Study Design

Objective

As in RMR1, the ambition for RMR2 is to achieve "Robust Mouse Rejuvenation". We define this as an intervention or treatment program that:

• is applied to mice of a strain with a well-documented mean lifespan of at least 30 months

• is initiated at around 12 months younger than the mean lifespan

• increases both mean and maximum lifespan by at least 12 months The primary endpoint for the study is to determine the interactions between the various interventions, as revealed by differences between treatment groups (receiving different subsets of the interventions), on overall lifespan. However, we are also investigating aging and morbidity trajectories, causes of death, and functional decline. In this way we will add greatly to the understanding of which benefits these interventions confer and how they synergize, or possibly antagonize.

Age at study initiation

As in RMR1, interventions will begin in mid-late life, between 18-20 months of age, in order to assess the repair/rejuvenation capacity of interventions. The study will run through the remaining lifespan of all mice with the exception of animals selected for cross-sectional analysis at timepoints, as in RMR1.

Mouse Strain

For this second RMR study, we have two well-validated mouse strains to choose from. One option is to use the same pre-aged C57Bl/6J mice as in RMR1. There are a number of practical advantages to using this strain, including that it is the most common strain for biomedical research on mouse lifespan, used in approximately 90% of laboratory studies. As such, naturally aged animals are readily available from Jackson Laboratory (JAX, Bar Harbor, Maine) at a range of advanced ages, permitting studies such as our own, investigating interventions begun in late life. Extensive research has been conducted on C57Bl/6J mice, leading to well- established baseline data for various age-related parameters, which can facilitate comparisons and benchmarking in aging studies. Many interventions of interest to RMR have been developed and tested for efficacy in his strain. However, due to their inbred ancestry and consequent genetic uniformity, C57Bl/6J have disadvantages which are likely relevant for translation of therapies to humans, as they may not exhibit as diverse aging phenotypes as outbred strains, failing to capture the full genetic complexity of aging and age-related diseases seen in human populations, as well as a more limited range of intervention responses. This is a significant tradeoff, and one which remains under careful consideration.Alternatively, we may opt for HET3 mice, which for the first time are now available at scale, pre-aged from JAX. HET3 mice are generated through a four-way intercross (BALB/cByJ × C57Bl/6J F1 females to C3H/HeJ × DBA/2J F1 males), and are the strain utilized by the NIH’s well-established Interventions Testing Program (ITP), which aims to identify and systematically test dietary and drug interventions that can extend healthspan and lifespan in mice with the potential for translation to human aging research. The primary, but significant advantage of conducting aging studies in HET3 mice over inbred strains is because of their increased genetic diversity, which more closely mimics the genetic heterogeneity seen in human populations -- making them a valuable model for studying complex traits and diseases related to aging. HET3 mice also often exhibit slightly longer average lifespans compared to other strains, permitting research to track age-related changes and diseases which develop later in life (although they are somewhat shorter-lived than CL57Bl6/J).While the ITP has been collecting data on HET3 mice for over a decade, some differences in mean and maximum lifespan and intervention responses have been observed across testing sites, and baseline values for parameters like reference blood counts, chemistries, and functional performance are significantly more limited than for C57Bl/6 mice. Additionally, because the HET3 strain is only very recently available pre-aged and at scale, few of the interventions of interest to us have yet been studied in this model. The Study of Longitudinal Aging in Mice (SLAM) conducted by the National Institute on Aging currently aims to assess normative mouse aging and investigate potential differences in aging phenotypes between C57Bl/6J and HET3 mice of both sexes to identify and characterize phenotypic and biological predictors of mouse aging. Results from this study are pending, however, as is phenotypic data on aged HET3 cohorts being collected by JAX labs, so known baselines remain limited. Furthermore, genetically heterogeneous HET3 mice present a challenge when considering interventions such as cell therapies due to immune incompatibility. For these reasons, we expect that as in RMR1, RMR2 will also be conducted in CL57Bl/6J mice.

Treatment Groups

RMR2 is planned to include 10 groups, as in RMR1, including groups receiving just one intervention as to validate that we are successfully recapitulating effects reported in prior work. We continue to reason that very little additional information would result from also including the six possible combinations of two out of four interventions. Three out of four, on the other hand, gives key information, especially on the existence of any antagonistic interactions.

Controls

We again plan to use two types of control for each intervention. Mock treatment controls closely resemble an experimental treatment but lack the active ingredient or activity. They are administered in the same fashion as an active treatment. This might include, for example, a saline injection, giving a gene therapy vector lacking the experimental gene or with the code “scrambled” (nonfunctional), or spiking animal chow with an inactive drug. The other type of control is termed “naïve”, where animals receive only the experimental treatments (if any) assigned to their group, without any additional mock treatments. Comparison between mock and naïve controls allows us to discriminate treatment outcomes from effects which might be related to the act of administration or the vehicle composition. For example in RMR1, animals in the “No mTERT - Naive” group received only rapamycin, HSCs, and a senolytic, while those in the “No mTERT – Mock” group received all three experimental treatments, plus an AAV9 gene therapy vector lacking mTERT. Comparison of outcomes between mock and naïve groups is necessary to distinguish real from placebo effects – in this example, if treatment with the AAV9 vector itself has an effect on the animals, independent of the mTERT gene.Results from the RMR1 study, however, indicate that it is likely not necessary to have equally as many naïve and mock control animals in each group, as little to no effect is observed across groups from administration or vehicle treatment alone. Reducing the size of naïve controls in a given intervention group, for example, would still provide necessary data about possible administration effects, without sacrificing statistical power for in-group analysis.

Scale of study

We aim to conduct RMR2 on a similar scale as RMR1, including 500 male and 500 female mice. In the event of funding limitations, one suboptimal possibility is to conduct RMR2 with only 500 animals to start, which cuts the study size in half, while maintaining statistical power for individual treatment groups. While instead using a single sex would enable us to initiate RMR2 more expediently, it remains a very undesirable option due to significant known sexual dimorphism with respect to both lifespan and intervention effects in rodents and humans alike, which indeed has been observed in RMR1. Other choices for cutting down the number of mice would significantly impact the ability to draw statistically significant conclusions. To elaborate on this: typical “simple” studies in the literature with just two treatment groups rarely use fewer than the 50 mice per treatment group that we are using in RMR1. The many-group, multiplexed nature of our studies affects this in two main ways, which essentially cancel out: on the one hand we can ask about the effect of an intervention across all the five groups that receive it versus those that do not, so effectively the group size for each sex is 250 rather than 50; but on the other hand, the complexity introduces a “multiple hypothesis problem” whereby one expects to see a large difference between SOME pair of groups purely from random chance because there are lots of pairs, meaning that the level of statistical significance required to draw conclusions is higher. Thus, it would be inadvisable to drop below 50 mice per treatment group. We of course get far more from this design of study than the above, not least in terms of synergy information, but that does not change the basic group size requirement just outlined.