Changes in β cell number and whole glucose levels during β cell cluster regeneration
To understand how and to what point the regeneration of β cell clusters proceeds, we observed the changes in the morphology of pancreatic islets and the cell number of β cell clusters after β cell ablation. In this study, β cell ablation was performed using the nitroreductase/metronidazole (NTR/Mtz) method 1 2. First, we ablated β cells from ins:Switch/gcga:GFP/ins:NTR lines via treatment with 24 h of Mtz between 3 and 4 days post-fertilization (dpf), and then observed the morphological changes of β cell clusters until 15 dpa (19 dpf) (Fig. 1A). We found that the first β cells appeared by 2 dpa and almost the same size and morphology as the β cell clusters of controls were regenerated by 15 dpa, (Fig. 1A). Next, we monitored changes in cell number in the β cell clusters using ins:H2BGFP/ins:NTR transgenic lines. Zebrafish have two types of islets, a principal islet, which is a single, huge islet, and secondary islets, which are multiple smaller islets (Fig. 1B and C). The principal islet develops during embryogenesis, followed by multiple secondary islets during postembryonic development 9 10. When we first counted the β cell number in the principal islet, we found the first β cells on 2 dpa (6 dpf). Although the β cell number increased until 3 dpa (7 dpf), the increase temporarily paused at 7 dpa (11 dpf). After 7 dpa, β cells started increasing again. Finally, the number recovered to the same level as that of the developing principal islet by 13 dpa (17 dpf) (Fig. 1D). When we next counted the β cell number in the secondary islet, we found no clear difference in the β cell number between the developing and regenerating pancreas during our period of observation (Fig. 1E). On the other hand, we found that the total number of β cells in the pancreas (sum of the number of β cells in the principal and secondary islets) showed the same pattern as that in the principal islet (Fig. 1D). Together, our results indicate that zebrafish β cell clusters can recover cell numbers to a normal level by 13 dpa. In addition, under our experimental conditions, we did not find that β cell ablation affected secondary islet development. Therefore, in the remainder of our experiments, we analyzed phenotypes in only the principal islet, but not secondary islets, after β cell ablation.
It has previously been reported that whole glucose levels recovered within several days after β cell regeneration 5. To confirm these results and to estimate when β cell cluster function was recovered, we next monitored whole glucose levels in ins:H2BGFP/ins:NTR lines treated with or without Mtz, from 3 to 17 dpf (Fig. 1G). We found that whole glucose levels became high immediately after β cell ablation. However, whole glucose levels peaked at 2 dpa (6 dpf), followed by recovery by 5 dpa (9 dpa) (Fig. 1G). These results imply that the functionality of the β cell clusters was recovered by 5 dpa. Interestingly, there is a gap in the recovery timing between glucose levels and cell number after β cell ablation (Fig. 1D, F, and G). This may suggest that β cell clusters undergo a two-step regeneration process, first regenerating functionality in a morphologically incomplete state by 5 dpa and then regenerating morphology by 13 dpa.
Regenerating β cells arise from cells in contact with α cells
It has previously been reported that some β cells arise from α cells (gcga:GFP positive glucagon-expressing cells) after β cell ablation, albeit at a low frequency 5 11. To confirm the relationship between regenerating β cells and α cells, we decided to investigate changes in α cell numbers after β cell ablation. As a result, the number of α cells itself was not significantly different from that of the control, although the size of the islet became somewhat smaller after β cell ablation (Fig. 2A and B). Using the gcga:GFP/ins:Switch/ins:NTR line, we next examined the correlation between regenerating β cells and α cells after β cell ablation. These results demonstrated that approximately 15% of ins:Switch-expressing regenerating β cells were gcga:GFP-expressing cells (Fig. 2C), but that the remaining 85% of the ins:Switch-expressing cells were adjacent to gcga:GFP-expressing cells (Fig. 2D). Interestingly, ins:Switch-expressing cells with gcga:GFP-expression were always adjacent to other gcga:GFP-expressing cells (Fig. 2C). These results indicated that all regenerating β cells arise from cells adjacent to α cells.
All regenerating β cells arise from Neurod1-expressing cells
Neurod1, which plays an important role in islet development, is known as a pan-endocrine marker 10 12 13, 14. To investigate the relationship between Neurod1-expressing cells and regenerating β cells, we examined changes in neurod1:EGFP expression and ins:Switch expression using neurod1:EGFP/ins:Switch/ins:NTR lines after β cell ablation. ins:Switch expression was always observed in neurod1:eGFP-expressing cells at both 2 and 3 dpa (6 and 7 dpf; Fig. 3A and B). These results suggest, as one possibility, that Neurod1-expressing cells are the main source of regenerating β cells. Therefore, we next generated neurod1:Cre transgenic lines and performed cell lineage-tracing experiments using the neurod1:Cre/ins:Switch/ins:NTR triple transgenic line. In the neurod1:Cre/ins:Switch (Tg(insulin:loxP:mCherrySTOP:loxP:H2B-GFP;cryaa:Cerulean)) line, β cells that arose from non-Neurod1-expressing cells expressed only mCherry, while β cells that arose from Neurod1-expressing cells expressed H2BGFP. These phenotypic analyses showed that all regenerating β cells expressed H2BGFP in a loxp-dependent manner at 13 dpa (17 dpf; Fig. 3C). These results support the possibility that regenerating β cells always arise from Neurod1-expressing cells (N1 cells).
The cell lineage of N1 cells that contribute to early and late regeneration is different
There have already been abundant N1 cells in the principal islet just prior to and after β cell ablation. To know if the number of N1 cells sufficient for β cell cluster regeneration already existed in the islets just after β cell ablation, we next generated a neurod1:CreERT2 transgenic line, and established neurod1:CreERT2/ins:Switch/ins:NTR lines for lineage tracing experiments. For these experiments, neurod1:CreERT2/ins:Switch/ins:NTR lines were treated with or without Mtz from 3 to 4 dpf and with 4-Hydroxy Tamoxifen (4OHT) from 4 to 5 dpf, and then their phenotypes were analyzed (Fig. 4A). We first observed phenotypes of developing β cells in neurod1:CreERT2/ins:Switch/ins:NTR lines without Mtz. We found that most of the β cells in the principal islet expressed H2BGFP at both 9 and 17 dpf, although faint mCherry signals remained in some H2BGFP-positive cells (Fig. 4B and C). In contrast, in secondary islets, β cells expressed mCherry, but not H2BGFP (Fig. 4 C). These results indicated that N1 cells, which are necessary for β cell development in the principal islet, are already present in the pancreas by 5 dpf, but that N1 cells, which are the source of β cell development in secondary islets, develop after 5 dpf. Next, we analyzed phenotypes of these transgenic lines after β cell ablation. After β cell ablation, all regenerating β cells expressed H2BGFP at 5 dpa in the principal islet, although mCherry signals remained in some cells (Fig. 4E and H). However, single mCherry-expressing cells appeared by 7 dpa in the pancreas of some zebrafish. The number of single mCherry-expressing cells increased after 9 dpa (Fig. 4F–H). On the other hand, the number of H2BGFP-expressing cells did not change between 5 and 13 dpa (Fig. 4E–H). In addition, the total number of H2BGFP- and single mCherry-expressing cells was shown to have a similar pattern to the results of Fig. 1D (Fig. 4H). These results suggest that all regenerating β cells are generated from N1 cells by 5 dpa (9 dpf), which are already present in the principal islet by 1 dpa (5 dpf), and that N1 cells, which are the source of β cells after 5 dpa, are newly generated after 1 dpa.
Most new Neurod1-expressing cells generate after 7 dpa
To investigate when new N1 cells, which become the source of β cells after 5 dpa, are generated we treated neurod1:CreERT2/ins:Switch/ins:NTR lines again with 4OHT at a time point between 4 and 11 dpa after treatment with Mtz and 4OHT, as previously investigated, and then analyzed mCherry and H2BGFP expression at 17 dpf (Fig. 5A). In these transgenic lines treated with a second treatment of 4OHT at 4 to 5 or 6 to 7 dpa, the number of H2BGFP-expressing cells and the area of mCherry-expressing cells did not significantly change compared to controls which did not receive a second 4OHT treatment (Fig. 5B, C, F, and G). However, in transgenic lines treated with a second 4OHT treatment at 8 to 9 or 10 to 11 dpa, the number of H2BGFP-expressing cells increased and the area of mCherry-expressing cells was reduced significantly (Fig. 5D, E, F, and G). These results suggest that most of the new N1 cells for regenerating the morphology of β cell clusters are produced after 7 dpa.