Abstract
Establishing reliable islet potency assay is a critical and unmet issue for clinical islet transplantation. Recently, we reported that islets contained high levels of high mobility group box 1 (HMGB1) and damaged islets released HMGB1 in a mouse model. In this study, we hypothesized that the amount of released HMGB1 could reflect the degree of islet damage, and could predict the outcome of islet transplantation. Four groups of damaged mouse islets and three groups of damaged human islets were generated by hypoxic conditions. These islets were assessed by in vivo (transplantation) and in vitro (released HMGB1 levels, released C-peptide levels, PI staining, TUNEL staining, ATP/DNA, and glucose-stimulated insulin release test) assays. In addition, the ability of each assay to distinguish between noncured (n = 13) and cured (n = 7) mice was assessed. The curative rates of STZ-diabetic mice after receiving control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 100%, 40%, 0%, and 0%, respectively. Only amounts of released HMGB1 and ratio of PI staining significant increased according to the degree of damages in both human and mouse islets. In terms of predictability of curing diabetic mice, amounts of released HMGB1 showed the best sensitivity (100%), specificity (100%), positive (100%), and negative predictive values (100%) among all the assays. The amount of released HMGB1 reflected the degree of islet damage and correlated with the outcome of islet transplantation in mice. Hence, released HMGB1 levels from islets should be a useful marker to evaluate the potency of isolated islets.
Introduction
Islet transplantation is a promising therapy for type 1 diabetes mellitus (26) and is preferred by patients over insulin injection therapy (6). However, major issues still prevent the use of islet transplantation as a standard therapy (10,27), including limitations of the current potency assays (20). The potency assay for islet characterization prior to clinical transplantation should be reliable, operator independent, reproducible, and transferable to other laboratories (20). It should also be able to work with relatively small yet representative islet numbers without requiring islet handpicking (which may bias the results) and should be able to provide real-time results (i.e., completed within hours) (20). Development of such an assay has been mandated by the US Food and Drug Administration (34).
Currently, the in vivo islet transplantation assay using diabetic mice is thought to most reliably correlate with clinical islet transplantation (20). However, results from this assay can be obtained only after several days and therefore cannot be used for islet prerelease testing. To develop a potency assay for an islet prerelease test, the most important endpoint is the correlation between the results of the in vitro potency assay and the in vivo potency assay using a transplantation model. Currently, the cell membrane integrity test (2), ATP assay (4,8,29), glucose-stimulated insulin secretion (GSIS) test (13,26,27), and detection of apoptosis by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and/or Annexin V staining (1,18) are widely performed as potency assays of isolated islets. However, these assays are limited in their ability to predict clinical outcomes and the results of the in vivo transplantation assay (20). Recently, tests for cellular composition and β-cell viability (7), mitochondrial membrane potential (5,16), and oxygen consumption rate of islets (19,30,31,33) were developed and proved their reliability for predicting in vivo transplant results. However, those assays require specialized equipment and techniques (20) and therefore are not considered the standard method.
High mobility group box 1 (HMGB1) is a chromatin-binding protein that regulates transcription and chromosome architecture (17,25,28). Recently, it was shown that HMGB1 plays a crucial role in responding to tissue damage, indicating that HMGB1 is a prototype of the emerging damage-associated molecular pattern molecule (9,25,28). We reported that HMGB1 was uniquely abundant in pancreatic islets, and damaged islets released HMGB1 outside the cells in a mouse model (15). Based on those results, in the present study we hypothesized that the amount of released HMGB1 could reflect the degree of islet damage and could predict the outcome of islet transplantation.
Materials and Methods
Animals
Male C57BL/6 mice (8–12 weeks old) were purchased from Harlan Laboratories (Houston, TX) and used as diabetic recipients and islet donors. Diabetes was induced by intravenous injection of streptozotocin (STZ, 180 mg/kg; Sigma, St. Louis, MO). When blood glucose levels exceeded 400 mg/dl 2–3 days after STZ injection and the mice remained hyperglycemic at the time of islet transplantation, then they were used as the diabetic recipients. The experiments were approved by the institutional animal care and use committee.
Mouse Islet Isolation
Islets were isolated by the static digestion method using collagenase and the density gradient purification method (15). Islets 150–250 μm in diameter were hand-picked for each experiment.
Human Islet Isolation
Two research-grade human pancreases from brain-dead donors were provided from a local organ procurement organization (Southwest Transplant Alliance, Dallas, TX) for this study. Pancreases were procured using a standardized technique, and pancreatic ductal injection was performed (11). Human islet isolation was conducted as described previously (11,12,14) in the standard Ricordi technique (22) with modifications introduced in the Edmonton protocol (21,26).
Islet Culture and Exposure to Hypoxia
Mouse and human isolated islets were cultured in DMEM (Sigma-Aldrich, St. Louis, MO) and CMRL1066 (Mediatech, Inc., Manassas, VA), respectively, at 37°C in 95% air and 5% CO2 for 24 h after islet isolation. In the mouse model, islets were divided into four groups: control, hypoxia-3h, hypoxia-6h, and hypoxia-24h islets. In the control group, islets were washed twice with culture medium after initial culture and were cultured at 37°C in 95% air and 5% CO2 for 24 h. In the hypoxia-3h, hypoxia-6h, and hypoxia-24h groups, islets were washed twice with culture medium after initial culture and placed into modular incubator chambers (Billups-Rothenberg, Inc., Del Mar, CA). The chambers were flushed with the hypoxic (1% O2, 5% CO2, and 94% N2) gas, closed to maintain the hypoxic condition, and then put into an incubator at 37°C for 3 (hypoxia-3h), 6 (hypoxia-6h), or 24 h (hypoxia-24h). In the human model, islets were divided into three groups including control (normoxia-48h), hypoxia-24h, and hypoxia-48h. The human islets were subjected to the same hypoxic conditions as the mouse islets.
Histological Study
The islets in the each group were fixed in 10% formaldehyde solution, processed, and embedded in paraffin. The sections were stained with anti-mouse insulin antibody (Sigma-Aldrich, St. Louis, MO) (15).
Propidium Iodide (PI) Staining and PI+ Area Assay
Islets were stained with Hoechst 33342 (HO342) and PI, as previously described (24). The PI+ area and islet area were measured by the digital imaging software Image J (provided by NIH). Thirty samples were obtained by three independent experiments in each group. The percentage of PI+ areas in the islets was expressed as mean ± SD.
TUNEL Assay
The TUNEL assay was performed to detect apoptosis in the islets. The ApopTag Fluorescein in Situ Apoptosis Detection Kit (Chemicon International, Temecula, CA) was used according to the manufacturer's instructions. Fifteen samples were obtained by three independent experiments in each group. TUNEL+ cells and all cells were counted to calculate the ratio of DAPI+ cells.
In Vivo Assessment
Two hundred mouse islets in the each group were transplanted under the left kidney capsule of STZ-induced diabetic mice; this number was chosen based on Sakata et al.'s report of STZ-induced diabetic mice became normoglycemic after receiving 200 islets under the kidney capsule (23). Five cases of transplantation were performed in each group. The nonfasting blood glucose levels were measured using Accu-Chek Aviva (Roche Diagnostics, Indianapolis, IN) three times a week in all the recipients for 30 days after islet transplantation. Normoglycemia after transplantation was defined as two consecutive blood glucose levels reading <200 mg/dl. The data were expressed as mean ± SE.
Assay of Medium HMGB1 and C-Peptide Levels
At the end of normoxic or hypoxic culture, islet culture media from each well were collected. The amount of HMGB1 in the medium was measured using an HMGB1 ELISA kit II (Shino-test, Kanagawa, Japan) (35), and the amount of C-peptide in the medium was measured using a C-peptide ELISA kit (Mercodia, Uppsala, Sweden) according to the manufacturer's instructions. Fifteen samples were obtained by three independent experiments in each group. The amounts of HMGB1 and C-peptide in media were normalized to the total DNA of cultured islets, and the data were expressed as mean ± SD.
GSIS
Islets in each group were incubated with low (2.8 mmol/L) and high (20.0 mmol/L) concentrations of glucose solution in Functionality/Viability Medium CMRL1066 (Mediatech, Inc.) for 1 h at 37°C. Insulin concentrations were measured by an insulin ELISA kit (ALPCO Diagnostics, Salem, NH). The stimulation index was calculated by determining the ratio of insulin concentrations in the high-glucose solution to that in the low-glucose solution. The insulin secretion levels were normalized to the total DNA of islets. Fifteen samples were obtained by three independent experiments in each group. The data were expressed as mean ± SD.
Assay of ATP Content
To measure the ATP content of islets in each group, islets were washed twice and sonicated in 1 ml of PBS. The amount of ATP was measured using an ATP assay system (ATP-lite, Perkin Elmer, Groningen, Netherlands) according to the manufacturer's instructions. Fifteen samples were obtained by three independent experiments in each group. The data were normalized to total DNA and expressed as mean ± SD.
DNA Measurement and Unit Conversion
DNA amounts were measured immediately after the normoxic and hypoxic culture (dsDNA Assay Kit, Molecular Probes, Inc., Eugene, OR). To compare the measured molecule amounts among the different samples, each amount of examined molecule was converted to μg DNA of total cultured islets.
Statistical Analysis
The statistical significance of the amount of released HMGB1 levels, the amount of released C-peptide levels, PI+ area assay, TUNEL assay, ATP/DNA assay, insulin secretion levels, and stimulation index was determined by one-way ANOVA and Tukey post hoc test. The statistical significance between the noncured and cured group was determined by Student's t-test. Receiver operating characteristics (ROC) analyses were employed to examine the ability to discriminate curable and non-curable results in the in vivo assay for each molecule and to determine the cut-off values. The cut-off values were chosen to obtain a good balance between sensitivity and specificity based on ROC curves. The positive and negative predictive values were determined according to the standard definition. All statistical analysis was performed using PASW statistics 18.0.2 (SPSS Inc, Chicago, IL). Differences were considered significant when values were p < 0.05.
Results
Morphological Appearance of Four Degrees of Damaged Mouse Islets
We made four groups of islets, including control, hypoxia-3h, hypoxia-6h, and hypoxia-24h islets in a mouse model. These four groups of mouse islets were morphologically examined. The islet surface became rougher, the dark spots more noticeable, and the PI and TUNEL staining positivity greater as the amount of hypoxia increased (Fig. 1A).
Islet morphology and in vivo islet transplantation assay. (A) Control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were examined by phase-contrast microscopy, Hoechst33342 (HO342)/propidium iodide (PI) staining and insulin/terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining. Scar bars: 50 μm. Two hundred control, hypoxia-3h, hypoxia-6h, or hypoxia-24h mouse islets were transplanted beneath the kidney capsule of streptozotocin (STZ)-induced diabetic mice. The mean blood glucose levels (B) and curative rates (C) were measured. The solid black line indicates control group (n = 5), dashed line indicates hypoxia-3h group (n = 5), gray solid line indicates hypoxia-6hr group (n = 5), and dotted line indicates hypoxia-24hr group (n = 5). The blood glucose levels were expressed as mean ± SE.
Transplant Outcomes in Mice
All five of the STZ-induced diabetic mice in the control group became normoglycemic after receiving 200 islets under the kidney capsule, whereas in the hypoxia-3h group, two of five of the diabetic mice became normoglycemic, and in the hypoxia-6h and -24h islet groups, none did (Fig. 1B, C). A nephrectomy of the islet-bearing kidney in all cured mice confirmed the reelevation of the blood glucose level, ensuring that the reversal of diabetes was due to the transplanted islets.
Differences of In Vitro Assays Among the Four Groups of Mouse Islets
The amount of released HMGB1 from the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets was 0.13 ± 0.03, 0.32 ± 0.06, 1.01 ± 0.67, and 2.46 ± 1.11 ng/μg DNA, respectively (Fig. 2A). The amounts of released HMGB1 significantly increased according to the degree of hypoxia-induced damages (Fig. 2A). The amount of released C-peptide from the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets was 0.40 ± 0.10, 0.29 ± 0.03, 0.92 ± 0.28, and 8.13 ± 3.14 pmol/μg DNA, respectively (Fig. 2B). Only the most severely damaged group (hypoxia-24h) significantly increased the amount of released C-peptide. PI+ areas/islet areas in the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 1.6 ± 1.3%, 8.6 ± 3.9%, 21.1 ± 8.4%, and 37.0 ± 11.2%, respectively (Fig. 2C). The percentages of TUNEL+ cells in the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 1.9 ± 0.7%, 2.7 ± 0.9%, 24.8 ± 1.2%, and 30.4 ± 3.1%, respectively (Fig. 2D). Both PI+ area/islet assay and TUNEL assay significantly increased positive cells according to the degree of damages. The ATP/DNA ratios in the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 18.3 ± 7.9, 25.4 ± 5.7, 20.4 ± 8.4, and 9.3 ± 4.6 pmol/μg DNA, respectively (Fig. 2E). Only the most severely damaged group (24-h hypoxia group) significantly decreased the ATP/DNA ratio. Insulin levels in response to low-glucose (2.8 mM) in the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 0.22 ± 0.14, 0.99 ± 0.51, 2.08 ± 1.19, and 1.00 ± 0.65 ng/μg DNA, respectively (Fig. 2F). Hypoxia-6h showed significantly the high value compared to all other groups. Insulin levels in response to high-glucose (20.0 mM) in the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 3.90 ± 2.93, 4.45 ± 0.81, 5.61 ± 3.45, and 2.76 ± 1.12 ng/μg DNA, respectively (Fig. 2G). The only significant difference was found between the hypoxia-6h and the hypoxia-24h islets (*p < 0.05). Stimulation indices in the control, hypoxia-3h, hypoxia-6h, and hypoxia-24h mouse islets were 22.4 ± 13.3, 5.2 ± 1.8, 2.8 ± 0.9, and 3.5 ± 2.1, respectively (Fig. 2H). Only the control group had significant high value compared to all other groups.
In vitro assays for four groups of damaged mouse islets. Four groups of damaged mouse islets were analyzed by the amounts of released high mobility group box 1 (HMGB1; A) and C-peptide (B) levels, PI+ area assay (C), TUNEL assay (D), ATP/DNA (E), insulin release against 2.8 mM glucose (F), insulin release against 20.0 mM glucose (G), and stimulation index (H). The data were expressed as mean ± SD. *p < 0.05, **p < 0.01, and ***p < 0.001.
Assay Results in Cured Versus Noncured Mice
To determine which assay could best predict the outcome of islet transplantation in a mouse model, we analyzed the predictive values of each assay. The cut-off value, sensitivity, and specificity were defined by ROC analysis (Fig. 3). As shown in Table 1, the HMGB1 level had the highest sensitivity, specificity, and predictive values.
Sensitivity, Specificity, and Predictive Values of Different Assays When Comparing Results Between Cured and Noncured Mice After Islet Transplantation
HMGB1, high-mobility group box 1; PI, propidium iodide; GSIS, glucose-stimulated insulin secretion. TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling.
Significant at p < 0.05 between noncurend and cured.
Significant at p < 0.001 between noncured and cured.
Receiver operating characteristic curves for determination of cut-off values are shown. The cut-off values were set to be well balanced between sensitivity and specificity and expressed as block nodes. The areas under the curves (variable, mean ± SE, p-value) were as follows: (A) HMGB1/DNA, 1.0 ± 0.0, p < 0.001; (B) C-peptide/DNA, 0.791 ± 0.106, p = 0.04; (C) PI+ area assay, 0.956 ± 0.043, p = 0.001; (D) TUNEL assay, 0.956 ± 0.047, p = 0.001; (E) ATP/DNA, 0.659 ± 0.126, p = 0.25; (F) stimulation index, 0.956 ± 0.042, p = 0.001; (G) insulin release against 2.8 mM glucose, 0.956 ± 0.042, p = 0.001; (H) insulin release against 20.0 mM glucose, 0.582 ± 0.142, p = 0.55.
Release of HMGB1 After Glucose Stimulation
Next we examined whether HMGB1 was released in response to glucose stimulation. After stimulating 100 intact mouse islets with 2.8 and 20.0 mmol/L glucose for 1 h, total released C-peptide levels were 0.0002 ± 0.0000 and 0.21 ± 0.04 pmol/μg DNA, respectively, a significant difference (p < 0.001). However, HMGB1 levels were not detected with either 2.8 or 20.0 mmol/L glucose stimulation, although a sufficient number of islets was used for the experiments (Fig. 4).
C-peptide was released by intact islets by glucose stimulation but HMGB1 was not released. After glucose-stimulated insulin secretion (GSIS) test using 100 intact mouse islets in each well, C-peptide levels (A) and HMGB1 levels (B) were measured (n = 5). The data were expressed as mean ± SD. *p < 0.001.
In Vitro Assays for Damaged Human Islets
At the end of this study, we analyzed each assay using three groups of human islets. The amount of total released HMGB1 in the control, hypoxia-24h, and hypoxia-48h human islets was 0.34 ± 0.07, 2.76 ± 0.56, and 5.01 ± 0.45 ng/μg DNA, respectively (Fig. 5A). The amount of released HMGB1 significantly increased according to the damage of islets.
Human damaged islets were analyzed by released HMGB1 levels, released C-peptide levels, PI+ area assay, ATP/DNA and GSIS tests. Three degrees of damaged human islets were analyzed by released HMGB1 levels (A), released C-peptide levels (B), PI+ area assay (C), ATP/DNA (D), and GSIS test (E, F). The data were expressed as mean ± SD. *p < 0.01, **p < 0.001.
The total amount of released C-peptide in the control, hypoxia-24h, and hypoxia-48h human islets was 10.8 ± 2.6, 8.3 ± 0.4, and 22.3 ± 6.7 pmol/μg DNA, respectively (Fig. 5B). Only the most severely damaged islets significantly increased the amount of released C-peptides. The PI+ area/islet area of control, hypoxia-24h, and hypoxia-48h human islets was 1.1 ± 0.4%, 17.1 ± 5.22%, and 26.9 ± 8.4%, respectively (Fig. 5C). The ratio of PI+ area/islet area significantly increased according to the damage of islets.
The ATP/DNA of control, hypoxia-24h, and hypoxia-48h human islets was 14.2 ± 2.2, 6.8 ± 2.3, and 10.5 ± 2.6 pmol/μg DNA, respectively (Fig. 5D). The only significant difference was found between the control islets and the hypoxia-24h islets (**p < 0.001).
The stimulation indices of control, hypoxia-24h, and hypoxia-48h human islets were 7.1 ± 1.6, 1.8 ± 0.3, and 1.2 ± 0.3, respectively (Fig. 5E). Only the control group had significantly high stimulation index compared to the other groups.
The insulin levels in response to low-glucose (2.8 mM) of control, hypoxia-24h, and hypoxia-48h human islets were 13.9 ± 0.8, 28.7 ± 4.5, and 42.0 ± 5.2 μIU/μg DNA, respectively (Fig. 5F, white columns). The basal insulin releases significantly increased according to the damage of islets.
The insulin levels in response to high glucose (20.0 mM) of control, hypoxia-24h, and hypoxia-48h human islets were 93.9 ± 24.3, 50.1 ± 8.7, and 51.4 ± 12.6 μIU/μg DNA, respectively (Fig. 5F, black columns). Only the control group had significantly higher stimulated insulin release compared to the other groups.
Discussion
In this study, we generated islet damage by hypoxia. We selected the hypoxic damage model because it mimics the damage occurring during the islet isolation process, because islets are frequently exposed to hypoxic conditions during organ procurement, organ transportation, islet isolation, and islet storage before transplantation. Mouse islets were divided into four groups: one control and three with different hypoxic conditions.
First, we examined the usefulness of the amount of released HMGB1 and C-peptide from mouse islets as an in vitro potency assay. The reason HMGB1 was chosen as a marker for detecting islet damage is that we have demonstrated mouse islets contained much higher levels of HMGB1 than that of other organs, including pancreas (15). Pancreas tissues contain about only 1% of endocrine cells, and most of the remained populations are the exocrine cells. Therefore, HMGB1 contents in pancreas tissues reveal nearly about that of exocrine cells. These results suggested that islets contain much higher HMGB1 levels than that of exocrine cells. Furthermore, we have demonstrated that proinflammatory cytokines-induced damaged mouse islets released HMGB1 extracellular milieu (15). For that reasons, we used HMGB1 as a damage marker for islets in the present study. The reason C-peptide was chosen as a marker for detecting islet damage is that Tiernberg et al. (32) and Eriksson et al. (3) have reported that released C-peptide levels were correlated to the islet damage in humans in vivo and in vitro. The amount of released HMGB1 significantly increased with time under hypoxic conditions; however, the amount of released C-peptide levels did not differ significantly between the control and hypoxia-6h groups (Fig. 2A, B). Taken collectively, these results demonstrate that the amount of released HMGB1 levels could reflect the degree of islet damage more sensitively than C-peptide levels. Furthermore, when comparing the noncured and cured group, there was a significant difference in the amount of released HMGB1 but not in the amount of released C-peptide (Table 1). Both HMGB1 and C-peptide were thought to be released from damaged islets; therefore, we speculated that HMGB1 would be more potent than C-peptide because C-peptide but not HMGB1 was also released by glucose stimulation. In this study, we demonstrated that C-peptide but not HMGB1 was released from intact islets by glucose stimulation (Fig. 4). Therefore, in islet culture medium, HMGB1 should be released only from damaged islet cells, while C-peptide should be released from both damaged islets and intact glucose-stimulated islets. For that reason, released HMGB1 levels are more specific for detecting the degree of damaged islets than released C-peptide levels. Furthermore, proinsulin should be a candidate as a marker for detecting islet damage. Therefore, we will compare the amount of released HMGB1 and proinsulin levels in the future study.
The PI+ area assay also demonstrated significant elevations based on the degree of hypoxia-induced damage (Fig. 2C), as well as excellent sensitivity, specificity, and predictive values (Table 1). However, the PI+ area assay requires skill in preparing histology samples and measuring the areas and thus is less objective than a test of the amount of released HMGB1. The TUNEL assay showed excellent sensitivity, specificity, and predictive values as well (Table 1), but has the same issues as the PI+ area assay. Therefore, the amount of released HMGB1 should be more suitable for a potency assay than the two staining assays.
The ATP/DNA assay could not detect a significant difference between the control and hypoxia-3h or hypoxia-6h groups (Fig. 2E). Furthermore, this assay could not detect a significant difference between the noncured and cured groups (Table 1). It is well known that ATP concentrations fluctuate rapidly (because of their short half-life) and are sensitive to transient changes in local conditions (20). This might be why the ATP/DNA assay showed poor results in detecting the degree of damaged islets.
For the GSIS test, the stimulation index dramatically decreased only after 3-h hypoxic conditions (Fig. 2H), resulting in relatively poor sensitivity, specificity, and predictive values to detect transplant outcomes (Table 1). The stimulation index seemed too sensitive to predict transplant outcomes.
The high-glucose (20.0 mmol/L) test yielded almost identical insulin levels among the four groups of damaged islets (Fig. 2G) and between the cured and noncured groups (Table 1). Therefore, stimulated insulin levels did not have enough sensitivity to detect transplant outcomes. Kim et al. reported that there was no significant difference in the high-glucose stimulated insulin release between cured (intact) and noncured (damaged) porcine islets, which concurs with our current study (8). On the other hand, basal insulin levels were highest at 6-h hypoxic injury and decreased at 24-h hypoxic injury (Fig. 2F). When intact islets are exposed to hypoglycemic conditions, insulin secretion is minimized to prevent further hypoglycemia. Damaged islets compromised this mechanism and then released insulin paradoxically. However, when islets suffered further severe damage, they were no longer able to release the insulin, as shown in the 24-h hypoxic damage model. Therefore, basal insulin levels during GSIS could not correlate with the degree of islet damage.
At the end of this study, we evaluated human islets using the same assays. We used the results of the PI+ area assay to determine the duration of hypoxic conditions for human islets. According to our results of PI+ area assay of intact human islets and hypoxia-6h human islets, there were no significant differences (data not shown). This means human islets are tougher against hypoxic conditions than that of mouse islets. When comparing mouse and human islets, the PI+ area assay showed almost the same levels of positivity between hypoxia-6h mouse islets and hypoxia-24h human islets, as well as between hypoxia-24h mouse islets and hypoxia-48h human islets. Therefore, for the study of human cells, we used three groups: control (normoxia-48h), hypoxia-24h, and hypoxia-48h.
Similar to the mouse experiments, only the amount of released HMGB1 and PI+ area assay demonstrated significant elevations based on the degree of hypoxia-induced damage. However, HMGB1 measuring has more objectivity than that of PI assay, because counting the number of PI+ cells and/or measuring PI+ areas are somewhat operator dependent. Therefore, released HMGB1 levels should be the most reliable and objective method for predicting clinical outcomes for islet transplantation. However, it is well known that human islets have huge variability. Therefore, we will check the correlations between the amount of released HMGB1 and in vivo assay in human islet preparations from different donors in the future study. HMGB1 measurement with clinical islet transplantation will reveal the actual cut-off value for the amount of released HMGB1 from human islets, and this is our ongoing project. In the present study, we used pure hand-picked islets, because we just wanted to focus on the islets specific at first. However, for the clinical use, the assessment methods need to evaluate impure human islets. Therefore, we will use impure islets and check the differences between pure and impure human islets in the future study.
The concentrations of HMGB1 can be measured by the standard ELISA technique; therefore, the other desirable features of a potency assay such as simplicity and operator independence are achievable. Considering these features, the measurement of released HMGB1 levels in islet culture medium should be one of the best potency assays for islet transplantation.
Footnotes
Acknowledgments
This study was supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (1R21DK090513-019) (to M.F.L.) and the Juvenile Diabetes Research Foundation (#3-2011-447 to M.T.). The authors thank Ms. Yoshiko Tamura, Ms. Ana M. Rahman, and Ms. Anne-Marie Brun for technical support and Ms. Cynthia Orticio for professional editing. The authors declare no conflicts of interest.