A senolytic immunotoxin eliminates p16 INK4a -positive T cells and ameliorates age-associated phenotypes of CD4 + T cells in a surface marker knock-in mouse

Senescent cells were recently shown to play a role in aging-related malfunctions and pathologies. This consensus has been facilitated by evidence from senolytic model mice capable of eliminating senescent cells in tissues using well-characterized senescent markers, such as p16 INK4a (hereafter p16). However, since the incomplete or arti- ficial gene expression regulatory regions of manipulated marker genes affect their cognate expression, it currently remains unclear whether these models accurately reflect physiological senescence. We herein describe a novel approach to eliminate p16 -expressing cells from mice at any given point in time, generating a new type of knock-in model, p16 hCD2 mice and a toxin-conjugated anti-human CD2 antibody (hCD2-SAP) as an inducer. p16 hCD2 mice possess an intact Cdkn2a locus that includes a p16 coding region and human CD2 (hCD2) expression unit. We confirmed cognate p16 -associated hCD2 expression in mouse embryonic fibroblasts (MEFs) and in several tissues, such as the spleen, liver, and skin. We detected chronological increases in the hCD2-positive population in T lymphocytes that occurred in a p16 -dependent manner, which reflected physiological aging. We then confirmed the high sensitivity of hCD2-SAP to hCD2 and validated its efficacy to remove p16 - positive cells, particularly in T lymphocytes. The multiple administration of hCD2-SAP for a prolonged p16 - positive cell deficiency partially restored aging-related phenotypes in T lymphocytes, such as the contraction of the CD4 + naïve population and expansion of senescence-associated T cells. Our novel approach of targeting p16- positive senescent cells will provide novel insights into the mechanisms underlying physiological aging in vivo .


Introduction
Irreversible growth-arrested senescent cells accumulate in various tissues with normal aging. These cells arise from several stress signals, such as DNA damage, oxidative stress, and oncogene activation (Herranz and Gil, 2018;Gorgoulis et al., 2019), and also contribute to the promotion of tissue repair and suppression of tumorigenesis (Demaria et al., 2014;Sharpless et al., 2004). On the other hand, these cells represent a senescence-associated secretory phenotype (SASP), in which senescent cells produce proinflammatory factors, matrix metalloproteases, and (caption on next page) Y. Sugiyama et al. angiogenic factors (Coppé et al., 2008(Coppé et al., , 2010. The continuous exposure of older organisms, including senescent cells, to SASP causes age-related dysfunctions and diseases, such as cancer, diabetes, and arthritis (Malaquin et al., 2016;Rea et al., 2018). The Cdkn2a locus, which encodes p16 and p19 ARF (hereafter p19) that exhibit tumor suppressor activity, has recently been identified as a reliable marker of cellular senescence (Krishnamurthy et al., 2004). Several groups created genetically modified mice that eliminate p16-or p19-expressing senescent cells in vivo and revealed that the clearance of these cells from aged mice attenuated age-related disorders and promoted a healthy life span (Baker et al., 2011(Baker et al., , 2016Hashimoto et al., 2016). These model mice are manipulated by introducing a gene cassette designed to express drug-induced cell ablation genes and reporter genes driven by a p16 or p19 promoter. However, transgene expression may be affected by reporter alleles, which causes variations in the data obtained from these mice. For example, reporter genes did not function as expected in several tissues, such as the liver, colon, and T lymphocytes, in one of these mouse models (Baker et al., 2016). The use of a knock-in allele under the control of a cognate promoter enables more faithful transgene expression to the physiological p16 or p19 expression pattern. It has not yet been established whether the expression of knock-in products is too low to exert drug-induced cell death. Two groups created new reporter mice in which Gt(ROSA)26Sor or CAG-promoterdriven tdTomato and diphtheria toxin A or diphtheria toxin receptors are expressed under the control of Cre inserted into the endogenous p16 locus. These mice enable the monitoring and elimination of p16expressing senescent cells under more physiological conditions than in other transgenic models (Grosse et al., 2020;Omori et al., 2020). However, the expression levels of reporters may not accurately reflect cognate p16 expression in these reporter mice because of interventions by different promoters.
Antigen-antibody reactions are a widely accepted tool in molecular biology due to their high specificity and sensitivity. Especially in cancer biology or therapeutic antibodies in medical practice, toxin-conjugated antibodies, immunotoxins, are promising therapeutic strategies. More than twenty immunotoxins that target cancer cell-specific surface antigens have been assessed in clinical trials and some have already been approved, such as moxetumomab pasudotox (anti-CD22), denileukin diftitox (targeting CD25), and tagraxofusp (targeting CD123) (Li et al., 2022). In contrast, senescent cell research in aging has not yet ventured into this field, even in basic studies with animal models. Recent studies revealed rapid increases in senolytic approaches that target surface antigens on senescent cells, including chimeric antigen receptor T cells and vaccinations (Amor et al., 2020;Yoshida et al., 2020;Suda et al., 2021;Rossi and Abdelmohsen, 2021). These advances will contribute to the future use of "senolytic immunotoxins".
We herein generated a new reporter mouse model, in which the gene encoding human CD2 (hCD2) was knocked into the endogenous p16, Cdkn2a locus. This mouse possesses the hCD2 transgene connected to the 2A peptide with the cognate p16 gene under the control of an endogenous p16 promoter and displays hCD2 on the surface of senescent cells in addition to p16 expression. We confirmed that the level of hCD2 on the surface of MEFs prepared from this mouse increased with that of p16 expression in a manner that depended on the passage number. We also demonstrated that hCD2-positive senescent cells were selectively removed by a toxin-conjugated antibody in vitro and in vivo, allowing us to visualize p16-positive T cells and investigate the age-associated phenotypes of splenic CD4 + T cells in vivo.

Materials and methods
2.1. Mouse model p16hCD2 mice were generated by the Combi-CRISPR method (Yoshimi et al., 2021) in C57BL/6N fertilized eggs using microinjection with Cas9, gRNA, and a targeting vector including exon 3 of the Cdkn2a, 2A peptide sequence and the hCD2 gene. Actually, we planned and confirmed the insertion of the hCD2 expression unit coding the GPIlinked human CD2-CD52 fusion protein at a location posterior to exon 3 of the Cdkn2a gene by sequence reading, since this fusion protein has been established as one of the exogenous functional surface markers (Komatsu et al., 2009;Kendal et al., 2011).
All mice were maintained under specific pathogen-free conditions. All animal experiments were performed with the approval of the Animal Care and Use Committee of the National Center for Geriatrics and Gerontology (NCGG) (Obu, Japan).

Total RNA extraction and quantitative RT-PCR
MEFs/tissues were lysed in TRI Reagent (Molecular Research Center, Cincinnati, OH, USA), followed by total RNA extraction based on the acid guanidinium thiocyanate-phenol-chloroform extraction method according to the manufacturer's instructions. A DNase treatment and reverse transcription to cDNA were performed using ReverTra Ace qPCR RT Master Mix with gDNA Remover (Toyobo, Osaka, Japan). Quantitative RT-PCR amplification was performed with Luna Universal qPCR Master Mix (New England BioLabs, Beverly, MA, USA) or THUNDER-BIRD Next SYBR qPCR Mix (Toyobo). Expression levels were normalized In vivo p16 and hCD2 expression in tissues from p16hCD2 mice. (A) Comparisons of p16 (left) and hCD2 (right) expression between young (5 weeks old, n = 4) and mature (10 months old, n = 3) mice in the indicated organs by a quantitative RT-PCR analysis. (B) Representative percentage of hCD2-positive cells in splenic CD3 + T cells from young (5 weeks old), adult (4 months old), and mature (10 months old) p16hCD2 mice by FACS analyses. Percentages indicate the proportion of hCD2-positive CD3 + cells in each gate. (C) Frequencies of hCD2-positive CD3 + cells in (B). Each symbol represents an individual mouse. Horizontal lines represent means. (D) Comparison of p16 expression in sorted CD3 + T cells with or without surface hCD2 expression by a quantitative RT-PCR analysis. Each dot represents an individual mouse (13-20 months old). Horizontal lines represent means. ND, not detected. *p < 0.05, **p < 0.01, ***p < 0.001. ns, not significant. The primers used were as follows: 5 ′ -CTACTGG CGCTGCCAAGGC-3 ′ and 5 ′ -GTGGGTGTCGCTGTTGAAGTC-3 ′ for Gapdh; 5 ′ -AATCCAGTGTC-GAGCCTGTCA-3 ′ and 5 ′ -CCGCTTATGTTGCTGGATGCT-3 ′ for hCD2/52. In evaluations of p16, we used two sets of primers: 5 ′ -GCCGTGTGCAT-GACGTG-3 ′ and 5 ′ -GCACCGTAGTTGAGCAGAAGAG-3 ′ for set A and 5 ′ -GCCGTGTGCATGACGTG-3 ′ and 5 ′ -GCACCGTAGTTGAGCAGAAGAG-3 ′ for set B. Primer set B was only used to evaluate sorted T cells in Fig. 2D.

TS2/18.1.1 antibodies and the development of hCD2-SAP
A hybridoma cell line producing anti-human CD2 antibodies (TS2/ 18.1.1) was purchased from the Developmental Studies Hybridoma Bank (Iowa, USA). Using this cell line, we obtained mouse ascites by a consignment service (Immuno-Biological Laboratories Co., Ltd., Japan). The TS2/18.1.1 monoclonal antibody was purified from ascites using the protein A-based purification column system, Ab-Rapid PuRe EX (ProteNova Co., Ltd., Japan). After purification, we utilized the laboratory service of Cytologistics (USA) to conjugate TS2/18.1.1 with saporin.

Trypan blue dye exclusion test
MEFs cultured with hCD2-SAP for the indicated time, as described elsewhere, were harvested in a tube. Each respective cell suspension was mixed with the same volume of 0.4 % trypan blue solution (Sigma-Aldrich), followed by live cell counting using a hemocytometer.

SA-βgal staining
MEFs prepared from p16hCD2 mice were cultured in glass-bottomed chambers with or without hCD2-SAP. Adherent MEFs were treated with bafilomycin A1 and incubated with SPiDER-βGal solution (Cellular Senescence Detection Kit, Dojindo, Kumamoto, Japan) containing Hoechst33342 (Dojindo) according to the manufacturer's instructions. Based on the preliminary evaluation of MEFs, only the treating concentration of SPiDER-βGal was five times higher than that of the manufacturer's recommendation in order to get enough signals. Fluorescence was detected using a fluorescence microscope BZ-X810 (Keyence, Osaka, Japan).

Administration of hCD2-SAP to mice
The hCD2-SAP solution was stocked at − 30 • C as small aliquots.
Prior to its administration, thawed solution was diluted to 100 μg/mL in PBS and subjected to filter sterilization using 0.22 μm MILLEX GV (Merck Millipore, Darmstadt, Germany). Ten micrograms of hCD2-SAP (100 μL of solution) per mouse were injected into the tail vein using a 27G needle.

FACS analyses of T cells
The spleen was dissected and mashed in cold PBS to dissociate splenocytes. After filtration with 100 μm disposable filters (Sysmex, Kobe, Japan), tissue was lysed in ACK lysis buffer on ice for 2 min and washed in cold PBS. Lymphocytes were treated with a FcR blocker (BioLegend, Cat#101319, TruStain FcX antibody) on ice for 30 min, followed by antibody staining on ice for 30 min, washing, and analysis/ sorting by FACS CantoII/AriaII (BD Bioscience). The following antibodies were used in FACS analyses and cell sorting: Cat#103005), and anti-CD62L-PE (BioLegend, Cat#104407).

Statistical analysis
The significance of differences was evaluated using the Student's ttest or Dunnett's test, where appropriate. A p value <0.05 was considered to be significant.

Establishment and validation of the knock-in p16hCD2 mouse model
To express an immunotoxin target molecule on the surface of p16positive senescent cells without affecting the transcription of p16 or p19, we inserted a self-cleaving viral 2A peptide sequence juxtaposed to the stop codon of p16 located in exon 3 of the Cdkn2a locus, followed by a gene encoding the hCD2 expression unit (Fig. 1A). After the generation of this knock-in mouse by gene editing, we prepared and cultured heterozygous MEFs to validate knock-in hCD2 expression in addition to cognate p16 expression. As expected, p16-dependent hCD2 mRNA expression increased with serial passages (Fig. 1B). Similarly, hCD2 mRNA expression levels strongly correlated with p16 levels measured at various passages (Fig. 1C). We then investigated hCD2 expression on the surface of p16hCD2 MEFs using a FACS analysis. Surface hCD2 expression strongly correlated with intracellular p16 expression and also increased with passage numbers, with 0 % of the double-positive population at passage number 1 (P1) increasing to 27.6 % by P5 (2 weeks of culture) (Fig. 1D). These results indicate that surface hCD2 expression in cells derived from p16hCD2 mice faithfully reflected the expression of the senescent cell marker p16.
To examine functional hCD2 expression with physiological aging, we assessed the expression of p16 and hCD2 in several tissues of young (5 weeks old) and mature (10 months old) mice. As shown in Fig. 2A, p16 expression levels in the spleen, liver, lungs, and skin were significantly higher in mature mice than in young mice. Similar to p16 expression, hCD2 expression levels tended to be higher in these tissues, with significant increases being observed in the spleen, liver and skin. Due to the apparent induction of p16 and hCD2 in the spleen with aging, we examined lymphocytes in the spleen in order to clarify whether the hCD2-positive population increased with aging. The frequency of hCD2- positive cells in CD3 + T cells increased with aging (Fig. 2B), with the percentage of hCD2-positive cells being 4-and 8-fold higher in 4-monthold adults and 10-month-old mature mice, respectively, than in 5-weekold young mice (Fig. 2C). A quantitative RT-PCR analysis revealed that the expression of p16 was restricted to the hCD2-positive population (Fig. 2D). These results indicate that surface hCD2 expression in vivo is detectable by a FACS analysis.

Development of an immunotoxin that targets hCD2-positive cells
Since the p16-dependent surface hCD2 expression in vivo tissue was demonstrated in our novel senescence monitor mouse model, we investigated whether p16-expressing senescent cells may be not only visualized, but also eliminated by a target immunotoxin strategy using the anti-hCD2 monoclonal antibody, clone TS2/18.1.1. We initially obtained immortalized p16hCD2 MEFs constitutively expressing hCD2 on their surface and validated the binding specificity and target toxicity of the monoclonal antibody, TS2/18.1.1 as an immunotoxin by a FACS analysis. As shown in Fig. 3A, the anti-hCD2 antibody strongly bound to immortalized p16hCD2 MEFs, but did not bind to hCD2-negative WT MEFs, confirming the specificity of this antibody for recognizing and binding to hCD2 on the cell surface. We then examined the availability of the ribosome-inactivating protein, saporin, also known as a stable toxic molecule, for the target immunotoxin with this anti-hCD2 antibody. We assessed the indirect effects of the immunotoxin with saporinconjugated anti-mouse IgG (anti-mIgG-SAP), which binds the monoclonal anti-hCD2 antibody. Immortalized p16hCD2 MEFs were incubated with the anti-hCD2 antibody and then with the anti-mIgG-SAP indirect second immunotoxin treatment, and we confirmed the significant efficacy of the target immunotoxin by enumerating living cell numbers (Fig. 3B, C).
After confirming this indirect binding and toxic specificity, we directly conjugated this antibody to saporin to produce an immunotoxin against hCD2 (anti-hCD2-saporin: hCD2-SAP). After incubation with hCD2-SAP, the number of living p16hCD2 MEFs markedly decreased, which was significantly different compared with the same dose of the above second immunotoxin. In contrast, the isotype control antibody conjugating saporin (mouse IgG-Saporin) did not exhibit cytotoxicity (Fig. 3B, C).
To examine the specificities and efficacies of hCD2-SAP in more physiological senescence processes, we cultured p16hCD2 MEFs at P5, consisting of cells with various hCD2 expression levels, with variable hCD2-SAP concentrations for 5 days and then subjected cells to a FACS analysis. We speculated that hCD2-SAP specifically eliminated the hCD2-expressing population in a dose-dependent manner among annexin V-negative viable MEF populations (Fig. 4A). We also showed that hCD2-SAP was less cytotoxic to p16hCD2 MEFs in an early passage (P3) because of fewer hCD2-positive cells, whereas it damaged MEFs in a late passage (P9), in which the majority of cells expressed hCD2 (Fig. 4B). Furthermore, since we used the p16hCD2 mouse model to visualize and eliminate hCD2 on the cell surface in a p16 expressiondependent manner, we confirmed whether our target hCD2-expressing cells were also clarified by senescence-associated β-galactosidase (SAβgal), which is another cellular senescence marker. As shown in Fig. 4C, SA-βgal activity, a well-known phenotype of senescent cells, was markedly higher in MEFs in a late passage (P5) than in an early passage (P1). On the other hand, hCD2-SAP reduced SA-βgal-positive MEF numbers in P5 due to its specific immunotoxin effects against p16hCD2 MEFs, but not wild-type MEFs (Fig. 4D). Collectively, these results strongly suggest that hCD2-SAP selectively eliminated senescent cells in p16hCD2 MEFs.

A hCD2-SAP treatment eliminates hCD2-positive cells in p16hCD2 mice
We investigated whether hCD2-SAP effectively eliminated hCD2positive cells in vivo. We intravenously injected hCD2-SAP three times into 9-~10-month-old p16hCD2 mice at a dose of 10 μg per mouse every two days, as previously reported for a saporin-conjugated immunotoxin (Fig. 5A). A few days after this continuous injection, we validated the efficiency of the immunotoxin for the removal of p16-positive cells from the spleen and lymph nodes using a FACS analysis. By hCD2-SAP injections, hCD2-positive lymphocytes in the spleen (Fig. 5B, C) and lymph nodes ( Supplementary Fig. 1) were excluded from their T cell population. Moreover, the administration of hCD2-SAP decreased p16 and hCD2 mRNA levels in the liver and spleen (data not shown). Staining with the anti-hCD2 antibody consistently showed cell shape-like fluorescence (Fig. 5E), indicating hCD2 expression on the cell surface, and the intensity of fluorescence was weaker in specimens from hCD2-SAPtreated mice than in those from PBS-treated mice (Fig. 5D, F). These results strongly suggest that hCD2-SAP also functioned as an effective immunotoxin against p16hCD2 mice and eliminated p16-positive senescent cells, which is consistent with the results of in vitro MEF experiments.

Treatments with hCD2-SAP ameliorated aging-associated T cell phenotypes in p16hCD2 mice
The results obtained on the hCD2-SAP-mediated elimination of hCD2-positive T cells prompted us to assess the effects of hCD2-SAP on aging-associated phenotypes in the T cell population. We aimed for the long-term depletion of hCD2-positive T cells because of the life span of lymphocytes. To achieve this, we performed multiple consecutive intravenous injections of hCD2-SAP once a week for 3 weeks (Fig. 6A). The hCD2-positive T cell population was significantly smaller in p16hCD2 mice injected with hCD2-SAP than in PBS-treated mice or wild-type mice (Fig. 6B). Quantitative RT-PCR consistently showed significant reductions in the expression of p16 and hCD2 in the spleen and liver from hCD2-SAP-injected p16hCD2 mice (Fig. 6C, Supplementary Fig. 2).
We then investigated whether this treatment improved aginginducible phenotypes, which exert a negative impact on the immune system. We examined the frequency of naïve T cells (CD62L + and CD44 − ), which chronologically contract with aging. Typical results from splenic naïve T cell fractions in p16hCD2 mice using a FACS analysis are shown in Fig. 6D. As expected, the percentages of both CD4 + and CD8 + naïve T cells from 10 to 11-month-old mature mice were lower than those of 2-month-old mice, whereas the mature mice administered hCD2-SAP showed a slight increase in the percentages. The naïve CD4 + T cell ratio was significantly smaller in mature mice than in young mice, and was partially restored by approximately 15 % on average by the immunotoxin treatment (Fig. 6E). The frequency of hCD2-positive cells among CD4 + T cells in mature mice decreased by only approximately 1 % with or without the immunotoxin treatment (Fig. 6F).

Relative mRNA levels to
Gapdh (caption on next page) We also investigated whether the alterations observed were involved in a unique memory-phenotype CD4 + T cell subpopulation, namely, senescence-associated T cells (SA-T cells) (Shimatani et al., 2009;Tahir et al., 2015) because the aging-associated T cell phenotype was partially ameliorated after the immunotoxin treatment. SA-T cells progressively accumulate with age and exhibit the characteristic features of cellular senescence, including an increase in the activity of the cellular senescence marker, SA-β-galactosidase in addition to the constitutive expression of PD-1 and CD153. SA-T cells corrupt not only T cell receptor-mediated proliferation, but also typical cytokine production, which is crucial for immune responses. They secrete abundant amounts of pro-inflammatory factors, such as osteopontin or chemokines, which induce persistent tissue inflammation. As shown in Fig. 6G, the hCD2-SAP treatment markedly reduced the percentage of SA-T cells among CD4 + T cells. Collectively, these results are consistent with the hCD2-SAP treatment reducing the senescent phenotypes of T cells in p16hCD2 mice in a p16 expression-dependent manner, leading to the amelioration of immunosenescence, including reductions in SA-T cells and increases in naïve T cells.

Discussion
In the present study, we developed a novel knock-in mouse model in which hCD2 is recognized on the surface of p16-associated senescent cells. We demonstrated that it was possible to reduce senescent cells in these mice using a saporin-conjugated anti-hCD2 monoclonal antibody. In the past decade, cellular senescence research has exponentially increased with both in vitro and in vivo mouse models that target cell cycle regulator-derived cellular senescence markers, thereby providing a more detailed understanding of the heterogeneity and diversity of senescence biology. In transgenic models, such as INK-ATTAC, utilizing p16 expression, their enhanced transgene expressions preclude the precise evaluation of the physiological functions of senescent cells. Alternatively, a knock-in system allows for the expression of an external gene product that is faithful to cognate promoter activity. However, the transcriptional activity of endogenous p16 is generally too low and weak to sufficiently induce the inserted knocked-in gene products in order to eliminate senescent cells. p16-Cre knock-in mice crossed with floxed reporter strains carrying strong stable promoters were recently reported to enhance the transcriptional activity of intrinsic p16 (Grosse et al., 2020;Omori et al., 2020). "Antigen-antibody reactions" are a widely accepted biological technique due to their high sensitivity, specificity, and reliable quantitative capability. Therefore, we confirmed the efficiency and accuracy of hCD2-SAP as an immunotoxin conveying the protein synthesis blocker, saporin to targeted senescent cells without genetic enhancements in cognate p16 and linked hCD2 expression. The population with surface hCD2 expression that strongly correlated with that of p16 ( Fig. 1) was selectively eliminated after the hCD2-SAP treatment in the chronological p16CD2 MEF culture in vitro (Fig. 4). hCD2-expressing senescent cells also selectively decreased in the spleen and liver of mature p16hCD2 mice following intravenous injections of the immunotoxin in vivo (Fig. 5). The present results revealed that the expression of p16 in splenocytes markedly increased in 10-month-old mice, and this was partially due to the accumulation of p16-positive T cells (Fig. 2). Although the expansion of T cells expressing p16 is a well characterized in physiological aging, the function of this T cell subset in age-related phenotypes remains unclear (Lemster et al., 2008;Liu et al., 2009;Pustavoitau et al., 2016;Baker et al., 2016).
The present results demonstrated that the frequency of p16-positive T cells in the spleen and lymph nodes was markedly reduced by the hCD2-SAP injections (Figs. 5B, C, 6B, Supplementary Fig. 1). The elimination of hCD2-positive cells with hCD2-SAP partially rescued the population size of CD4 + naïve cells in the spleen by approximately 15 % (Fig. 6E). On the other hand, the CD4 + CD153 + PD-1 + subset was recently identified as SA-T cells that accumulate with aging and represent common senescent cell phenotypes, such as the arrest of the cell cycle and the secretion of inflammatory factors (Shimatani et al., 2009;Tahir et al., 2015). Therefore, we investigated the frequency of hCD2positive cells among the splenic SA-T cell subset to clarify the involvement of p16 − positive senescent T cells in SA-T cell-related T cell aging. However, it was only 1-2 % of splenic SA-T cells, which was equivalent to that of hCD2-positive cells in the splenic T cell population (Fig. 6B, Supplementary Fig. 3A, B). Since SA-T cells expressed the cyclindependent kinase (CDK) inhibitors, Cdkn1a and Cdkn2b (Fukushima et al., 2018), Cdkn2a, which encodes p16, does not appear to be proactively involved in SA-T cell-related T cell aging. Furthermore, we showed that the contraction of the SA-T population was observed in hCD2-SAP-injected mice than in PBS-injected mice (Fig. 6G). Consequently, the present results suggest that p16-positive T cells are present and collaborate with other subtypes of T cells in immunosenescence, including T cell aging.
Despite the result showing that the reduction in hCD2-positive cells was only approximately 1 % (Fig. 6F), other indirect mechanisms may be contributing to increases in the CD4 + naïve subset besides the reduction of senescent CD4 + T cells. A previous study reported that irradiation-induced p16-positive splenocytes were eliminated in p16-3MR transgenic mice, which ameliorated irradiation-triggered phenotypes, including SASP, and decreased the proliferation of splenic T cells (Palacio et al., 2019). Our results suggest the non-cell autonomous effects, such as SASP, of p16-positive cells on the homeostasis of CD4 + naïve T cells. Thus, p16-positive cells potentially attenuate T cellmediated acquired immunity in aging.
In conclusion, we successfully created a method to clear p16-positive cells in vivo using a toxin-conjugated antibody that targets a surface molecule expressed on these cells. Our system not only provides a valuable tool to elucidate the physiological influence of senescent cells in age-related disorders and declining function, but also paves the way to develop new therapeutic methods to remove senescent cells by targeting molecules that are selectively expressed on the surface of senescent cells. Several groups have identified molecules that are preferentially expressed on the plasma membrane of senescent cells (Rossi and Abdelmohsen, 2021). Among these molecules, DPP4 and B2M have been proposed as potential targets, and senescent cells expressing these proteins were cleared from a cultured cell population by antibody-dependent cell-mediated cytotoxicity and an antibody-drug conjugate, respectively (Kim et al., 2017;Poblocka et al., 2021). These findings strongly suggest the applicability of immunotoxins as a novel type of senolytic strategy, and, thus, further studies are needed to identify the best endogenous cell surface markers that are applicable for clinical usage in the near future.

CRediT authorship contribution statement
A.N. and M.M. conceived and designed the research. T.M., designed, Fig. 6. The continuous administration of hCD2-SAP partially recovered age-associated phenotypes of T cells in p16hCD2 mice. (A) The experimental scheme of the continuous administration of hCD2-SAP. (B) The percentage of the hCD2-positive population in CD3 + splenocytes after the continuous administration of hCD2-SAP to young (2 months old) and mature (10-11 months old) p16 hCD2 mice and mature (13 months old) WT mice. (C) p16 and hCD2 expression in the spleen after the continuous administration of hCD2-SAP by quantitative RT-PCR. (D) Representative percentage of naïve T cells in splenic CD4 + (left) or CD8 + (right) T cells from young and mature p16hCD2 mice by FACS analyses. Percentages indicate the CD62L + CD44 − naïve T cell population in each gate. (E) Frequencies of CD62L + CD44 − naïve T cells in (D). (F) The percentages of hCD2-positive cells in CD4 + T cells in (D). (G) The percentages of SA-T (CD153 + and PD-1 + ) populations in CD4 + T cells in (D). Young mice (2-3 months old at the sampling point). Mature mice (10-13 months old at the sampling point). In (B), (C), (E-G), each symbol represents an individual mouse. Horizontal lines represent means. *p < 0.05, **p < 0.01, ****p < 0.0001. ns, not significant. generated, and validated the p16hCD2 mouse model. Y.S., Y.K., and T.H, designed and performed experiments and analyzed the results. T.Y. planned and supervised the research and A.N., M.M., and Y.S. wrote the manuscript. All authors contributed to the discussion in the final manuscript.