Introduction

TGFβ (transforming growth factor beta) is the archetype of the transforming growth factor superfamily1. TGFβ (3 isoforms, TGFβ1-2-3) is secreted as a latent complex in which the mature entity remains non-covalently bound to its pro-domain called Latency-Associated Peptide (LAP). During physiological (e.g. embryonic development, wound healing, immunosuppression) and pathological processes (e.g. fibrosis, cancer), bioactive TGFβ is released from the latent complex to further modulate the activity or the differentiation of surrounding cells. The molecular processes regulating this activation are scarcely understood, the greatest proportion of secreted TGFβ being inactive2. After activation, TGFβ signaling occurs through a heterotetrameric receptor complex composed of two subunits, the type I and type II TGFβ receptors (TβRI and TβRII, respectively). Upon binding to its receptors, TGFβ enables TβRII to transphosphorylate TβRI, which in turn activates the canonical Smad pathway (by phosphorylating SMAD2 and SMAD3 transcription factors that further interact with SMAD4 to accumulate inside the nucleus) and other signaling pathways (MAPK, RhoA, and Pi3K/Akt)3,4. The effects of a constitutive inhibition of TGFβ signaling has been studied by many approaches using genetically-engineered mice with genes knocked out, deleted, mutated, or dominant-negative mutated. These alterations result in severe phenotypes during development or after birth usually leading to death as observed in TGFβ-KO, TβR-KO and Smad-KO mice5. Organ-specific inactivation of TGFβ signaling using the Cre/lox system circumvents these developmental defects, unveiling its crucial role in cell homeostasis and pathogenesis. For instance, during tumorigenesis, loss-of-function of TGFβ signaling results mostly in increased proliferation and tumorigenesis5. Conversely, TGFβ gain-of-function has scarcely been explored. Organ-specific activation of TGFβ signaling (constitutively active ligand overexpression or expression of a constitutively active receptor) leads to developmental defects in the mammary gland and skin6,7 and to fibrosis in the liver and lungs8,9,10. Few years ago, we created the first conditional constitutively active TβRI receptor allele in mice (LSLTβRICA allele)11,12). While constitutive activation of this transgene is embryonically lethal13, we showed that it could potentiate KRASG12D-driven pancreatic transformation when targeted in pancreatic epithelial cell lineages13. Targeted expression of TβRICA in other organs such as the ovaries14,15, uterus16, T cells17, testis18 and liver19 results in a large panel of defects in homeostasis and differentiation. Considering the crucial role of TGFβ as a secreted factor in the microenvironment, we created in the present study an original mouse model using the Flp/Frt recombination system20 to conditionally express constitutively active TGFβ (FSFTGFβCA) in the fibroblastic compartment (Fsp1-Flpo), a cell population present in the microenvironment of almost all tissues. We chose the Flp/Frt approach with the final goal of combining this system with the Cre/lox system in dual recombination system (DRS models) in order to explore the effect of increased microenvironmental bioactive TGFβ as observed in many physiopathological contexts.

Results

We engineered the [FSFTGFβCA] mouse strain by homologous recombination in ES cells at the ROSA26 locus. Details of the experimental procedure to generate this strain are available in the experimental methods section (see Supplementary Figs. S1,2,3,4). Briefly, we constructed a recombinant targeting vector expressing the FSFTGFβCA transgene and encoding a Flp-conditional (FSF, Frt-STOP-Frt cassette) constitutively active (CA) TGFβ1 mutant (TGFβCA) (Fig. 1a). TGFβCA encodes a modified secreted polypeptide in which two cysteine residues (Cys223 and Cys225) are substituted with serine residues21, preventing the formation of the inactive LAP-TGFβ complex, thus driving the direct secretion of a bioactive TGFβ. In the presence of a recombinase of the Flp family, the transcriptional STOP signal can be excised and the TGFβCA transgene expressed under the control of the ubiquitous cytomegalovirus (CMV) early enhancer/chicken β-actin promoter (CAG). To facilitate the detection of cells expressing the transgene, a DNA sequence coding a fluorescent protein (eYFP) under the control of an internal ribosome entry site (IRES) was fused to the 3′ end of TGFβCA cDNA. In order to validate the conditional expression of the TGFβCA transgene ex vivo, we prepared ear skin primary fibroblasts from [FSFTGFβCA] mice. These cells were transiently transfected with an empty plasmid or a pSICO-Flpo plasmid expressing the Flpo recombinase under the control of the CMV promoter. We observed the recombined transgene (excision of the STOP cassette) by PCR on genomic DNA exclusively when the Flpo recombinase was expressed (Fig. 1b). A RT-PCR was performed on total RNA prepared from these transfected cells and revealed an increase in TGFβCA (Fig. 1c, left panel) and eYFP (Fig. 1c, right panel) mRNA levels in cells transfected with the pSICO-Flpo plasmid.

Figure 1
figure 1

Generation of the [FSFTGFβCA] mouse strain. (a) Transgenesis strategy to generate the [FSFTGFβCA] mouse strain. Site-directed TGFβCA transgene integration by homologous recombination into the ROSA26 locus and Flp-mediated excision of the transcriptional “Stop” allowing TGFβCA transgene expression are represented. Primers (p) used for PCR (panel b) and RT-PCR (panel c) are represented by grey arrowheads and detailed in Table 1. CAG, composite constitutive human cytomegalovirus enhancer and chicken beta-actin promoter; Neo, neomycin antibiotic resistance cassette; IRES, internal ribosome entry site; eYFP, enhanced yellow fluorescent protein; bGH polyA, bovine growth hormone polyadenylation signal. (b) PCR on genomic DNA prepared from [FSFTGFβCA] murine primary ear skin fibroblasts transfected either with the pSICO-Flpo plasmid or an empty plasmid to detect the Flpo sequence, the unrecombined FSFTGFβCA, and the recombined FSFTGFβCA alleles. (c) Quantitative RT-PCR on total mRNA prepared from [FSFTGFβCA] murine primary ear skin fibroblasts transfected either with the pSICO-Flpo plasmid or an empty plasmid to detect TGFβCA (left panel) and eYFP (right panel) mRNA. In b and c, one representative experiment performed 3 times with fibroblasts from different mice is presented. In c, Prism 7.0 (Graphpad) was used to create graphs and the error bars represent the standard deviation from technical duplicates.

Full size image

In order to assess the activity of the TGFβCA transgene in vivo, we developed a strategy aiming at expressing the transgene at an early stage during embryonic development in all cell lineages. To that end, we bred [FSFTGFβCA] mice with [Act-Flpe] mice (expressing the Flpe recombinase in all cell lineages from the blastocyst stage of development)22 to generate [Act-Flpe; FSFTGFβCA] compound individuals (Fig. 2a). Since TGFβ plays a crucial role in both development and adult homeostasis, it was expected that early-stage ubiquitous TGFβCA expression would result in notable deleterious effects. From 42 litters, 231 individuals were obtained (Fig. 2b). Among the four expected genotypes in the offspring, [Act-Flpe; FSFTGFβCA] individuals were significantly underrepresented in comparison with [WT], [Act-Flpe], [FSFTGFβCA] individuals. From the 21 [Act-Flpe; FSFTGFβCA] mice genotyped after birth, 5 had died close to or at birth. Thus, at weaning, only 16 [Act-Flpe; FSFTGFβCA] mice were alive and 2/3 of the expected [Act-Flpe; FSFTGFβCA] mice were missing. [Act-Flpe; FSFTGFβCA] survivors did not present any obvious external defects, even though 5 mice were sacrificed beyond the age of one year (Fig. 2c and sup Table 1). We verified by PCR on genomic DNA prepared from tail samples that the excision of the STOP cassette was detectable (Recombined FSFTGFβCA transgene) in these surviving mice (Fig. 2d). A significant increase in the expression of both TGFβCA (Fig. 2e, left panel) and eYFP (Fig. 2e, right panel) mRNA were observed ex vivo by RT-PCR performed on total RNA prepared from [Act-Flpe; FSFTGFβCA] mice ear skin samples compared to [FSFTGFβCA] mice. Western blot performed on whole protein extracts from skin revealed the presence of the eYFP protein only in extracts prepared from [Act-Flpe; FSFTGFβCA] mice (Fig. 2f). Altogether these data demonstrate the robustness of this mouse model.

Figure 2
figure 2

In vivo functional validation of the FSFTGFβCA allele. (a) Breeding strategy ([Act-Flpe] x [FSFTGFβCA]) to generate [Act-Flpe; FSFTGFβCA] individuals. Primers (p) used for DNA genotyping (panel d) and RT-PCR (panel e) are represented by grey arrowheads. (b) [FSFTGFβCA] were crossed with [Act-Flpe] mice. Total numbers (black) and expected numbers (grey) of litters, pups, and offspring genotype distribution are presented. A χ² test was performed to statistically confirm the “loss” of a significant proportion of [Act-Flpe; FSFTGFβCA] individuals after birth. The Chi2 were calculated using Excel workbook developed by Montoliu36. (c) Kaplan Meyer disease free survival curve of mice of indicated genotypes. n represents the number of animals for each genotype. (d) PCR on genomic DNA prepared from tail snips of indicated genotypes to detect the ROSA26, Act-Flpe, the unrecombined FSFTGFβCA, and the recombined FSFTGFβCA alleles. (e) Quantification of TGFβCA (left panel) and eYFP (right panel) mRNA by RT-PCR on total RNA prepared from [FSFTGFβCA] and [Act-Flpe; FSFTGFβCA] ear skin samples. (f) Western blot analysis of eYFP and β-tubulin on whole protein extracts prepared from skin samples of indicated genotypes. In e and f, one representative experiment performed 3 times with skin samples from different mice is presented. In c and e, Prism 7.0 (Graphpad) was used to create graphs. In e, the error bars represent the standard deviation from technical duplicates.

Full size image

To understand the reasons underlying the lack of [Act-Flpe; FSFTGFβCA] individuals at weaning, we generated 153 embryos at different stages between E11.5 and E18 (Fig. 3a). We confirmed that a proportion (p = 0.094) of [Act-Flpe; FSFTGFβCA] embryos was missing prior to birth. Hence, we precisely analyzed embryos at different developmental stages and observed that approximately half of the [Act-Flpe; FSFTGFβCA] embryos had died by the E18 stage, according to the expected Mendelian proportions (Fig. 3b). We also observed macroscopically (gross anatomy) and by CT-scan (computerized tomography) scanner imaging, that half of the [Act-Flpe; FSFTGFβCA] embryos presented a severe growth retardation at the different stages (Fig. 3c). Brain (white arrowhead) and spinal column (grey arrowhead) defects were observed by CT-scan at stages E13 and E16. Altogether, these data demonstrate that the TGFβCA transgene can be conditionally expressed in the presence of Flp recombinase, and that the TGFβCA protein is functional in vivo as attested by i- the missing of a proportion of [Act-Flpe; FSFTGFβCA] late embryos and adults, and ii- the severe growth retardation observed in many [Act-Flpe; FSFTGFβCA] embryos.

Figure 3
figure 3

Developmental defects in [Act-Flpe; FSFTGFβCA] embryos. (a) [FSFTGFβCA] were crossed with [Act-Flpe] mice. Total numbers (black) and expected numbers (grey) of litters, embryos, and offspring genotype distribution are presented. A χ² test was performed to test the “loss” of a significant proportion of [Act-Flpe; FSFTGFβCA] individuals just before birth. The Chi2 were calculated using Excel workbook developed by Montoliu36. (b) Distribution of the four expected genotypes in the offspring of [Act-Flpe] x [FSFTGFβCA] at [E11-E13], [E14-E15], [E16-E18] and after birth. (b) Representative embryos of indicated genotypes at different stages of development (E11, E13, E16) visualized under the binocular (gross anatomy) and medial sagittal section by CT-Scan (computerized tomography scanner); 3D, three dimensions; 2D, two dimensions.

Full size image

[Fsp1-Flpo] mice were crossed with [FSFTGFβCA] mice to generate [Fsp1-Flpo; FSFTGFβCA] individuals (Fig. 4a). [Fsp1-Flpo; FSFTGFβCA] mice were obtained at the expected Mendelian ratio and did not present obvious phenotypic alterations. Some mice lived up to 2 years, similarly to control mice (Fig. 4b and sup Table 1). Recombination was verified on tail biopsies by PCR (Fig. 4c). As expected, RT-PCR revealed that Flpo was detected in immortalized ear skin fibroblasts prepared from [Fsp1-Flpo] and [Fsp1-Flpo; FSFTGFβCA] individuals (undetectable expression in [WT] and [FSFTGFβCA] fibroblasts) (Fig. 4d). TGFβCA mRNA was undetectable in [WT] and [Fsp1-Flpo] cells, barely detectable in [FSFTGFβCA] cells and highly expressed in [Fsp1-Flpo; FSFTGFβCA] cells (Fig. 4e). eYFP was not detected by Western blot analysis performed on whole protein extracts from skin (Fig. 4f, left panel) but on extracts from immortalized fibroblasts isolated from the skin of [Fsp1-Flpo; FSFTGFβCA] mice eYFP was detected (Fig. 4f, right panel).

Figure 4
figure 4

Generation of [Fsp1-Flpo; FSFTGFβCA] compound mice. (a) Breeding strategy ([Fsp1-Flpo] x [FSFTGFβCA]) to generate [Fsp1-Flpo; FSFTGFβCA] individuals. Flp-mediated excision of the transcriptional “Stop” leading to TGFβCA expression is represented. Primers used for DNA genotyping (panel c), RT-PCR (panel d, e and Fig. 5b–d) are represented by grey arrowheads. CAG, composite constitutive human cytomegalovirus enhancer and chicken β-actin promoter; Neo, neomycin antibiotic resistance cassette; IRES, internal ribosome entry site; eYFP, enhanced yellow fluorescent protein; bGH polyA, bovine growth hormone polyadenylation signal. Fsp1, Fibroblast specific promoter 1; Flpo, Flpo recombinase; SV40, Simian Virus 40; polyA, polyadenylation signal. (b) Kaplan Meyer disease free survival curve of mice of indicated genotypes. n represents the number of animals for each genotype. (c) PCR on genomic DNA prepared from tail snips of indicated genotypes to detect the ROSA26, Fsp1-Flpo, the unrecombined FSFTGFβCA, and the recombined FSFTGFβCA alleles. (d) Detection of Flpo mRNA by RT-PCR performed on total mRNA prepared from ear skin immortalized fibroblasts of indicated genotypes. (e) Detection of TGFβCA mRNA by RT-PCR performed on total mRNA prepared from ear skin immortalized fibroblasts of indicated genotypes. (f) Western blot analysis of eYFP and β-tubulin on whole protein extracts prepared from skin samples (left panel) and immortalized fibroblasts extracted from skin (right panel) of indicated genotypes. In (c–f) one representative experiment performed 3 times with biological material from different mice is presented. In b, d and e, Prism 7.0 (Graphpad) was used to create graphs. In d and e, the error bars represent the standard deviation from technical duplicates.

Full size image

In order to determine whether the mutant TGFβ is efficiently secreted as an active form, we measured the activity of the cytokine in the conditioned medium (CM) of immortalized ear skin fibroblasts. To that end, we transiently transfected HepG2 cells with a TGFβ-sensitive luciferase reporter and observed that luciferase activity was significantly higher in cells cultured in the presence of CM from [Fsp1-Flpo; FSFTGFβCA] fibroblasts compared to [WT] (Fig. 5a). We isolated fibroblasts from [WT], [FSFTGFβCA], [Fsp1-Flpo; FSFTGFβCA] and [Act-Flpe; FSFTGFβCA] mice back skin via cell sorting (Fig. 5b). [Act-Flpe; FSFTGFβCA] (ubiquitous expression of the transgene) mice were used as positive controls. RT-PCR confirmed the presence of PDGFRα mRNA in unsorted skin cells from all genotypes, and a clear enrichment after cell sorting using the PDGFRα antibody in PDGFRα-positive cells. Flpo recombinase was detected only in the skin of [Fsp1-Flpo; FSFTGFβCA] mice, and was clearly enriched in PDGFRα-positive cells after cell sorting (Fig. 5c). TGFβCA mRNA was detected in cells from [Fsp1-Flpo; FSFTGFβCA] mice and was enriched in PDGFRα-positive cells after cell sorting (Fig. 5d, left panel). As expected from the ubiquitous expression of the Flpe recombinase in [Act-Flpe; FSFTGFβCA] mice (not restricted to PDGRα-positive cells), TGFβCA was highly expressed in all cell lineages from [Act-Flpe; FSFTGFβCA] mice before and after cell sorting (Fig. 5d, right panel). Overall, these data confirm that [Fsp1-Flpo; FSFTGFβCA] mainly express TGFβCA in fibroblasts.

Figure 5
figure 5

Functional validation of [Fsp1-Flpo; FSFTGFβCA] mice. (a) Reporter assay on HepG2 cells transiently transfected with a TGFβ-sensitive luciferase reporter plasmid (CAGA9-Luc) and cultured with the conditioned medium (CM) of immortalized skin fibroblasts from the indicated genotypes. (b–d) Quantification of PDGFRα (b), Flpo (c) and TGFβCA (d) mRNA by RT-PCR on total RNA prepared from cells with indicated genotypes present in back skin before and after cell sorting of PDGRα-positive cells. In (a–d), one representative experiment performed 3 times with biological material from different mice is presented. Prism 7.0 (Graphpad) was used to create graphs and the error bars represent the standard deviation from technical duplicates.

Full size image

Discussion

To explore in vivo the effect of increased local activity of TGFβ, we created a conditional transgenic mouse model (Flpo/Frt system) that expresses bioactive TGFβ in fibroblasts. We created a new mouse allele, i.e. the FSFTGFβCA allele encoding a constitutionally active mutant form of TGFβ (TGFβCA) under the control of a Frt-STOP-Frt (FSF) sequence, with the Fsp1-Flpo allele encoding the Flpo recombinase under the control of the Fsp1 promoter. We validated the expression and the function of the FSFTGFβCA allele by several in vitro, ex vivo and in vivo approaches. Particularly, we generated [Act-Flpe; FSFTGFβCA] individuals (expressing the transgene ubiquitously) and observed that a proportion of them was missing at weaning likely as a result of the severe growth retardation observed in many [Act-Flpe; FSFTGFβCA] embryos at different stages of development. The few [Act-Flpe; FSFTGFβCA] adult survivors looked normal. We next generated [Fsp1-Flpo; FSFTGFβCA] mice which presented no obvious phenotype either despite the fact that the FSFTGFβCA allele was efficiently recombined and expressed in cells expressing the Flpo recombinase.

We were able to detect by RT-PCR very low levels of TGFβCA mRNA in the absence of Flp recombinase in [FSFTGFβCA] mice. However, we demonstrated that this weak leak was not sufficient to induce phenotypical disorders, as attested by the absence of phenotype in [FSFTGFβCA] mice, contrasting with the severe developmental defects we observed in [Actin-Flpe; FSFTGFβCA] embryos. Transgenic constitutively active TGFβ cannot be detected as the mutations do not occur in mature TGFβ21. Indeed, the mutations of cysteine residues 223 and 225 are present in the pro-domain thus preventing its interaction with mature TGFβ in a noncovalent manner to form a latent (inactive) complex. To circumvent this detection issue and be able to indirectly monitor the expression of the TGFβCA transgene, a sequence encoding a fluorescent reporter protein (eYFP) under the control of an IRES sequence was added to the transgene. eYFP mRNA was detected by RT-PCR in total RNA extracts from [Act-Flpe; FSFTGFβCA] mouse skin samples. We also detected the eYFP protein by Western blot analysis in whole protein extracts prepared from [Act-Flpe; FSFTGFβCA] mouse skin samples. Although we were unable to detect this protein in [Fsp1-Flpo; FSFTGFβCA] skin samples, Western blot analysis performed on immortalized fibroblasts prepared from [Fsp1-Flpo; FSFTGFβCA] mouse skin samples highlighted the protein. Of note, the eYFP protein was undetectable in situ both in cell culture and fixed or frozen tissues. Indeed, we were unable to visualize the eYFP protein, either directly (direct immunofluorescence) or indirectly (using a fluorescent or a biotinylated anti-YFP antibody). Even though we cannot rule out technical issues, the absence of detection likely results from a low level of protein expression due to a mechanism that remains to be determined (no translation or unstable protein for instance).

In order to validate the in vivo function of the TGFβCA transgene, we generated [Act-Flpe; FSFTGFβCA] individuals expressing TGFβCA in all cell lineages from the early embryonic stages. We observed that two thirds of these individuals were missing at weaning, and that most of the surviving embryos presented severe growth defects by the E15.5 stage. Such developmental retardations are not compatible with life after birth, likely explaining the absence of a great proportion of [Act-Flpe; FSFTGFβCA] individuals at weaning. The precise developmental defects at the origin of these abnormalities have not been addressed. However, we observed morphological features in accordance with an increased TGFβ signaling. Indeed, [Act-Flpe; FSFTGFβCA] embryos presented an abnormal forebrain development as described previously after injection of TGFβ in the brain of embryos23.

To express TGFβCA transgene in the fibroblasts, we generated [Fsp1-Flpo; FSFTGFβCA] individuals. Fsp1 is expressed at a late stage of embryonic development preventing embryonic lethality24,25,26. We were able to observe the expression of Fsp1 protein in fibroblasts directly with an anti-Fsp1 antibody, and, Fsp1 promoter activity with an in vivo reporter system ([Fsp1-Flpo; FSFhPLAP]) (under review elsewhere). The Fsp1 promoter has been successfully used by other groups to generate [Fsp1-Cre x TβRIILoxP/LoxP] mice carrying homozygous inactivation of type II TGFβ receptor in the stromal compartment26,27. These authors reported that abrogation of the canonical TGFβ signaling pathway in Fsp1-positive cells was compatible with full term embryonic development, though the males were sterile and both males and females died around 2 months of age, males developing preneoplastic lesions in the prostate epithelium and both sex developing squamous carcinoma of the forestomach. Fibroblast-specific activation of TGFβ signaling (constitutively active ligand overexpression or expression of a constitutively active receptor) has never been explored before and the present work is the first study to address this question, to our knowledge. We observed the absence of an obvious phenotype in [Fsp1-Flpo; FSFTGFβCA], contrasting with the phenotype observed in other models reporting an over-activation of TGFβ in specific tissues such as the mammary gland6 or head and neck epithelium8 and in other cell types such as hepatocytes9 or lung cells10. These observations may indicate that fibroblasts are not a sufficient source of bioactive TGFβ to induce defects observed in these models or may result from a transgene expression level too low to induce defects, or the presence of too few fibroblasts in organs in physiological conditions, or a lower bioactivity of the simian TGFβ used in this study compared to the porcine version used in other studies. Finally, and most likely, the [Fsp1-Flpo; FSFTGFβCA] model needs to be challenged with external or internal stresses to unveil a clear phenotype. For instance, it will be of particular interests to challenge these mice with ionizing radiations, pro-inflammatory stimuli or oncogenic chemicals or transgenes.

Cre-lox models do not easily permit independent temporal or spatial manipulation of different targeted genes. The combination of Flpo-Frt and Cre-loxP recombination systems or DRS bypasses this limitation28,29. Tissue-specific Flp drivers and Frt-flanked transgenes remain rare compared to tissue-specific Cre drivers and Floxed transgenes. Consequently, the development of DRS has been hindered. The Fsp1-[Flpo/FSFTGFβCA] mouse model that we created in the present study, represents a significant contribution in the development of DRS in both fields of TGFβ signaling (FSFTGFβCA) and microenvironment (Fsp1-Flpo). Indeed, our system paves the way for studies attempting to express bioactive TGFβ by fibroblasts (or another compartment if another specific Flp driver is used) in any Cre/lox-based model. Uncoupled spatiotemporal regulation of different genetic alterations using DRS should enable scientists to develop models better mimicking the complexity and heterogeneity of human diseases.

Methods

Biological material

Cells

HepG2 cells (ATCC® HB-8065™) were cultured in complete medium (Dulbecco’s Modified Eagle’s Medium, supplemented with 0.03% L-glutamine and containing 10% fetal bovine serum, a mix of 100 U mL−1 penicillin and 100 µg mL−1 streptomycin sulfate, 1% MEM non-essential amino acids) and propagated at 37 °C under 5% (v/v) CO2 atmosphere. Primary ear fibroblasts were isolated and cultured as follows: mouse ears were rinsed with 70% ethanol and samples of a few square-millimeters were harvested. Primary ear fibroblasts were isolated using mechanical dilacerations, followed by enzymatic dissociation (600 µL of a mix of collagenase D (4 mg mL−1, COLLD-RO Roche) /Dispase II (4 mg mL−1, Roche) in RPMI medium) at 37 °C for 1 h. The reaction was stopped by adding 5 mL of complete medium (Dulbecco’s Modified Eagle’s Medium, supplemented with 0.03% L-glutamine and containing 10% fetal bovine serum, a mix of 100 U mL−1 penicillin and 100 µg mL−1 streptomycin sulfate) and cells were then incubated at 37 °C overnight. The following day, after filtration through a 100 µm pore cell strainer, cells were pelleted and reseeded in complete medium (Dulbecco’s Modified Eagle’s Medium, supplemented with 0.03% L-glutamine and containing 10% fetal bovine serum, a mix of 100 U mL−1 penicillin and 100 µg mL−1 streptomycin sulfate, 1% MEM non-essential amino acids and 50 µM β-mercaptoethanol). Medium was changed after 48 h. All cells were propagated at 37 °C under 5% (v/v) CO2 atmosphere. Immortalized fibroblasts were prepared as follows: primary fibroblasts were cultured to reach confluence in 100 mm cell culture dishes, split in two and grown in a 75 cm² flasks. At confluence, cells were divided, retaining only one third. This step was repeated until the cells entered into senescence. The medium was then changed every 2/3 days until immortalized clones grew. After immortalization, cells were cultured without β-mercaptoethanol. Sorted skin cells were prepared as follows: mice were sacrificed and shaved, and a piece of skin from the back of the animals was harvested and fat was removed. 5 mL of digestion medium (RPMI1640, 20% FBS, 1% PS, 1% Hepes, 1% Glutamine) was added to the skin in a petri dish. The skin was dilacerated using scissors and a cutter. Another 5 mL of digestion medium was added to help harvest the mixture. The mixture was then transferred to 50 mL tubes under agitation with a magnetic bar. 1 mL of collagenase type IA (10 mg/mL, C2674-1G (NA56)) and 200 µL of DNase1 (10 mg/mL, 11284932001 125,10) were added. The mix was agitated 30–90 min at 37 °C. The insoluble debris were eliminated by filtration onto a nylon grid. The digestion product was then centrifuged 5 min at 1,300 rpm and suspended in cell sorting buffer (1X PBS, 1% BSA, 0.5 mM EDTA). Cells were labelled in cell sorting buffer with PE anti-mouse CD140a/ PDGFRα antibody (BioLegend Cat. No. 135905 1:50) and PE Rat IgG2a, κ Isotype Control Antibody (BioLegend Cat. No. 400507 1:50). Cells were sorted using the BD FACS Aria III SORP (BECTON DICKINSON (BD™)).

Mice

The experiments were performed in compliance with the animal welfare guidelines of the European Union and with the French legislation. All experimental protocols (CECCAPP protocol #CLB-2012-012 #CLB-2017-007 #CLB-2018-025) were approved by the CECCAPP Région Rhône-Alpes ethical committee https://www.sfr-biosciences.fr/ethique/ceccapp/ceccapp under the control of the Ministère de l’Enseignement Supérieur et de la Recherche (#C2EA15) http://www.enseignementsup-recherche.gouv.fr/cid70597/l-utilisation-des-animaux-a-des-fins-scientifiques.html.

The FSFTGFβCA allele was genetically engineered as follows by the ICS (Institut Clinique de la Souris, Strasbourg, France): we obtained the simian TGFβCA cDNA clone inside the πH3M plasmid21 from Pr. H. L. Moses, Vanderbilt-Ingram Cancer Center. We verified that the coding sequence contained the two point mutations C223S/C225S (data not shown). We functionally validated the TGFβCA-expressing vector as attested by a 400-fold activation of the luciferase activity of a TGFβ-sensitive reporter gene transfected in HepG2 cells (see Supplementary Fig. S1). TGFβCA cDNA was further inserted into a targeting vector plasmid (see Supplementary Fig. S1) and fully sequenced (see Supplementary Fig. S2). [FSFTGFβCA] mouse strain was generated by homologous recombination in ES cells after vector injection into C57BL/6 J blastocysts. 93 screened clones were injected and 3 positive clones (#51; #62; #77), containing the FSFTGFβCA transgene inserted at the ROSA26 locus, were identified by PCR on genomic DNA (see Supplementary Fig. S3). Primers used for PCR are detailed in Table 1. At that point, 3 positive clones (#51; #62; #77) were validated by Southern blot with internal probes (see Supplementary Fig. S3). Digestions sites used to validate the 5′ and 3′ insertions are given in Table 2. 6 different digests are used to validate correct HR event. 3 digests validate the 5′ insertion, 3 other digests validate the 3′ insertion. Then, the 3 positive clones (#51; #62; #77) were validated by Southern blot with external probes (see Supplementary Fig. S3). Digestions used to validate with 5′ and 3′ probes are detailed in Table 2. One digest was used with 5′ probe and 3 different digests were used with 3′probe. After blastocyst injection, 2 out 3 ES clones (#51 and #62) gave rise to 8 chimeras (4 per ES clone). 2 heterozygous males and 2 heterozygous females were obtained after the breeding #51 chimera with C57BL/6 J mice, arguing in favor of an efficient germinal transmission of the transgene. The 2 heterozygous males were used for rederivation. Among the 9 pups obtained, 6 were positive for the transgene (2 males and 4 females) (see Supplementary Fig. S4). Primers used are positioned on gene sequence in Supplementary Fig. S3. These mice were then imported into our mouse facility (AniCan).

Table 1 List of primers.
Full size table
Table 2 Restriction enzymes sites on DNA.
Full size table

We generated the Fsp1-Flpo allele (in review elsewhere). Briefly, it is composed of a ~3,427 bp Fsp1 promoter fragment (Exon_1/Intron_1/Partial_Exon_2; ENSMUSG00000001020) corresponding to the region previously described to drive the expression of Fsp1 in fibroblasts25,26,30. The Flpo transgene is a mouse codon-optimized Flp (Flpo) site-specific recombinase (SSR), which recombines DNA Frt-sites31.

The Act-Flpe allele was previously described22 and was obtained from the Jackson lab (B6;SJL-Tg(ACTFLPe)9205Dym/J, Stock No: 003800).

Mice were housed and bred in the “AniCan” specific pathogen-free animal facility of the CRCL (Centre de Recherche en Cancérologie de Lyon), France. The experiments were performed in compliance with the animal welfare guidelines of the European Union and with the French legislation (CECCAPP protocol #CLB-2012-012 #CLB-2017-007 #CLB-2018-025).

Embryos

Pregnant mice were killed by cervical dislocation. E11.5 to E18 embryos were removed from [Act-Flpe] or [FSFTGFβCA] females previously impregnated by [FSFTGFβCA] or [Act-Flpe] males, respectively, put on a 100 mm petri dishes on ice, and sacrificed by hypothermia in cold PBS. Neonates were decapitated. A piece of tail was sectioned for genotyping. Embryos were fixed 24 h in formalin and then stocked in PBS at 4 °C before use. Macroscopic images were acquired using SteREO Lumar.V12 coupled with AxioCam MRc5 (Zeiss).

For CT (computerized tomography)-scan analysis, formalin-fixed embryos were incubated 24 h under agitation in rotation at room temperature in 0.0125 M lugol (Sigma) (renewed every 8 h), then included in 1% agarose in a 11 mm in diameter centrifuge tube (Beckman coulter). Tubes with embryos in agarose were scanned (55kVp 181 µA, exposure time: 12.25 s, 0.5 mm aluminum filter) using the Perkin Elmer Quantum FX®. Medial sagittal sections were obtained using a Caliper analyze (AnalyzeDirect). 3D reconstruction was performed using 3DViewer. Gimp was used for improving the quality of the pictures obtained.

Cell biology

Cell transfection

At day 1,200,000 ear fibroblasts were plated in 12-well plates. At day 2, cells in 1 mL of complete medium (Dulbecco’s Modified Eagle’s Medium, supplemented with 0.03% L-glutamine and containing 10% fetal bovine serum, a mix of 100 U mL−1 penicillin and 100 µg mL−1 streptomycin sulfate, 1% MEM non-essential amino acids) were transfected with 100 µL of transfection mix. These mix were maintained for 20 min at room temperature before transfection (800 ng of pSICO-Flpo plasmid (Addgene, 24969; Tyler Jacks), 4 µL of Lipofectamine® 2000 Transfection Reagent (Invitrogen), Opti-MEM medium (Thermofisher) to 100 µL). Medium was changed 6 h after transfection and replaced by a complete medium (Dulbecco’s Modified Eagle’s Medium, supplemented with 0.03% L-glutamine and containing 10% fetal bovine serum, a mix of 100 U mL−1 penicillin and 100 µg mL−1 streptomycin sulfate, 1% MEM non-essential amino acids). 48 h after transfection, wells were rinsed with PBS. For gDNA extraction, cells were lysed 30 min at 95 °C using 100 µL of Lysis buffer (25 mM NaOH, 0.2 mM EDTA) and 100 µL of neutralization buffer was added to stop the reaction (40 mM Tris-HCl).

Conditioned medium production and concentration

For conditioned medium (CM) production from cell cultures of primary fibroblasts, 300,000 cells were seeded onto a 6-well culture plate in complete DMEM. Following overnight culture for cell attachment, the medium was replaced by DMEM containing 0.5% FBS. After 24 h of culture, conditioned media were centrifuged at 1,200 rpm and cell supernatants were harvested and stored at −80 °C.

Luciferase assay

To measure the activity of TGFβ produced by fibroblasts from [Fsp1-Flpo; FSFTGFβCA] mice, and test the functionality of the TGFβCA transgene, on day 1, 30,000 HepG2 cells were seeded onto a 12-well culture plate in complete medium (Dulbecco’s Modified Eagle’s Medium, supplemented with 0.03% L-glutamine and containing 10% fetal bovine serum, a mix of 100 U mL−1 penicillin and 100 µg mL−1 streptomycin sulfate, 1% MEM non-essential amino acids). On day 2, HepG2 cells were transfected with 500 ng CAGA9-Luc (pGL3 MLP CAGA32) and 10 ng renilla (phRL-CMV) plasmids. For the functionality test of TGFβCA transgenes, we also transfected 800 ng of empty vector (pCMV5B WT, derived from pCMV5B hTbR2 addgene)33 or TGFβ plasmid (pTGFβCA πH3M simianTGFbeta1 CA C223S/C225S, Harold L. Moses (Vanderbilt-Ingram Cancer Center, Nashville, TN, USA). On day 3, HepG2 cells were deprived 1 h at 37 °C using 0.5 mL MEM 0% FBS. We added 0.5 mL/well of conditioned medium and treated or not with TGFβ purchased from Abnova (10 ng/mL final concentration). On day 4, cells were rinsed twice with PBS 1X, lysed 20–30 min with 50 µL of Passive Lysis Buffer 1X (Promega) under agitation and centrifuged 12,500 rpm for 1 min at room temperature. 50 µL of Luciferin substrate (Promega) and 10 µL of supernatant were mixed in a 96-well white plate and Firefly activity was read using the microplate reader infinite M1000 pro (TECAN). 50 µL of StopandGlo® reagent (Promega) was added and the renilla activity was read using the microplate reader.

Molecular biology

Genomic and recombination PCR

DNA extraction and PCR were performed as previously described12 by using Taq DNA polymerase (Life Technologies) and primers cited in Table 1.

RNA/DNA extraction

RNA extraction from sorted skin cells: cells from cell sorting where lysed and RNA extracted using RNeasy Micro Kit (Qiagen).

Total RNA was extracted from transfected cells and purified from cells using lysis buffer from RNeasy Mini Kit (Qiagen).

RNA extraction from organs

Total RNAs were extracted from organs using ultraturax to mix them in a home-made RNA extraction solution (Guanidine thiocyanate 5 M (Sigma G6639), Citrate de sodium 0.5 M pH 7.0, Lauryl sarcosine 10%, βmercaptoethanol 1%). Purification was performed using the RNeasy Mini Kit (Qiagen).

Reverse-transcription PCR

First-strand cDNAs were prepared using 250 ng of RNA and SuperScript II Reverse Transcriptase in the presence of random primers (Invitrogen)34. Semi-quantitative PCR on cDNA was performed as previously described35 and using an Applied Biosystems StepOnePlus Real-Time PCR System with MESA GREEN qPCR MasterMix Plus (Eurogentec). All the real-time values were averaged and compared using the threshold cycle (CT) method, in which the amount of target RNA (2−ΔΔCT) was normalized against the endogenous expression of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (ΔCT). The amount of target mRNA in control cells or tissues was arbitrarily set at 1.0. Primers used for PCR are listed in Table 1.

Western blots

For protein analysis, cells were washed once with cold phosphate-buffered saline (PBS), and lysed with RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% sodium deoxycholate, 0.1% SDS, 1% Nonidet) containing a protease and phosphatase inhibitor cocktail. For tissue samples (skin), the tissue was homogenized (by grinding) in RIPA buffer. After protein quantification, 100 μg of protein was used for total lysate samples. Samples were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and detected by immunoblot (WB) using Amersham ECL Detection Reagent (GE Healthcare). We used anti-GFP to detect YFP protein (Cell Signaling #2956) and anti-β-tubulin (Sigma T5293) primary antibodies and anti-rabbit or anti-mouse HRP-coupled antibodies from Dako (#P0260 and #P0448, respectively).