Mobilization‐based transplantation of young‐donor hematopoietic stem cells extends lifespan in mice

Abstract Mammalian aging is associated with reduced tissue regeneration and loss of physiological integrity. With age, stem cells diminish in their ability to regenerate adult tissues, likely contributing to age‐related morbidity. Thus, we replaced aged hematopoietic stem cells (HSCs) with young‐donor HSCs using a novel mobilization‐enabled hematopoietic stem cell transplantation (HSCT) technology as an alternative to the highly toxic conditioning regimens used in conventional HSCT. Using this approach, we are the first to report an increase in median lifespan (12%) and a decrease in overall mortality hazard (HR: 0.42, CI: 0.273–0.638) in aged mice following transplantation of young‐donor HSCs. The increase in longevity was accompanied by reductions of frailty measures and increases in food intake and body weight of aged recipients. Young‐donor HSCs not only preserved youthful function within the aged bone marrow stroma, but also at least partially ameliorated dysfunctional hematopoietic phenotypes of aged recipients. This compelling evidence that mammalian health and lifespan can be extended through stem cell therapy adds a new category to the very limited list of successful anti‐aging/life‐extending interventions. Our findings have implications for further development of stem cell therapies for increasing health and lifespan.


| INTRODUC TI ON
Stem cells are critical to tissue regeneration and homeostasis during aging and disease (Signer & Morrison, 2013). As a hallmark of aging (Lopez-Otin, Blasco, Partridge, Serrano, & Kroemer, 2013), stem cell dysfunction is critical to improving the quality of life for people with advanced age (Fontana, Kennedy, Longo, Seals, & Melov, 2014). Stem cell-based therapy holds considerable promise for treating aging-related diseases (Ikehara & Li, 2014), with hematopoietic stem cells (HSCs) being the most widely used for stem cell therapies (Daley & Scadden, 2008). It is becoming increasingly clear that age-related changes in the niche space can induce alterations in hematopoiesis, including myeloid lineage skewing (Guidi et al., 2017;Moerman, Teng, Lipschitz, & Lecka-Czernik, 2004;Stier et al., 2005;Stolzing, Jones, McGonagle, & Scutt, 2008). However, extrinsic stimulation of HSCs with cytokines is highly dependent on intrinsic determinants (de Haan & Van Zant, 1997). To date, the "gold standard" measure of HSC functionality remains an in vivo repopulating assay to determine their ability to re-establish lineage cell production in recipients during hematopoietic stem cell transplantation (HSCT; Kwarteng & Heinonen, 2016;Rossi et al., 2012). Unfortunately, conventional HSCT procedures require harsh cytotoxic conditioning-irradiation and/or chemotherapy-that alters HSC niches in the bone marrow, permanently damaging bone architecture (Green & Rubin, 2014;Naveiras et al., 2009). These limitations have confounded efforts to assess health-associated benefits of HSC replacement and rejuvenation. Further, it has been impossible to determine the extent of which extrinsic factors drive age-related decline of the hematopoietic system.
The majority of HSCs reside in specialized niches within the bone marrow, although some HSCs leave these niches and migrate into the blood, ~1%-5% of total HSCs each day (Bhattacharya et al., 2009). Mobilization of HSCs into the peripheral blood can be achieved through administration of G-CSF (Teipel et al., 2015), an effect that is dramatically increased when G-CSF is administered in combination with other mobilizers, such as AMD3100 (Pusic & DiPersio, 2010). This HSC mobilization strategy constitutes the basic mechanism underlying collection of peripheral blood donor stem cells in the clinic. Critically, this increased mobilization also creates temporarily empty niches in the bone marrow, opening a window of opportunity for donor cell engraftment. Here, we use a novel mobilization-based HSCT procedure to investigate the health-associated benefits of replacing HSCs from aged recipients with young-donor HSCs. Additionally, we take advantage of the niche-preserving properties of this mobilization-based HSCT to investigate the influence of aged niche signaling upon a low percentage of young-donor HSCs.

| Long-term donor chimerism was achievable following mobilization-based conditioning
To reduce the adverse effects of cytotoxic conditioning agents, we developed a mobilization-based conditioning procedure, eliminating the need for irradiation, followed by transplantation of donor HSCs.
G-CSF and AMD3100 (complementary mobilizing agents) were used to mobilize HSCs in ten-week old mice. After peak mobilization (day 5), mice were transplanted with 2.0 × 10 6 lineage-negative, agematched, GFP + bone marrow cells ( Figure S1). A total of seven transplantation cycles were performed for each recipient, with donor chimerism (GFP + ) increasing with every cycle of transplant, reaching ~90% at 1 month after the 7th cycle and stabilizing at ~77% by 4 months posttransplantation (Table 1).

| Nontoxic hematopoietic reconstitution with young HSCs increases the longevity of aged recipient mice
Next, we used this novel HSCT method to investigate the impact of replacing aged (19-month) HSCs with young-donor (2-month) HSCs.  Table S2 Figure 1).
These results confirm the increase in longevity that we previously observed in aged GFP + recipients receiving GFPyoung-donor HSCs-17% increase in median lifespan and HR of 0.14 (95% CI, 0.054 to 0.348, p < .0001, Figure S2). Additionally, aged recipients exhibited a reduction in age-specific mortality rates into late life ( Figure 2b).
Further, we found no significant differences between survival curves and age-specific mortality rates of mobilized and non-mobilized control groups, suggesting that mobilization-based HSCT does not negatively affect the longevity of aged recipients ( Figure 2c). Together, these data suggest that young-donor HSCs are able to extend the longevity of aged recipients.
To monitor donor chimerism, aged-matched (19-month) female recipients received 2-month-old GFP + young-donor female HSCs in parallel with aged recipients of wild-type young-donor HSCs. Since only LT-HSCs are capable of long-term self-renewal and continued contribution to hematopoiesis four months posttransplant (Dykstra et al., 2007), we obtained peripheral blood samples at both one and four months posttransplant to assess donor chimerism and longterm donor cell reconstitution. Gating on GFP + cells (Figure 2d), donor chimerism reached 91.2 ± 1.6% at 1 month after the last HSCT cycle and stabilized at 74.8 ± 2.6% by 4 months posttransplantation, comparable with donor chimerism observed in irradiation-based HSCT recipients (80.3 ± 5.6%, Figure 2e).

| Nontoxic hematopoietic reconstitution delays the accumulation of health-associated deficits in aged recipient mice
We monitored the progression and accumulation of age-related health deficits of all surviving mice on a monthly basis (Table   S2; Rockwood et al., 2017;Whitehead et al., 2014). Deficit accumulation index (DAI) ratios were generated by evaluating 31 potential age-associated deficits under a noninvasive frailty index (FI; Howlett & Rockwood, 2013;Parks et al., 2012;Rockwood et al., 2017) and plotted as a function of age (Figure 3a), and mean DAI scores were generated for each group at each time point analyzed (Table S3). Strikingly, recipients of young-donor HSCs accumulated fewer age-related health deficits compared with nontransplanted controls at any given time during advanced life; the difference in health deficits ranged from −0.08 (95% CI: 0.026-0.129, p = .001) at 26 months to −0.30 (95% CI: 0.08-0.51, p = .004) after 38 months of age ( Figure 3b). Moreover, the average daily food intake of recipients of young-donor HSCs improved significantly, peaking at 3.45 ± 0.15 grams per day ( Figure 3c).
Given that these food intake measurements were obtained on a per-cage basis, the differences in survival of HSC-recipient versus control mice brings up potential cage effects that might influence eating behaviors, especially at later ages. Additionally, recipients of young-donor HSCs maintained body weight later into advanced age compared with nontransplanted controls (Figure 3d), consistent with previous NIA-derived cohorts (Turturro et al., 1999).
Most time points showed no statistically significant differences between mobilized and nonmobilized controls, except at 29month nonmobilized controls had a −0.09 (p = .048) lower DAI score than mobilized controls. Food intake and body weight of

| Replacement of aged HSCs with young-donor cells reverses age-associated lineage skewing in aged recipients
Age-associated lineage skewing has been described by observing increased contributions of myeloid cell lineages in the peripheral blood at the expense of lymphoid cell lineages, both upon transplantation and steady state (Benz et al., 2012;Dykstra et al., 2007;Franceschi et al., 2007;Kovtonyuk, Fritsch, Feng, Manz, & Takizawa, 2016;Makinodan, 1998). Thus, we investigated whether young-donor HSCs preserved youthful phenotypes within the aged stroma ( Figure S3a). Young (2-month-old) and aged (19-month-old) recipients (all female) of a single mobilization-based HSCT cycle were assessed for GFP + young-donor HSCs contribution in the peripheral blood (see Figure S4 for gating strategy and representative flow plots). Four months post-HSCT, donor chimerism stabilized at 9.4 ± 0.8% (young recipients, 6 months old) and 15.2 ± 1.4% (aged recipients, 23 months, Figure 4a). Remarkably, white blood cell populations (WBCs) derived from young-donor HSCs within aged recipients (GFP + ) retained their youthful phenotypic distribution ( Figure 4b; Table S4; Rossi et al., 2005). Limited replacement with young-donor HSCs partially ameliorated the overall aged phenotypic distribution F I G U R E 1 Young-donor hematopoietic stem cells (HSCs) extend the lifespan of aged female recipients. Survival of wild-type, female mice (19-month-old) receiving mobilization-based conditioning followed by infusion of either young-donor (2-month-old) HSCs (young to old, blue) or PBS (mobilized control, red) were compared with aged, wild-type, female nonmobilized control (

| Young-donor cells maintain a youthful distribution of LSK cell subtypes in aged recipients
Next, we investigated young-donor HSC expansion into LSK (Lin -, Sca-1 + , c-Kit + ) cells within aged recipients (summarized in Table S5; Dykstra et al., 2007). ing that young-donor HSCs do not influence aged HSCs through cell nonautonomous traits. Further, no significant differences were found in mice receiving mobilization factors followed by sham transplants (Fig. S3j-p), despite potential myeloid cell proliferation (Knudsen et al., 2011). Together, these data suggest that actively proliferating aged HSCs were replaced with young-donor HSCs and that young-donor HSCs maintain a youthful phenotypic distribution among LSK subtypes following transplantation into aged recipients.

| D ISCUSS I ON
Importantly, this study includes the first successful HSCT, in which severe adverse effects such as rapid declines in body F I G U R E 3 Young-donor hematopoietic stem cells (HSCs) delay the accumulation of age-related health deficits in aged female recipients. Long-term effects of young-donor (2-month-old) HSCs in aged (19-month-old), female mice-identical numbers (N) to lifespan assessment. (a) Spaghetti plot of the frailty index scores obtained from each surviving mouse from each group each month, FI scores were fit to a mixed model, and the overall FI score was generated for each group for each time point (26, 29, 32, and 35 months In an ongoing study, toxicity profiles are being compared for this method versus conventional HSCT procedures (data not shown).
Others are designing transplantation regimens that limit toxicity by eliminating the use of irradiation or chemotherapeutic drugs, however these methods require depletion of endogenous HSCs (Chhabra et al., 2016;Palchaudhuri et al., 2016). In the current studies, we observed nonsignificant differences within all health span parameters investigated or within any cell lineage investigated in the peripheral blood or bone marrow in mice receiving mobilization factors followed by sham transplants, providing strong evidence of a lack of long-term adverse effects, despite potential myeloid cell proliferation (Knudsen et al., 2011).
To the best of our knowledge, this study is the first to achieve significant extension of the healthy lifespan of mice using a cell-based therapy approach. This was evidenced in aged female recipients (19-month-old) of young-donor (2-month-old) HSCs by extension of lifespan, reduction in age-related health deficits, and increases in food intake and body weight. One limitation was that all mice used in this study were female. Ideally, we would have included equal-sized cohorts of both male and female mice; however, to achieve statistical power, we needed a minimum of 30 aged mice. We chose to do the studies in females because some interventions have been shown to be applicable only to males (Austad & Bartke, 2015). We therefore reasoned that an effect on lifespan shown in females would be a more robust finding than one shown in males.  shorter host lifespans (Guest et al., 2015) and increased myeloid lineage skewing due to "defective" HSCs (Lee, Yoon, Choi, & Jung, 2019).
We speculate that these health-associated benefits are a consequence of replacing actively proliferating myeloid-biased HSCs known to accumulate in aged mice (reviewed in (Kovtonyuk et al., 2016)), subsequently decreasing their contributions to LSK frequen-

| HSC mobilization-based conditioning
Two-or 19-month-old C57BL/6NIA female mice were obtained through the NIA on a monthly basis and randomly assigned to their respective groups. HSCs were mobilized by administration

| Longevity assessment
Longevity assessment was initiated two weeks after arrival at UTHSCSA from the NIA, to remove any animals that did not handle the acute stress of transportation or acclimate to the new environment. Upon arrival, 150 animals were separated randomly into one of four groups (maximum of five animals per cage). Once chosen, animals remained with the same cage-mates, and no others, until end of life. Subjects removed from the study were those that did not survive past two weeks upon arrival from the NIA. Subjects censored were those that experienced experiment-related mortality. To determine the time and type of death, mice were inspected at least twice daily. If aged mice appeared to be too weak to obtain food, a mush of ground pellets and water was placed on the cage bottom so that they did not succumb to dehydration/starvation.
Moribund mice were euthanized if judged that they would not survive past another 48 hr. A mouse was considered severely moribund if it exhibited more than one of the following six clinical signs: inability to eat or drink; abnormally low body temperature; severe lethargy (reluctance to move when gently prodded with forceps); severe balance or gait disturbance; rapid weight loss for a week or more; an ulcerated or bleeding tumor. The age at which a moribund mouse was euthanized was taken as the best available estimate of its natural lifespan. A total of eight animals were censored from this study (seven transplanted, one mobilized control) as a result of procedure-associated error during administration of cells or saline. Additionally, a total of six animals were removed from this study (three transplanted, one mobilized control, two nonmobilized controls) as a result of failure to acclimate to housing conditions.
Kaplan-Meier analysis was used to generate survival curves to assess median and overall lifespans. Survival curves were compared using the log-rank test to generate hazard ratios between the groups.

| Age-specific mortality
The instantaneous rate of mortality at each age was computed using a piecewise polynomial B-spline hazard model assuming a Poisson distribution (Lambert & Eilers, 2005)

| Quantification of age-related health deficits
Starting at 26 months of age, mice from each group were evaluated under a noninvasive frailty index (FI) based on the clinical assessment of 31 potential deficits, as previously described (Clegg, Young, Iliffe, Rikkert, & Rockwood, 2013;Howlett & Rockwood, 2013;Parks et al., 2012;Whitehead et al., 2014). Clinical assessments were performed between 10 a.m. and 2 p.m. at monthly intervals. The rater was blinded to the group from which the animals derived. A second rater assessed a subset of randomly selected mice and there were compared with initial assessments to maximize inter-rater reliability.
Assessments with a >10% difference in DAI scores between raters were re-scored. Mice were placed individually in a fresh cage under a sterile flow hood in a procedure room designed for behavioral testing.
The room, located on a quiet hall in the UTHSCSA Laboratory Animal Resources facility, had no other occupants (mice or humans). Clinical assessment included evaluation of the integument, the musculoskeletal system, the vestibulocochlear/auditory systems, the ocular and nasal systems, the digestive system, the urogenital system, and the respiratory system, as well as signs of discomfort. Body weight, an additional means to assess frailty, was performed separately. The hearing test used a clicker of the type used to train dogs. The deficit accumulation index (DAI) score was computed using a deficit rating scale. For each parameter, a score of 0 was given if there was no sign of a deficit, a score of 0.5 denoted a mild deficit, and a score of 1 indicated a severe deficit. The DAI scores for each of the 31 items on the checklist were added, and the total was divided by the number of deficits measured to yield a DAI score between 0 (no deficits) and 1 (all possible deficits) for each animal. Demographic characteristics for each cohort were expressed using the mean DAI scores ± standard deviation. Differences in DAI scores over time among the groups were estimated using a mixed model with group by time interactions (see Statistics).

| Food intake assessment
Starting at 19 months of age, the average food intake was calculated for each group of mice on a monthly basis until animals expired.
Average food intake was measured by recording the initial total mass of food available per cage followed by measuring the mass of food remaining after twenty-four hours. The difference in mass between the initial available food and food remaining after twenty-four hours was divided by the total number of mice present per cage before averaging with mice from all cages measured within each group. The mass of food was measured with a CS200 Compact Scale (Ohaus). In addition, cages were inspected before each recording for food crumbs. Crumbs were cleared from cages during the initial food measurement. Crumbs found twenty-four hours after the initial measurement were added to the total food available to ensure accurate recording of food intake. If a mouse expired within the twenty-four hour window in which food was measured, the calculation was discarded and a new measurement was initiated with the new number of mice available.

| Statistical analyses
The rationale for the numbers (n) of mice in each group to provide adequate power to obtain significant results for the survival study was based on preliminary survival data obtained in our laboratory (Fig. S3). Assuming that differences in survival curves of experimental and control groups would be determined within the same order, a minimum of 30 mice in each group was estimated to have a power of >99% with an effect size of 1.311 (α = 0.01).
Differences in DAI scores over time among the groups were estimated using a mixed model with group by time interactions.
The model had random intercepts and random slopes to account for the correlations within mice, as well as for mouse-specific trajectories. Time was treated as a quadratic effect to accommodate curvilinear age-related changes identified through model selection with the Akaike Information Criterion (AIC). Linear contrasts were used to estimate confidence intervals and test for significant differences at each time point. For these studies, statistical tests were two-sided at significance level 0.05. These analyses were performed in the R environment for statistical computing 3.3.0 (R Core Team, 2016) within an accountable data analysis process (Gelfond, Goros, Hernandez, & Bokov, 2018). Statistical analyses of daily food intake and body mass assessment data were performed using GraphPad Prism 6.02 (GraphPad Software). All data are expressed as mean ± standard error of the mean. Multiple group comparisons were analyzed by two-way ANOVA, followed by post hoc analyses using Bonferroni posttest or one-way ANOVA, followed by Tukey's posttest. Differences among treatment groups were considered statistically significant at *p < .05, **p < .01, ***p < .001. For all other measures, significance was assigned using the Student's t test. Statistical analyses were performed by the UT Health San Antonio Nathan Shock Center Statistics Core.

ACK N OWLED G M ENT
We thank E. Kraig and M. Jazwinski for their help and advice on this project.

CO N FLI C T O F I NTE R E S T S
The authors have declared that no additional conflict of interest exists.

E TH I C A L A PPROVA L
The procedures for all animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Texas Health San Antonio.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data associated with this study are present in the paper or Supplementary Materials.