All mouse experiments were approved by the Institutional Animal Care and Use Committee at Shengjing Hospital of China Medical University (Animal Welfare Assurance). Surgeries were performed under ketamine/xylazine anesthesia, according the Principles of Laboratory Care, supervised by a qualified veterinarian. All efforts were made to minimize pain and suffering. Female Balb/C mice of 12 weeks of age were used in the current study. Four mice were analyzed in each experimental condition. 50 mg/kg Bromodeoxyuridine (BrdU, Sigma, China) was intraperitoneally injected two hours before sacrifice for labeling proliferating beta cells.

Bone marrow cells were isolated as has been previously described [31], [32].The mouse pancreas was perfused with 30 mg/dl collagenase (Sigma, China) from the common bile duct, and then incubated in a 37°C shaker for 30 minutes at a speed of 200 times per minute. After centrifugation, the pellet was resuspended in Histopaque (Sigma) of a gravity of 1.12 for a subsequent gradient centrifugation at 1200 rpm for 20 minutes. The suspension fraction was used for serial islet hand-pickings. Islet purity was assured by absence of exocrine cell markers Sox9 and Amylase. Purified islets were further digested with 10 µg/ml trypsin (Sigma) for 25 minutes to prepare single cell fraction for flow cytometry. The Aldefluor Kit (StemCell Technologies, China) was applied according to the manufacturer's instructions, to identify high ALDH enzymatic activity. Flow cytometry was performed using a FACSAria (Becton Dickinson) flow cytometer. Immunostaining on cytospun cells were performed as has been previously described [33].

Mouse pancreas were dissected out and fixed with 4% Paraformaldehyde for 10 hours, and then cyro-protected in 25% sucrose overnight. Samples were then sectioned in 6 µM and immunostained with guinea pig polyclonal insulin, rat polyclonal BrdU and Ki-67, rabbit polyclonal ALDH (ab23375), phosphohistone 3 (PHH3) and Ki-67 antibodies (all purchased from Abcam). BrdU staining needs antigen retrieval, which was performed by incubation of the slides with 1 mol/l HCl at room temperature for 45 minutes. Secondary antibodies were Cy3- and Cy2- conjugated antibodies for corresponding species (Jackson Labs, USA).

Four mice were measured in each experimental condition. Insulin staining was used to identify beta cells. The quantification of ALDH+ cells in all Ki-67+ beta cells, or ALDH+ cells in all BrdU+ beta cells, or ALDH+ cells in all PHH3+ beta cells, was based on 5 sections that were 100 µm apart from each other. 1500∼3000 beta cells were counted in each animal (Table S1).

All values are depicted as mean ± standard deviation from 4 individuals and are considered significant if p<0.05. All data were statistically analyzed using one-way ANOVA with a Bonferoni correction.

Aldefluor fluorescence represents precisely the ALDH activity, and has been widely used for the identification, evaluation, and isolation of stem and progenitor cells. Therefore, we first examined the aldefluor fluorescence in the isolated islets from pregnant mice at 3, 6, 9, 12, 15 and 18 days after pregnancy (G3, G6, G9, G12, G15 and G18, respectively), compared with non-pregnant mice (G0) of the same age (12 weeks of age) at the same period. Bone marrow cells were used as a positive control, while bone marrow cells pre-treated with 1.6 mM diethylaminobenzaldehyde (DEAB), a specific ALDH inhibitor, was used as a negative control (Fig. 1A). Isolated aldeflouor+ cells were immunostained positive for ALDH (Fig. 1B). While no aldefluor+ cells were detected in G0 islets, aldefluor+ cells were readily detected in the islets from the mice at different times in the pregnancy (Fig. 1C: G3: 0.89±0.14% in total islet cells, G6: 1.88±0.32%, G9: 3.45±0.65%, G12: 1.25±0.32%, G15: 0.82±0.21%, G18: 0.52±0.16%). To our knowledge, our study is the first to show that mouse islet cells upregulate ALDH activity during pregnancy.

To confirm the findings from the analysis on aldefluor fluorescence, and to identify which cell types in the islets from pregnant mice have increased ALDH activity, we performed immunostaining for ALDH and insulin on G9 mouse pancreas, since our data show the highest aldefluor fluorescence in G9 islets. G0 mouse pancreas was used as a control. We found that 4.5±0.7% of beta cells (identified by insulin staining) were positive for ALDH (Fig. 2A). In control G0 pancreas, although ALDH positivity was readily detected in some centroacinar and terminal ductal cells, we did not find any ALDH+ cells in the islets (Fig. 2B). These data suggest that aldefluor+ cells in the islets are predominantly beta cells.

We then aimed to examine the difference between ALDH+ beta cells and ALDH- beta cells. We first tested whether ALDH activity may relate to cell cycle activity, since it is well-known that beta cells substantially increase proliferation during pregnancy to increase beta cell mass in response to metabolic demand. Thus, we co-stained ALDH, insulin with a proliferation maker Ki-67, which labels all cells in G1, S and M phases of an active cell cycle. We found that 99.7±1.3% of ALDH+ beta cells were positive for Ki-67. Moreover, in all Ki-67+ beta cells, 32.5±2.2% were positive for ALDH (Fig. 2A-B). These data suggest that aldefluor+ cells in the islets are predominantly proliferating beta cells.

To determine precisely at which phase of a cell cycle, beta cells activate ALDH activity and thus become aldeflour-positive, we co-stained ALDH, insulin with additional proliferation markers, phosphohistone 3 (PHH3, a marker for M-phase proliferating cells) and Bromodeoxyuridine (BrdU, a marker for S-phase proliferating cells). BrdU was given to the mice two hours prior to sacrifice, since 2 hours is not long enough for a beta cell in late S-phase to pass M phase, split, then pass G0 phase and enter G1 phase again. Only a cell that does so within 2 hours in the current setting may affect our interpretation of data. This is very unlikely to occur, according to previous work that has demonstrated long G0 phase for beta cells [34]–[36]. We did not detect PHH+ALDH+ beta cells (less than 0.5%) (Fig. 3A), suggesting that beta cells may lose ALDH activity when they ender M phase of a cell cycle.

Similarly, we detect very few BrdU+ALDH+ beta cells (1.25±0.2%) (Fig. 3B), suggesting that beta cells essentially loss ALDH activity when they ender S phase of a cell cycle. Taken together, our data suggest that beta cells activate ALDH and become Aldefluor+ when they enter G1-phase of active cell cycle, but may downregulate ALDH when they leave G1-phase and enter S phase. These findings were then summarized and illustrated in Fig. 4.