Male rat leukocyte population dynamics predict a window for intervention in aging
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
Yanai et al. used flow cytometry and methlyation profiling to characterize populations of immune cells in the peripheral blood of male rats, finding age-dependent differences in cell composition and DNA methylation profiles, with marked changes occurring at specific time points (e.g., at 15 months and 24 months of age). This raises the possibility that interventions to modify blood aging may be most effective if done prior to these inflection points. This manuscript will be of broad interest to scientists in the geroscience realm and in particular to those using the aging rat as a model for the aging human hematopoietic system.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Many age-associated changes in the human hematopoietic system have been reproduced in murine models; however, such changes have not been as robustly explored in rats despite the fact these larger rodents are more physiologically similar to humans. We examined peripheral blood of male F344 rats ranging from 3 to 27 months of age and found significant age-associated changes with distinct leukocyte population shifts. We report CD25 + CD4 + population frequency is a strong predictor of healthy aging, generate a model using blood parameters, and find rats with blood profiles that diverge from chronologic age indicate debility; thus, assessments of blood composition may be useful for non-lethal disease profiling or as a surrogate measure for efficacy of aging interventions. Importantly, blood parameters and DNA methylation alterations, defined distinct juncture points during aging, supporting a non-linear aging process. Our results suggest these inflection points are important considerations for aging interventions. Overall, we present rat blood aging metrics that can serve as a resource to evaluate health and the effects of interventions in a model system physiologically more reflective of humans.
Article activity feed
-
-
Author Response
Reviewer #2 (Public Review):
- Using a panel of just 8 monoclonal antibodies authors managed to fit a model performing well on the training data (with r2=0.89), although it is unclear how well it works on a test set.
We are also (pleasantly) surprised that we had such a nice fitting model with only 8 antibodies. We had performed this analysis using a neural network with a training and a validation set. While using the training set did result in increased predictive power of age (r2=0.90) which persisted in the validation set – we were unable to discriminate between healthy and sick animals, as almost all sick animals were aged. When we excluded sick animals, we did not have enough aged animals to partition both a training and validation set. However, we did find that the model had predictive power when validating it …
Author Response
Reviewer #2 (Public Review):
- Using a panel of just 8 monoclonal antibodies authors managed to fit a model performing well on the training data (with r2=0.89), although it is unclear how well it works on a test set.
We are also (pleasantly) surprised that we had such a nice fitting model with only 8 antibodies. We had performed this analysis using a neural network with a training and a validation set. While using the training set did result in increased predictive power of age (r2=0.90) which persisted in the validation set – we were unable to discriminate between healthy and sick animals, as almost all sick animals were aged. When we excluded sick animals, we did not have enough aged animals to partition both a training and validation set. However, we did find that the model had predictive power when validating it on independent experiments using fresh and fixed samples of 4 different age groups (and now also on female data!). This validation is included in Figure 3 – Figure supplement 2 and more robustly discussed in the manuscript.
- Authors bring an important point of the effect of the difference in sample preparation (fresh vs fixed samples) and show (Supplementary Fig. 4) that there is indeed a shift. But it is unclear from the description whether the model was refitted including the new data (which presumably has paired fresh and fixed samples) or if it was the original model applied to these samples.
The original model using only fixed samples was used to plot the new data. Suppl. Fig 4b has now been changed to be more clear regarding this point (now Figure 3 – Figure supplement 2).
- The actual model for calculating age from the cell counts is not in the paper preventing it from being applied by the other groups. In addition, these animals are encoded differently for the data on health and cell counts. Taken together, it is impossible to verify the results provided in this part of the paper.
Our apologies for the confusing way we presented the initial submission. We have changed the Suppl. Table labels to be clearer and have also included the formula used to calculate the model so it will be more useful to the community. (see Figure 3 – Figure supplement 1)
Reviewer #3 (Public Review):
- This study has used only male mice. This is an important limitation that has not been acknowledged in this work. This is a key limitation as the generalizability of their findings to females is uncertain. The work should be extended to include female animals.
We agree that this is a weakness of the study. Unfortunately, the number of animals required to represent each month of age was too high to include both sexes in this initial experiment. We chose one sex in part to minimize confounding parameters of sex-specific differences for this experiment. Given the reported sex differences in human blood aging, it is indeed likely that a sex-specific model would need to be generated for females. However, given the importance of addressing the sex-specificity of the model we present, we did a small additional experiment to evaluate 15 female rats (5 from young, mid-age, and old ages) to examine if there are indeed sex-specific differences in females (yes) and if the male-generated model can be used for female data (also surprisingly yes!). We are cautious to not over-interpret this small data set but suggest that this model may have utility for females as well, and include the identification of sex-specific differences.
- The abstract is not well written and is quite vague. It does not give the reader a clear idea of the rationale for the work. The key findings are not clearly presented, and the claims made go quite far beyond the data presented in the study.
Thank you for the frank comment. We have changed the abstract significantly to more accurately reflect the key findings.
- The authors use the term fragility in the abstract but never again. Potentially they mean frailty, which is a more common term in the geroscience literature. A role for frailty, as a validated measure of overall health in aging humans and preclinical models, has not been considered in this study. It would have been interesting to have measured frailty in the aging rats they investigate.
In the abstract revision we have rephrased the statement that had included fragility. In retrospect, we agree that frailty measurements would have of great interest to measure. We revisited all parameters collected for these studies, but unfortunately, the comprehensive analyses/measurements needed for quantification of frailty were not performed. We have added a statement to the discussion to advocate for this in future studies.
- The authors note that they consider the "health status" of all rats used in the study and indeed they have included a table with some health outcomes. As noted above, a measure of frailty would have been very useful to quantify health in these rats. However, one issue that arises in this study is that the authors have excluded rats with overt sickness from the analysis. This would seem to bias their sample quite considerably. If the authors removed all the animals with overt sickness, then they are looking at blood aging from only the least frail rats in their sample. There is ample evidence that pathology does not equal disease expression. For example, pathology alone does not predict dementia risk in the absence of frailty (PMID: 30663607). Known cardiovascular disease risk factors are more potent in the face of frailty (PMID: 31986990; PMID: 32353205; PMID: 33951158). Similarly, biomarkers and genes do not equal disease expression (PMID: 34933996; PMID: 33210215). The work would be more impactful if the authors also included analysis of blood aging in samples from the rats with overt illness.
We apologize for the phrasing used to describe the excluded animals. The animals that were excluded were moribund and had to be euthanized for humane purposes before their designated cross-sectional time point and blood samples were not collected at the time of euthanasia. Retrospectively, the 13 moribund animals excluded would have potentially provided insight to our model by adding an additional layer of phenotypes. However, we hope that the work we present here could provide a tool for future longitudinal studies to predict pathology, and thus allow the researchers to potentially adjust experimental schedules.
-
Evaluation Summary:
Yanai et al. used flow cytometry and methlyation profiling to characterize populations of immune cells in the peripheral blood of male rats, finding age-dependent differences in cell composition and DNA methylation profiles, with marked changes occurring at specific time points (e.g., at 15 months and 24 months of age). This raises the possibility that interventions to modify blood aging may be most effective if done prior to these inflection points. This manuscript will be of broad interest to scientists in the geroscience realm and in particular to those using the aging rat as a model for the aging human hematopoietic system.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The …
Evaluation Summary:
Yanai et al. used flow cytometry and methlyation profiling to characterize populations of immune cells in the peripheral blood of male rats, finding age-dependent differences in cell composition and DNA methylation profiles, with marked changes occurring at specific time points (e.g., at 15 months and 24 months of age). This raises the possibility that interventions to modify blood aging may be most effective if done prior to these inflection points. This manuscript will be of broad interest to scientists in the geroscience realm and in particular to those using the aging rat as a model for the aging human hematopoietic system.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)
-
Reviewer #1 (Public Review):
In this manuscript, Yanai et al. characterized the age-associated changes in population dynamics of immune cells in the peripheral blood of male rats. Using flow cytometry, authors described distinct characteristics in terms of leukocyte composition associated with young, middle-aged and old peripheral blood, such as increased myeloid bias in the older groups. Also, authors identified a novel aging prediction marker, CD25+ T cell frequency, for age prediction in rats. Additionally, authors performed and analyzed methylation profiles of whole peripheral blood and observed age-associated shifts in the methylome of peripheral blood.
The manuscript presents interesting findings and provides important resources for the aging research community to utilize rats as a relevant model to study blood aging. Only minor …
Reviewer #1 (Public Review):
In this manuscript, Yanai et al. characterized the age-associated changes in population dynamics of immune cells in the peripheral blood of male rats. Using flow cytometry, authors described distinct characteristics in terms of leukocyte composition associated with young, middle-aged and old peripheral blood, such as increased myeloid bias in the older groups. Also, authors identified a novel aging prediction marker, CD25+ T cell frequency, for age prediction in rats. Additionally, authors performed and analyzed methylation profiles of whole peripheral blood and observed age-associated shifts in the methylome of peripheral blood.
The manuscript presents interesting findings and provides important resources for the aging research community to utilize rats as a relevant model to study blood aging. Only minor edits would be needed to improve rigor and clarity of the manuscript.
-
Reviewer #2 (Public Review):
The paper presents an assessment of blood aging in rats using two approaches: evaluating the peripheral blood cell composition and performing whole blood DNA methylation analysis.
Although the overall changes in cell composition have been studied previously, here H Yanai, C Dunn, et al. look at it from a different perspective, focusing on the timing of drastic shifts identifying several key points along with the lifespan. They also train an age predictor based purely on the blood cell composition, determine which cell types contribute to it the most and show that this measure, as well as a simple myeloid/lymphoid ratio, reflect the disease state. Using a dataset previously used to construct rat DNA methylation-based age predictor, H Yanai, C Dunn, et al. focused on previously overlooked aspects of aging in …
Reviewer #2 (Public Review):
The paper presents an assessment of blood aging in rats using two approaches: evaluating the peripheral blood cell composition and performing whole blood DNA methylation analysis.
Although the overall changes in cell composition have been studied previously, here H Yanai, C Dunn, et al. look at it from a different perspective, focusing on the timing of drastic shifts identifying several key points along with the lifespan. They also train an age predictor based purely on the blood cell composition, determine which cell types contribute to it the most and show that this measure, as well as a simple myeloid/lymphoid ratio, reflect the disease state. Using a dataset previously used to construct rat DNA methylation-based age predictor, H Yanai, C Dunn, et al. focused on previously overlooked aspects of aging in rat blood epigenome such as identification of regions differentially methylated in aging and detection of the most affected genes. Similarly, to the cell count approach, authors estimate DNA methylation-based gene breakpoints, detecting two distinct waves of changes and identifying genes, promoter regions of whose were the most affected by this.
H Yanai, C Dunn, et al. estimate the main change in cell composition happening at 15 months, while the young age DNA methylation breakpoint - at 12 months. The fact that the epigenetic effect is observed first is impressive considering the huge effect of cell composition on bulk methylation.Although the paper provides new insights into the trajectory of rat blood aging and the claims are mostly supported by the data, some of the points remain unclear or questionable.
- Using a panel of just 8 monoclonal antibodies authors managed to fit a model performing well on the training data (with r2=0.89), although it is unclear how well it works on a test set.
- Authors bring an important point of the effect of the difference in sample preparation (fresh vs fixed samples) and show (Supplementary Fig. 4) that there is indeed a shift. But it is unclear from the description whether the model was refitted including the new data (which presumably has paired fresh and fixed samples) or if it was the original model applied to these samples.
- The actual model for calculating age from the cell counts is not in the paper preventing it from being applied by the other groups. In addition, these animals are encoded differently for the data on health and cell counts. Taken together, it is impossible to verify the results provided in this part of the paper.
-
Reviewer #3 (Public Review):
Age-related changes that occur in human blood have also been characterized in mouse models. However, one limitation to the mouse model is that mice are not a compelling model of the aging human blood and immune systems. Also, they do not develop spontaneous blood cancers that commonly occur in older people. By contrast, rats and humans share many genes involved in immunity and hematopoiesis that are absent in mice and older rats can develop age-related leukemias as in humans. Here the authors use flow cytometry to investigate changes in peripheral blood composition across the life course in aging male rats. They show that the composition of blood changes during aging and that many of these changes are like those observed in people. Further, they show that these changes are not linear but exhibit clear …
Reviewer #3 (Public Review):
Age-related changes that occur in human blood have also been characterized in mouse models. However, one limitation to the mouse model is that mice are not a compelling model of the aging human blood and immune systems. Also, they do not develop spontaneous blood cancers that commonly occur in older people. By contrast, rats and humans share many genes involved in immunity and hematopoiesis that are absent in mice and older rats can develop age-related leukemias as in humans. Here the authors use flow cytometry to investigate changes in peripheral blood composition across the life course in aging male rats. They show that the composition of blood changes during aging and that many of these changes are like those observed in people. Further, they show that these changes are not linear but exhibit clear inflection points, most prominently at 15 and 24 months of age. DNA methylation changes also exhibit clear inflection points. These findings suggest that rat blood aging is not continuous but occurs in phases. This raises the possibility that interventions to modify blood aging may be the most beneficial if administered before these inflection points.
Strengths:
Some previous studies have examined blood aging in the mouse model, but the mouse is not a compelling model of human blood aging. An advance made here is that the authors show that the peripheral blood from aging male rats shows similar changes to those seen in older humans. This supports the idea that rats are an important model to use in studies of the aging blood and immune systems. Other strengths of this work include: 1) the use of a very large sample size to explore this question; 2) replication of their findings in both fixed cells and fresh blood; and 3) the demonstration of "inflection points" in blood aging with several different experimental approaches. These studies provide strong preclinical data that blood aging is non-linear and suggest there may be optimal windows throughout the aging process where interventions may be most effective.
Weaknesses:
The authors have been reasonably cautious in their conclusions, and most are supported by their data. Still there are some weaknesses in the study.
This study has used only male mice. This is an important limitation that has not been acknowledged in this work. This is a key limitation as the generalizability of their findings to females is uncertain. The work should be extended to include female animals.
The abstract is not well written and is quite vague. It does not give the reader a clear idea of the rationale for the work. The key findings are not clearly presented, and the claims made go quite far beyond the data presented in the study.
The authors use the term fragility in the abstract but never again. Potentially they mean frailty, which is a more common term in the geroscience literature. A role for frailty, as a validated measure of overall health in aging humans and preclinical models, has not been considered in this study. It would have been interesting to have measured frailty in the aging rats they investigate.
The authors note that they consider the "health status" of all rats used in the study and indeed they have included a table with some health outcomes. As noted above, a measure of frailty would have been very useful to quantify health in these rats. However, one issue that arises in this study is that the authors have excluded rats with overt sickness from the analysis. This would seem to bias their sample quite considerably. If the authors removed all the animals with overt sickness, then they are looking at blood aging from only the least frail rats in their sample. There is ample evidence that pathology does not equal disease expression. For example, pathology alone does not predict dementia risk in the absence of frailty (PMID: 30663607). Known cardiovascular disease risk factors are more potent in the face of frailty (PMID: 31986990; PMID: 32353205; PMID: 33951158). Similarly, biomarkers and genes do not equal disease expression (PMID: 34933996; PMID: 33210215). The work would be more impactful if the authors also included analysis of blood aging in samples from the rats with overt illness.
Despite these shortcomings, in general the authors' claims and conclusions are justified by their data.
-