Development and Evaluation of a Laser Scanning Cytometry Analysis of Isolated Human Islets
Human islet preparations of 1000 IEQ allowed the preparation of more than 50 serial sections, giving opportunity to quantitate beta cells, apoptotic beta cells, islet hormone-producing cells, and non-islet cells in duplicate. The remaining sections/blocks provided opportunities for additional future analyses. The stained preparations were scanned with an iCys laser scanning cytometer. For detecting apoptotic beta cells, slides were stained for insulin and TUNEL. TUNEL-positive nuclei were detected by green fluorescence using the 488-nm laser, while insulin cytoplasmic/peripheral red fluorescence was detected using the 633-nm laser. shows a merged mosaic image of a whole islet preparation stained for TUNEL and insulin. A corresponding XY scattergram is shown in . The TUNEL marker was gated based on green “Max pixel”, and insulin was gated on red “Peripheral max” parameters to construct scattergram (), where cells double-stained for both TUNEL and insulin are shown in yellow, indicating beta cells undergoing apoptosis. The staining of cells was confirmed using the iCys “Gallery” module which allows visualizing individual cells (). A percentage of apoptotic beta cells against the total beta cell number is shown by the histogram () constructed from the scattergram ().
Evaluation of beta cell specific apoptosis
To compare results obtained by LSC with those obtained by conventional visual examination, images of samples analyzed by LSC were also captured by a conventional camera. TUNEL and insulin double-positive cells were counted visually, and the percentage of apoptotic beta cells calculated. Two human islet preparations that differed in beta cell and apoptotic beta cell numbers were selected in this study. Results were remarkably close: beta cell percentages obtained by visual vs. LSC () were 38.28±2.55% vs. 37.08±1.94% in sample HI-1 and 68.22±0.26 vs. 69.58±2.96% in sample HI-2, respectively (p=0.997), and those of apoptotic beta cells were 4.88±0.63% vs. 5.44±0.51% in HI-1 and 2.71±0.50% vs. 2.41±0.23% in HI-2, respectively (p=0.837), demonstrating the reliability of the LSC data.
Results of the LSC apoptotic beta cell assessments for 102 islet preparations from the SC-ICRC are summarized in . The mean apoptotic beta cell percentage was 2.85±0.4%, with a range between 0.27% and 18.3%. Eighty-five out of 102 preparations contained less than 5% apoptotic beta cells.
To test whether LSC analysis is capable of detecting cell damage of the same islet sample, apoptosis was induced by culturing human islets with a combination of inflammatory cytokines (TNFα, IFNγ, and IL1β) for 16 h (). This treatment reproducibly increased beta cell apoptosis: the percentage of apoptotic beta cells increased to 13.22±0.49 % from 4.45±1.03 % in untreated islets (n=3, p<0.002; ).
Detection of increased beta cell apoptosis by LSC following treatment with inflammatory cytokines
Analysis of cell composition in isolated human islets by LSC
To evaluate cells contained in islet preparations, sets of serial histology sections of each islet preparation were stained for the following sets of markers: insulin and amylase, glucagon and pancreatic polypeptide, and somatostatin and cytokeratin 19. Fluorescein(FITC/Cy2)-conjugated secondary antibodies were used for insulin, pancreatic polypeptide, and cytokeratin 19, while conjugates with Cy5/Alexa647 was used for amylase, glucagon, and somatostatin. Slides were scanned as described (). XY scattergrams of the corresponding scans are shown on the far left panels in . The markers were gated on red vs. green “Peripheral max” scattergrams in the second column (), and the corresponding histograms for each channel are shown in the third and fourth columns (). Individual positive cells were visualized using the “Gallery” module in the last column ().
Evaluation of cell compositions in islet cell preparations
Cell composition in the whole islet preparations, or calculated “per islet” was analyzed in a total of 102 islet preparations and the percentages of each cell type are shown in . In islets, the percentage of insulin-positive cells ranged 36.4%-72.4% with a mean value of 54.5±1.2%; glucagon-positive cells: 12.4%-48.2 % with a mean of 33.9±1.2%; somatostatin-positive cells: 2.9%-29.5% with a mean of 12.1±0.7%; and pancreatic polypeptide-positive cells: 0.5-4.2% with a mean of 1.5±0.2%.
These results were compared with the data for the islet cell composition obtained using LSC analysis of dissociated islets (). In this set of experiments, islets were dissociated, fixed on slides, immunostained for islet hormones and evaluated by LSC. The comparison of data between dissociated and undissociated islets of the same preparations showed a very close correlation (p=0.9936) for all islet hormones, with the only exception being pancreatic polypeptide (3.75±0.79% in dissociated islets vs. 1.52±0.53% in undissociated islets). A possible explanation for this difference could be the low number of pancreatic polypeptide-positive cells in the islet preparations.
Validation of reproducibility and reliability of LSC analysis
Reproducibility and reliability of LSC results were evaluated further using scanning and analysis of multiple sections of the same islet preparation, and analysis on the same section multiple times. Both tests were performed on islet sections stained for islet cell markers, as well as for insulin and TUNEL. shows LSC analysis results obtained from six randomly selected sections of one human islet sample (HI-3) stained for islet cell markers (4A), and from two islet samples (HI-3 and HI-4) stained for insulin and TUNEL (4B). There were no marked differences in results between slides, as indicated by the small coefficients of variation (CV<39% with ranges of 2.2% to 38.5%) of six readings for each staining , demonstrating that the LSC results reliably represent the islet cell composition and beta cell apoptosis in the islet preparation.
Reproducibility of the results obtained by LSC analysis
One potential problem with fluorescent staining is the decay of intensity, caused by long and multiple-excitation light exposures. As shown on the fluorescence intensity of all markers was found to remain stable throughout three separate scans of the same slide with low variability between the individual scans (CV<10% with ranges of 0.30% to 9.6%). Thus, our results indicate that multiple scans of the same slide are possible for repeated analysis.
Correlation between islet quality assessed by LSC and in vivo islet functionality
To determine whether there is a correlation between islet quality (as assessed by LSC) and islet functionality, LSC analysis results were compared with in vivo islet function following transplantation of 1200-1600 IEQ in diabetic NODscid mice. In vivo data were divided into two groups depending on average blood glucose levels measured between weeks 3-5: islet preparations that reversed diabetes (blood glucose<200) and islet preparations that did not reverse diabetes (blood glucose>200). Comparison of beta cell apoptosis and diabetes reversal vs. non-reversal showed highly significant differences (p<0.0001; n=59; ). In all of the islet preparations that achieved diabetes reversal in NODscid mice had less than 5% apoptotic beta cells. The predictive power of the beta cell apoptosis for transplant efficiency in mice was assessed by Receiver Operating Curves (ROC) analysis (), and was found to be highly significant. The area under the curve (AUC) was 0.8561 (sensitivity 100% and specificity 75%; 95% confidence interval: 0.7474-0.9647), where 1 is equivalent full predictability and 0.5 indicates no predictive relationship. None of the preparations with >4.2% apoptotic beta cells (16 out of 59) reversed diabetes in mice (negative predictive value 1.00); whereas over 80% of the preparations with <2.5% apoptotic beta cells (26 out of 59) reversed diabetes (positive predictive value 0.85). Analyses of other sets of LSC results, however, including beta cell content, did show much weaker correlations with the mouse transplant data (not shown).
Figure 5 Correlation of percentages of beta cell apoptosis and in vivo functionality of transplanted human islets into streptozotocin-diabetic NODscid mice (n=59): Laser scanning cytometry data for beta-cell-specific apoptosis were plotted against blood glucose (more ...)