Reduced Ki67 Staining in the Postmortem State Calls Into Question Past Conclusions About the Lack of Turnover of Adult Human β-Cells

Because a reduction in β-cell mass is fundamental to the pathogenesis of diabetes, there has been great interest in the replicative capacity of human β-cells (1). A marked expansion of β-cell number occurs in childhood such that much of the β-cell mass found in adults is achieved by the age of 10 years (2). Many have argued that there is virtually no replication of β-cells in the adult pancreas (2–4), and pathologists have rarely found mitotic figures in β-cells of autopsied pancreases (5).

To examine β-cell replication, one study used autopsy samples from individuals who had received thymidine analogs in clinical trials before death; tissue sections were immunostained for these analogs to search for replication. The same study used isolated islets from organ donors to estimate the genomic DNA integration of atmospheric C¹⁴. It concluded that no new β-cells were born after the age of 30 years (6). Another approach examined the accumulation of intracellular lipofucsin bodies in β-cells, concluding that human β-cells live for many years and have very little, if any, replication in adult life (7). Other studies used samples taken from pancreases at autopsy or after cadaveric organ donation. Studies that used Ki67, a nuclear protein expressed by cells in the active cell cycle, found that the β-cells of adult pancreases usually had no Ki67 positivity (2,4,8). However, evidence of active replication was found in many of the pancreases of children younger than the age of 10 years (2,4). A puzzling aspect of these studies is that the β-cells of a number of pancreases from young children had no Ki67 positivity.

Others suggest that adult human β-cells are able to replicate. β-Cells in cultured human islets replicate when stimulated by lactogenic hormones, albeit considerably less so than rodent β-cells (9). More recent studies have also found a low rate of replication, which can be considerably enhanced by genetic manipulation of cell cycle genes (10). In our study of islets obtained from adult organ donors and transplanted into immunocompromised mice, there was clear Ki67 positivity of β-cells in the range of 0.2–0.7% (11). Others have reported similar results (10). Also, samples taken at the time of surgery have been reported to have Ki67 staining, ranging from 0.1% to 0.5%, of their β-cells (11,12).

Because of these discrepancies, we postulated that pancreases subjected to warm and cold ischemia during the process of autopsy and cadaveric donation would have a reduction in the appearance of replication markers. Therefore, we subjected young mice and newborn pigs to autopsy conditions of warm ischemia, followed by cold ischemia, and evaluated their pancreatic tissue using immunostaining for the replication markers Ki67 and proliferating cell nuclear antigen (PCNA).

The Joslin Institutional Animal Care and Use Committee approved all animal experiments.

Autopsy conditions that typically occur in humans were replicated in 4-week-old C57BL/6 mice. Mice were killed by CO₂ in a fed state or after a 24-h fast. The cadavers were left at room temperature (RT) for 3 h and then moved to 4°C. Pancreases were removed and fixed in 4% paraformaldehyde (PFA) immediately after death, after 3 h at RT, or after 3, 9, or 21 h at 4°C (Fig. 1).

Piglets (6-day-old Yorkshire; Parsons Farm, Hadley, MA) were killed by anesthetic overdose (ketamine and xylazine) and exsanguination of the carotid and jugular arteries. The cadavers were left at RT for 3 h, at which point the first samples were taken, and then moved to 4°C for 21 h, after which the second sample was taken from each pig.

Samples of mouse and pig pancreas were fixed in 4% PFA for 6 h and then stored in PBS at 4°C until processing and paraffin embedding. Immunostaining was performed on adjacent sections for 1) Ki67 and insulin and 2) PCNA and insulin. Sections were washed in 0.3% Triton X for 15 min, followed by antigen retrieval in citrate buffer (pH 6.0) for 15 min in the microwave. After blocking with an avidin-biotin blocking kit (Vector Laboratories, Burlingame, CA) and normal donkey serum at 1:50 (Jackson ImmunoResearch, West Grove, PA), rabbit anti-Ki67 (Abcam, Cambridge, MA) and rabbit anti-PCNA at 1:200 (Dako, Carpenteria, CA), primary antibodies were added. Sections were incubated overnight at 4°C. To identify the β-cells, sections were costained with guinea pig anti-insulin (1:500, Dako) for 1 h at RT. Incubation with secondary antibodies (biotinylated anti-rabbit IgG, Alexa Fluor 488 streptavidin; Alexa Fluor 594 anti-guinea pig IgG, 1:200; all Jackson ImmunoResearch) was for 1 h at RT. Slides were mounted with Vectashield with DAPI (Vector Laboratories).

Photographs of insulin-positive cells were systematically taken across the whole section using a LSM 710 confocal microscope (×25 objective, Zeiss). Zen software was used to quantify all cells stained for insulin or costained for insulin and Ki67 or PCNA on blinded images; at least 500 β-cells were counted per sample. To assess the peri-insular acinar cells under mouse autopsy conditions, the same images containing islets in at least 20 cells in cross section were blindly counted for Ki67 positivity and total acinar cells (determined by nuclear morphology visualized with DAPI); more than 1,000 peri-insular acinar cells were counted per sample. Glucagon-rich areas of the pancreas were counted as confirmed by immunostaining for glucagon and pancreatic polypeptide.

Data are presented as mean ± SEM. Statistical significance was tested with unpaired two-tailed Student t tests. For comparison of multiple time points, ANOVA, followed by the Bonferroni test, was used.

To replicate human autopsy conditions, we followed the schema shown in Fig. 1. With delayed fixation of the pancreas, there was a rapid marked fall in the frequency of Ki67 staining in β-cells compared with those fixed immediately (Fig. 2). At the time of death in the fed state, frequency was 9.3 ± 0.5% (n = 3) and was 6.9 ± 0.9% after a 24-h fast (n = 6, no significant difference). However, the frequency of Ki67-positive β-cells fell to 4.1 ± 0.9% (n = 6) after just 3 h at RT, to 2.7 ± 0.7% at 6 h (n = 5), to 1.6 ± 0.6% at 12 h (n = 5), and to 2.9 ± 0.8% (n = 5) at 24 h (ANOVA P < 0.001).

The replication marker PCNA was also evaluated on the adjacent sections of nonfasted mice after immediate fixation and after 24 h of warm/cold ischemia. For the immediate fixation group, values were 7.9 ± 1.4% (n = 8), very similar to the Ki67 values. At 24 h the PCNA was 5.2 ± 1.3% (n = 5), which failed to be statistically lower than after immediate fixation (P = 0.176).

To determine if the declines in Ki67 positivity were restricted to β-cells, acinar cells were evaluated on the same images. Peri-insular acinar cells were considered to be the acinar cells surrounding islets of at least 20 cells in cross section and no more than 5 acinar cells in any direction from islets of this size; this selection of acinar cells was done to provide consistency. Similar to what we saw in β-cells, Ki67 positivity in the peri-insular acinar cells fell from 8.26 ± 1.5 after immediate fixation to 2.69 ± 0.78 after 24 h of warm/cold ischemia (P = 0.006).

Pancreases taken from 6-day-old neonatal pigs were subjected to 3 h of warm ischemia at RT, followed by 21 h of cold ischemia at 4°C. Pancreatic tissue was not taken for immediate fixation, but only after 3 h of warm ischemia, at which point the β-cell Ki67 positivity was higher in the tail of the pancreas than in the body/head portion (17.9 ± 0.7 vs. 11.8 ± 0.7, n = 4 each; P = 0.001) (Fig. 3). We do not have an explanation for the difference between the tail and body/head regions, but because of glucagon staining of the islets in the body/head region, the difference cannot be attributed to sampling from the ventral lobe. Comparing 3 h of warm ischemia with the following 21 h of cold ischemia, the Ki67 positivity of β-cells fell markedly from 17.9 ± 0.7 to 5.2 ± 1.7 in the tail region (P = 0.002) and from 11.8 ± 0.7 to 3.6 ± 1.1 for the body/head region (P = 0.002). As shown in Fig. 4, the marked decline in Ki67-positive β-cells was similar in mice and newborn pigs. When the replication marker PCNA was evaluated on adjacent sections, the positivity of β-cells fell from 23.4 ± 1.9 at the end of the 3 h warm ischemia period to 3.7 ± 1.8 after 21 h of cold ischemia (P = 0.0003). The difference between the PCNA positivity of the β-cells in the tail and body/head portions for either condition was not significant.

Our data indicate that exposure of the pancreas to cold and warm ischemia, as occurs with autopsies and cadaveric pancreas collection, leads to reduced Ki67 protein staining in β-cells and acinar cells and PCNA staining in β-cells. Pathologists have long known that delays of fixation for even less than 2 h can lead to a marked reduction in the frequency of mitotic figures (13) and more recently to loss of Ki67 (14–16). These findings suggest that estimates of β-cell proliferation using pancreatic tissue from autopsies or cadaveric donors are unreliable.

If the measurements of Ki67 positivity in autopsied and cadaveric donor pancreases are unreliable, valid information should be obtained from freshly fixed pancreatic tissue. Indeed, data from surgical specimens, transplanted human islets, and cultured human islets support the concept that adult human β-cells have Ki67 positivity in the range of 0.1–0.7% (9,11,12,17).

Major questions for diabetes are: How much β-cell turnover occurs in the adult human pancreas and can this be harnessed to regenerate meaningful numbers of new β-cells? In addition to β-cell replication, there is growing evidence that new β-cells are produced from neogenesis. Some argue that the presence of β-cells in or adjacent to the duct epithelium supports this concept (18), but these cells could have been there since fetal life or early childhood. Others point out the increase in β-cells as singlets or in small clumps that can be seen with pregnancy (19) or insulin resistance (17), but it is possible that these resulted from replication. A stronger argument comes from the findings of increased numbers of cells costained for duct markers and islet hormones, in particular, the combination of insulin and cytokeratin, which suggests an active process (11,20,21).

What can we say about the rate of β-cell turnover in the adult human pancreas if replication and neogenesis are both taking place? We know there is some rate of β-cell death as seen in studies using markers such as TUNEL (22). This should be balanced by an equivalent rate of β-cell birth. However, TUNEL staining represents an event and not a rate. Ki67 staining might provide some insight, although positive staining does not necessarily represent the birth of a new cell. The appearance of Ki67 means that a cell has entered the cell cycle, yet division may not occur. Moreover, there can be death of replicating cells (23,24). Despite these caveats, because the length of the cell cycle may be ∼27 h (25), Ki67 staining might be seen for a slightly shorter time. Thus, if β-cells in an adult human had 0.4% Ki67 positivity and if there were no neogenesis or cell death, β-cell mass could double in less than a year.

In summary, the current study finds that the frequency of Ki67-stained β-cells falls markedly when mouse and neonatal pig pancreases are subjected to warm and cold ischemia. These findings suggests that pancreases obtained from autopsies or organ donors will have reduced Ki67 staining of β-cells, leading to the mistaken conclusion that adult human β-cells do not replicate. The rates of β-cell turnover in the human pancreas and the capacity for β-cell regeneration remain to be determined.

Funding. This study was supported by grants from JDRFhttp://dx.doi.org/10.13039/100000901 (G.C.W.), National Institutes of Healthhttp://dx.doi.org/10.13039/501100000272 grants R01-DK-066056 and DK-093909 (S.B.-W.) and P30-DK-036836 (Joslin Diabetes Research Center and its Advanced Microscopy Core), and the Diabetes Research & Wellness Foundation.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. B.A.S. and J.H.-L. researched data. S.B.-W. and G.C.W. conceived the project. B.A.S. and G.C.W. wrote the manuscript. All authors reviewed and edited the manuscript. G.C.W. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented as a poster at the 74th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 13–17 June 2014.