|Home | About | Journals | Submit | Contact Us | Français|
Heme oxygenase-1 (HO-1) and its catabolic byproducts have potent anti-inflammatory activity in many models of disease. It is not known, however, if HO-1 also plays a role in the homeostatic control of T cell activation and proliferation. We demonstrate here that the HO-1 inhibitor, tin mesoporphyrin (SnMP), induces activation, proliferation, and maturation of naïve CD4+ and CD8+ T cells via interactions with CD14+ monocytes in vitro. This response is dependent upon interactions of T cells with MHC Class I and II on the surface of CD14+ monocytes. Furthermore, CD4+CD25+FoxP3+ regulatory T cells (Tregs) were able to suppress this proliferation, even though their suppressive activity was itself impaired by SnMP. Given the magnitude of the antigen-independent T cell response induced by SnMP, we speculate that HO-1 plays an important role in dampening non-specific T cell activation. Based on these findings, we propose a potential role for HO-1 in the control of naïve T cell homeostatic proliferation.
Heme oxygenase-1 (HO-1) catalyzes the breakdown of heme from intracellular hemoproteins and erythrocyte-derived hemoglobin into biliverdin, ferrous iron (Fe+2), and carbon monoxide (CO) (1, 2). In addition to its crucial role as a catabolic enzyme, HO-1 is also a potent stress-response protein, induced by free-radical and oxidative stress caused by heavy metal exposure, bacterial lipopolysaccharide (LPS), inflammatory cytokines, hypoxia, and hyperoxia (2–8). Under varying circumstances, it has anti-proliferative, anti-oxidant, and anti-apoptotic effects (2, 9–11). The ability of HO-1 to be induced by such a great number and variety of stimuli underscores its importance as a cytoprotective and homeostatic factor.
Recent studies have focused on the role that HO-1 serves as an immune mediator. In keeping with its anti-oxidant and anti-proliferative effects in certain cell types, HO-1 appears to be predominantly suppressive of immune responses and to have anti-inflammatory effects. For example, in the NOD mouse model of type-1 diabetes, induction of HO-1 with cobalt protoporphyrin (CoPP) reversed established autoimmune pancreatitis and ongoing islet destruction, restored insulin production and normoglycemia, and was associated with fewer infiltrating CD11c+ dendritic cells and cytotoxic CD8+ T cells (12). These effects were specifically associated with HO-1 activity and were blocked by the HO-1 inhibitor, tin mesoporphyrin (SnMP). Such immunosuppressive effects have been observed in multiple other animal models, including those demonstrating the importance of HO-1 in allograft tolerance (13). In addition, there is evidence that HO-1 may play a role in the maintenance of immune homeostasis. For example, mice deficient in HO-1 develop an inflammatory state characterized by splenomegaly, lymphadenopathy, abnormal CD4/CD8 cell ratios, and T cell hyper-responsiveness to ex vivo stimuli (14–16). Likewise, HO-1 deficiency in humans is associated with leukocytosis, abnormalities in secondary lymphoid organs (e.g., asplenia and lymphadenopathy), and evidence of severe persistent endothelial damage (17, 18). There is also a large body of work now demonstrating that relatively common promoter polymorphisms in the HO-1 gene, thought to influence the magnitude and rate of induction, can have a profound effect on a wide variety of inflammatory disorders [reviewed in ref. (19)].
HO-1 is upregulated upon T cell activation, and both HO-1 and CO can inhibit T cell proliferation, suggesting that the induction of HO-1 may play an important role in the regulation of T cell activation and homeostasis (11, 20). Previous studies have shown that proliferation of CD3+ T cells stimulated though the T cell receptor (TCR) can be inhibited by exposure to low concentrations of CO and that this effect acts through p21cip-dependent activation of caspase 8 (11). Other investigators have shown that the anti-proliferative effects of CO on CD4+ T cells depend upon inhibition of the extracellular related kinase (ERK) pathway, leading to decreased production of interleukin-2 (IL-2) (20). Though these studies demonstrate the ability of HO-1 and its products to prevent cellular activation, it remains unclear whether HO-1 exerts an anti-proliferative effect at baseline and/or whether relief of such inhibition leads to T cell activation.
SnMP is a potent inhibitor of HO-1 mediated heme catabolism that has now been provided to many patients for the treatment of both neonatal jaundice and inherited hyperbilirubinemia syndromes (21). It was developed to possess unique structural and photophysical properties that make it a particularly potent and bioavailable in vivo inhibitor suitable for clinical use in newborns (22, 23), and studies to date have revealed a very favorable therapeutic profile with no significant adverse side effects. Given the potential immunomodulatory effects of HO-1 in health and disease, we tested the possibility that pharmacologic inhibition of HO-1 by SnMP would also lead to the activation of human T cells. Specifically, we hypothesized that inhibition of HO-1 in peripheral blood mononuclear cells (PBMCs) in vitro by SnMP would result in T cell activation and proliferation.
Human PBMCs were isolated from healthy adult donors by density gradient centrifugation of whole blood on cell separation medium (Histopaque-1077; Sigma-Aldrich, St. Louis, MO). PBMCs were collected, washed in phosphate buffered saline (PBS; Life Technologies, Rockville, MD), counted, and re-suspended in RPMI-10 [RPMI 1640 media (Gibco) with 10% heat inactivated fetal bovine serum (FBS; Hyclone Laboratories, Rockford, IL), 2 mM L-glutamine (Mediatech, Washington, DC), 100 U/mL penicillin/streptomycin (Mediatech)]. Only freshly-isolated cells were used for primary culture experiments. All samples were obtained in accordance with guidelines and under protocols approved by the Committee on Human Research (CHR) at the University of California, San Francisco. Cobalt protoporphyrin (CoPP) and SnMP were purchased in powdered form from Frontier Scientific (Park City, Utah), dissolved in 0.1 mM NaOH, and titrated to a pH of 7.6. The following fluorophore-conjugated monoclonal antibodies (mAbs) were used for detection of cell surface markers: CD3 (SP34-2, Alexa700 or Pacific Blue), CD4 (RPA-T4, Alexa-700 or Pacific Blue), CD11c (B-ly6, allophycocyanin [APC] or V450), CD16 (B73.1, Pecy7), CD19 (H1B19, A700), CD20 (2H7, PE), CD38 (HB7, APC), CD25 (M-A251, PE-Cy7 or APC-Cy7), CD56 (B159, A700), CD69 (HB50, PE-Cy7), CD86 (FUN-1, APC), CD127 (hIL-7R-M21, PE), HLA-DR (L243, APC-Cy7) (all from BD Biosciences, San Jose, CA), CD8 (3b5, PE-Cy5.5, Caltag Laboratories, Burlingame, CA), CD14 (RMO52, ECD, Beckman Coulter, Fullerton, CA), CD45RA (2H4, ECD, Beckman Coulter), CD27 (O323, APC-Alexa750, eBioscience, San Diego, CA), CD163 (6H1, PE, eBiosciences), and BDCA-2 (AC144, FITC, Miltenyi Biotec, Auburn, CA). The following mAbs were used alone or in combination for detection of intracellular antigens: FoxP3 (PCH101, APC, eBioscience), Ki-67 (B56, FITC, BD Biosciences), and HO-1 (rabbit polyclonal (ab13243, unconjugated, Abcam, Cambridge, MA). HO-1 primary antibody was detected using F(ab′)2 anti-rabbit IgG conjugate (Q-11401MP, Qdot605, Invitrogen, Carlsbad, CA).
PBMCs were cultured on Upcell™ 96F MicroWell plates (Nunc, Rochester, NY) under various treatment conditions for indicated periods of time, and adherent cells were detached from the plates by incubating the plates at 25°C for 20 minutes. For flow cytometry analysis, cells were washed in staining buffer [PBS with 2% FBS and 2 mM EDTA (Sigma-Aldrich, St. Louis, MO)], incubated at 4°C in the presence of directly-conjugated fluorescent mAbs for 30 minutes, washed in staining buffer, and then fixed in 2% paraformaldehyde (PFA). All cells were stained with a live/dead marker (Amine-Aqua/Am-Cyan or Amine-Violet/Pacific Blue Live/Dead; Invitrogen) so that dead cells could be excluded from analysis. FoxP3 staining was carried out according to the manufacturer’s protocol, with slight modifications (eBioscience, San Diego, CA). Briefly, cells were washed after incubation with primary antibodies, re-suspended in FoxP3 fixation/permeabilization buffer (eBioscience), and then incubated for 1 hour at 4°C, washed twice in FoxP3 permeabilization buffer (eBioscience), and stained with anti-FoxP3 mAb in FoxP3 permeabilization buffer for 1 hour at 4°C. Cells were then washed twice in FoxP3 permeabilization buffer and re-suspended in 2% PFA. Data were acquired with an LSR-II flow cytometer (BD Biosciences, San Jose, CA) and analyzed with FlowJo software (Treestar, Ashland, OR). HO-1 and Ki-67 staining were carried out using the BD Cytofix/Cytoperm kit according to the manufacturer’s protocols (BD Biosciences).
For in vitro assays involving cell depletion or selection, cells were washed and re-suspended in MACS® buffer (PBS with 0. 5% BSA and 2 mM EDTA). For depletion of cells expressing CD25, CD14, CD4, CD8, or CD45RA, PBMCs were incubated with the appropriate beads (e.g., CD25 Microbeads II, CD14 Microbeads, CD4 Microbeads, CD8 Microbeads, or CD45RA Microbeads, respectively) for 30–45 minutes at 4°C. Labeled cells were washed with staining buffer and run through magnetic columns (MS or LS Columns, Miltenyi Biotec). The unbound fraction was kept as the “depleted fraction” and cells that were retained in the column were isolated as the “positive fraction.” In the case of CD25 beads, this fraction was found to be consistently 50–70% FoxP3+ and was used for Treg add-back assays. For mock depletions, cells were processed in parallel, incubated with anti-biotin magnetic beads (Miltenyi), and isolated using the same procedure. For isolation of purified CD3+ T cells and CD14+ cells, PBMCs were incubated with cocktails of biotin-conjugated antibodies (Miltenyi) (designed to bind to all PBMCs except those positive for CD3 and/or CD14) for 30 minutes at 4°C, washed in MACS® buffer, and then incubated with anti-biotin magnetic beads (Miltenyi), according to the manufacturer’s protocol. Labeled cells were washed once more in staining buffer and passed over magnetic columns (MS or LS columns, Miltenyi). The unbound fraction (i.e., cells positive for CD3 and/or CD14) was retained for further use. All cells enriched in this manner by magnetic separation were monitored for purity by flow cytometry, comparing them against mock-depleted and unfractionated PBMCs using an appropriate phenotyping panel. Sorted cells were counted with a hemacytometer by trypan blue exclusion to determine the number of live cells and re-suspended in appropriate buffer.
PBMC were cultured with vehicle control, CoPP or SnMP (10 μM) for 7 days, and then harvested as described above. For some experiments, unmanipulated harvested cells were lysed and used for protein analysis, while in other experiments, CD3 or CD14 isolation/depletion was performed prior to cell lysis. Cells were washed in PBS, then lysed in radioimmuneprecipitation assay (RIPA) buffer containing PMSF (1mM), pepstatin A (1 μg/mL), aprotinin (2 μg/mL), and leupeptin (5 μg/mL) (Sigma-Aldrich, St. Louis, MO). Protein was quantified using BCA assay as per kit manufaturer’s instructions (Pierce, Rockford, IL). Cell protein lysates mixed with sample loading buffer (NuPAGE® LDS Sample Buffer, NuPAGE® Reducing Agent; Invitrogen) according to the manufacturer’s instructions, and heated at 70°C for 10 minutes. Samples (20 μg protein) were loaded onto gels (NuPAGE® Novex® 4–12% Bis-Tris; Invitrogen) and subjected to electrophoresis (200 V, 45 min, (XCell SureLock Mini Cell; Invitrogen) under reducing conditions. Proteins were then transferred to polyvinilidene fluoride (PVDF) membranes (Immobilon P; Millipore, Billerica, MA) and blocked for 1 hour in 5% nonfat milk in Tris buffered saline (TBS; 100mM Tris-Cl, pH 7.5; 0.9% NaCl) containing 0.1% Tween (TBST), followed by incubation for 1 hour at with anti-HO-1 (1:2000; SPA895; rabbit anti-human polyclonal; Assay Designs) and anti-GAPDH (1:2000; mAbcam 9484; mouse anti-human monoclonal; AbCam) primary antibodies for 1 hour. After 3 washes with TBST, the blots were incubated for 45 mins with anti-rabbit and anti-mouse secondary antibodies (Dako, Carpinteria, CA) conjugated with polymeric horseradish peroxidase (HRP). Finally, the bands were detected using a luminescence detection system (Amersham™ ECL Plus; GE Healthcare, Piscataway, NY) and autoradiographic film (Amersham Hyperfilm™ ECL, GE Healthcare).
For proliferation assays, cells were first labeled with 5 μM carboxy-fluorescein diacetate succinimidyl ester (CFSE) (Sigma-Aldrich, St. Louis, MO) in PBS at 37°C for 10 minutes, followed by three washes in RPMI-10. Cells were typically plated in RPMI-10 at a density of 1–2 million cells/mL in 96 well U-bottom plates or 1–4 million per/mL in 6 well flat-bottom plates. Cells were exposed to metalloporphyrins in culture for 5–7 days at 37°C, 5% CO2 with no additional stimuli or growth factors. Upon harvesting, cells were washed in MACS® buffer, labeled for flow cytometry, and analyzed as previously described (24). The frequency of CFSElow cells was used as a measurement of total T cell proliferation. For CD14+ add-back assays, negatively-selected CFSE-labeled CD3+ T cells were incubated with a range of dilutions of negatively-selected CD14+ cells (to avoid stimulation of CD14+ cells by positive selection with CD14-beads). For CD25 and CD45RA-depletion experiments, PBMCs were labeled with CFSE and then subjected to either CD25, CD45RA, or mock-depletion as described above, using anti-CD25, anti-CD45RA, or anti-biotin microbeads (Miltenyi). Depleted and mock-depleted cells were cultured with or without metalloporphyrins for 7 days.
96-well flat-bottom culture plates were coated with anti-CD3 mAb (SP34-2, BD Biosciences) at a concentration of 5 μg/mL for 4 hours at 37°C. After washing coated plates thoroughly in PBS, 150,000 CFSE-labeled CD25-negative (CD25−) responder cells were incubated with a range of dilutions of enriched CD25+ cells. In different experiments, CD25+ cells were isolated either from PBMCs or from PBMCs that had been cultured with SnMP or CoPP (10 μM) for 7 days. Anti-CD3 stimulated cells were collected after 5 days in culture, washed in MACS® buffer, labeled for flow cytometry, and analyzed as described above. Control wells with no anti-CD3 mAb and no CD25+ Tregs were used for all stimuli.
For Transwell (Corning, Corning, NY) experiments, CFSE-labeled responder T cells were placed in the upper well of the chamber of a 1 μm pore size cell culture insert (Corning), and negatively-selected CD14+ cells were placed either in the upper or lower chamber of the well. Cells were incubated with or without SnMP (10 μM) in RPMI-10 at 37° for 7 days, harvested, labeled with florescent mAbs, and analyzed for proliferation by flow cytometry. For proliferation assays in the presence of MHC-blocking Abs, the following purified mAbs were used without azide and endotoxin: anti-HLA-A, -B, -C (W6/32) (obtained from Biolegend, San Diego, CA), anti-HLA-DR, DP, DQ (Tü39), and isotype controls (obtained from BD PharMingen, San Diego, CA). The mAbs were first added to CFSE-labeled PBMCs 30 min before SnMP (10 μM) and were then added again on day 3 of culture. Cells were harvested on day 7, stained for flow cytometry, and analyzed as described above.
PBMCs were isolated and CD3+ T cells were purified by negative immunomagnetic selection, as previously described (Miltenyi). CD3+ T cells were then labeled with CFSE and stained with a live/dead marker as well as with fluorescent mAbs against CD3, CD20, CD14, CD4, CD8, CD45RA and CD27. Live naïve (N: CD20−CD14−CD3+CD45RA+CD27+) and central memory (CM: CD20−CD14−CD3+CD45RA−CD27+) T cells were sorted by FACS on a BD FACSAria (BD Biosciences). Sorted N, CM, and non-sorted cells were then co-incubated with CD14+ cells isolated by negative immunomagnetic selection, using a dose range of SnMP. Cells were harvested at the end of 7 days, stained for flow cytometry, and analyzed as described above.
Previous studies have shown that HO-1 expression in T cells inhibits CD3-dependent activation and proliferation (20). To investigate whether inhibition of HO-1 might result in activation of T cells, we cultured PBMCs from healthy adult donors either in the presence of the HO-1 inhibitor, SnMP, or of the HO-1 inducer, cobalt protoporphyrin (CoPP). CoPP was chosen as a HO-1 inducer rather then heme because, unlike heme, it cannot be broken down enzymatically, allowing for a more constant concentration in culture. In the absence of any other activating stimulus, both CD4+ and CD8+ CD3+ T cells were found to proliferate after incubation with SnMP (Figure 1A), an effect that was reproducible in PBMC cultures from multiple donors (Figure 1B). The magnitude of proliferation was positively correlated with the concentration of SnMP, reaching maximal levels at approximately 50 μMol for both CD4+ and CD8+ cells (Figure 1C). There was no significant change in T cell viability at concentrations used in these experiments (0–50 μM; Supplemental Figure 1). Using the FlowJo proliferation software platform, it was calculated that up to 25% of the original population of T cells underwent proliferation at an SnMP concentration of 50 μMol. SnMP exposure was associated with increased T cell expression of the proliferation marker, Ki-67, and of the activation markers, CD38 and CD25 (Figures 1D and 1E, respectively). Of note, the HO-1 inducer CoPP did not induce T cell proliferation but conversely, reduced baseline levels of CD4+ cell proliferation and overall CD3+ cell Ki-67 expression seen in control samples (Figures 1A, B, and D).
In contrast to the proliferative response observed in PBMC cultures, isolated CD3+ T cells did not become activated or proliferate when exposed to SnMP (Figure 2A). Based on published evidence that HO-1 can alter the stimulatory activity of myeloid cells (25), we hypothesized that SnMP-induced T cell proliferation may involve interaction with cells of the myeloid lineage. We used CD14+ peripheral blood monocytes (prepared by negative selection) as a representative myeloid cell type, and confirmed by flow cytometry that these enriched cells expressed CD14, CD11c, and HLA-DR (MHC Class II), consistent with the phenotype of human peripheral blood monocytes (ref. 26; data not shown). When isolated T cells were co-cultured with such CD14+ monocytes, SnMP-induced T cell activation and proliferation was restored (Figure 2A, right panels). Similar to PBMC cultures, proliferation of T cells in these T cell:monocyte co-cultures was dependent on the concentration of SnMP (Figure 2B). To establish whether there is a relationship between the frequency of CD14+ cells in culture and the extent of T cell proliferation, add-back experiments were carried out with varying ratios of T cells and monocytes. These experiments showed that the percentage of CD14+ cells added back into culture correlates directly with the magnitude of SnMP-induced T cell proliferation (Figure 2C), reaching maximal levels at a T cell:monocyte ratio of approximately 10:1.
To explore what phenotypic changes were induced in monocytes by exposure to SnMP and CoPP, we performed flow cytometric analysis of multiple cell surface proteins known to be important markers of myeloid differentiation and activation, as well as intracytoplasmic staining for HO-1 (Supplemental Figures 2 and 3). In preliminary experiments, T cell proliferation was not observed until day 4 of PBMC culture (data not shown), leading us to postulate that phenotypic changes in monocytes associated with T cell activation might be present prior to day 4. We found that on day 3, HO-1 protein was upregulated in monocytes by CoPP and, to a lesser extent, by SnMP, which is an expected result based on previous experiments using metalloporphyrins in myeloid cells (25). HO-1 protein expression was not significantly altered in T cells either by SnMP or CoPP. SnMP resulted in decreased expression of CD11c and CD16, while expression of the C-type lectin BDCA-2 and the co-activating molecule CD86 (B7-2) were increased. Like SnMP, CoPP reduced CD16 expression. In contrast to SnMP, CoPP had no effect on CD11c expression, and decreased expression of CD86 and BDCA-2. CoPP also decreased expression of HLA-DR (MHC II). In control cells, there was a broad range of expression of the heme scavenger receptor, CD163. CoPP decreased CD163 expression, while SnMP increased CD163 expression, such that the difference in CD163 expression was significant between the two conditions.
To determine whether SnMP-induced T cell proliferation requires direct contact between CD3+ T cells and CD14+ monocytes, enriched preparations of these two populations were cultured either on the same or the opposite side of semi-permeable transwell membranes. As shown in Figure 3A, cell-to-cell contact was required for the SnMP effect on T cell proliferation. Since peripheral blood monocytes express MHC Class I and II antigens that might interact with the T cell receptor (TCR) on T cells, monocytes and T cells were cultured together with SnMP in the presence of monoclonal anti-MHC antibodies known to block such interactions. There was a trend toward reduced proliferation with either MHC Class I or II blockade, and a statistically significant ablation of the proliferative response occurred even at low antibody concentrations when both MHC Class I and II were blocked (Figure 3B). This observation suggests that the SnMP-induced T cell proliferative response involves engagement of both MHC I and II on CD14+ monocytes.
To identify the cell populations that proliferate in response to SnMP, PBMC cultures were stained for multiple cell-surface maturation markers after a week in culture, with or without SnMP. Flow cytometry plots (Figure 4A; left) demonstrate that SnMP induced a notable decrease in the percentage of naïve (TN: CD45RA+CD27+) CD4+ and CD8+ Tcells and a reciprocal increase in the percentage of central memory (TCM: CD45RA− CD27+) CD4+ and CD8+ T cells, many of which (delineated by circles in the flow plots) had high levels of CD27 expression (TCMCD27high). Meanwhile, the relative fractions of effector memory (TEM: CD45RA−CD27−) and CD45RA-positive effector memory T cells (TEMRA: CD45RA+CD27−) remained unchanged. The SnMP-induced decrease in TN and increase in TCM cells was dose dependent (Figure 4A; right). Phenotypic analysis was then carried out on CFSE-labeled cells exposed to SnMP to determine the maturational phenotype of proliferating (CFSElow) cells. No proliferating CD4+ or CD8+ T cells had a naïve phenotype, but were rather found to be predominantly of the TCM (CD27+ or CD27high) phenotype, with fewer proliferating TEM cells (Figure 4B).
The concomitant decrease in the frequency of naïve cells and increase in the frequency of proliferating cells with memory phenotypes suggested that the proliferating cells were derived from the CD45RA+ naïve T cell compartment. To test this possibility, PBMCs were either depleted of CD45RA+ T cells or mock depleted, prior to culture with SnMP. As shown in Figure 4C, minimal proliferation was detected in CD45RA-depleted cultures. To confirm that CD45RA+CD27+ TN cells represent the predominant cell type proliferating in response to SnMP, these cells and TCM cells were isolated by FACS and then cultured with CD14+ monocytes in the presence or absence of SnMP. Both CD4+ and CD8+ TN cells were found to proliferate and to up-regulate CD25 in response to SnMP, whereas TCM cells did not (Figure 4D; left), and the proliferation of TN cells was dose dependent (Figure 4D; right). In aggregate, these experiments show that SnMP induces activation and maturation of naïve CD45RA+CD27+ CD4+ and CD8+ T cells, leading to their maturation into memory cells.
We next evaluated the function of Tregs in SnMP-treated PBMC cultures. When Tregs were depleted from PBMCs prior to culture with SnMP, a statistically significant 2- to 3-fold increase in proliferation was observed (Figure 5A), suggesting that Tregs are able to suppress the SnMP-induced proliferative response. This was confirmed by Treg add-back assays, demonstrating that SnMP-induced proliferation could be suppressed when Tregs were present at higher frequencies prior to SnMP exposure (Figure 5B).
Given recent studies demonstrating that normal Treg development and suppressive function requires the activity of HO-1 in antigen presenting cells (APCs)(26), we next investigated whether SnMP had an inhibitory effect on Treg function. CD25+ cells were isolated from PBMCs that had been cultured for a week with vehicle control, CoPP, or SnMP. The frequency of purified CD4+CD25+ T cells expressing FoxP3 was similar in all treatment groups (52–59% FoxP3+; Figure 5C). The suppressive capacity of these purified metalloporphyrin-treated Tregs was then evaluated by adding them to cultures of CFSE-labeled, anti-CD3 stimulated autologous T cells. SnMP-treated Tregs were significantly less effective at suppressing responder T cell proliferation than Tregs isolated from cultures treated with CoPP or vehicle control (Figure 5D). Together, these findings suggest that Tregs have a suppressive effect on SnMP-induced T cell activation but that SnMP counteracts this effect by reducing the suppressive capacity of Tregs.
Numerous studies have demonstrated the importance of HO-1 and its enzymatic products as anti-inflammatory mediators in various disease states (12, 27–32). Compared to wild-type mice, HO-1 knockout mice develop a progressive inflammatory state and splenocytes from these mice respond to TCR activation with increased production of pro-inflammatory cytokines such as IL-2, IFN-γ, TNF-α, GM-CSF, and IL-6 (16). Such effects have been observed in human cells as well, where HO-1 and CO inhibit T cell proliferation in response to activation through the TCR in vitro (11, 20). Furthermore, HO-1 activity in APCs such as dendritic cells (DCs) and cells of the monocyte/macrophage lineage can significantly influence the outcome of their interactions with T cells (25,35). For example, splenocytes from HO-1 knockout mice display enhanced production of IL-6, TNF-α, IFN-γ, and IL-12 in response to LPS stimulation ex vivo (16). Chauveau et al. showed that induction of HO-1 expression in rat and human DCs led to impaired LPS-induced activation and maturation, and impaired ability to stimulate allogeneic T cell proliferation (25). Recent work by Tzima et al has also demonstrated a role for myeloid-expressed HO-1 in triggering innate immunity (35): mice carrying a myeloid-specific ablation of the HO-1 gene had impaired production of IFN-β in response to viral and bacterial infection, and a more severe disease course after induction of experimental autoimmune encephalomyelitis. Together, these results suggest that HO-1 in myeloid cells may play a complex and important role in T cell activation and differentiation. Given that HO-1 over-expression and CO exposure can inhibit T cell activation via the TCR (11, 20), and that HO-1 can inhibit T cell activation by APCs (25), we considered whether HO-1 might exert some function at baseline in maintaining T cell quiescence. Specifically, we asked whether exposure to the potent pharmacologic HO-1 inhibitor, SnMP, would result in T cell activation.
Here, we demonstrate that the HO-1 inhibitor, SnMP, induces activation, proliferation, and maturation of naïve CD4+ and CD8+ T cells via interactions with CD14+ monocytes in vitro. Notably, SnMP did not induce proliferation in isolated T cells, but only in cultures where CD14+ cells were also present. While this observation does not rule out a direct effect of HO-1 on T cells, it does indicate that such an effect is not sufficient to induce activation. Proliferation occurred in the presence of very few monocytes, and there was a direct correlation between the frequency of monocytes present in culture and the extent of proliferation. Experiments using transwells and blocking antibodies demonstrated that SnMP-induced T cell activation requires direct cell-to-cell MHC Class I and II-dependent interactions between T cells and monocytes. Both MHC Class I and II blockade inhibited SnMP-induced proliferation, and there was an amplified effect of dual blockade, with abrogation of proliferation even at low antibody concentrations.
Given that MHC-dependent interaction of monocytes with T cells plays a crucial role in this in vitro system, we analyzed the phenotypic changes that occur in CD14+ cells on day 3 of culture, prior to observed T cell proliferation. In the absence of additional cytokines, monocytes in PBMC culture normally stick to plastic plates and differentiate into monocyte-derived macrophages (MDM), which we see occurring in vehicle control samples, where CD14+ cells are also CD11c+, CD16+, and HLA-DR+. We noted several differences in CD14+ cells that were cultured in SnMP. Among the effects noted were a decrease in the expression of both CD11c and CD16. Most notably, SnMP induced upregulation of the co-activating molecule CD86 (B7-2), which plays an important role in the MHC-TCR immunological synapse by providing crucial secondary signals that modulate T cell activation. We postulate that this upregulation may enhance the ability of monocyte-derived CD14+ cells to activate T cells via TCR-MHC interactions. The C-type lectin BDCA-2, which is typically expressed on plasmacytoid dendritic cells, was also significantly upregulated in SnMP-treated CD14+ cells. Together, these changes demonstrate that the CD14+ population undergoes several phenotypic changes in response to HO-1 inhibitor exposure, some of which have the potential to confer activating function, while HO-1 induction by CoPP is associated with changes that are associated with a non-inflammatory phenotype (i.e., a decrease in the expression of CD86 and HLA-DR).
We found that SnMP-induced T cell proliferation can be inhibited by CD25+FoxP3+ Tregs but that, reciprocally, SnMP can inhibit the suppressive function of Tregs. Tregs from HO-1 deficient mice have no intrinsic defect in their ability to suppress T cell activation, but it is now known that their ability to do so maximally and efficiently requires interactions with wild-type APCs that have intact HO-1 activity (26, 36). Accordingly, we suggest that inhibition of HO-1 activity in APCs in human PBMC cultures results in impaired Treg function. It is widely accepted that Tregs can induce changes in APCs to down-regulate their antigen presenting functions (37, 38).
Conversely, both immature DCs and alternatively activated macrophages can induce Treg development de novo, while classically activated macrophages and mature dendritic cells can have negative effects on Treg function (39). The mechanism by which HO-1 activity supports Treg function remains a matter for speculation. CO produced by HO-1 in APCs could act in a paracrine fashion to support Tregs by suppressing T cell proliferation or by inducing transcriptional changes in the Tregs themselves, leading to enhanced survival or suppressor activity. Catabolic products of heme breakdown could also work in an autocrine fashion to drive APC differentiation toward a phenotype that supports Treg survival or function.
Based on the results of our experiments, we suggest the model shown in Figure 6 to describe the interactions leading to T cell activation and proliferation in PBMC cultures upon HO-1 inhibition by SnMP. In this model, unopposed baseline endogenous HO-1 activity supports the quiescent state of monocytes. There may be an endogenous effect of HO-1 in both naive T cells and Tregs, but it is also likely that the effects of HO-1 are exerted via interactions with quiescent monocytes that promote Treg survival and function. In this baseline state, anti-proliferative signals prevail, and interaction with self-MHC allows for T cell survival and low-level baseline rates of proliferation. Exposure to SnMP results in HO-1 inhibition, leading to pro-activating phenotypic changes in monocytes, naïve T cells, or both. The primary observed effect resulting from this is the MHC-dependent induction of T cell proliferation. HO-1 inhibition also results in monocyte-mediated impairment of Treg function, indirectly augmenting the extent of naïve T cell proliferation. Together, these effects are sufficient to induce proliferation of a surprisingly large fraction of T cells present in PBMC cultures.
While we base our model on the assumption that the enzymatic activity of HO-1 is responsible for the observed effect, it is important to consider the alternative possibility that non-enzymatic functions of HO-1 play a role. Recent work has shown that HO-1 possesses important transcriptional modifying activity that is completely independent of its catalytic function. In NIH 3T3 cells exposed to hypoxia or heme, HO-1 underwent cleavage of a C-terminal domain, leading to nuclear translocation of the N-terminal domain of HO-1 and subsequent transcriptional regulation by this cleaved portion (40). Furthermore, HO-1 protein that has been made to be catalytically inactive through site-directed mutagenesis participates directly in its own transcriptional autoregulation despite the absence of an active catalytic site (41). Thus, the phenotypic changes observed in response to SnMP may also be related to transcriptional changes induced by the presence of non-catalytically active (i.e. inhibited) HO-1 protein. This possibility is especially intriguing since HO-1 expression is induced by SnMP, resulting in a relative excess of inhibited HO-1. Of note, we attempted to carry out spectrophotometric HO-1 enzyme activity assays to confirm induction and inhibition of HO-1 (data not shown), but were limited by the number of cells available from an individual donor. Normally this assay is carried out on tissue or cell-line lysates, from which protein yield is not normally limiting. We were unable to generate sufficient quantities of microsomal protein from single volunteer human donors to carry out this assay successfully, and so were unable to directly demonstrate that SnMP inhibits HO-1 activity in our system. There is ample evidence from the literature that synthetic metalloporphyrin inducers and inhibitors of HO-1 are active in hematopoetic cells, and specifically in cells of the monocyte/macrophage lineage (42–44), and so it is reasonable to assume that SnMP acts as an inhibitor in our system.
The findings of this in vitro study suggest that HO-1 plays a role in controlling naïve T cell activation, maturation, and proliferation, and in vivo studies are clearly warranted to validate the physiological relevance of our findings. The goal of such studies would be to determine if HO-1 activity represents a safeguard mechanism to prevent non-specific T cell activation by APCs, and whether removal of this safeguard by HO-1 down-regulation or inhibition plays a role in promoting T-cell activation and maturation under physiologic or pathologic circumstances. Activation of T cells in vitro by SnMP required interaction with MHC Class I and II, presumably via the TCRs on responding T cells. This is notable in that the observed response almost certainly represents a TCR-mediated response to self-MHC. Normally, T cells do not undergo widespread activation and proliferation in response to self-MHC, which is crucial for the maintenance of tolerance to self and prevention of autoimmunity. There are many mechanisms in place to ensure that T cell activation occurs only in appropriate settings (e.g., upon exposure to dangerous pathogens or upon detection of malignancy) and not in response to self-antigens. Chief among these mechanisms is the thymic deletion of autoreactive T cells through negative selection (45) and, in the peripheral immune system, the maintenance of tolerance by regulatory cells such as CD4+CD25+FoxP3+ Tregs (46, 47). These regulatory cells also participate in the tuning and modulation of immune responses to ensure their appropriate activation and termination. In the absence or relative paucity of these regulatory mechanisms, the immune response may proceed unchecked, causing collateral damage to the host (46, 47). Given the extent of proliferation observed after HO-1 inhibitor exposure, we posit that HO-1 may also serve as a safeguard mechanism to prevent inappropriate T-cell activation. In many animal disease models, absence of HO-1 activity results in excess inflammation that contributes to pathology, including models of diabetes, asthma, multiple sclerosis, cerebral malaria, and transplant rejection (12, 27–32). The work presented here provides further support for a potential role of HO-1 in preventing inappropriate T cell activation in humans.
Naïve T cells in the periphery undergo low levels of homeostatic proliferation until they encounter cognate antigen in the context of activating signal, at which point they go on to become effector and memory cells (48). This homeostatic proliferation is now thought to occur almost exclusively in lymph nodes, where T cells move through the parenchyma and come into contact with fibroblastic reticular cells (49). Among the signals that are crucial for naïve T cell survival and proliferation, one of the most important is contact with MHC molecules on supporting accessory cells (48). In addition to the influence of critical growth factors, low-avidity interactions between the TCR on naïve T cells and MHC provide survival signals that allow these cells to continue to proliferate at low levels, thereby maintaining a diverse and appropriately quiescent naïve T cell population. Our experiments suggest that myeloid HO-1 activity may represent a “braking” mechanism for naïve T cell proliferation, allowing for low-level proliferation in response to self-MHC while preventing uncontrolled activation and proliferation. If so, its absence could lead to dysregulated homeostasis. Indeed, mice that are deficient in HO-1 have clear evidence of dysfunctional lymphocyte homeostasis, including splenomegaly, lymphadenopathy, altered CD4/CD8 ratio, and disorganized lymph node and splenic architecture (14–16). They also have a higher frequency of activated T cells (15). This suggests that HO-1 plays a role in the regulation and maintenance of the peripheral T cell pool, and/or in the prevention of inappropriate activation.
Our study suggests that HO-1 plays a role in T cell homeostasis, and support for this hypothesis is found most convincingly in our examination of the maturational profile of cells treated with SnMP. The experiments shown in Figure 4 clearly demonstrate that proliferating cells are primarily naïve T cells that adopt memory cell phenotypes, a phenomenon which is observed in some models of homeostatic T cell proliferation (50–53). For example, naive T cells transferred into syngeneic lymphopenic hosts repopulate the host’s peripheral immune system by undergoing robust self-MHC dependent proliferation, during which they take on the phenotype and characteristics of memory cells (50–53). It may be that the naïve cells that become proliferating memory cells in our experiments are undergoing a similar homeostatic proliferative response. It remains to be seen whether comparable responses may occur in vivo during inhibition of HO-1. Certainly, this type of response would seem more likely to occur in secondary lymphoid organs, where T cells come in contact with myeloid cells for an extended period of time. Furthermore, variations in HO-1 activity that would theoretically lead to more or less T cell proliferation could occur in specialized subanatomic regions, possibly influenced by concentration gradients of natural HO-1 inducers such as heme.
HO-1 has been shown in many instances to be anti-proliferative and to down-regulate potentially harmful inflammatory responses. The experiments presented here raise another possible role for HO-1 in T cell homeostasis. Namely, that unopposed HO-1 in myeloid cells provides homeostatic signals to T cells, allowing them to remain in a non-activated state. In the absence of HO-1, or in the presence of inhibited HO-1, a different set of signals (or perhaps merely the absence of anti-proliferative signals) may then lead to T cell activation and proliferation. This effect may also represent a mechanism to alleviate suppression of T cells in settings where activation is needed, such as infection or malignancy. In aggregate, these findings demonstrate that HO-1 can alter human T cell activation, maturation, and proliferation in vitro, and suggest that this multifunctional protein may play a role in controlling lymphoid development and homeostasis in vivo. They also suggest the possibility that SnMP, or other pharmacologic HO-1 inhibitors, could be used as clinical modulators of T cell maturation, which would have potential use in settings requiring immune reconstitution such as chemotherapy and following initiation of highly active antiretroviral therapy for HIV.
We thank the volunteer blood donors who participated in this study; Dr. Nader Abraham for helpful technical advice; Terrence Ho and the Division of Experimental Medicine Flow Cytometry Core for assistance with flow cytometry and cell sorting; Drs. Jeffrey Bluestone and Dr. Mark Anderson for helpful discussions regarding antibody blockade experiments; and Drs. Jennifer Babik and Louise Swainson for thoughtful discussion, and for careful review of the manuscript.
During the period in which this work was carried out, TB was a fellow of the Pediatric Scientist Development Program, and was supported by the American Pediatric Society, the American Academy of Pediatrics and the March of Dimes. LS was supported by a California HIV Research Program Dissertation Fellowship Award D09-SF-313. AK was supported by a grant from the Beatrice Renfield Foundation. This work was also supported in part by National Institutes of Health Awards U01 AI43641 and R37 AI40312 (to J.M.M.), who is the recipient of the Burroughs Wellcome Fund Clinical Scientist Award in Translational Research and the National Institutes of Health Director’s Pioneer Award Program, part of the National Institutes of Health Roadmap for Medical Research, through Grant DPI OD00329.