Precommitment low-level Neurog3 expression defines a long-lived mitotic endocrine-biased progenitor pool that drives production of endocrine-committed cells

Abstract

The current model for endocrine cell specification in the pancreas invokes high-level production of the transcription factor Neurogenin 3 (Neurog3) in Sox9(+) bipotent epithelial cells as the trigger for endocrine commitment, cell cycle exit, and rapid delamination toward proto-islet clusters. This model posits a transient Neurog3 expression state and short epithelial residence period. We show, however, that a Neurog3(TA.LO) cell population, defined as Neurog3 transcriptionally active and Sox9(+) and often containing nonimmunodetectable Neurog3 protein, has a relatively high mitotic index and prolonged epithelial residency. We propose that this endocrine-biased mitotic progenitor state is functionally separated from a pro-ductal pool and endows them with long-term capacity to make endocrine fate-directed progeny. A novel BAC transgenic Neurog3 reporter detected two types of mitotic behavior in Sox9(+) Neurog3(TA.LO) progenitors, associated with progenitor pool maintenance or derivation of endocrine-committed Neurog3(HI) cells, respectively. Moreover, limiting Neurog3 expression dramatically increased the proportional representation of Sox9(+) Neurog3(TA.LO) progenitors, with a doubling of its mitotic index relative to normal Neurog3 expression, suggesting that low Neurog3 expression is a defining feature of this cycling endocrine-biased state. We propose that Sox9(+) Neurog3(TA.LO) endocrine-biased progenitors feed production of Neurog3(HI) endocrine-committed cells during pancreas organogenesis.

Keywords: Neurog3; endocrine-biased; mitotic; progenitor.

Figure 1.

The Neurog3^Protein+ population is comprised of mitotic Neurog3^pLO and post-mitotic Neurog3^pHI cells. E12.5 (A) and E14.5 (B) pancreatic epithelium showing Muc1, Sox9, and Neurog3. White and red arrowheads indicate Sox9⁺ Muc1⁺ Neurog3^pLO cells and Sox9⁻ Muc1⁻ Neurog3^pHI, respectively. (C) Average fluorescence intensity of nuclear Neurog3 signal in Sox9⁺ Neurog3^Protein+ versus Sox9⁻ Neurog3^Protein+ cells. n = 9 cryosections; N = 3 pancreata at E12.5 and E14.5. (*) P = 2 × 10⁻⁵; (**) P = 0.0002. (D) Percentage of Sox9⁺ Neurog3^pLO versus Sox9⁻ Neurog3^pHI cells at E12.5 (n = 1114; N = 3), E14.5 (n = 3797; N = 3), and E16.5 (n = 4374; N = 3). (*) P = 0.0895; (**) P = 3 × 10⁻⁵; (***) P = 0.0001. E12.5 (E,G) and E14.5 (F,H) pancreatic epithelium showing pHH3, Neurog3, Muc1, and DAPI or pHH3, Neurog3, and E-cadherin. Red arrowheads indicate intraepithelial pHH3⁺ Neurog3^pLO cells. (I) Mitotic index of Muc1⁺ Neurog3^pLO cells versus Muc1⁻ Neurog3^pHI cells at E12.5 (n = 1546) and E14.5 (n = 10080). (J) Average fluorescence intensity of nuclear Neurog3 signal in Muc1⁺ Neurog3^Protein+ pHH3⁺ cells versus Muc1⁻ Neurog3^Protein+ pHH3⁻ cells at E12.5 and E14.5, calculated as in C. Data are mean ± SEM. (*) P = 0.0005; (**) P = 0.0087.

Figure 2.

Neurog3^TA Sox9⁺ cells, comprising Neurog3^TA.pLO and Neurog3^TA.pUD cells, represent a mitotic Neurog3^TA.LO endocrine-biased progenitor pool. E12.5 (A,C) and E14.5 (B,D) pancreatic epithelium showing EYFP, Neurog3, and Sox9 or EYFP, Neurog3, Muc1, and pHH3. (A,B) Red and green arrowheads indicate Sox9⁺ EYFP⁺ Neurog3^TA.pLO cells and Sox9⁺ EYFP⁺ Neurog3^TA.pUD cells, respectively. (C,D) Red and green arrowheads indicate pHH3⁺ Muc1⁺ Neurog3^TA.pLO and pHH3⁺ Muc1⁺ Neurog3^TA.pUD cells, respectively.

Figure 3.

The Neurog3^RG1 transgenic reporter marks the mitotic Neurog3^TA.LO progenitor population. E12.5 (A,C) and E14.5 (B,D) pancreatic epithelium showing H2B^mCherry, GFP^GPI, Sox9, and Neurog3 or H2B^mCherry, GFP^GPI, Muc1, and pHH3. (A,B) Red and green arrowheads indicate Sox9⁺ Neurog3^TA.pLO and Sox9⁺ Neurog3^TA.pUD cells (seen by low H2B^mCherry signal from Neurog3^RG1), respectively. (C,D) Red arrowheads indicate Muc1⁺ Neurog3^RG1+ pHH3⁺ Neurog3^TA.LO cells. (E) The percentage of Neurog3^RG1+ Sox9⁺ Neurog3^TA.LO versus Neurog3^RG1+ Sox9⁻ Neurog3^TA.HI cells at E12.5 (n = 585; N = 3) and E14.5 (n = 6334; N = 3). Data are mean ± SEM. (*) P = 1 × 10⁵.

Figure 4.

Division of Neurog3^TA.LO progenitors in real time. Division of a Neurog3^TA.LO cell into two delaminating Neurog3^TA.HI cells (A) or two Neurog3^TA.LO progenitors (B). Images are screen captures from Fig. 4B; Supplemental Movie 2. White and red arrowheads indicate a Neurog3^TA.LO division event and apical attachment, respectively.

Figure 5.

The mitotic Neurog3^TA.LO population is prevalent under Neurog3 heterozygous and strong hypomorphic conditions. E12.5 (A,C,E,G) and E14.5 (B,D,F,H) pancreatic epithelium showing EGFP, Neurog3, and Sox9 or EGFP, Muc1, and pHH3 from Neurog3^EGFP/+ and Neurog3^EGFP/FL embryos. (A–D) Red and green arrowheads indicate Sox9⁺ Neurog3^TA.pLO and Sox9⁺ Neurog3^TA.pUD cells (marked by low EGFP signal from Neurog3^EGFP), respectively. (E–H) Red arrowheads indicate Muc1⁺ EGFP⁺ pHH3⁺ Neurog3^TA.LO cells.

Figure 6.

Limiting Neurog3 expression expands the Neurog3^TA.LO population at the expense of Neurog3^TA.HI cells. (A) The percentage of Sox9⁺ Neurog3^TA.pLO versus Sox9⁺ Neurog3^TA.pUD cells in wild-type (Neurog3^RG1) (see Fig. 3E for n), Neurog3^EGFP/+ (E12.5, n = 168, N = 3; E14.5, n = 2425, N = 3), and Neurog3^EGFP/FL (E12.5, n = 1237, N = 3; E14.5, n = 5818, N = 3) pancreata. (*) P = 0.0523; (**) P = 0.00067; (***) P = 8.3 × 10⁻⁵; (****) P = 4 × 10⁻⁶; (*****) P = 2.5 × 10⁻⁵. (B) The percentage of Sox9⁺ Neurog3^TA.LO versus Sox9⁻ Neurog3^TA.HI cells in Neurog3^EGFP/+ and Neurog3^EGFP/FL (see A for n) pancreata. (*) P = 8 × 10⁻⁵; (**) P = 8 × 10⁻⁸; (***) P = 2.4 × 10⁻⁷. (C) The percentage of Sox9⁺ cells that are Neurog3^TA.LO in wild-type (Neurog3^RG1) (E12.5, n = 4497, N = 3; E14.5, n = 11930, N = 4), Neurog3^EGFP/+ (E12.5, n = 4718; E14.5, n = 15903), and Neurog3^EGFP/FL (E12.5, n = 4656; E14.5, n = 15671) pancreata. (*) P = 0.00015; (**) P = 0.0009; (***) P = 0.0022; (****) P = 2 × 10⁻⁷; (*****) P = 1 × 10⁻⁶. Data are mean ± SEM.

Figure 7.

Reducing Neurog3 expression increases the mitotic index of Sox9⁺ Neurog3^TA.LO progenitors. (A, left) The mitotic index (percentage pHH3⁺) of Muc1⁺ Neurog3^TA.LO cells under wild-type (Neurog3^RG1) (E12.5, n = 569, N = 6; E14.5, n = 2401, N = 6), Neurog3^EGFP/+ (E12.5, n = 362, N = 3; E14.5, n = 904, N = 3), and Neurog3^EGFP/FL (E12.5, n = 783, N = 4; E14.5, n = 3905, N = 3) conditions. (Right) The mitotic index of Muc1⁺ Neurog3⁻ cells in wild-type (Neurog3^RG1) (E12.5, n = 1200 cells, N = 3; E14.5, n = 7490, N = 3), Neurog3^EGFP/+ (E12.5, n = 1003, N = 3; E14.5, n = 2737, N = 3) and Neurog3^EGFP/FL (E12.5, n = 2226, N = 4; E14.5, n = 5217, N = 3) pancreata. Data are mean ± SEM. (*) P = 0.0138; (**) P = 0.0207; (***) P = 0.0018. (B) Model depicting the behavior of Neurog3^TA.LO cells in the pancreatic epithelium.