Lfng-expressing centroacinar cell is a unique cell-of-origin for p53 deficient pancreatic cancer

Lfng expression in the adult pancreas is restricted to the centroacinar compartment

To characterize Lfng-expressing cells in the adult pancreas, we made use of a Lfng-RFP/CreERT2 mouse strain expressing RFP and tamoxifen-inducible Cre under the Lfng promoter. Anti-RFP immunostaining in Lfng-RFP/CreERT2 mice showed similar but more restricted Lfng expression in the exocrine pancreas compared to X-gal staining in Lfng^lacZ knock-in mice [10]. The vast majority of RFP⁺ cells are located near the centroacinar compartment and display centroacinar rather than acinar cell morphology (Fig. 1A, B). Bearing CD24⁺CD44^– surface marker, RFP⁺ cells account for about 3% of lineage depleted pancreatic cells (Fig. 1C–E). We performed lineage tracing for Lfng-expressing cells. While anti-YFP staining was completely negative in R26^YFP pancreas (Fig. 1F), R26^YFP;Lfng-RFP/CreERT2 pancreas showed YFP⁺ cells (descendants of Lfng-expressing cell) either remaining centroacinar (Fig. 1G) or becoming acinar (Fig. 1H) under homeostatic conditions. Quantitative lineage tracing with or without caerulein treatment showed comparable number of RFP⁺ cells as well as YFP⁺ cells, suggesting that caerulein-induced exocrine injury did not alter self-renewal or differentiation of Lfng-expressing cells. Next, we tested the effect of Lfng deletion on the fate of Lfng-expressing cells. R26^YFP;Lfng^fl/fl;Lfng-RFP/CreERT2 mice exhibited significantly increased number of RFP⁺ cells and almost a 10-fold increase of YFP⁺ cells compared to the R26^YFP;Lfng-RFP/CreERT2 mice, irrespective of caerulein treatment (Fig. 1I–R). Many of the YFP⁺ cells in R26^YFP;Lfng^fl/fl;Lfng-RFP/CreERT2 mice exhibited acinar location and morphology (Fig. 1N, P). Thus, deletion of Lfng appeared to enhance proliferation and differentiation of Lfng-expressing centroacinar cells to acinar cells. Progenitor subpopulations in the centroacinar and terminal duct cells express high level of ALDH1 enzymatic activity [27, 28]. We found no overlapping of either RFP or YFP with Aldh1a1 (Fig. 1S, T), though co-staining of YFP and Aldh1b1 was observed in a subset of centroacinar cells in R26^YFP;Lfng^fl/fl;Lfng-RFP/CreERT2 mice (Fig. 1U, V). Interestingly, Lfng-expressing cells express high level of CD24 (Fig. 1E), a surface marker for pancreatic progenitors identified using human embryonic stem cells [29]. Additionally, CD24^high but not CD24^low subpopulation in the mouse pancreatic ducts was able to form organoids [30]. These results suggest that Lfng-expressing cells may represent a progenitor subpopulation in the centroacinar compartment.

A, B Anti-RFP immunostaining in the pancreas from wild type and Lfng-RFP/CreER mice at 3 months of age. Arrow: positive staining in a centroacinar cell. Representative flow cytometry analysis for RFP⁺ cells in lineage-depleted pancreatic cells from adult wild type (C) and Lfng-RFP/CreER (D, E) mice. F–H Anti-YFP immunostaining in the pancreas from R26^YFP and R26^YFP;Lfng-RFP/CreER mice 6 weeks after tamoxifen induction at 6 weeks of age. Arrows: positive staining in a centroacinar cell (G) and acinar cell (H). I–L Anti-RFP immunostaining in the pancreas from R26^YFP;Lfng-RFP/CreER and R26^YFP;Lfng^fl/fl;Lfng-RFP/CreER mice 6 weeks after tamoxifen induction at 6 weeks of age, with or without caerulein treatment, which started at one week post-tamoxifen and lasted for 3 weeks. Arrows: positive staining in centroacinar cells. M–P Anti-YFP immunostaining in the pancreas from R26^YFP;Lfng-RFP/CreER and R26^YFP;Lfng^fl/fl;Lfng-RFP/CreER mice 6 weeks after tamoxifen induction, with or without caerulein treatment. Q Quantitation of RFP⁺ cells depicted in I-L. *p < 0.05; **p < 0.01; ns: non-significance. R Quantitation of YFP⁺ cells depicted in M-P. ****p < 0.0001; ns: non–significance. S Double immunofluorescent staining for RFP (red) and Aldh1a1 (green) in R26^YFP;Lfng^fl/fl;Lfng-RFP/CreER pancreas. Arrows: an Aldh1a1⁺ cell (left) and an RFP⁺ cell (right). T Double immunofluorescent staining for YFP (green) and Aldh1a1 (red) in R26^YFP;Lfng^fl/fl;Lfng-RFP/CreER pancreas. Arrows: an YFP⁺ cell (top) and an Aldh1a1⁺ cell (bottom). U Double immunofluorescent staining for YFP (green) and Aldh1b1 (red) in R26^YFP;Lfng^fl/fl;Lfng-RFP/CreER pancreas. Arrowhead: co-staining of Aldh1b1 and YFP in a centroacinar cell. V Quantitation of cells with co-staining of YFP and Aldh1b1. Scale bars: 25 μm.

Full size image

Lfng-expressing centroacinar cell can be targeted to form PanIN and progress to PDAC

Stem/progenitor-like cells are thought to be the cellular origin for pancreatic tumors. We wondered whether Lfng-expressing cells could serve as a cell-of-origin for PDAC. Lfng-RFP/CreERT2-directed Kras^G12D alone rarely caused PanIN, but in conjunction with caerulein treatment, resulted in extensive PanIN (Fig. 2A–E). PanIN lesions in R26^lacZ;Kras;Lfng-RFP/CreERT2 mice stained positive for LacZ, confirming they were derived from Lfng-expressing cells (Fig. 2G). LacZ⁺ stromal cells (arrows in Fig. 2H) may represent descendants from Lfng-expressing cells that have undergone epithelial-mesenchymal transition (EMT) or PanIN-associated fibroblasts that have turned on Lfng expression, whereas LacZ⁻ PanIN-like lesions (arrowhead in Fig. 2H) could be pseudo-PanIN lesions induced by caerulein. RFP as a surrogate for Lfng expression was detected in a subset of PanIN cells (Fig. 2I) and in limited number of stromal cells (Fig. 2J). Next, we generated p53^fl/fl;Kras;Lfng-RFP/CreER (hereinafter referred to as PKC-Lfng) mice to drive Kras^G12D expression coupled with p53 deletion in Lfng-expressing cells. These mice developed pancreatic tumors with metastasis in less than 2 months following tamoxifen induction (Fig. 2K–M). Anti-YFP staining of the R26^YFP;p53^fl/fl;Kras;Lfng-RFP/CreER (R26^YFPPKC-Lfng) pancreas was positive in all tumor cells (Fig. 2N) and negative in native duct (arrowhead in Fig. 2O), confirming that tumor cells had originated from Lfng-expressing cells. There were YFP⁻ PanIN-like lesions (arrow in Fig. 2O), likely representing pseudo-PanIN lesions induced by adjacent PDAC [31]. YFP staining was noted within the islet in R26^YFPPKC-Lfng mice (Fig. 2P), but not in R26^YFP;Lfng-RFP/CreER mice (data not shown). These cells were insulin-negative (Fig. 2Q), suggesting tumor cell invasion into the islet rather than conversion of targeted cell to endocrine cell. Anti-RFP staining revealed Lfng expression in PanIN and ductal tumor cells, and occasionally in mesenchymal-like tumor cells (Fig. 2R–T).

A–D Hematoxylin and eosin staining of pancreas from Lfng-RFP/CreER and Kras;Lfng-RFP/CreER mice 20 weeks after tamoxifen injection at 4 weeks of age, with or without 4 weeks of daily caerulein treatment immediately following tamoxifen administration. E Quantification of ADM/PanIN lesions in Kras;Lfng-RFP/CreER mice depicted in (B, D). **p < 0.01. F–H X-Gal staining of the pancreas from R26^LacZ and R26^LacZ;Kras;Lfng-RFP/CreER mice at 9 weeks after tamoxifen injection at 4 weeks of age. These mice also received 4 weeks of daily caerulein treatment immediately following tamoxifen administration. Arrows: positive staining in the stroma. Arrowhead: pseudo-PanIN. I, J Anti-RFP immunostaining in the pancreas from Kras;Lfng-RFP/CreER mice 16 weeks after tamoxifen injection followed by 4 weeks of caerulein treatment. K Representative gross pathology of the pancreatic tumor in p53^fl/fl;Kras;Lfng-RFP/CreER (PKC-Lfng) mice at 7 weeks post-tamoxifen. Arrowhead: tumor of the entire pancreas. Arrow: metastatic growth on the liver. L, M Histology of the pancreatic tumor and liver metastasis shown in (K). * area of hepatocytes. N–P Anti-YFP immunostaining in the pancreas of 10-week-old R26^YFP;p53^fl/fl;Kras;Lfng-RFP/CreER (R26^YFPPKC-Lfng) mice treated with tamoxifen at 4 weeks old. Arrow: YFP-negative PanIN-like lesion. Arrowhead in O: YFP-negative native duct. Arrowhead in P: YFP-positive cell within an islet. Q Double immunofluorescent staining for insulin (red) and YFP (green) in the R26^YFPPKC-Lfng pancreas. Arrows: YFP-positive cells. R–T Anti-RFP immunostaining in pancreatic tumors from R26^YFPPKC-Lfng mice. S is high-magnification image of the square area in R. Arrows: RFP-positive tumor cells in poorly differentiated area. Scale bars: 50 μm.

Full size image

We treated PKC-Lfng mice with caerulein immediately following tamoxifen induction at one month of age. Almost two-thirds of PKC-Lfng mice died during one month of daily caerulein treatment. Surprisingly, PKC-Lfng mice that survived the caerulein treatment lived much longer compared to those without caerulein treatment (Supplemental Fig. 1A). Congruently, 3 out of 7 such mice showed very few PanINs (no PDAC) at 3 months post tamoxifen (Supplemental Fig. 1B). Of note, caerulein treatment started at 1 month post tamoxifen, a time point when PanIN had formed, had no effect on tumor onset (Supplemental Fig. 1C). Thus, while caerulein treatment accelerates Kras-induced PanIN from p53-intact Lfng-expressing cells, it may prevent PanIN formation from p53-deficient Lfng-expressing cells.

We also generated p53^R172H;Kras;Lfng-RFP/CreER mice to drive Kras^G12D and p53^R172H expression in Lfng-expressing cells. These mice developed very few PanIN without caerulein treatment (Supplemental Fig. 2A–C) but succumbed to PDAC with caerulein treatment following tamoxifen induction (Supplemental Fig. 2D–H). Descendants of Lfng-expressing cells in these mice were seen in normal acini, PanIN, and invasive tumors (Supplemental Fig. 2J–K). Like PKC-Lfng mice, this model showed Lfng expression in a few PanINs and limited number of tumor cells (Supplemental Fig. 2L–N). Collectively, these results demonstrate that Lfng-expressing cells in adult pancreas are readily targeted by combinatory Kras and p53 mutations to form PDAC.

Lfng-expressing centroacinar cell serves as a unique cell-of-origin for loss-of-p53 PDAC

As noted above, Lfng-expressing cells account for only 3% of lineage depleted pancreatic cells. Lfng-RFP/CreER-mediated recombination occurs in less than 1% of pancreatic cells, compared to 80% recombination efficiency of Mist1^CreER/+ in acinar cells (Fig. 3A, B). Despite the very low percentage of pancreatic cells being targeted, PKC-Lfng mice developed PDAC shortly after tamoxifen induction, suggesting that Lfng-expressing cells may serve as an important cell-of-origin for PDAC. We compared tumor development in mice targeted with the same set of oncogenic mutations directed by Lfng-RFP/CreER versus Mist1^CreER/+. Kras^G12D alone readily induced PanIN under Mist1^CreER/+, but almost none under Lfng-RFP/CreER (Fig. 3C, D). Combination of Kras^G12D and p53^R172H under Mist1^CreER/+ caused formation of PanIN/PDAC affecting almost entire pancreas but resulted in very few precursor lesions when driven by Lfng-RFP/CreER (Fig. 3E, F). Strikingly, combination of Kras^G12D and p53 deletion through Lfng-RFP/CreER induced rapid PanIN development and progression to PDAC, comparable to those through Mist1^CreER/+ (Fig. 3G–I), and even slightly shortened survival (Fig. 3J). Mist1 is expressed in acinar but not centroacinar cells [32], whereas Lfng expression is mostly confined to the centroacinar compartment. These results indicate that Lfng marks a centroacinar subpopulation serving as a unique cell-of-origin for PDAC when targeted by oncogenic Kras coupled with p53 deletion. To understand why Lfng-expressing cells are uniquely susceptible to transformation when p53 is lost, we examined the effect of p53 deletion on the fate of Lfng-expressing cells by staining for RFP and YFP in R26^YFP;Lfng-RFP/CreERT2 versus R26^YFP;p53^fl/fl;Lfng-RFP/CreERT2 following tamoxifen induction. Deleting p53 in Lfng-expressing cells caused a modest expansion of this population (Fig. 3K, L) and a drastic increase of their progenies (Fig. 3M, N), supporting that p53 constrains the propagation of Lfng-expressing cells in the adult pancreas.

A Anti-YFP immunostaining in the pancreas from R26^YFP;Lfng-RFP/CreER and R26^YFP;Mist1^CreER/+ mice 6 weeks after tamoxifen injection. B Quantitation of YFP⁺ cells depicted in (A), presented as the percentage of YFP⁺ cells among all cells in the acinar compartment. ****p < 0.0001. C Representative photomicrographs of the pancreas from Kras;Lfng-RFP/CreER and Kras;Mist1^CreER/+ mice at 12 weeks after tamoxifen injection. D Quantitation of ADM and PanIN lesions in Kras;Lfng-RFP/CreER and Kras;Mist1^CreER/+ mice at 12 weeks post-tamoxifen, presented as the percentage of total area affected by ADM and PanIN lesions. **p < 0.01. E Representative photomicrographs of the pancreas from p53^R172H;Kras;Lfng-RFP/CreER and p53^R172H;Kras;Mist1^CreER/+ mice at 12 weeks after tamoxifen injection. F Quantitation of PanIN and PDAC lesions in p53^R172H;Kras;Lfng-RFP/CreER and p53^R172H;Kras;Mist1^CreER/+ mice at 12 weeks post-tamoxifen, presented as the percentage of total area affected by PanIN and PDAC lesions. ****p < 0.0001. G Representative photomicrographs of the pancreas from p53^fl/fl;Kras;Lfng-RFP/CreER and p53^fl/fl;Kras;Mist1^CreER/+ mice at 4 weeks after tamoxifen injection. H Quantitation of PanIN and PDAC lesions in p53^fl/fl;Kras;Lfng-RFP/CreER and p53^fl/fl;Kras;Mist1^CreER/+ mice at 4 weeks post-tamoxifen, presented as the percentage of total area affected by PanIN and PDAC lesions. ns: non-significance. I Representative histology of the pancreatic tumors from p53^fl/fl;Kras;Lfng-RFP/CreER and p53^fl/fl;Kras;Mist1^CreER/+ mice at 7 weeks after tamoxifen injection. J Kaplan-Meier survival analysis for p53^fl/fl;Kras;Lfng-RFP/CreER and p53^fl/fl;Kras;Mist1^CreER/+ mice following tamoxifen injection. Log-rank test p = 0.0608. K, L Anti-RFP immunostaining and quantitation of RFP⁺ cells in the pancreas from R26^YFP;Lfng-RFP/CreER and R26^YFP;p53^fl/fl;Lfng-RFP/CreER mice 6 weeks after tamoxifen injection at 6 weeks of age. *p < 0.05. M, N Anti-YFP immunostaining and quantitation of YFP⁺ cells in the pancreas from R26^YFP;Lfng-RFP^/CreER and R26^YFP;p53^fl/fl;Lfng-RFP/CreER mice 6 weeks after tamoxifen injection at 6 weeks of age. **** p < 0.0001. Scale bars: 50 μm.

Full size image

PDAC derived from Lfng-expressing cell is enriched in Aldh1⁺ progenitor-like cells and contains endocrine tumor cells

We performed immunohistochemistry on PKC-Lfng tumors side-by-side with acinar-derived p53^fl/fl;Kras;Mist1^CreER (hereinafter referred to as PKC-Mist1) and ductal-derived p53^fl/fl;Kras;Sox9-CreER (hereinafter referred to as PKC-Sox9) tumors. Both PKC-Lfng and PKC-Mist1 tumors are positive for ductal marker CK19 and epithelial marker E-Cadherin along with high expression of vimentin, whereas PKC-Sox9 tumors exhibit lower expression of these markers (Fig. 4A–I). Tumors derived from Lfng-expressing centroacinar cells may be enriched in progenitor-like cells. Indeed, PKC-Lfng tumors showed more Aldh1a1⁺ cells compared to PKC-Sox9 tumors, whereas PKC-Mist1 tumors were completely negative for Aldh1a1 (Fig. 4J–L). For comparison, all three tumor types showed robust Aldh1b1 expression (Fig. 4M–O). Consistent with our previous finding that pancreatic cancer triggers selective depletion of β-cells [33], immunostaining revealed various insulin-negative spaces within the islets in three PDAC models (Fig. 4P–R). Interestingly, pancreas of PKC-Lfng mice contain morphologically normal acinar cells as well as tumor cells that express insulin (Fig. 4S–U), which are rarely seen in PKC-Mist1 and PKC-Sox9 mice. Lfng-expressing cells are CD24⁺ (Fig. 1E). CD24⁺, but not CD24⁻, human pancreatic progenitor cells can differentiate into insulin-producing cells [29]. These results suggest that Lfng-expressing centroacinar cells are multipotent and may adopt an endocrine fate under oncogenic transformation.

A–C Immunostaining for cytokeratin 19 (CK19) in pancreatic tumors from p53^fl/fl;Kras;Lfng-RFP/CreER (PKC-Lfng), p53^fl/fl;Kras;Mist1^CreER/+ (PKC-Mist1), and p53^fl/fl;Kras;Sox9-CreER (PKC-Sox9) mice. D–F Immunostaining for E-cadherin in pancreatic tumors from PKC-Lfng, PKC-Mist1, and PKC-Sox9 mice. G–I Immunostaining for vimentin in pancreatic tumors from PKC-Lfng, PKC-Mist1, and PKC-Sox9 mice. J–L Immunostaining for Aldh1a1 in pancreatic tumors from PKC-Lfng, PKC-Mist1, and PKC-Sox9 mice. M–O Immunostaining for Aldh1b1 in pancreatic tumors from PKC-Lfng, PKC-Mist1, and PKC-Sox9 mice. P–R Anti-insulin immunostaining in pancreatic tumors from PKC-Lfng, PKC-Mist1, and PKC-Sox9 mice. Arrows: insulin-negative area within the islet. S–U Anti-insulin immunostaining in the pancreas from PKC-Lfng mice. Arrows: insulin-positive solitary tumor cells (T) and ductal tumor cells (U). Scale bars: 50 μm.

Full size image

Lfng is required for the initiation of PDAC from Lfng-expressing centroacinar cell

To determine whether Lfng plays a role in PDAC arising from Lfng-expressing cells, we examined tumor development in Lfng^fl/fl;p53^fl/fl;Kras;Lfng-RFP/CreER (LPKC-Lfng) in comparison with PKC-Lfng mice. At 4 weeks post-tamoxifen, PKC-Lfng mice had developed PanIN lesions covering about 8% of the pancreatic tissue, whereas the pancreas from LPKC-Lfng mice was lesion-free (Fig. 5A, B). Consistent with this observation, LPKC-Lfng mice had significantly prolonged survival as compared to PKC-Lfng mice (Fig. 5D). At pathological endpoint, all PKC-Lfng mice had advanced PDAC, whereas the vast majority of LPKC-Lfng remained pancreatic lesion-free (Fig. 5C). Hemangioma-like lesions were found at various sites in LPKC-Lfng mice (data not shown), which may be the cause of morbidity. Since Lfng is expressed in venous endothelial cells [34] (Supplemental Fig. 4D), vascular lesions found in LPKC-Lfng mice were more likely primary tumors than metastasis from pancreas. Thus, deletion of Lfng blocks PDAC arising from Lfng-expressing centroacinar cells in adult mice. LPKC-Lfng pancreas stained negative for cleaved caspase 3, indicating that apoptosis was not the underlying mechanism of tumor suppression (Fig. 5E). Like Kras;Mist1^CreER [32], PKC-Lfng tumors showed upregulation of Hes1 compared to the pancreas from Kras^LSL-G12D control mice. Interestingly, LPKC-Lfng pancreas showed Hes1 expression comparable to that of Kras^LSL-G12D mice, suggesting that deletion of Lfng might revert Notch activation in PKC-Lfng pancreas (Fig. 5F). Next, we traced Lfng-expressing cells by anti-YFP staining in R26^YFP;Lfng^fl/fl;p53^fl/fl;Kras;Lfng-RFP/CreER (R26^YFPLPKC-Lfng) mice. The vast majority of YFP⁺ cells were found in the acinar compartment, resembling centroacinar and acinar cells (Fig. 5G, H), while rare YFP⁺ cells were detected in morphologically normal duct (Fig. 5I). Immunofluorescence staining showed many YFP⁺ cells co-expressing the acinar marker amylase (Fig. 5J, L), whereas those remained in centroacinar location were amylase⁻ (arrows in Fig. 5J). Interestingly, a few YFP⁺ cells residing in the islet expressed insulin (Fig. 5K, O-O”), suggesting conversion of Lfng-expressing cells to β-cells. We also noted YFP⁺ acinar-like cells with lower amylase expression compared to genuine acinar cells at early (Fig. 5M-M”), but not late (Fig. 5N-N”), time points, suggesting progressive transition to acinar cells. Taken together, inactivation of Lfng drives Kras^G12D,p53^Deletion-targeted centroacinar cells to adopt acinar and, in rare cases, ductal and endocrine fates, deviating from the ADM and PanIN courses. While deleting p53 in Lfng-expressing centroacinar cells drastically increased their descendants (Fig. 3N), deletion of both p53 and Lfng reverted this phenotype (Fig. 5P, Q), suggesting that Lfng may be required for the hyperproliferation of p53-deficient centroacinar cells. Indeed, while Ki67 immunostaining were completely negative in Lfng-RFP/CreER and Lfng^fl/fl;Lfng-RFP/CreER pancreas, p53^fl/fl;Lfng-RFP/CreER pancreas showed Ki67⁺ cells residing in the centroacinar compartment, and Lfng^fl/fl;p53^fl/fl;Lfng-RFP/CreER pancreas had significantly decreased number of Ki67⁺ cells compared to p53^fl/fl;Lfng-RFP/CreER (Fig. 5R, S). Finally, Lfng deletion had no effect on the number of RFP⁺ cells (Fig. 5P, Q), ruling out the possibility of decreased Lfng-RFP/CreER promoter activity in LPKC-Lfng mice.

A, B Representative photomicrographs of the pancreas from p53^fl/fl;Kras;Lfng-RFP/CreER (PKC-Lfng) and Lfng^fl/fl;p53^fl/fl;Kras;Lfng-RFP/CreER (LPKC-Lfng) mice at 4 weeks post tamoxifen induction (A) and quantitation of ADM/PanIN lesions in these mice presented as the percentage of total area affected by precancerous lesions (B). **p < 0.01. C Representative photomicrographs of the pancreas from PKC-Lfng and LPKC-Lfng mice at the pathological endpoint. D Kaplan-Meier survival analysis for PKC-Lfng and LPKC-Lfng mice following tamoxifen injection. Log-rank test p < 0.0001. E Immunostaining for cleaved caspase 3 in the PKC-Lfng and LPKC-Lfng pancreas at 6 weeks post tamoxifen induction. Arrowhead: an apoptotic cell positive for cleaved caspase 3. F Relative mRNA level of Hes1 in the Kras^LSL-G12D, PKC-Lfng, and LPKC-Lfng pancreas, determined by quantitative RT-PCR. *p < 0.05. G–I Anti-YFP immunostaining in the pancreas of R26^YFP;Lfng^fl/fl;p53^fl/fl;Kras;Lfng-RFP/CreER (R26^YFPLPKC-Lfng) mice 45 days after tamoxifen injection. Arrowhead: YFP-positive cell in a normal duct. J–L Double immunofluorescence staining for amylase (red) and YFP (green) in the pancreas from R26^YFPLPKC-Lfng mice at 45 and 75 days post-tamoxifen. Arrows in J: YFP⁺ centroacinar cells. M-M” Immunofluorescence staining for YFP (M), amylase (M’), and merged (M”) in the R26^YFPLPKC-Lfng pancreas at 45 days post-tamoxifen. Arrowheads: YFP⁺ acinar cells showing lower amylase expression. N-N” Immunofluorescence staining for YFP (N), amylase (N’) and merged (N”) in the R26^YFPLPKC-Lfng pancreas at 75 days post-tamoxifen. O-O” Immunofluorescence staining for YFP (O), insulin (O’), and merged (O”) in the R26^YFPLPKC-Lfng pancreas at 45 days post-tamoxifen. P Anti-YFP and anti-RFP immunostaining in the pancreas of R26^YFP;p53^fl/fl;Lfng-RFP/CreER and R26^YFP;Lfng^fl/fl;p53^fl/fl;Lfng-RFP/CreER mice 45 days after tamoxifen injection. Arrowheads: positive RFP staining in centroacinar cells. Q Quantitation of YFP⁺ cells and RFP⁺ cells depicted in (P). ****p < 0.0001. ns: non-significance. R Anti-Ki67 immunostaining in the pancreas of indicated genotypes at 45 days after tamoxifen injection. S Quantitation of Ki67⁺ cells depicted in (R). *p < 0.05, ***p < 0.001^, ns: non-significance. Scale bars: 50 μm.

Full size image

We also compared precursor lesions in Kras;Lfng-RFP/CreER and Lfng^fl/fl;Kras;Lfng-RFP/CreER mice after tamoxifen induction followed by caerulein treatment. Lfng deletion had no effect on ADM/PanIN formation under this condition (Supplemental Fig. 3A–C). R26^YFP;Kras;Lfng-RFP/CreER and R26^YFP;Lfng^fl/fl;Kras;Lfng-RFP/CreER mice showed similar YFP and RFP staining, indicating that Lfng deletion did not alter the fate of Lfng-expressing centroacinar cells targeted by Kras^G12D alone (Supplemental Fig. 3D–I).

Notch3 is a functional Notch receptor for the initiation and progression of PDAC originating from Lfng-expressing centroacinar cell

To understand Notch regulation by Lfng during PDAC initiation and progression, we performed immunohistochemistry for all four Notch receptors in the PKC-Lfng and LPKC-Lfng pancreas at 2 time points; e.g., 28 and 45 days post-tamoxifen. Notch1 and Notch2 were undetectable in these mice (Fig. 6A–H). To the contrary, some centroacinar cells (Fig. 6R), all PanIN lesions (Fig. 6I) and tumor cells (Fig. 6K) showed robust Notch3 expression in PKC-Lfng mice. For comparison, Notch3 was detected only in blood vessels and weakly in islets in LPKC-Lfng mice (Fig. 6J, L). Notch4 expression was barely detectable in a few PanIN/PDAC cells in PKC-Lfng mice (Fig. 6M, O), whereas in LPKC-Lfng mice it was completely negative on day 28 post-tamoxifen (Fig. 6N) and weak in islets on day 45 post-tamoxifen (Fig. 6P). Co-expression of Notch3 and RFP was observed in a subset of centroacinar cells (Fig. 6Q-Q”) and in many tumor cells (Fig. 6S-S”) in PKC-Lfng mice. Thus, Notch3 is the sole Notch receptor that is upregulated during tumor initiation and progression from Lfng-expressing centroacinar cells, suggesting an essential role for Lfng-dependent Notch3 signaling in this type of tumor. Indeed, breeding of a Notch3 null mutation [34] into PKC-Lfng mice significantly prolonged survival (Fig. 6T).

A–D Immunostaining for Notch1 in the pancreas from p53^fl/fl;Kras;Lfng-RFP/CreER (PKC-Lfng) and Lfng^fl/fl;p53^fl/fl;Kras;Lfng-RFP/CreER (LPKC-Lfng) mice at 28 and 45 days post tamoxifen injection. E–H Immunostaining for Notch2 in the PKC-Lfng and LPKC-Lfng pancreas at 28 and 45 days post-tamoxifen. I–L Immunostaining for Notch3 in the PKC-Lfng and LPKC-Lfng pancreas at 28 and 45 days post-tamoxifen. Arrows in I: positive staining in ADM. Arrowhead in I: positive staining in PanIN. M–P Immunostaining for Notch4 in the PKC-Lfng and LPKC-Lfng pancreas at 28 and 45 days post-tamoxifen. Q-Q” Double immunofluorescence staining for RFP (red) and Notch3 (green) in the PKC-Lfng pancreas at 28 days post-tamoxifen. Arrows: co-staining of RFP and Notch3 in centroacinar cells. R Anti-Notch3 immunostaining in the PKC-Lfng pancreas at 28 days post-tamoxifen. Arrow: positive staining in centroacinar cells. S-S” Double immunofluorescence staining for RFP (red) and Notch3 (green) in the PKC-Lfng pancreas at 45 days post-tamoxifen. T Kaplan-Meier survival analysis for PKC-Lfng and Nocth3^-/-;p53^fl/fl;Kras;Lfng-RFP/CreER (NPKC-Lfng) mice following tamoxifen induction. Log-rank test p = 0.0008. Scale bars: 50 μm.

Full size image

Lfng exerts an oncogenic role in acinar- and ductal-derived PDACs

To determine whether Lfng plays any role in acinar cell-derived PDAC, we first crossed a Lfng-eGFP reporter into Kras;Mist1^CreER/+, an acinar-derived PADC precursor model [32]. GFP expression was detected in very few pancreatic cells with centroacinar location (Supplemental Fig. 4A, B) and in endothelial cells (Supplemental Fig. 4C, D) in Lfng-eGFP mice, but was turned on specifically in ADM and PanIN lesions in Lfng-eGFP;Kras;Mist1^CreER/+ mice (Fig. 7A, B). Consistent with the Lfng upregulation in precursor lesions, deletion of Lfng decelerated lesion formation in this model (Fig. 7C–E). Notably, pancreas in 2 out of 10 of Lfng^fl/fl;Kras;Mist1^CreER/+ mice showed complete loss of acini, leaving islets and desmoplasia-associated ducts embedded within a fat pad (Supplemental Fig. 5). Lfng is also upregulated in acinar-derived tumor cells as shown by robust GFP expression in Lfng-eGFP;p53^fl/fl;Kras;Mist1^CreER/+ mice (Fig. 7F), and deletion of Lfng in p53^fl/fl;Kras;Mist1^CreER/+ significantly prolonged survival (Fig. 7G). Both R26^YFP;Mist1^CreER/+ and R26^YFP;Lfng^fl/fl;Mist1^CreER/+ mice showed about 75% recombination efficiency among acinar cells (Supplemental Fig. 6), indicating that suppression of PanIN/PDAC in Lfng^fl/fl;Kras;Mist1^CreER/+ and Lfng^fl/fl;p53^fl/fl;Kras;Mist1^CreER/+ mice was not due to decreased Mist1^CreER/+ activity. Double immunofluorescence staining for GFP and individual Notch receptors in Lfng-eGFP;p53^fl/fl;Kras;Mist1^CreER/+ mice showed that Notch3 was highly expressed in tumor cells, some of which co-expressing Lfng or adjacent to Lfng-expressing cells (Supplemental Fig. 7A–C). Notch1 and Notch2 were undetectable, whereas Notch4 was noted in limited number of tumor cells (Supplemental Fig. 7D–F). These results suggest an oncogenic role for Lfng through Notch3 in acinar-derived PDAC.

A, B Anti-GFP immunostaining in the pancreas of Lfng-eGFP;Kras;Mist1^CreER/+ mice at 14 weeks post-tamoxifen injection. C, D Representative photomicrographs of pancreas from Kras;Mist1^CreER/+ and Lfng^fl/fl;Kras;Mist1^CreER/+ mice at 12 weeks after tamoxifen injection. E Quantitation of ADM and PanIN lesions depicted in (C, D), presented as the percentage of total area affected by the precancerous lesions. **p < 0.01. F Anti-GFP immunostaining in the pancreas of Lfng-eGFP;p53^fl/fl;Kras;Mist1^CreER/+ mice at 7 weeks post-tamoxifen induction. G Kaplan-Meier survival analysis for p53^fl/fl;Kras;Mist1^CreER/+ and Lfng^fl/fl;p53^fl/fl;Kras;Mist1^CreER/+ mice following tamoxifen injection at 1 month of age. Log-rank test p < 0.0001. H, I Anti-GFP immunostaining in the pancreas of Lfng-eGFP;p53^fl/fl;Kras;Sox9-CreER mice 6 months after tamoxifen injection. J Anti-GFP immunostaining in the pancreas of Lfng-eGFP;p53^fl/fl;Kras mice. K Kaplan-Meier survival analysis for p53^fl/fl;Kras;Sox9-CreER and Lfng^fl/fl;p53^fl/fl;Kras;Sox9-CreER mice following tamoxifen injection at 1 month of age. Log-rank test p = 0.0246. L Genetic alterations of LFNG in patients with PDAC. M Overall survival of PDAC patients with or without LFNG alterations. Log-rank test p = 0.0182. Data derived from The Cancer Genome Atlas Program (TCGA) pancreatic cancer dataset. N, O Immunostaining for LFNG in tumor-adjacent near normal tissues from PDAC patients. Arrows: centroacinar cells. P–U Representative photomicrographs of anti-LFNG immunostaining on a human pancreatic tissue microarray containing specimens of adenocarcinoma (n = 38), acinic cell carcinoma (n = 1), and squamous cell carcinoma (n = 1), duplicate cores per case. ACC: acinic cell carcinoma. V Quantitation of LFNG immunostaining in adenocarcinomas of Grade 1, 2, and 3. Staining scores: 0 = completely negative, 1 = weak, 2 = intermediate, and 3 = strong. ANOVA p < 0.0001. Scale bars: 50 μm in (A–D, F, and H–J); 25 μm in (N, O); 100 μm in (P–U).

Full size image

Next, we determined whether Lfng also plays a role in ductal cell-derived PDAC using Sox9-CreER-mediated model [35, 36]. Although R26^YFP;Sox9-CreER pancreas contained very few YFP⁺ ductal cells (data not shown), R26^YFP;p53^fl/fl;Kras;Sox9-CreER pancreas showed YFP staining in normal and tumorous ducts (Supplemental Fig. 8A–D) and in PDAC cells (Supplemental Fig. 8E, F), confirming the ductal origin of PDAC in this model. Lfng-eGFP;p53^fl/fl;Kras;Sox9-CreER showed GFP in metaplastic ductal cells and tumor cells (Fig. 7H, I), whereas Lfng-eGFP;p53^fl/fl;Kras pancreas showed no GFP in normal ducts (Fig. 7J). Moreover, deletion of Lfng in p53^fl/fl;Kras;Sox9-CreER mice significantly delayed tumor onset and improved survival (Fig. 7K). Like acinar-derived PDAC, these tumors also exhibited robust Notch3 expression, co-localizing or juxtaposing with Lfng (Supplemental Fig. 7G–I). These data suggest that Lfng plays an oncogenic role in ductal-derived PDAC as well.

High LFNG expression is associated with high grade and poor survival in human PDAC

Alterations of LFNG, predominantly in the forms of mRNA upregulation and gene amplification, were detected in 9% PDAC patients and were associated with shortened overall survival (Fig. 7L, M). Consistent with its expression in mouse pancreas, LFNG protein was detected in a subset of cells at the centroacinar vicinity in normal and adjacent normal pancreatic tissues from human patients (Fig. 7N, O). Staining for LFNG in a pancreatic tissue microarray showed varied intensity, with the staining score positively correlating with tumor grade in adenocarcinoma specimens (Fig. 7P–T, V). Of note, acinic cell carcinoma of the pancreas stained negative for LFNG (Fig. 7U). These results suggest that LFNG may play an important role in human PDAC, consistent with our investigation in murine model.

• SEO Powered Content & PR Distribution. Get Amplified Today.
• PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
• PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
• PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
• PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
• Source: https://www.nature.com/articles/s41388-024-03226-7