Carbon Monoxide Protects Pancreatic β-Cells From Apoptosis and Improves Islet Function/Survival After Transplantation

Male C57BL/6 mice (Charles River Laboratories, Wilmington, MA) were housed in accordance with guidelines from the National Institutes of Health. The experiments were approved by the Institutional Animal Care and Use Committee.

The murine insulinoma cell line βTC3 (DSMZ, Braunschweig, Germany) was cultured in Dulbecco’s Modified Eagle’s Medium (Life Technologies, Grand Island, NY) supplemented with 2 mmol/l l-glutamine, 100 units/ml penicillin G, 100 units/ml streptomycin, and 10% FCS (Life Technologies).

βTC3 cell lines were seeded at 2 × 10⁵ cells, washed once with 500 μl PBS, and stained with 200 μl 0.05% crystal violet in 20% ethanol for 10 min at room temperature. To elute stain from cells, 100 μl 50% acetic acid was added to each well, and 50 μl were transferred into 96-well microtiter plates and read with a plate reader (EL 340 biokinetics reader; Bio-Tek Instruments, Winooski, VT) at 562 nm.

The β-galactosidase expression vector (Clontech Laboratories, Palo Alto, CA) was cloned into the pcDNA3 vector (17) (Invitrogen, Carlsbad, CA). A 1.0-kbp XhoI-HindIII fragment encoding the full-length rat HO-1 cDNA was cut from the prHO-1 vector (18) and subcloned into the pcDNA3 vector.

βTC3 cell lines were seeded at 3 × 10⁵ cells in 16-mm wells and transfected 15–20 h later using Lipofectamine plus reagents (Life Technologies) according to the manufacturer’s instructions. Total DNA was maintained constant using empty pcDNA3 vector. The percentage of viable cells was assessed by normalizing the percentage of viable cells to the number of control-transfected cells without the apoptotic stimulus (100% viability) (7,17).

βTC3 cultures were incubated with recombinant tumor necrosis factor-α (TNF-α) (500 units) (R&D Systems, Minneapolis, MN) for 24 h, and islets were stimulated with TNF-α (5,000 units/ml) (R&D Systems) and cycloheximide (CHX) (Sigma, St. Louis, MO) (50 μg/ml) for 48 h (6). βTC3 cells or islets were harvested, dispersed, fixed in 70% ethanol, and suspended into DNA staining buffer (PBS, pH 7.4, containing 0.1% Triton X-100, 0.1 mmol/l EDTA, 50 μg/ml propidium iodide, and 50 mg/ml RNase A). DNA content was analyzed on a FACScan (Becton Dickinson, Palo Alto, CA). Cells with DNA content ≥2 n were scored as viable; cells with a hypoploid DNA content (<2 n, termed A°) were scored as apoptotic. To exclude debris and cell fragments, all events with a FL-2 area profile below that of chicken erythrocyte nuclei were excluded from analysis.

Murine recombinant TNF-α (R&D Systems) was dissolved in PBS and 1% BSA and added to the culture medium (17.5 ng/ml = 500 units) 24 h after transfection. The caspase-3 inhibitor Z-DEVD-FMK and the caspase-8 inhibitor IETD-CHO (Calbiochem, San Diego, CA) were dissolved in DMSO (Sigma) and added to the culture medium (10 and 1 μmol/l, respectively) 2 h before treatment with TNF-α. Tin protoporphyrin IX (SnPPIX) (Porphyrin Products, Logan, UT) was dissolved (10 mmol/l) in 100 mmol/l NaOH and added 6 h after transfection to the culture medium (50 μmol/l). The guanylyl cyclase inhibitor 1H[1,2,4]oxadiazolo[4,3-α]quinoxalin-1 (ODQ) (Calbiochem) was dissolved in DMSO and added to the culture medium (100 μmol/l) 6 h after transfection. The cGMP analog 8-bromoguanosine-3′-5′-cyclic-monophosphate (8-Br–cGMP) (Sigma) was dissolved in water and added to the culture medium (10 μmol/l) 30 min before induction of apoptosis. The protein kinase G inhibitor KT5823 (Calbiochem) was dissolved in DMSO and added to the culture medium (1.6 μmol/l) 6 h after transfection.

Cells and islets were exposed to 1% CO in compressed air balanced with 5% CO₂, as described elsewhere (15). Islets were incubated in RPMI medium presaturated with CO (4°C overnight, 1% CO, 5% CO₂) for 2 h at 37°C, while treatment with 1% CO, 5% CO₂ was continued.

Recipient C57BL/6 mice (8 weeks of age) were rendered diabetic by a single intraperitoneal injection of 220 mg/kg (19) streptozotocin (Sigma) freshly dissolved in citrate buffer. Mice were transplanted if two consecutive nonfasting blood glucose levels >350 mg/dl were obtained from whole blood.

Pancreatic islets of Langerhans (C57BL/6 mice) were provided by the Islet Core Laboratory of the Juvenile Diabetes Research Foundation Center for Islet Transplantation at Harvard Medical School and isolated as described previously (20).

Using a dissecting microscope, 250 islets of 150–250 μmol/l in diameter were hand-picked. Islets were transplanted under the kidney capsule as described previously (19). The same numbers of control and treatment animals were transplanted from each islet preparation. Animals were anesthetized with ketamine (0.9 mg/20 g body wt) and xylazine (0.1 mg/20 g body wt) administered as a single intraperitoneal dose.

Graft function was defined when the first of 3 consecutive days of nonfasting blood glucose levels <200 mg/dl was reached. The primary end point of the experiment was defined as time to normoglycemia.

Blood glucose data are summarized as the means ± SD of mice receiving untreated or treated islets. Time to recovery of islet function was calculated using Kaplan-Meier life tables and differences between groups tested using a log-rank test, with the three islet preparations treated as separate strata in the analysis and the median time to recovery, with 95% CI reported.

TNF-α induced high levels of cell death in the insulinoma cell line βTC3 (21) (Fig. 1A). DNA fragmentation was demonstrated (Fig. 1B), suggesting that TNF-α induces β-cell death through apoptosis. TNF-α–mediated apoptosis was strictly dependent on the activation of caspase-8 and partially dependent on that of caspase-3 (Fig. 1C).

Overexpression of HO-1 protected βTC3 cells transiently transfected with a HO-1 expression vector from TNF-α–mediated apoptosis (6) (87% survival vs. 33% in control) (Fig. 2A). Blocking HO-1 activity by SnPPIX (22) suppressed the antiapoptotic effect (Fig. 2A), suggesting that one or more of the end products of heme catabolism by HO-1, i.e., iron, bilirubin and/or CO, is required for its antiapoptotic function.

When the action of HO-1 was suppressed by SnPPIX, CO exposure suppressed TNF-α–mediated apoptosis to a similar extent as HO-1 (Fig. 2A). Exposure to exogenous CO alone protected from TNF-α–induced apoptosis (11.7 vs. 20.3% in apoptotic cells vs. controls not exposed to CO) (Fig. 2B, DNA fragmentation analysis). Similarly, β-cell apoptosis induced by etoposide or serum starvation was suppressed by CO exposure (Fig. 2C).

Exposure to CO for 24 h protected isolated murine (C57/BL6) islets of Langerhans from TNF-α plus CHX-mediated apoptosis (11.7 vs. 20.3% in apoptotic cells vs. controls not exposed to CO), as assayed by DNA fragmentation (Fig. 3A) or histological analysis (Fig. 3B).

Inhibition of soluble guanylate cyclase (sGC) activity by oxadiazoloquinoxalin (ODQ) suppressed the antiapoptotic effect of CO, suggesting that sGc is a mediator for CO (Fig. 4A), as also found in fibroblasts (12). The cGK activator/cGMP analog, 8-Br–cGMP, suppressed βTC3 apoptosis to a similar extent as CO (Fig. 4B). Inhibition of cGKs by KT5823 suppressed the antiapoptotic effect of exogenous CO (Fig. 4C), suggesting involvement of one or more cGKs.

We exposed βTC3 cells to CO for different time periods (1–24 h) immediately after the addition of TNF-α and tested for apoptosis 24 h later. One hour of CO exposure was sufficient to prevent β-cell apoptosis (Fig. 5A).

We exposed β-cells to CO for 1 h, 0.5–12 h after induction of apoptosis by TNF-α. CO still suppressed β-cell apoptosis 2 h after TNF-α stimulation (Fig. 5B).

Further, 1 h of preincubation with CO prevented β-cell apoptosis (data not shown). Lastly, 1 h of preincubation with CO still prevented β-cell apoptosis in cells stimulated with TNF-α 2–3 h after the end of the 1-h treatment with CO (Fig. 5C). Thus, relatively brief treatment with CO provides an antiapoptotic effect that lasts for an extended time period.

We evaluated whether pre-exposure of islets to CO can improve islet survival and/or function after transplantation. To test this, we transplanted a marginal islet mass under the kidney capsule of diabetic syngeneic recipients. A total of 500 handpicked transplanted C57/BL6 islets led rapidly to normoglycemia [1.5 ± 0.5 days (n = 4)], whereas 250 transplanted islets resulted in a significant delay [14.2 ± 2.9 days (n = 9), data not shown]. Using a 250-islet marginal mass in this system does not involve rejection or recurrence of autoimmune disease (2). The time needed to reach normoglycemia was reduced in a highly significant manner (P = 0.0011), when islets were preincubated for 2 h in medium presaturated with CO [7 days (95% CI 6–8)] as compared with control islets not pre-exposed to CO [14 days (12–18)] (Figs. 6A and B). In total, three different islet preparations were used for these experiments. There was no statistically significant difference in the time to normoglycemia for CO-treated and untreated islets among these three preparations (P > 0.25).

Islet transplantation suffers from the need for high numbers of donor islets (23). Islet apoptosis after transplantation is thought to be caused by stress factors such as hypoxia, pro-inflammatory cytokines, and free radicals released from macrophages in the microenvironment of the transplanted islets (4,19,24,25). The observation that islets undergo primary graft nonfunction when transplanted into syngeneic recipients (26) strongly suggests that nonspecific inflammation plays a major role in this process.

Preventing apoptosis might improve islet survival/function after transplantation, allowing a significant reduction in the number of islets required for this procedure. In the present studies, we have investigated TNF-α–mediated β-cell apoptosis based on the assumption that TNF-α is a potential mediator of β-cell death after transplantation (24,27,28).

Freshly isolated rat islets do not express HO-1, whereas cultured rat islets have increased HO-1 expression (29,30). HO-1 induction by hemin (31) or CoCl₂ (30) can protect islets from the inhibitory effects of interleukin (IL)-1β on glucose-stimulated insulin release and glucose oxidation. HO-1 expression protects these cells from TNF-α–mediated apoptosis (Fig. 2A) (6), a finding that may have important therapeutic implications (2,28,32).

We show here that CO protects pancreatic β-cells from apoptosis, as it does in endothelial cells (7,33). CO is an important signaling molecule in several cell types, including islets of Langerhans, where CO has been suggested to stimulate insulin and glucagon release (34). Compared with NO, CO is only a weak activator of sGC (35). However, CO may exert its effects for longer times and thus over longer distances than NO (34,36).

Our findings (Figs. 4A and B) suggest that cGMP is the main messenger of the antiapoptotic action of CO, as it is in fibroblasts (12). In monocytes (15) and endothelial cells (33), CO effects involve the p38 mitogen-activated protein kinase pathway, which does not appear to be involved in the present studies (data not shown).

We (17) and others (15,34,37) suggest that the protective function of CO may involve both antiapoptotic and anti-inflammatory effects. Low concentrations of CO in macrophages inhibit expression of the lipopolysaccharide-induced pro-inflammatory cytokines, while increasing the expression of IL-10 both in vivo and in vitro (15). In islets of Langerhans, the release of pro-inflammatory cytokines by activated macrophages upregulates inducible nitric oxide synthase in β-cells (25), which is highly cytotoxic for β-cells (38). Exogenous CO also suppresses NO production in islets (37), a potential direct protective effect.

We show here that pre-exposure of islets to CO for 2 h increases their survival and/or function after transplantation in a highly significant manner presumably based on the antiapoptotic and anti-inflammatory effects of CO. Preventing apoptosis that occurs after islet transplantation (32) would improve function of the transplanted islets and reduce the number of islets needed to treat diabetes. Given the above, CO might be a useful therapeutic tool for improving islet transplantation in humans.

This work was supported in part by the Juvenile Diabetes Research Foundation International and by a National Institutes of Health grant (HL58688) awarded to F.H.B. L.G. is supported by a fellowship of the Deutsche Forschungsgemeinschaft (DFG), and P.O.B. is supported by a fellowship of the Swiss National Science Foundation. F.H.B. is a paid consultant for Novartis Pharma, Basel.

We thank Shane Gray and Takahashi Maki for technical help and Christiane Ferran for discussions.