1 Senescent, dysfunctional human cardiac progenitor cells (CPCs) accumulate in the aged heart and elimination of senescent cells enhances CPC activation and cardiomyocyte proliferation in aged mice

Rationale: Aging leads to increased cellular senescence and is associated with decreased potency of tissue-specific stem/progenitor cells. Objective: To determine the impact of ageing and senescence on human cardiac stem/progenitor cell (CPC) biology and regenerative potential, and investigate whether elimination of senescent cells in aged mice enhances CPC activation and cardiomyocyte proliferation. Methods and Results: CPCs were isolated from the right atrial appendage (~200mg) of human subjects with cardiovascular disease (n=119), aged 32-86 years, and assessed for expression of senescence-associated markers (p16 INK4A , SA- β -gal, DNA damage γ H2AX, telomere length), Senescence-Associated Secretory Phenotype (SASP), cell growth, differentiation, and regenerative potential following transplantation into the infarcted mouse heart. Senescent cells were eliminated in aged mice (22 – 32 months) in vivo either genetically, using INK-ATTAC mice, which results in inducible elimination of p16 Ink4a -expressing senescent cells upon the administration of the drug AP20187, or pharmacologically using intermittent oral administration of combined senolytics, Dasatinib (D) and Quercetin (Q). In aged subjects (>74 years old) over half of CPCs are senescent, unable to replicate, differentiate, regenerate or restore cardiac function following transplantation into the infarcted heart. Aged-senescent CPCs secrete SASP factors, which renders otherwise healthy, cycling-competent CPCs to senescence. CPCs SASP and debilitative . aged (INK-ATTAC or vs. elimination in vivo using or with administration of results in activation and increased number of small, immature, proliferating in the mouse isolated from the failing human heart develop a senescent phenotype with age exhibited by increased expression of senescence-associated markers (p16 INK4A , SA- β -gal), DNA damage, shortened telomere length and a SASP. Aged human hearts with dilated cardiomyopathy showed greater numbers of p16 INK4A –positive CPCs and cardiomyocytes with shorter telomeres than age-matched controls 47 . Similarly, CPCs isolated from failing, aged hearts show increased p16 INK4A and inflammatory factor expression 48 . Reliably detecting senescent cells in vivo is an ongoing challenge, and it is important that a combination of senescent cell biomarkers are used for detection as any one marker used in isolation is prone to false positives. Our study used a combined panel of the senescence-associated biomarkers, p16 INK4A , γ H2AX, telomere length, SA β -gal activity and SASP expression, to detect senescent CPCs. Our findings demonstrate that CPCs accumulate in the failing hearts of elderly subjects (>76 years) and are dysfunctional, showing impaired proliferation, clonogenicity, spherogenesis, and differentiation, compared to CPCs isolated to improved myocardial contractility and amelioration of ventricular remodelling (decreasing fibrosis, hibernation, and stunning), inhibition of the inflammatory response, increased cardiomyocyte survival and angiogenesis/neovascularisation 60 . Most importantly, the transplanted cells’ secretome activates the heart’s endogenous myocardial regenerative reserve, which then gives rise to new vasculature and cardiomyocytes, leading to endogenous (and autologous) myocardial regeneration 68,69 . Results from the SCIPIO clinical trial showed that CPCs injected into subjects with ischemic cardiomyopathy who had post-infarction + 20µM) or vehicle for 48 hours decreased the number of TAF-, p16 Ink4a - and SA β -gal-positive cells, and decreased the secretion of key SASP components, PAI-1, GM-CSF, IL-6, IL-8 and MCP-1 20 . Clearance of senescent human CPCs using D+Q in the present study also abrogated the SASP, and the deleterious impact of the SASP on impairing proliferation and inducing senescence to healthy, cycling competent CPCs in a co-culture environment was abolished with D+Q treatment. Thus D+Q can kill human senescent cells, including tissue specific myocardium restoring LV function. The present work demonstrates approaches that eliminate senescent cells may be useful for treating age-related cardiac deterioration and rejuvenating the regenerative capacity of the aged heart. The next steps would be to determine whether senolytic approaches could be used in conjunction with cell therapy interventions to improve the environment (the ‘soil’) the cells (the ‘seeds’) are being transplanted into and the intrinsic reparative mechanistic processes that are compromised with age. Indeed, targeting senescent cells could also impact the potency of resident stem/progenitor populations in other aged organs. The present findings provide new insights

Recently, the Kirkland lab has demonstrated that transplanting relatively small numbers of senescent preadipocyte cells into young (6 month old) mice causes persistent physical dysfunction, measured through maximal speed, hanging endurance and grip strength, 1 month after transplantation. Transplanting even fewer senescent cells into older (17 month old) recipients had the same effect and reduced survival, indicating the potency of senescent cells in shortening health-and lifespan. Intermittent oral administration of the senolytics, D and Q to senescent cell-transplanted young mice and naturally aged mice alleviated physical dysfunction and increased post-treatment survival by 36% while reducing mortality hazard to 65% 20 . Altogether these data indicate that cellular senescence is causally implicated in generating age-related phenotypes and that systemic removal of senescent cells can prevent or delay tissue dysfunction, physical dysfunction and extend health-and lifespan.
Mammalian aging is associated with gradual loss of the capacity of the tissue-specific stem/progenitor cells to maintain tissue homeostasis or to repair and regenerate tissues after injury or stress 29 . Indeed, in most tissues there is an overlap between aging and stem cell impairment [30][31][32][33] . Function of tissue-specific stem cells declines with age due to several factors including telomere shortening, increased senescence and elevated expression of p16 Ink4a and other cyclin-dependent kinase inhibitors (CDKIs) 34 , DNA damage and external influences affecting stem cell niche homeostasis 30,[35][36][37][38] .
The discovery that the adult mammalian heart possesses a pool of tissue-specific, resident cardiac stem/progenitor cells (CSCs or CPCs) was 15 years ago 39 . These cells express stem cell surface receptors, c-kit, Sca-1, which allows their isolation and purification from deep within the cardiac tissue 40 . They possess properties of stem cells, being clonogenic, selfrenewing, and multipotent in vitro and in vivo [39][40][41][42] . Using a cardiac diffuse damage rodent model, which is in the presence of a patent coronary circulation, has a drop out of only ~10% cardiomyocytes, and recapitulates muscle wear-and-tear, we have demonstrated that the adult heart has intrinsic regenerative capacity 41 . Indeed, resident CSCs spontaneously restore . CC-BY-NC-ND 4.0 International license certified by peer review) is the author/funder. It is made available under a The copyright holder for this preprint (which was not this version posted August 21, 2018. . https://doi.org/10.1101/397216 doi: bioRxiv preprint cardiac function by regenerating lost cardiomyocytes, and when CSCs and their expansion were ablated, no cardiac regeneration or functional recovery was apparent leading to overt heart failure. However, the regenerative process was completely restored by replacing the ablated CSCs with the progeny of one CSC. Conversely, selective suicide of these exogenous CSCs and their progeny abolished regeneration, severely impairing ventricular performance.
Thus, we showed that when tested in the appropriate model, CSCs are necessary and sufficient for the regeneration and repair of the damaged heart 41 .
Recently considerable controversy, confusion and debate has surrounded the role and significance of CSCs in cardiac homeostasis and repair. Most of this confusion stems from using a single marker -c-Kit -to genetically fate map c-kit-positive cells in mice. Such studies have indicated that these cells only minimally contribute cardiomyocytes during ageing and following injury 43,44 . However, these studies have not specifically tagged or lineage traced the CPCs, or isolated them and characterised their stem cell properties.
Moreover, these studies test the regenerative potential of the heart using the nonphysiological ischemic myocardial infarction model that has a drastic drop out of cardiomyocytes (~30% of LV). Furthermore, several limitations of the use of Kitcre-KnockIn strategies for CPC identification and cell-fate mapping have recently been revealed 45 . Indeed, the very low number of endogenous c-Kit pos CPC-generated cardiomyocytes detected in the Kitcre mice simply reflects the failure to recombine the CPCs to track their progeny and the severe defect in CPC myogenesis produced by the Kitcre allele 45 .
We have since re-assessed and refined the phenotype, characteristics, and regenerative potential of the CSCs, and showed that the majority (~90%) of the resident c-kit-positive cardiac cells are blood/endothelial lineage committed CD45-positive, CD31-positive cells.
Among the cardiac c-kit-positive cell cohort, only a very small fraction (1-2%) has the phenotype and differentiation/regenerative potential of true multipotent CSCs; the rest are progenitor cells. We show that single CD45-, CD31-negative, c-kit-positive cell-derived clones, when stimulated with TGF-β/Wnt molecules, acquire full transcriptome and protein expression, sarcomere organisation, spontaneous contraction, and electrophysiological properties of differentiated cardiomyocytes in vitro and in vivo 42 . Therefore, the adult mammalian heart fits squarely with other organs and tissues as possessing a resident multipotent stem cell and progenitor population. These cells are rare but can regenerate cardiomyocytes lost to wear and tear, but understandably not large, ischemic segmental loss of tissue, as occurs with a myocardial infarction.
Over the years, accumulated evidence from human and mouse studies signified that cardiac aging and pathology affects the activity and potency of the resident cardiac stem/progenitor cells (abbreviated hereafter as CPCs) [46][47][48][49] . This translates into a diminished capacity of the aged and diseased myocardium to maintain homeostasis, and repair and regenerate following injury [50][51][52][53][54][55] . The aging milieu might therefore limit the success of cell transplantation therapies where the outcome is direct cardiogenic differentiation of transplanted cells and/or stimulation of endogenous regenerative mechanisms. As the majority of cardiovascular disease patients in need of regenerative therapies are of advanced age, regulation of CPC and cardiovascular aging/senescence is mission critical.
Here we provide new information about the existence and biology of CPCs and further show their importance and relevance in the aged and diseased human heart, which we hope will help resolve the controversy plaguing this important field. Indeed, we have carried out an extensive analysis of CPCs in the human failing heart with advanced age and show the accumulation of senescent-CPCs, which exhibit diminished self-renewal, differentiation, and regenerative potential in vivo. We show that Senescent-CPCs have a SASP that negatively affects healthy non-senescent, cycling-competent CPCs, rendering them senescent. Clearing the senescent-CPCs using combinations of senolytic drugs attenuates the SASP and its effect on promoting senescence in vitro. The effects of elimination of senescent cells on the heart and its regenerative capacity have not been elucidated. We report novel data that show systemic elimination of senescent cells in vivo in aged mice using senolytics (D+Q) or using the 'suicide' transgene, INK-ATTAC with administration of AP20187, results in CPC activation and increased number of small, immature, proliferating cardiomyocytes in the aged mouse heart. These findings are directly translational and transformative, which have great potential to inform the development and design of future clinical trials.

Methods
Expanded Methods are provided in the Online Data Supplement.
Myocardial samples (~200mg each) were obtained from the right atrial appendage (n=119) of human subjects with cardiovascular disease, aged 32-86 years. All subjects gave informed consent before taking part in the study (NREC #08/ H1306/91). Cardiac tissue was minced . CC-BY-NC-ND 4.0 International license certified by peer review) is the author/funder. It is made available under a INK-ATTAC mice aged 3 months (n=10) and 22 months (n=10) were randomly assigned to treatment groups and injected intraperitoneally (i.p.) with either vehicle or AP20187 for 2 consecutive days every 2 weeks for 2 months. These mice were injected (i.p) with EdU (123 mgkg -1 ) 4 days and 2 hours prior to sacrifice.

CPCs exhibit a senescent phenotype with increased age
Human CPCs were isolated from biopsies of right atria, obtained from subjects who had given informed consent before undergoing cardiac surgery (aortic disease, valve disease, coronary artery bypass graft (CABG), or multiple diseases), using sequential enzymatic digestion and dissociation, Optiprep density gradient to remove large debris, followed by On average, 22±9%, 31±4%, 48±9%, and 56±16% of CPCs expressed p16 Ink4a isolated from 50-59, 60-69, 70-79, and 80-89 year old subjects, respectively. We also found an increase (P<0.05) in the number of senescence-associated β -galactosidase-(SA-β-gal; ~60%) and DNA damage marker, γ H2AX-positive CPCs (~20%) isolated from old (71-79 years), compared to middle-aged (54-63 years) subjects (Figure 1b,c). Moreover, p16 INK4A -positive CPCs co-expressed γ H2AX (Figure 1c). Further interrogation by Q-FISH revealed that, while the average telomere length of CPCs isolated from old and middle-aged subjects' hearts were comparable, CPCs isolated from old (78-84 years) subjects' hearts contained a 12% subpopulation with telomere length of <6Kb, which is regarded as being critically short (Figure 1d) 56 . Approximately 2% of the CPCs isolated from human hearts were Ki67-. CC-BY-NC-ND 4.0 International license certified by peer review) is the author/funder. It is made available under a The copyright holder for this preprint (which was not this version posted August 21, 2018. . https://doi.org/10.1101/397216 doi: bioRxiv preprint positive, reflective of their mainly dormant, quiescent phenotype 41 . There were no differences between middle-aged and old subjects in number of Ki67-positive CPCs, and we did not see any Ki67-positive CPCs that were p16 INK4A -positive (Supplementary Figure 2c). These findings indicate that the aged human heart contains an increased proportion of aged senescent-CPCs, which could translate to their dysfunctionality.  Even though CPCs isolated from old hearts showed decreased proliferation, clonogenicity, and differentiation potential, only ~50% of CPCs are senescent in old myocardium (Figure   1a), therefore these data imply that a functionally cycling-competent CPC population still exists in old myocardium. Indeed, single CPC-derived clones from young, middle-aged, and old subjects were indistinguishable in terms of morphology, senescence, multipotency, selfrenewing transcript profile, and differentiation (Supplementary Figure 3). These findings suggest that CPCs age and become senescent in a stochastic, non-autonomous manner. This resembles what was seen in rat preadipocytes 57 .

Aged-senescent CPCs lose their regenerative capacity in vivo
To purify for a senescent population of CPCs we utilised the C 12 -5- The copyright holder for this preprint (which was not this version posted August 21, 2018. . https://doi.org/10.1101/397216 doi: bioRxiv preprint derived cells; containing 86±5% cardiac fibroblasts, 13±3% vascular smooth muscle, 1±1% endothelial cells 41 ). Sham animals were treated the same way, except ligation of LAD coronary artery was not performed and they did not receive cells but were injected with the same volume of PBS. Mice were administered BrdU via osmotic mini pumps for 14 days after MI and cell injection to track new cell formation (Figure 3a). All cell populations were labelled prior to injection with PKH26 lipophilic membrane dye, which exhibited high labelling efficiency and label dye retention over population doublings in cycling-competent CPCs and c-kit neg cardiac-derived cells in vitro (Supplementary Figure 7). We sacrificed a sub-set of MI-mice that had been injected with 5 x 10 5 SA-β-gal-negative cycling-competent Removal of p16 Ink4a senescent cells can delay the acquisition of age-related pathologies in adipose tissue, skeletal muscle, heart, blood vessels, lung, liver, bone, and eye 9,11,12,[16][17][18][19][20]24,25,28 . Recent studies have documented the use of senolytic drugs for the selective clearance of senescent cells from 'aged' tissues [16][17][18][19][20][21][22][23][24][25][26][27]58 . We tested the potential of 4 senolytic  (Figure 5j,k). Co-culture of cycling-competent CPCs with senescent-CPCs led to increased (P<0.05) secretion of SASP factors into the medium, but the level of SASP factors was reduced (P<0.05) with application of D+Q (Figure 5l). These findings document that senescent CPCs have a SASP, and clearance of senescent CPCs using a combination of D+Q senolytics abrogates the SASP and its detrimental senescenceinducing effect on healthy, cycling-competent CPCs.

Elimination of senescent cells in vivo activates resident CPCs and increases number of small, proliferating cardiomyocytes in the aged heart
Eliminating a relatively small proportion (~30%) of senescent cells using a 'suicide'  (Figure 6a). Tissues in which p16 Ink4a expression and/or senescent cells are decreased by D+Q in wild type mice as well as by AP20187 in INK-ATTAC mice include the aorta, adipose tissue, cardiac and skeletal muscle, lung, liver, and bone 9,16-28 . We showed p16 Ink4a mRNA expression was decreased (P<0.05) in the heart following D+Q or AP20187 treatment in aged INK-ATTAC or wildtype mice (Figure 6b).
Previously we have shown an improvement of heart function in old mice after D+Q treatment 24 . Analysis of cardiac cross-sections revealed significantly higher (P<0.05) CPC (Sca-1 + /c-kit + /CD45 -/CD31 -/CD34 -) 42 numbers (Figure 6c; Supplementary Figure 10a (Figure 6d), suggesting these myocytes to be immature and newly formed, compared to vehicle-treated mice, which exhibited only rare small myocytes but a greater proportion of hypertrophied myocytes (Figure 6d). We found an increase (P<0.05) of small, proliferating Ki67-positive myocytes (~0.25%) in old hearts following AP20187-or D+Qtreatment, compared to vehicle-treated control (0.03±0.03%) (Figure 6e,f). To corroborate these data we injected EdU 4 days and 2 hours prior to sacrifice of old (22m) and young (3m) AP20187-treated INK-ATTAC mice (Supplementary Figure 10c). We found increased  (Figure 6f). Finally, we detected a decrease (P<0.05) in fibrosis in the LV following AP20187-and D+Q-treatment, compared to vehicle-treated control (Figure 6i). In contrast to the treatment of aged mice, treatment of young adult (2-3 Months) INK-ATTAC or wild-type mice with AP20187 or D+Q, respectively, did not alter EdU-positive myocyte number (Figure 6h), CPC numbers or myocyte diameter (data not shown). These findings show that clearance of senescent cells leads to stimulation of CPCs and cardiomyocyte proliferation and that this strategy is specific to the aged heart.

Discussion
In a large sample cohort our study shows that CPCs isolated from the failing human heart develop a senescent phenotype with age exhibited by increased expression of senescenceassociated markers (p16 INK4A , SA-β-gal), DNA damage, shortened telomere length and a SASP. Aged human hearts with dilated cardiomyopathy showed greater numbers of p16 INK4A -positive CPCs and cardiomyocytes with shorter telomeres than age-matched controls 47 .
Similarly, CPCs isolated from failing, aged hearts show increased p16 INK4A and inflammatory factor expression 48 . Reliably detecting senescent cells in vivo is an ongoing challenge, and it is important that a combination of senescent cell biomarkers are used for detection as any one marker used in isolation is prone to false positives. Our study used a combined panel of the strategies to activate the regenerative capacity of the aged heart through delivery of growth factors or cell therapy will also be sub-optimal. Therefore, the success of cardiac regenerative therapeutic approaches thus far tested for treating patients with heart failure and disease could be of limited efficacy in promoting myocardial regeneration because of the increased number of senescent, dysfunctional CPCs and cardiomyocytes 47,48 and the resultant presence of a cardiac SASP in the aged and failing heart that impairs the function of the remaining nonsenescent CPCs.
The present study found that CPCs age in a stochastic non-autonomous manner and it is possible to clonally select for a cycling-competent population of CPCs even from diseased or aged hearts. There are individual CPCs in older individuals that have replicative and functional capacities resembling those of CPCs in younger subjects. A similar scenario was found in the case of rat fat cell progenitors 57 . While the abundance of progenitors cloned from adipose tissue that had restricted capacities for replication and differentiation into adipocytes or that were non-replicative but viable (i.e., senescent) increased progressively with aging in rat fat, there remained cells that had the capacities for replication and adipogenic differentiation characteristic of clones derived from young rats. Together, these findings indicate that: (1) it may be feasible to isolate CPCs even from older individuals that are functional, capable of supporting cardiac regeneration if removed from their toxic milieu, and that could be therapeutically relevant in treating patients, especially if they were autologously generated 59 and (2) that by clearing senescent CPCs with a toxic SASP from the aged heart, there remains a tissue-resident population of CPCs with capacity to regenerate damaged heart tissue.
When we purified for a homogenous SA-β-gal-positive, senescent CPC population, we showed that these cells had poor engraftment and survival, and were unable to contribute to cardiac regeneration, repair or restoration of cardiac function following transplantation into the infarcted myocardium. This is contrary to in vitro-selected, SA-β-gal-negative, The copyright holder for this preprint (which was not this version posted August 21, 2018. . https://doi.org/10.1101/397216 doi: bioRxiv preprint the number of proliferating cardiomyocytes present in the old (22 month old) mouse heart is 0.07±0.00% of total cardiomyocytes. Elimination of senescent cells lead to double the amount of the proliferating cardiomyocytes found in a young heart, and triple the number found in an old heart. Therefore, the present data represent a significant and physiologically relevant increase and activation of the resident CPC compartment and cardiomyocyte proliferation following clearance of senescent cells.
Although clearing senescent cells using a genetic approach is not feasible in humans, the senolytic (D+Q) pharmacological approach described here is clearly translatable and can be used to target a fundamental aging mechanism present in most tissues, including the heart. Pharmacologically eliminating senescent cells or inhibiting the production of their SASP has been shown to improve cardiovascular function 16,24 , physical function 20 , enhance insulin sensitivity 11 , prevent age-related bone loss 17 , reduce frailty, and increase lifespan and health span 13,20 . Indeed, D+Q administration over 3 months decreased senescent cell markers (TAF+ cells) in the media layer of the aorta from aged (24 months) and hypercholesterolemic mice, which was met with improved vasomotor function 16 . The approaches used in the present study, and thus far in previous studies, target global elimination of all senescent cells.
Therefore, whether cell-specific elimination or local delivery of senolytics results in a greater effect in eliminating senescent cells and rejuvenating tissue regenerative capacity is yet to be determined.
Previous work has shown that senescent human primary preadipocytes as well as human umbilical vein endothelial cells (HUVECs) develop a SASP with aging, and conditioned medium from senescent human preadipocytes induced inflammation in healthy adipose tissue and preadipocytes 13  The anthracycline doxorubicin (DOX) is an effective chemotherapeutic agent used to treat pediatric cancers, but is associated with progressive and dose-related cardiotoxicity that may not manifest itself until many years after treatment. In a juvenile mouse model of anthracycline-mediated late onset cardiotoxicity, Huang et al. 73 reported decreased number of CPCs, which showed impaired differentiation into the cardiomyocyte and endothelial lineage in the MI-border zone of animals that had been exposed to DOX as pups. DOX treatment also led to significantly less CPCs in the juvenile heart, and those that remained showed decreased proliferation and increased expression of p16 Ink4a at 12 days of age. Moreover, 72 hours of DOX treatment (10nM or 100nM) on isolated CPCs in vitro led to attenuated proliferation, reduced telomerase activity and induced expression of p16 Ink4a56 . The present study showed that DOX treatment (0.2µM) for 24 hours to CPCs induced p16 Ink4a , γ H2AX and SA β -gal expression evidencing that CPCs had become senescent following DOX exposure. Future work should focus on the regenerative capacity of the heart following DOX exposure and whether senolytics could be used following DOX treatment and in later life to eliminate senescent cells and rejuvenate CPCs and the regenerative potential of the heart.
In conclusion, the present work demonstrates that in the aged and failing human heart a large majority of its resident CPCs are senescent and dysfunctional exhibiting impaired proliferation, clonogenicity and differentiation potential. Importantly, senescent CPCs have a SASP that can negatively affect cell behavior and render neighbouring cells into senescence.
Elimination of senescent human CPCs in vitro attenuates the SASP and its deleterious effect.