Identification of Sox9-dependent acinar-to-ductal reprogramming as the principal mechanism for initiation of pancreatic ductal adenocarcinoma

Abstract

Tumors are largely classified by histologic appearance, yet morphologic features do not necessarily predict cellular origin. To determine the origin of pancreatic ductal adenocarcinoma (PDA), we labeled and traced pancreatic cell populations after induction of a PDA-initiating Kras mutation. Our studies reveal that ductal and stem-like centroacinar cells are surprisingly refractory to oncogenic transformation, whereas acinar cells readily form PDA precursor lesions with ductal features. We show that formation of acinar-derived premalignant lesions depends on ectopic induction of the ductal gene Sox9. Moreover, when concomitantly expressed with oncogenic Kras, Sox9 accelerates formation of premalignant lesions. These results provide insight into the cellular origin of PDA and suggest that its precursors arise via induction of a duct-like state in acinar cells.

Figure 1. SOX9 is expressed in human premalignant and malignant pancreatic lesions

(A–H) Immunohistochemistry for SOX9 and hematoxylin counterstain on a tissue microarray spotted with human pancreatic tissue cores. Representative images showing SOX9 expression in normal pancreatic ducts (A), chronic pancreatitis (B), mucinous cystic neoplasms (MCN) (C), intraductal papillary mucinous neoplasms (IPMN) (D), pancreatic intraepithelial neoplasia 1 (PanIN1) (E), PanIN2 (F), PanIN3 (G), and pancreatic ductal adenocarcinoma (PDA) (H). (I) Number of tissue cores within each phenotypic category displaying no, weak, or strong SOX9 staining intensity. Scale bars: 100μm.

Figure 2. Kras^G12D expression in acinar, but not in ductal/centroacinar cells, readily induces PanIN formation

(A) Sox9CreER;R26R^YFP and Ptf1a^CreER;R26R^YFP mice were injected once with tamoxifen (TM) at postnatal day (p) 10 and analyzed at p14 or at 8 to 17 months (mo) of age. (B) Quantification of Sox9⁺ or Cpa1⁺ cells expressing YFP at p14 (n=4). H&E (C, E, G) or Alcian blue and eosin (D, F, H) staining of pancreatic sections from 8–17-month-old control (C, D), Ptf1a^CreER;LSL-Kras^G12D;R26R^YFP (E, F), or Sox9CreER;LSL-Kras^G12D;R26R^YFP (G, H) mice reveals abundant Alcian blue⁺ PanINs only in Ptf1a^CreER;LSL-Kras^G12D;R26R^YFP mice. (I–L) Immunohistochemistry of YFP in 8–17-month-old mice shows expression of YFP in acinar cells in Ptf1a^CreER;R26R^YFP mice (I, arrows) and ductal/centroacinar cells (CACs) in Sox9CreER;R26R^YFP mice (K, arrowheads). PanINs in Ptf1a^CreER;LSL-Kras^G12D;R26R^YFP (J) and Sox9CreER;LSL-Kras^G12D;R26R^YFP (L) mice are YFP⁺, indicating an acinar or ductal/CAC origin, respectively. (M) Quantification of Alcian blue⁺ pancreatic area in 8–17-month-old mice (n=9 in Ptf1a^CreER;LSL-Kras^G12D mice; n=6 in Sox9CreER;LSL-Kras^G12D mice). The Alcian blue⁺ area in Sox9CreER;LSL-Kras^G12D;R26R^YFP mice was multiplied by 4.6 (normalized) to account for the greater total number of recombined cells in Ptf1a^CreER;R26R^YFP mice (see Figure S1L). (N) Schematic showing the predominantly acinar origin of PanINs after expression of oncogenic Kras. Values are shown as mean ± s.e.m. **P<0.01. Scale bars: 1 mm (C–H) and 100μm (I–L). See also Figure S1.

Figure 3. Acute pancreatitis promotes PanIN formation from Kras^G12D-expressing acinar cells, but not ductal/centroacinar cells

(A) Ptf1a^CreER;LSL-Kras^G12D;R26R^YFP, Sox9CreER;LSL-Kras^G12D;R26R^YFP and control mice were injected once with tamoxifen (TM) at postnatal day (p) 10. At 6 weeks (w) of age, mice were treated with two sets of caerulein (CA) or saline injections on alternating days and analyzed 48 hours (h) or 21 days (d) later. (B–D, F–H, J–L) H&E staining reveals occasional PanINs in 9-week-old Ptf1a^CreER;LSL-Kras^G12D;R26R^YFP mice (F, inset) and persistent ADM and PanINs 21d after CA (H), but normal pancreas morphology in Sox9CreER;LSL-Kras^G12D;R26R^YFP and control mice (D, L). Asterisks denote PanINs. (E, I, M) Alcian blue and eosin staining of pancreatic sections from mice 21d after CA treatment. (N) Quantification of Alcian blue⁺ pancreatic area reveals a significant increase in PanINs after CA in Ptf1a^CreER;LSL-Kras^G12D;R26R^YFP (n=4), but not in Sox9CreER;LSL-Kras^G12D;R26R^YFP mice (n=5). (O) Schematic showing that pancreatic injury promotes PanIN formation from Kras^G12D-expressing acinar, but not ductal/centroacinar cells. N.S., not significant. Values are shown as mean ± s.e.m. **P<0.01. Scale bars: 100μm. See also Figure S2.

Figure 4. Sox9 is necessary for and accelerates Kras^G12D-induced PanIN formation

(A, B) Immunohistochemistry shows Sox9 expression in ductal/centroacinar (CAC) (A, arrowheads), but not acinar cells in control mice. In 2-month-old Ptf1a^Cre;LSL-Kras^G12D mice, Sox9 is also detected in some acinar cells (B, arrows). (C–D) Co-immunofluorescence staining for Sox9, CK19 and Cpa1 confirms Sox9 expression in ductal/CACs (C′, arrowhead points to CAC) in control mice and shows Sox9⁺Cpa1⁺CK19⁻ acinar cells in Ptf1a^Cre;LSL-Kras^G12D mice (D′, arrowheads). (E–Q) Tamoxifen (TM) was administered to Ptf1a^CreER;Sox9^f/f;R26R^YFP, Ptf1a^CreER;LSL-Kras^G12D;Sox9^+/+;R26R^YFP and Ptf1a^CreER;LSL-Kras^G12D;Sox9^f/f;R26R^YFP mice at postnatal day (p) 21, 23 and 25 to simultaneously ablate Sox9 and induce Kras^G12D in acinar cells. Mice were analyzed at 8–12 months (mo) of age. H&E (F–H) and Alcian blue staining (I–K) shows almost no PanINs after Sox9 deletion. (L–Q) Immunohistochemistry reveals expression of YFP in PanINs in both Ptf1a^CreER;Kras^G12D;Sox9^+/+;R26R^YFP and Ptf1a^CreER;LSL-Kras^G12D;Sox9^f/f;R26R^YFP mice (M, N). Sox9 expression in PanINs in the Sox9^f/f background indicates lack of Sox9^f recombination (Q). H&E (R–T) and Alcian blue staining (U–W) reveals abundant ADM and PanINs in Ptf1a^Cre;LSL-Kras^G12D;Sox9^OE, but not in Ptf1a^Cre;LSL-Kras^G12D or control mice at 3–4 weeks of age. Immunohistochemistry for HA (X–Z) and Sox9 (AA–AC) shows Sox9⁺ PanINs originating from cells that recombined the Sox9^OE transgene. (AD, AE) Quantification of Alcian blue⁺ pancreatic area reveals a significant reduction of PanINs after Sox9 deletion (n=9) and conversely, an increase after Sox9 misexpression (n=5). (AF) Immunofluorescence staining shows abundant CK19 and little Cpa1 expression in 4-week-old Ptf1a^Cre;LSL-Kras^G12D;Sox9^OE mice. Arrow points to CK19⁺Cpa1⁺ cell. Values are shown as mean ± s.e.m. *P<0.05 and ***P<0.001. Scale bars: 25μm (A–B), 50 μm (C, D), 100 μm (L–Q, X–AC, AF), and 500μm (F–K, R–W). See also Figure S3.

Figure 5. Sox9 misexpression induces ductal genes

(A–F) Co-immunofluorescence staining for Cpa1 and CK19 shows CK19⁺Cpa⁺ cells in Ptf1a^Cre;Sox9^OE mice (B, D, F), but not in controls (A, C, E). (G–H) Immunohistochemistry for CK19 reveals greater staining intensity in 6-month-old Ptf1a^Cre;Sox9^OE than in control mice. (I) QRT-PCR analysis of Mist1, amylase and CK19 in whole pancreas RNA from Ptf1a^Cre;Sox9^OE and control mice (n=5). (J, K) H&E staining shows acinar clusters with dilated lumens in 6-month-old Ptf1a^Cre;Sox9^OE mice (K, arrows). Values are shown as mean ± s.e.m. *P<0.05 and **P<0.01. Scale bars: 50μm.

Figure 6. Sox9 promotes persistent acinar-to-ductal reprogramming (ADR) and formation of mucinous metaplastic lesions after acute pancreatitis

(A) Six-week-old Ptf1a^Cre;Sox9^OE and control mice were treated with two sets of caerulein (CA) or saline injections on consecutive days and analyzed 48 hours (h) or 7 days (d) later. Saline-treated mice were analyzed at 21d. (B–G) H&E staining reveals persistent acinar-to-ductal metaplasia (ADM) in Ptf1a^Cre;Sox9^OE mice (G). (H–M) Co-immunofluorescence staining for CK19 and Cpa1 shows a few CK19⁺Cpa1⁺ (L, arrow) 48h after CA and mainly CK19⁺Cpa1⁻ cells after 7d (M) in Ptf1a^Cre;Sox9^OE mice. (N–S) Immunohistochemistry for Sox9 and Alcian blue staining shows Sox9⁺Alcian blue⁺ mucinous metaplastic lesions in Ptf1a^Cre;Sox9^OE mice (S) but not in control mice (P) 7d after CA. (T) Schematic summarizing the phenotypes of Sox9 misexpressing mice in the presence and absence of Kras^G12D or acute pancreatitis. w, weeks. Scale bars: 100μm (B–G) and 50μm (H–S). See also Figure S4.

Figure 7. Sox9-deficient acinar cells expressing Kras^G12D can undergo persistent acinar-to-ductal reprogramming (ADR) after acute pancreatitis, but do not progress into PanINs

(A) Ptf1a^CreER;Sox9^+/+, Ptf1a^CreER;LSL-Kras^G12D;Sox9^+/+ and Ptf1a^CreER;LSL-Kras^G12D;Sox9^f/f mice were injected with tamoxifen (TM) at postnatal day (p) 21, 23 and 25. At 6 weeks (w) of age, mice were treated with two sets of caerulein (CA) or saline injections on alternating days and analyzed 48 hours (h) or 21 days (d) later. (B–D, G–I, L–N) H&E staining shows persistent acinar-to-ductal metaplasia (ADM) in Ptf1a^CreER;LSL-Kras^G12D;Sox9^+/+ and to a lesser extent also in Ptf1a^CreER;LSL-Kras^G12D;Sox9^f/f mice (I, N, arrows). PanINs are only present in Ptf1a^CreER;LSL-Kras^G12D;Sox9^+/+ mice (I, arrowhead). (E, J, O) Co-immunofluorescence staining for CK19, Cpa1 and Sox9 reveals CK19⁺Cpa1⁻ duct-like cell clusters in Ptf1a^CreER;LSL-Kras^G12D mice in the presence and absence of Sox9 (J, O, arrows). (F, K, P) Alcian blue and eosin staining shows PanINs in Ptf1a^CreER;LSL-Kras^G12D;Sox9^+/+ mice (K), but not in Ptf1a^CreER;LSL-Kras^G12D;Sox9^f/f mice (P). Arrows in K, P point to ADM and arrowheads in J, K to PanINs. (Q) Quantification of Alcian blue⁺ pancreatic area 21d after CA (n=4–5). (R) Schematic summarizing the phenotype observed in Sox9 loss-of-function experiments in the presence of Kras^G12D and acute pancreatitis. Values are shown as mean ± s.e.m. *P<0.05. Scale bars: 50μm. See also Figure S5.