Development of thymic tumor in [LSL:KrasG12D; Pdx1-CRE] mice, an adverse effect associated with accelerated pancreatic carcinogenesis

Pancreatic Ductal AdenoCarcinoma (PDAC) represents about 90% of pancreatic cancers. It is one of the most aggressive cancer, with a 5-year survival rate below 10% due to late diagnosis and poor therapeutic efficiency. This bad prognosis thus encourages intense research in order to better understand PDAC pathogenesis and molecular basis leading to the development of innovative therapeutic strategies. This research frequently involves the KC (LSL:KrasG12D;Pdx1-CRE) genetically engineered mouse model, which leads to pancreatic cancer predisposition. However, as frequently encountered in animal models, the KC mouse model also exhibits biases. Herein, we report a new adverse effect of KrasG12D mutation in KC mouse model. In our hands, 10% of KC mice developed clinical signs reaching pre-defined end-points between 100- and 150-days post-parturition, and associated with large thymic mass development. Histological and genetic analyses of this massive thymus enabled us (1) to characterize it as a highly proliferative thymic lymphoma and (2) to detect the unexpected recombination of the Lox-STOP-Lox cassette upstream KrasG12D allele and subsequent KRASG12D protein expression in all cells composing thymic masses. Finally, we highlighted that development of such thymic tumor was associated with accelerated pancreatic carcinogenesis, immune compartment disorganization, and in some cases, lung malignancies.


Results
In KC mouse model, LSL recombination occurs in pancreas but not in tail. KC mouse model were obtained by crossing the Pdx1-CRE and LSL:Kras G12D mouse strains, in order to study pancreatic carcinogenesis. In the LSL:Kras G12D strain, the endogenous Kras G12D allele is preceded by a Lox-STOP-Lox (LSL) cassette avoiding Kras G12D transcription in absence of recombination. In order to induce a conditional expression of the mutated oncogene in the pancreas, we used the Pdx1-CRE mouse strain, in which the CRE recombinase coding sequence is placed downstream of the Pdx1 gene promoter (Fig. 1a).
Mice were genotyped one-week post-parturition by a classical Polymerase Chain Reaction (PCR). All KC mice carried both the wild-type (WT) and the mutated (LSL:Kras G12D ) Kras alleles, while WT mice presented only the WT Kras band (Fig. 1b, bottom). In order to confirm correct pancreas-specific recombination, conditional Kras G12D PCR was performed on tail and pancreas of one KC mouse (Fig. 1b, top). Primers used for this PCR, which allows to detect LSL cassette recombination, are presented in Fig. 1b. In tail of KC mice, we observed wild-type (WT) and LSL:Kras G12D bands only, while in pancreas we noted the presence of a third band (corresponding to Kras G12D without the LSL cassette). The coexistence of the 3 bands in pancreas suggests that recombination is partial, not in all cells composing the tissue extract. As controls, we analyzed pancreatic DNA from (1) WT mouse pancreas, for which we detected the WT Kras allele only, with a large band, and (2)  www.nature.com/scientificreports/ mouse tumor tissue with almost total recombination (CTRL+), which showed mainly the presence of the WT and Kras G12D alleles without the LSL cassette. This result confirmed the correct pancreas-specific recombination of mutated Kras G12D oncogene.
10% of KC mice display severe clinical signs associated with thymic mass development. While constituting a cohort of KC mice invalidated or not for our gene of interest, we unexpectedly observed some KC mice suffering from severe clinical signs between 100-and 150-days post-parturition independently from this gene of interest. Those clinical signs included drastic weight loss and/or respiratory distress, and corresponded to pre-defined end-points, thus leading us to euthanize the animals before the scheduled date. These mice finally accounted for about 10% of the total cohort (9/89), which thus cannot be ignored. In the work presented here, we decided to characterize the 5 KC mice (4 males and 1 female) displaying such suffering signs in our cohort that we compared to 4 age-matched KC mice (100 and 150 days) and 1 older mouse (200 days), without apparent symptoms from the same cohort. Body weight monitoring has been performed on mice from 100 days post parturition, and show the rapid weight loss of two mice with thymic tumors (Fig. 2a). It also has to be noted that we observe a high body weight variation between each mouse of the two groups, that cannot be due to the sex of the mice since they are mostly males, but rather to the fact that the mice are mostly not littermates. Finally, we calculated cumulative survival proportion depending on the development or not of thymic mass (Kaplan-Meier method, Fig. 2b), and clearly showed that mice harboring thymic mass had poorer overall survival compared to unaffected mice, coherently with the observed end-points and subsequent euthanasia. During dissection, we observed that all mice reaching a humane end-point had a rib cage invaded by a huge mass clearly developed from thymus. This thymic mass caused the crushing of esophagus, heart and lung, explaining the observed clinical signs (Fig. 2c). It has to be noted that in the Fig. 2c, lung and heart were clearly visible in the mouse with a histologically normal thymus (left photo), but completely hidden by the thymic mass in the symptomatic mouse (right photo).  www.nature.com/scientificreports/ Pathological thymus exhibits total LSL cassette recombination, with subsequent KRAS G12D expression and thymic lymphoma development. We then suspected an artefactual recombination of LSL cassette in the developing thymic masses and consequently performed a conditional Kras G12D PCR on these tissues, (Fig. 3a, upper panel). Interestingly, we highlighted that normal thymus did not present any LSL cassette recombination, while the recombination was total in thymic masses. Indeed, in those tissues, we were not able to detect any LSL:Kras G12D band, suggesting that all cells proceeded to LSL cassette recombination and thus expressed Kras G12D allele. Moreover, such recombination seemed to be specific to the thymus, as no recombination of the LSL cassette was detected in the tail of mice suffering from thymic mass development. It has to be noted that WT Kras amplicon was always visible in normal thymus, while it was sometimes difficult to discriminate in thymic masses. Genotyping PCR was performed on tail from all analyzed animals to validate the heterozygosity of the Kras G12D allele for all KC mice (homozygous Kras G12D is lethal) (Fig. 3a, lower panel). Finally, we histologically confirmed the consequence of the LSL cassette excision on paraffin-embedded thymus To better characterize the consequence of such artefactual recombination and KRAS G12D expression in thymus, we further analyzed these thymic masses and compared them to normal thymus. Paraffin-embedded tissues were cut and sections were stained with Hematoxylin-Eosin or labelled by immunohistochemistry (IHC) for CD3e (T lymphocytes) and Ki67 (proliferation) (Fig. 3c). As expected, normal thymus was composed of (1) a connective tissue capsule, (2) the cortex (C), which is darker and contains lymphocytes (CD3e-positive cells) and (3) the medulla (M), containing larger lymphocytes (CD3e-positive cells) and Hassall's Corpuscles. In contrast, the thymic masses did not show such organization, and were very rich in lymphocytes (round cells predominantly occupied by the nucleus), reminiscent of a thymic lymphoma. This was confirmed by Ki67 and CD3e immunolabelings of the thymus demonstrating highly proliferative cells, positive for lymphocytic marker (Fig. 3c).
Therefore, LSL cassette recombination and subsequent mutated Kras G12D oncogene expression occurred in thymus of 10% of KC mice, leading to the development of thymic lymphoma.
Thymic lymphoma development is associated with accelerated pancreatic carcinogenesis, lung malignancies and immune disorder. Finally, we wanted to decipher if such thymic tumor devel- www.nature.com/scientificreports/ opment could have an effect on pancreatic carcinogenesis. We thus analyzed the pancreas of mice exhibiting or not a thymic mass, by classical H&E staining and CK19 (pancreatic ductal cell/malignant epithelial carcinoma marker) and Ki67 immunolabelings (Fig. 4a). Lesion extent (area occupied by lesions visible in H&E staining) and proportion of proliferating cells within lesions have been quantified thanks to QuPath software. We clearly noticed that mice with thymic tumor displayed larger area occupied by pre-tumoral lesions in three of the four mice ( Fig. 4a, left). The difference between the two groups is not significant, most likely due to low number of mice and the fact that one of the mice with thymic tumor did not present larger lesion extent yet. Nevertheless, we were able to detect a significant increase in proliferation, with more Ki67-positive cells in pancreatic lesions of mice suffering from thymic tumor, allowing us to affirm that mice displaying thymic tumors presented more aggressive lesions. In order to better understand KRAS G12D expression in KC mouse model, we immunolabelled pancreas and detected KRAS G12D mutated protein within preneoplastic lesions, as well as in Langerhans islets of all KC mouse pancreas (with and without thymic lymphoma), in contrast to WT mouse, which did not display any labeling. In addition, KRAS G12D was detected in acinar cells bordering pre-tumoral lesions, and that seemed to undergo Acinar-to-Ductal Metaplasia (ADM). In mice suffering from thymic lymphoma, KRAS G12D labeling area was more important, in view of the increased extent of lesions in their pancreas (Fig. 4a, right).
Interestingly, in three out of four mice exhibiting thymic mass development, we also noticed proliferative (Ki67-positive) masses in lungs, which were also positive for KRAS G12D (Fig. 4b). These lung malignancies were negative for CK19 and did not show any PDAC morphology. In one of them, such masses were even visible macroscopically at necropsy (Fig. 4c).
Thymus being one of the primary lymphoid organs, we also investigated the immune compartment, notably thanks to spleen analysis. Interestingly, mice with thymic tumor displayed highly enlarged spleen, which seemed to be histologically disorganized (Fig. 5a,b). Indeed, white pulp seemed larger, and less organized and dense, as it can be observed following H&E staining and Ki67 and CD3e immunolabelings. In addition, KRAS G12D protein was detected in white pulp of spleen from mice developing thymic mass, suggesting either the recruitment of lymphocytes from the thymus or recombination of the LSL cassette within the spleen too. In addition, two of the thymic tumor-suffering mice also showed an important immune infiltration. Indeed, in the pancreas of those mice, numerous highly proliferating immune cells were observed, not only around lesions but also around still normal acini, as showed by CD3e (T lymphocytes) and Ki67 (proliferation) immunolabelings (Fig. 5c). In addition, this was accompanied by the presence of large lymph nodes on section (Fig. 5d). Those immune cells observed in close proximity to pancreas were also KRAS G12D -positive, evoking recruitment of lymphocytes from pathological thymus. Both results delineate immune dysregulation within mice suffering from thymic mass development.

Discussion
Taken together, our results highlight a new adverse effect linked to genetic modification in the KC mouse model. Indeed, we pointed out the development of thymic tumor in 10% of the KC mouse cohort, which cannot be ignored. Interestingly, our observations were made in two different animal facilities, indicating that it would better rely on intrinsic mechanism than on environmental cues. Mice presenting thymic mass reached humane end-points between 100-and 150-days post-parturition, including rapid weight loss and respiratory distress, probably due to lung, heart and esophagus crushing under the tumor mass, since these organs were smaller than in normal KC mice. In addition, those mice were prostrated in their cage, suggesting abdominal pain. Such symptoms are close to the ones developed in consequence to PDAC development and are thus confounding for this mouse model at early age. It must be noted that presence of this thymic mass could not be predicted without opening rib cage, showing the importance to proceed to a complete necropsy.
We first investigated Kras G12D expression within these thymic masses. To do so, we first analyzed LSL cassette recombination upstream Kras G12D allele and showed that the transcriptional STOP was deleted in all thymic masses, while it was never the case in normal thymus. Interestingly, this recombination was total, with no remaining allele with the LSL cassette, suggesting that all cells would have recombined, in contrast to pancreas where the LSL sequence persisted. It is important to note that LSL cassette recombination is not general in the mice harboring thymic tumor; since it is not recombined in tail. This was confirmed by anti-KRAS G12D immunohistochemistry, as we detected labeling in all cells from thymic masses, and not in normal thymus. Histologically, thymic masses showed a disorganized structure, with the absence of distinguishable cortex and medulla, in contrast to normal thymus. Tumors appeared to be lymphocyte-rich and highly proliferative, evoking thymic lymphoma, with few epithelial cells [37][38][39] . This diagnosis is further reinforced by the strong similarities observed at the macroscopic and microscopic levels between the pathological thymus of KC mice and the thymic lymphoma developed in the p53 KO mice 40 . Thus, we can confidently hypothesize that the development of thymic tumors, would rely on the expression of mutated KRAS oncogene.
Initially, our hypothesis for such atypical Kras G12D expression was that Pdx1 could be transiently expressed by thymic antigen-presenting cells for lymphocyte education, with subsequent Kras G12D expression, which could lead to malignant transformation in some cases. However, the described observations suggest another mechanism. The lymphoma phenotype of thymic tumoral mass, as well as the expression of Kras G12D in all thymic cells (including lymphocytes) suggested that recombination could occur independently from Pdx1 expression. Indeed, in the original study using Kras G12D oncogene in mouse model, Johnson and colleagues studied the organs, which were sensitive to sporadic recombination (unequal sister chromatid exchange or intra-chromosomal recombination) leading to Kras G12D expression. Interestingly, they highlighted the development of lung tumors, thymic lymphomas and papillomas, suggesting that lung, thymus and skin were the organs in which sporadic recombination occurred 41   www.nature.com/scientificreports/ sporadic recombination. In addition, we frequently observed papillomas on KC mice, with no deleterious effect on animal well-being (data not shown), coherently with the study of Johnson and collaborators. In addition, we highlighted that three out of the four mice developing thymic mass exhibited some malignancies in lung, more or less visible, that were positive for KRAS G12D and highly proliferative, as expected in consequence to KRAS activation. The mouse showing no lung disorder was dissected as scheduled, at 100 days, and had not reached humane end-points, suggesting that lung mass development could arise later. Lung being a known metastasis site for pancreatic tumor cells, these malignancies could be metastases, but the hypothesize seems to be ruled out by the fact that they are clearly negative for CK19 and do not display PDAC morphology. In addition, no correlation existed between pancreatic lesion extent and level of lung damage. In regards with the literature, it appears that such lung malignancies could be explained by sporadic recombination, as pointed out by Johnson and colleagues in 2001, who described lung cancer development as a consequence of Kras G12D expression only mediated by sporadic recombination (no CRE) 41 . Therefore, we suggest here that Pdx1-CRE construct is not sufficient to restrain Kras G12D recombination to the pancreas, and that sporadic recombination occurs in thymus (in 10% of cases) and lung. This is coherent with the fact that no evidence of Pdx1 expression in the thymus has been found, and some articles even reported the absence of Pdx1 in the thymus 42,43 . However, Pdx1 has already been shown to be expressed in skin, linked to the development of papillomas in KC mice 44 . To finally confirm our hypothesis, it would be interesting to perform www.nature.com/scientificreports/ Pdx1 labeling on thymus and thymic masses, which will allow to discriminate between sporadic or Pdx1-induced recombination. However, Pdx1 expression in thymic cells would probably be transient, and could thus be difficult to highlight. In another hand, (1) thymocytes are the place of many genetic recombination (Variable (Diversity) Junction (V(D)J) recombination), leading to a large panel of variable regions on T cell receptors 45 , and (2) sporadic recombination is favored by the presence of repeated sequences 46,47 , as can be considered Lox sequences surrounding transcriptional STOP upstream Kras G12D allele added by Jackson and colleagues in 2001 24 . Moreover, lymphocyte population has recently been shown to be sensitive to Kras activation 48 . All these considerations thus encourage the sporadic recombination hypothesis. The KC mouse model is one of the most used models to study pancreatic carcinogenesis. The cohort we initially developed (89 mice) aimed at determining the role of a matrix glycoprotein during pancreatic carcinogenesis, study which was particularly impacted by the development of the adverse effect we describe here. By investigating the impact of thymic mass development on pancreatic carcinogenesis, we unfortunately highlighted that thymic tumor occurrence was associated with an accelerated tumorigenesis. Indeed, we noted that (1) three out of the four analyzed mice suffering from thymic tumor displayed a larger extent of pancreatic lesions, even if this result was non-significant and (2) those lesions significantly presented an increased proliferation, as assessed by Ki67-positive cell quantification. Altogether, those results confirm that mice exhibiting thymic mass presented higher grade pre-tumoral to tumoral lesions, and thus an accelerated pancreatic carcinogenesis. In addition, we detected some KRAS G12D -positive cells within the adjacent spleen, suggesting an invasive phenotype for pancreatic (pre)tumoral cells (Supplementary Figure S1). Thus, the presence of thymic tumor affects the study of the pancreatic carcinogenesis process.
The accelerated carcinogenesis could be due to immune disorder within these mice. This hypothesis is corroborated by (1) the disorganization we observed within the spleen, with the perturbation of white pulp structure, and (2) high level of immune infiltration within the pancreas of two out of the mice with thymic mass, as demonstrated by CD3e immunolabeling. Indeed, disorder in thymus could affect T cell compartment and lead to the absence or the inefficiency of tumor-suppressing immune response in first steps of pancreatic carcinogenesis 49 . In addition, in the mouse presenting high level of immune infiltration, it was interesting to note that lymph nodes and immune cells around pancreatic cells were positive for KRAS G12D protein, as well as in spleen white pulp, suggesting either (1) sporadic recombination or (2) the recruitment of immune cells from thymic mass, hypothesis that we favor. Finally, the labeling of KRAS G12D was also quite interesting to understand the appearance of lesions in pancreas of KC mice. Indeed, we detected KRAS G12D within preneoplastic lesions, as well as in Langerhans islets, as expected by the expression of Pdx1 in these structures 25,33,34 . The presence of KRAS G12D in the islets could explain the islet disorganization, as well as the lesions developing within islets that have been previously described 35,36 . Interestingly, we also highlighted KRAS G12D in acinar cells surrounding pre-tumoral lesions and evoking first steps of ADM (less contiguous acini and with more visible light), suggesting that Pdx1 expression could be induced by the lesion and lead to transformation of adjacent normal acinar cells.
Our study thus pointed out an adverse effect of the Kras G12D construct in the KC mouse model classically used to study pancreatic carcinogenesis. To our knowledge, this is the first report showing the development of such thymic tumors in the KC mouse model. The hypothesis of sporadic recombination in organs with high level of recombination (thymus notably) is favored here. This bias is associated with an accelerated pancreatic carcinogenesis, and can thus lead to misinterpretations, knowing that thymic mass can be missed if only the pancreas is recovered (without opening the rib cage).

Methods
Animals. LSL:Kras G12D (B6.129S4-Kras tm4Tyj /J strain, The Jackson Laboratory) and Pdx1-CRE (B6.FVB-Tg(Pdx1-cre)6Tuv/J strain, The Jackson Laboratory) mouse strains were previously described 24,25,41 and generously gifted by P. Bertolino and L. Bartholin, respectively. They were maintained in two pathogen-free animal facilities ('' ALECS-SPF" and "PBES" (Lyon, France)), and crossed in order to obtain [LSL:Kras G12D ;Pdx1-CRE] mice, predisposed to pancreatic cancer development. All procedures were conducted in accordance with the guidelines of the European Union and French laws and approved by the local animal ethic committee under regulatory of governmental authority (CECCAPP, Comité d'Evaluation Commun au Centre Léon Bérard, à l' Animalerie de transit de l'ENS, au PBES et au laboratoire P4 (n° C2EA15), APAFIS #16,242-201,804,271,428,498). The study was carried out in compliance with the ARRIVE guidelines (Animal Research : Reporting of In Vivo Experiments) 50 . Animals were genotyped one week after birth and body weight tracking was started at 100 days post-parturition and was performed twice a week. Mice were euthanized either at 100-, 150-or 200-days post-parturition or at reaching end-point (weight loss, signs of pain, etc.), and cumulative survival percentage was estimated using the Kaplan-Meier method, before comparing groups thanks to Log-Rank (Mantel-Cox) statistical test. At necropsy, several organs were harvested, including pancreas, spleen, lung, thymus, as well as the tail for control. After dissection, one part of each organ (except tail) was fixed in formalin before being processed for paraffin embedding. The other part was directly snap frozen in liquid nitrogen for DNA extraction.
Histological staining and immunohistochemistry. For histological analyses, formalin-fixed paraffin embedded (FFPE) tissues were cut into 3 µm sections and Hematoxylin and Eosin (H&E) staining was performed after dewaxing and rehydration.
Immunohistochemistry on FFPE tissues was performed on 3 µm sections using ImmPRESS Excel Amplified Polymer Staining Kit, Anti-Rabbit IgG, Peroxidase (VECTOR LABORATORIES, MP-7601) for Ki67, CD3e and KRAS G12D , and R.T.U. Vectastain Universal Elite ABC Kit (VECTOR LABORATORIES, PK-7200) for CK19, following manufacturer's instructions. After deparaffinization and rehydration, antigen retrieval was performed either in Tris-EDTA buffer (pH9) (CD3e) or Sodium Citrate buffer (pH 6) (Ki67 and CK19) for 20 min at (1/50, #14,429, CELL SIGNALING); CD3e (1/500, MA5-14,524, THERMOFISHER)) were incubated in blocking solution or 1% Horse Serum (KRAS G12D ) overnight at 4 °C. The next day, amplification and development were performed following manufacturer's instructions, nuclei were counterstained with Gill's Hematoxylin and slices were mounted in DEPEX. All slices were imaged using the slide scanner AxioScanSP5 X (ZEISS) at CIQLE (Lyon, France) in order to have a general view of all tissues. Representative images were then extracted and shown. Quantification of lesion extent and proliferation was performed thanks to QuPath software, by calculating area of lesions (H&E staining) and proportion of proliferating cells (Ki67-positive nuclei), respectively. Groups were compared thanks to Mann-Whitney statistical test (non-parametric t test).