The SenKiller Mouse: Enabling Cell Type–Specific Ablation of Senescent Cells

Senescent cells play a critical role in hepatic injury and regeneration.^1,2 Multiple senescent hepatic cell types have been implicated in liver fibrosis, including hepatocytes, hepatic stellate cells, and cholangiocytes. Senescent hepatocytes,³ as well as cholangiocytes,⁴ have been shown to accumulate in chronic liver disease and may contribute to the pathogenesis of fibrosis through the paracrine action of their secreted senescence-associated secretory phenotype factors, which activate hepatic stellate cells, which then secrete extracellular matrix components, resulting in fibrosis.³ Although the senescence of Kupffer cells has not been studied extensively,⁵ increased cytokine production by senescent Kupffer cells was recently implicated in D-galactose–induced aging in mice.⁶

Previously, the impact of senescent cells on liver health was studied in mouse models employing systemically acting senolytic agents or chimeric antigen receptor T cells, which target all senescent cells in the body. By definition, these studies could not pinpoint the relevant cell type(s) responsible for the beneficial effect of eliminating senescent cells. Therefore, cell type–specific senescent cell ablation is a critical next step in the mechanistic understanding of the impact of senescence in liver biology and in validating specific senescent cells as potential therapeutic targets for hepatic fibrosis. To address this limitation, we developed a novel, cell type–specific senolysis mouse model, the ‘SenKiller’ mouse (Figure 1). This mouse employs a fragment of the p16Ink4a (Cdkn2a) promoter, a loxP-stop-loxP cassette harboring an mCherry fluorescent reporter, and an enhanced green fluorescent protein-tagged diphtheria toxin receptor (DTR-eGFP) to effect cell death following Cre-mediated recombination and diphtheria toxin (DT) administration (Figure 1). In combination with the multitude of available cell type–specific Cre-driver lines targeting the major hepatic cell types, the SenKiller mouse enables the identification of the cell type(s) responsible for the positive effects of senolytic approaches. Before Cre-mediated excision, the senescent cells are marked by mCherry expression, but after Cre activation, the mCherry cassette is excised, and the p16/Cdkn2a promoter drives production of the DTR-eGFP fusion protein. Subsequently, eGFP-positive senescent cells can be eliminated through the administration of DT (Figure 1), which does not target native murine cells.⁷

Figure 1.

Derivation of the SenKiller mouse. (A) Mouse embryonic stem (mES) cells were electroporated with the SenKiller targeting vector and a flippase-containing vector (pCAGGS FLPe). Flippase mediates recombination between the targeting vector Frt site and the pair of Frt sites in the ‘homing cassette.’ Correct integration into the homing cassette results in the entire targeting vector being integrated into the Col1a1 locus, providing the hygromycin resistance gene (Hygro) with a promoter and ATG start codon. The correct integration renders the mES cells resistant to hygromycin. The clones that survived hygromycin treatment were confirmed with Southern blot and then injected into mouse blastocysts, which were subsequently implanted into a pseudo-pregnant female mouse. The resulting pups were chimeric and were screened for germline transmission by outcrossing to C57BL/6 mice. (B) A representative image demonstrating germline transmission and establishment of the SenKiller line. (C) A schematic representation of the SenKiller transgenic cassette and diphtheria toxin (DT)-mediated cell death. The SenKiller cassette employs a fragment of the p16Ink4a (p16, encoded by Cdkn2a) gene promoter to drive the expression of a LoxP-stop-LoxP cassette harboring an mCherry fluorescent reporter and a diphtheria toxin receptor (DTR)-enhanced green fluorescent protein (eGFP) fusion protein. Before Cre-mediated recombination, the p16 promoter drives the expression of mCherry in p16+ cells, marking p16+ senescent cells across the whole organism. After Cre-mediated recombination, p16+ cells express the DTR-eGFP fusion protein, which gets inserted into the plasma membrane. When DT is administered, it will bind to DTR, and the complex will be endocytosed. The A subunit dissociates from the B subunit–DTR complex and inhibits protein synthesis by catalyzing the ADP-ribosylation of eEF-2. WT, wild-type.

As an initial test of the potential usefulness of the SenKiller mouse for the analysis of fibrotic injury to the liver, we treated young adult mice retro-orbitally with the AAV8-Tbg-Cre virus to activate the DTR-eGFP cassette specifically in hepatocytes. To induce hepatic steatosis, fibrosis, and senescence, SenKiller mice were fed a choline-deficient, L-amino acid–defined, high-fat diet,⁸ and mice were treated with DT to eliminate senescent hepatocytes or were given saline (Supplementary Figure 1A). Recombinant senescent hepatocytes were identified in liver sections by immunostaining for eGFP (Figure 2A and B) or by staining for senescence-associated beta-galactosidase activity (Figure 2C). Administration of DT reduced the frequency of senescent hepatocytes (Figure 2A–C) and hepatic p16 levels (Figure 2D) as expected. Thus, the SenKiller model enables the cell type–specific killing of senescent cells. There was no difference in body weight between the 2 groups (Supplementary Figure 1B), and both groups exhibited liver injury after 8 weeks on the diet, as indicated by elevated alanine aminotransferase levels in the blood (Supplementary Figure 1C).

Figure 2.

The SenKiller model enables the cell type–specific ablation of senescent cells. (A–C) Administration of diphtheria toxin (DT) significantly reduced the frequency of senescent hepatocytes. (A) Representative images of immunohistochemical (IHC) staining for enhanced green fluorescent protein (eGFP). (B) Representative images of immunofluorescent (IF) costaining for eGFP and hepatocyte nuclear factor 4 α (HNF4α). (C) Representative images of β-galactosidase (β-gal) activity staining. (D) Western blot of the analysis of the senescence marker p16 in control- and DT-treated SenKiller mice. β-Tubulin was employed as a loading control. (E–L) The hepatocyte-specific ablation of p16+ senescent cells has a minimal impact on liver injury in the choline-deficient, L-amino acid–defined, high-fat diet–induced metabolic dysfunction–associated steatohepatitis paradigm. (E and F) Representative images and quantification steatosis from hematoxylin and eosin (H&E)-stained liver sections (n = 3). (G) Hepatic triglyceride (TG) content (n = 3). (H and I) Representative images and quantification of Sirius Red staining (n = 3). (J) The expression of fibrosis markers Col1a1 and Acta2 between the saline and DT groups as determined by quantitative reverse transcription polymerase chain reaction analysis (n = 3). (K) Representative images of IF costaining for FoxA1, a marker of hepatocytes, and Ki67. (L) Representative images of IF costaining for cytokeratin 19 (CK19), a marker of cholangiocytes, and Ki67. DAPI, 4′,6-diamidino-2-phenylindole; mRNA, messenger RNA.

Next, we asked whether the elimination of senescent hepatocytes led to an improvement in hepatic health in mice exposed to the choline-deficient, L-amino acid–defined, high-fat diet paradigm. Steatosis was assessed in tissue sections (Figure 2E and F) and by biochemical quantification of hepatic triglyceride content (Figure 2G) but showed no differences between the 2 groups. Fibrosis was evaluated using Sirius Red staining (Figure 2H and I), which also did not show a benefit of the elimination of senescent hepatocytes. In addition, messenger RNA levels of the key fibrosis markers, type 1 collagen (Col1a1) and α-smooth muscle actin (encoded by Acta2), were not different between the 2 experimental groups (Figure 2J). Hepatocyte proliferation, as assessed by dual immunofluorescence staining for FoxA1 and Ki67, did not differ by treatment (Figure 2K). Likewise, the ductular reaction was similar between the 2 groups (Figure 2L). Thus, in this experimental model of metabolic dysfunction–associated steatohepatitis, the elimination of senescent hepatocytes shows no benefit. It is possible that the ablation of other senescent cell types in the liver would improve health outcomes or that the killing of senescent hepatocytes in other paradigms of metabolic dysfunction–associated steatohepatitis would alleviate symptoms.

In sum, we have developed and validated an experimental model system that will improve the mechanistic understanding of the impact of senescence on liver injury by enabling the specific killing of any senescent cell type(s) of interest. These mice will be made available to the research community via the Mutant Mouse Resource and Research Centers (Catalogue number 71913). Of note, the SenKiller model can be used to study senescence in any cell type or disease context.