mice that received untreated primed splenocytes (figures 8C and 8D), and third-party mice that received PD-treated splenocytes (figure 8D) died of lethal GVHD. These results demonstrate that inhibition of OXPHOS by 2-Se-Cl selectively depletes pathogenic T cells and associated immune responses, and may be applied to effectively prevent GVHD after MHC-mismatched HSCT.

Figure 8. Ex-vivo photodepletion with 2-Se-Cl selectively depletes alloimune responses to prevents GVHD.

Donor C57BL/6 splenocytes were cocultured for 5 days with irradiated (20 Gy) Balb/c splenocytes, and then photodepleted. On the day of HSCT, 5 × 10⁶ PD – treated or untreated primed splenocytes were infused into irradiated same-party BALB/c (9 Gy) or third-party C3H/HeJ (9.5 Gy) recipients together with 10 × 10⁶ donor C57BL/6 T-cell–depleted bone marrow cells . Body weight (A) and survival (C) for first-party and the body weight (B) and survival (D) of third party mice are shown. Three to 5 mice were transplanted in each group in 3 independent experiments.

PD with 2-Se-Cl preserves antiviral immunity

Finally, to verify the retention of antipathogen immune responses after photodepletion, splenocytes were harvested from CD90.1⁺ P14 TCR transgenic mice and photodepleted. Eighteen hours after PD, 10⁵ naive GP33-specific PD or non-PD CD8⁺ T cells were then transferred i.v. into naïve C57BL/6 CD90.2⁺ hosts. Mice were then infected i.p. with 2 × 10⁵ pfu of the LCMV-Armstrong. After 8 days animals were sacrificed and FACS analysis performed on splenocytes for enumeration and function. Recipient mice of PD-treated cells had an equivalent expansion of GP33-specific CD8⁺ T cells in response to LCMV infection when compared to non-PD treated cell recipients (figure 9A), with an expected CD44^high activated/memory T cell phenotype. Cytokine production is critical for T cell function and control of virus. Consequently, to confirm T cell function, post-harvest splenocytes were treated with GP33-41 or NP396-404 peptides for 5 hours in vitro, and then stained for intracellular cytokines IFN-γ, TNF-α, MIP1-α, and IL-2. All PD and non-PD CD90.1⁺ transferred CD8⁺ T cells produced both IFN-γ, MIP1-α in response to GP33-41 peptide stimulation, and demonstrated an equivalent polyfunctional response with coproduction of IFN-γ, TNF-α, MIP1-α (figure 9B). Additionally, to determine the effect of PD treatment on viral clearance, virus titers of spleens were measured in recipients of PD and non-PD cells. In both groups low virus levels were measured 8 days postinfection (mean 2.53 vs. 2.58 × 10³ pfu/g, figure 9C), consistent with rapid clearance of the virus and full retention of antiviral immunity. These results confirm that antipathogen immune responses remain fully intact after PD with 2-Se-Cl, and may be efficiently transferred at time of transplant. Collectively, these results demonstrate that PD with 2-Se-Cl is highly selective for pathogenic T cells that cause GVHD after MHC-mismatched HSCT, and via P-gp stimulation, actively protects non-pathogenic T cells central to a successful immune reconstitution after transplant.

Figure 9. Antipathogen immunity is fully retained after ex-vivo photodepletion with 2-Se-Cl.

Splenocytes were isolated from naive P14 Thy1.1⁺ mice, photodepleted, and then rested overnight. Cells were then enumerated by staining with α-CD8α and D^bGP33–41, followed by i.v. transfer of 10⁵ naive GP33-specific CD8⁺ T cells into naïve C57BL/6 Thy1.2⁺ hosts. Mice were then infected i.p. with 2 × 10⁵ pfu of the LCMV-Armstrong. After 8 days animals were sacrificed, and FACS analysis performed on splenocytes for A) enumeration and B) function, and compared to control. Mean ± SE are plotted. C) Viral levels in spleens were determined by plague assays. Five mice were used in each group, in two independent experiments with similar results.

Discussion

We have designed and evaluated 24 photosensitive chalcogenorhodamines for the ability to selectively deplete activated T lymphocytes and associated immune responses. These photosensitizers specifically target T cell bioenergetics and accumulate in proportion to OXPHOS. In this study, we demonstrate that photosensitizers with increased intracellular residence times non-selectively impede OXPHOS and may be toxic to resting cells. Consequently, we selected a photosensitizer that potently stimulates P-gp to protect against toxicity occurring in resting cells. To our knowledge, this is the first time that a therapeutic has been designed with the dual properties of P-gp stimulation and OXPHOS inhibition. Applying this novel agent selectively depleted alloreactive T cells and prevented GVHD in a complete MHC-mismatched murine model of lethal GVHD, leaving intact resting cells with a robust response potential.

Having an OXPHOS-specific photosensitizer facilitates increased uptake in activated T cells relative to resting T cells, offering a unique target to enable selective depletion. In this report, we present a model describing the determinants of photosensitizer selectivity, and one that may be followed in the design of future therapeutics selective for OXPHOS inhibition. In this model, the intracellular concentrations of a selective therapeutic are at an equilibrium involving influx and efflux, and the cytoplasmic concentration is determined by a balance between the opposing forces of OXPHOS and P-gp activity. All of the photosensitizers investigated in this report are highly dependent on the mitochondrial IM potential for intracellular accumulation (figures 3 and 5). Consequently, we selected 2-Se-Cl due to a potent ability to stimulate P-gp to facilitate low intracellular concentrations and potential for toxicity in resting cells. With 2-Se-Cl, increased OXPHOS of activated T cells slows cell extrusion and is the main determinant of intracellular accumulation.

Employing Staphylococcal superantigens (SAgs) to illustrate pathogenic T cells provided a mechanistic model to efficiently evaluate and identify the determinants of selectivity of our novel photosensitizers. In this study, we employed SAgs to stimulate T cells with a restricted TCR repertoire. SAgs induce a robust T cell response by cross-linking the TCR with a MHC class II product on an APC (32). In this model, responding T cells are highly proliferative with uniform high CD25 expression, representing a terminal effector population (33). Using extracellular flux analysis we show that the effector cells undergoing clonal expansion in response to SEB stimulation increase both aerobic glycolysis and OXPHOS to support the rapid cell growth. This bioenergetic phenotype is similar to the Warburg Effect in cancer cells, and has recently been identified in pathogenic T cells in both autoimmune and alloimmune diseases (15). Alloreactive T cells in vivo minimally increase glycolysis above resting cells due to a limitation of substrate compared to in vitro models (34). However, the recent discovery that the increased OXPHOS enables the increased aerobic glycolysis suggests that a novel therapeutic with the ability to inhibit OXPHOS will impede both metabolic pathways and starve cells of ATP to induce apoptosis in both in vitro and in vivo systems (16). Furthermore, the specificity of our photodepletion technique selectively impedes OXPHOS in the effector cells to inhibit the ATP production associated with both SRC and basal OCR. The SRC of mitochondria is the extra capacity available in cells to produce energy in response to increased stimulation (35) and may be reduced after PD (figure 4E). However, only a decrease in basal OCR was associated with cell death, consistent with basal OCR reflecting the ATP production required for cell survival in our model.

Evaluating our small library of photosensitives to identify the determinants of selectivity enabled the rapid translation of findings into an in vivo system. For this purpose, a well-established, complete MHC-mismatched murine model of GVHD was employed (19). This model was selected for study due to the very high rate of lethal GVHD. In this model, same-party recipients that received PD-treated primed splenocytes survived > 100 days after HSCT without evidence of GVHD. In contrast, third-party mice that received an equivalent PD-treated product developed lethal GVHD and died within 35 days of HSCT. Similar to our in vitro findings, the high selectivity of 2-Se-Cl for pathogenic T cells was evidenced in vivo by the robust response potential retained in resting and non-pathogenic T cells after PD, which resulted in the rapid onset of lethal GVHD in the third-party system. Similarly, antiviral immunity was maintained after PD using a transgenic murine model specific for LCMV, underscoring the selectivity of our approach for stimulated and pathogenic T cells.

The goal of our work is to improve selective depletion and to prevent GVHD after HSCT. Despite prophylaxis with calcineurin inhibitors, acute GVHD occurs in up to 60% of patients receiving transplants from HLA-identical siblings, and up to 80% of patients receiving transplants from HLA-matched unrelated donors (36). High-dose corticosteroids are the first-line treatment for acute GVHD, but are associated with limited efficacy and significant toxicity. Only 50% of patients with severe acute GVHD respond to treatment, and only 10% of patients with steroid refractory GVHD survive long term (37). For many patients, this risk of GVHD prohibits the curative option of HSCT. This is further compounded by the fact that the risk of GVHD increases with HLA-disparity between the donor and patient, and many patients are unable to find a fully HLA-matched donor. Only 30% of patients have an HLA-identical sibling, requiring the majority of patients to consider higher risk alternative donor options (38). For many, especially for patients of ethnic minority backgrounds, a fully HLA-matched unrelated donor cannot be found; merely 6% of African Americans are expected to find an HLA-matched adult donor by high-resolution molecular typing (39, 40).

Haploidentical HSCT is an alternative donor option for patients without a fully HLA-matched donor, but is associated with a high risk of GVHD due to significant HLA disparity. Almost all patients have an available HLA-haploidentical familial donor, which may include a parent, sibling, or child. Early attempts at performing T cell–replete haploidentical transplantations using conventional preparative regimens have been associated with very high rates of GVHD and graft rejection (41). However, we propose that selective depletion with 2-Se-Cl may successfully prevent GVHD, and make haploidentical transplantations a safer curative option for more patients.

In conclusion we present a new class of photosensitizers with a high potential for selectivity, and present a model that may be followed to develop new therapeutics to target cell bioenergetics. We have identified 2-Se-Cl as a superior agent in this class with the ability to modulate P-gp to selectively deplete stimulated T lymphocytes and associated immune responses. The mitochondrial specificity of 2-Se-Cl impedes OXPHOS of stimulated cells to induce apoptosis. Future applications for our approach are broad and may include targeting OXPHOS in alloimmune, autoimmune, and malignant T cells to provide alternative treatment options and improve clinical outcomes for patients with diseases such as GVHD, systemic lupus erythematosus, or peripheral T cell lymphoma.