Abstract
Animal models are commonly used to investigate developmental processes and disease risk, but humans and model systems (e.g., mice) differ substantially in the pace of development and aging. The timeline of human developmental circuits is well known, but it is unclear how such timelines compare with those in mice. We lack age alignments across the lifespan of mice and humans. Here, we build upon our Translating Time resource, which is a tool that equates corresponding ages during development. We collected 1,125 observations from age-related changes in body, bone, dental, and brain processes to equate corresponding ages across humans, mice, and rats to boost power for comparison across humans and mice. We acquired high-resolution diffusion MR scans of mouse brains (n = 16) of either sex at sequential stages of postnatal development [postnatal day (P)3, 4, 12, 21, 60] to track brain circuit maturation (e.g., olfactory association, transcallosal pathways). We found heterogeneity in white matter pathway growth. Corpus callosum growth largely ceases days after birth, while the olfactory association pathway grows through P60. We found that a P3–4, mouse equates to a human at roughly GW24 and a P60 mouse equates to a human in teenage years. Therefore, white matter pathway maturation is extended in mice as it is in humans, but there are species-specific adaptations. For example, olfactory-related wiring is protracted in mice, which is linked to their reliance on olfaction. Our findings underscore the importance of translational tools to map common and species-specific biological processes from model systems to humans.
Significance Statement
Mice are essential models of human brain development, but we currently lack precise age alignments across their lifespan. Here, we equate corresponding ages across mice and humans. We utilize high-resolution diffusion mouse brain scans to track the growth of brain white matter pathways, and we use our cross-species age alignments to map the timeline of these growth patterns from mouse to humans. In mice, olfactory association pathway growth continues well into the equivalent of human teenage years. The protracted development of olfactory association pathways in mice aligns with their specialized sense of smell. The generation of translational tools bridges the gap between animal models and human biology while enhancing our understanding of developmental processes generating variation across species.
Introduction
The brain consists of a complex network of fiber bundles connecting different regions. In humans, some brain pathways continue to mature and rewire well into adulthood (Arain et al., 2013), while other pathways mature for only a few years after birth (Cohen et al., 2016). Fiber bundles that grow for an extended time are often associated with higher-level cognition, and their development is concomitant with behavioral changes during adolescence (Arain et al., 2013; Qu et al., 2015; Cohen et al., 2016). It is well established that human development, including brain growth, proceeds more slowly than in many other mammals (Clancy et al., 2001; Workman et al., 2013; Charvet et al., 2023). Whether fiber bundle growth and maturation are unusually protracted in humans relative to other species remains uncertain. We have yet to map the timetable of the white matter pathway maturation in mice and compare these timelines with those found in humans. Such a comparison would necessitate age alignments across postnatal ages between humans and mice.
Here, we build on a long-standing project called Translating Time, which is a compilation of time points defined as biological or behavioral processes recorded in at least two species. Examples include when neurons are generated and when eyes open. We use these data to identify corresponding ages across humans and various model systems (Figs. 1, 2; Clancy et al., 2001, 2007; Workman et al., 2013; Charvet, 2021). Previously, we developed an integrative approach to generate cross-species age alignments across the lifespan. We used anatomical, behavioral, and transcriptional information to equate ages across the lifespan of humans and great apes (Charvet et al., 2023). Here, we expand on this approach, and we use a range of metrics to generate age alignments across the lifespan of humans and mice. Corresponding time points are extracted from bone, brain, and tooth maturation, as well as from temporal changes in gene expression. Collected time points vary widely in age (humans, 1.5 d postconception to 65 years after birth; mouse, 1 d to ∼2 years after birth). We used these data to compare the timeline of mouse brain circuit maturation with that of humans.
We collected time points across pre- and postnatal development in humans, rats, and mice to generate age equivalences across species. A, According to these data, a P4 mouse equates to a human at Gestational Week (GW) 24, and a P10 mouse equates to a human within their first years. Moreover, a P60 mouse maps to a human in their teens, and a 2-year-old mouse maps to a human in their 70 s. B, We fit a smooth spline through time points expressed in years postconception in humans and mice with SE from PIs. Black circles represent age predictions from the polynomial equation. Here, data from rats were mapped onto mice when data in mice were not available. C, Time points in rats are plotted against those found in mice. We fit a smooth spline through these data to equate the corresponding ages of rats to mice. D, A bar plot shows the distribution of time point types collected across humans, mice, and rats. Most of the time points are from the bone and brain. E, Age ranges for these specific categories largely span the first year. See Table 1.
Time points (expressed in years postconception) are plotted in (A) mice versus humans and (B) rats versus humans, which are color-coded by process. We fit a smooth spline by process to evaluate whether some processes show evidence of acceleration relative to others in (C) mice and in (D) rats. Time points show evidence of acceleration during postnatal development. All identified biological processes show relative acceleration as is evidenced by the smooth splines fit through the data.
We used a 9.4 T MR scanner to collect diffusion MR scans of mouse brains, and we tracked postnatal growth of white matter pathways. Diffusion MR relies on the diffusion of water molecules to identify the location and orientation of tracts. This information can be used to quantify the maturation of brain microanatomy. For example, fractional anisotropy is a measure of fiber orientation and which varies across the lifespan (Yeatman et al., 2014; Slater et al., 2019; Conte et al., 2024). It can also be used to reconstruct the white matter pathways in three dimensions (Mori and Zhang, 2006; Wedeen et al., 2012; Vasung et al., 2019; Charvet et al., 2020; Cottam et al., 2023). Here, we used diffusion metrics and tractography to characterize the development of human and mouse brain pathways at postnatal ages. There is a particular sequence of white matter myelination in different pathways (Cohen et al., 2016), and different white matter pathways grow for variable lengths of time (Catani and Mesulam, 2008; Tanaka-Arakawa et al., 2015; Cohen et al., 2016; Tak et al., 2016; Mohammad and Nashaat, 2017; Wilkinson et al., 2017; Vasung et al., 2019; Wilson et al., 2021; Chen et al., 2022). Here, we focus on the growth of association white matter pathways in mice to establish developmental timelines in murine brain circuit maturation. Some pathways are homologous across humans and mice (e.g., cingulate bundle; Schmahmann and Pandya, 2006; Wu et al., 2016), but direct comparisons across species are challenging because many pathways are not homologous. For example, some white matter bundles (e.g., arcuate fasciculus) have no clear counterpart in mouse brains (Catani and Mesulam, 2008; Catani and Thiebaut De Schotten, 2008; Calabrese et al., 2015; Becker et al., 2022; Charvet et al., 2022). Conversely, mice exhibit a relatively expanded olfactory association pathway traversing white matter olfactory cortical areas (Figs. 3, 4; Collins et al., 2018), but this pathway is not distinctly observed in humans. We instead compare growth trajectories of multiple pathways across mice and humans, some of which are homologous (e.g., corpus callosum) and some that are not (e.g., olfactory association pathways). We supplement our studies of diffusion MR tractography with observations from tract-tracers and temporal patterns in transcriptional profiles to confirm the validity of tractographies. We focus on transcripts expressed by long-range projections because they provide a complimentary means to compare timelines of brain pathway maturation (Gutman et al., 2012; Jones et al., 2013; Chen et al., 2015; Johnson et al., 2019; Charvet et al., 2022; Crater et al., 2022; Cottam et al., 2023).
We used a 9.4 T MR scanner to acquire whole-brain diffusion MR tractography. These postmortem mouse brains were scanned at P 3, 4, 12, and 60 with the color coding representing the average direction of fibers. For example, green fibers course across the dorsal (d) to ventral (v) axis, and pathways that course across the rostral (r) to caudal (c) axis are coded in blue. In the lower panel, horizontal slices (1 slice thick) capture fibers coursing through that filter. For example, the cingulate bundle is evident, though small at birth, and expands postnatally. Cerebellar parallel fibers are evident through postnatal development (asterisk). The color-coding maps use the following abbreviations: D, dorsal; V, ventral; M, medial; L, lateral; P, postnatal day. See Table 2.
We compare diffusion MR tractography scans at P60 and 3 with tract-tracers. We found a concordance in the location and direction of fibers in the white matter across methods. Specifically, we found that P60 and 3 mouse brain diffusion MR scans show bundles of axons coursing across the rostral to caudal axis through olfactory cortices (white asterisk) and the cingulate cortex (beige asterisk). Tract-tracers label fibers that course through a similar location and direction as those found with diffusion MR tractography. Bottom panel, Additional set of tract-tracers show pathways connecting various regions and which identify white matter pathways coursing across the rostral to caudal axis. The position and orientation of fibers resemble those observed with diffusion MR tractography. A 1–1.5 mm filter was applied to remove ∼50% of the fibers to enable visualization of brain pathways. See Table 4.
Materials and Methods
Translating Time
The Translating Time dataset consists of time points and observations. A time point is a specific biological or behavioral process that is recorded in at least two species. An observation is an individual's datum point. We leveraged observations from rats to equate ages across humans and mice. Some observations are available for both rats and humans, but they may be available for both humans and mice. The inclusion of observations from rats increases the sample available for comparisons across humans and rodents. We extrapolated mouse observations from rat observations when data were available for both rats and humans, but not available for mice. We collected a total of 474 time points and 11,125 observations collected across humans, rats, and mice. We use time points collected from past studies, and we collected new time points at pre- and postnatal ages (Figs. 1, 2; Workman et al., 2013; Charvet et al., 2017, 2023; Charvet and Finlay, 2018; Table 1). Time points were extracted from multiple modalities (e.g., brain, dental, bone, locomotor, life history maturation; Workman et al., 2013; Cusack et al., 2024). We classified time points as belonging to the brain (e.g., dental eruptions), bone (e.g., humerus bone ossification onset), life history (e.g., sexual maturity), locomotion (e.g., locomotion), or others. The bar plot shows the number of time points classified according to these processes. Accordingly, most time points are recorded from brain and body maturation, and relatively few time points are from life history (Fig. 1D). When maximum lifespan is omitted, mouse and human ages range from the second day of prenatal life to 65 years after birth in humans and extend to ∼2 years after birth in mice (Fig. 1). These time points are collected from multiple sources, which means that some variation in observations may be due to differences in procedures in data acquisition, quality, and definitions. Some variation in extrapolated ages may be due to such factors. Nevertheless, this approach is amenable to the acquisition of relatively large samples as is evidenced by a total of 1,125 observations captured across rats, mice, and humans.
This dataset was used to find corresponding ages across species
Time points were extracted from abrupt and gradual changes in biological processes. We generated cross-species age alignments from temporal changes in normalized gene expression. RNA sequencing data are from the frontal cortex of humans (n = 6; 35 d to 55 years old) and mice of varying ages (n = 17; age range, embryonic day 11–22 months of age). These are from individuals varying in age (GSE47966; Lister et al., 2013) and were used in a past publication (Hendy et al., 2020). We selected 10 genes that could be used as markers of long-range projections (e.g., NEFH), synapse formation (e.g., ARC) as well as cell proliferation (e.g., SOX1) and neurogenesis markers common to both species. We captured the age at which genes peak in their expression in the frontal cortex of humans and mice. We fit nonlinear regressions to age and normalized gene expression to extrapolate the age of peaks or plateaus in gene expression in both species (i.e., R library package easynls; model, 2, 3). Comparing the temporal profiles in the expression of these genes provides an additional means with which to study the timeline of circuit maturation in mice and humans.
Translating Time model
We used time points from both rats and mice to translate ages across humans and mice. Some time points are available in rats and humans, but not in mice. We predicted age in mice from rats to increase the samples with which to equate ages across humans and mice. There were 361 time points common to mice and rats and 54 time points available in only rats. We first imputed data with the multiple imputations by chained equations algorithm in cases where there was one missing time point across rats and mice (Van Buuren and Groothuis-Oudshoorn, 2011). We then fit a smooth spline through the log-transformed age (expressed in days postconception) to predict mouse time points from those available for rats (Fig. 1C). The smooth spline fit, which is generated from nonimputed and imputed observations, accounted for 89.75% of the variance in these data so that we can with good certainty predict corresponding ages across rats and mice (n = 948 observations). In the present study, we compared the overall pace of development and aging across humans and mice. We do so by fitting a smooth spline to the age of mouse and human time points, and we fit a polynomial equation derived from the smooth spline to provide an equation that finds corresponding ages across humans and mice. We considered standard errors (SE) from prediction intervals (PIs) to quantify within species variation in our cross-species age alignments. This method deviates from past methods in that we do not fit linear models or quasi-Newton optimization-based models (Workman et al., 2013; Charvet et al., 2023). We do not test for specific heterochronies in behavioral and biological processes across humans and mice. We know that primates and rodents possess heterochronies in different structures during development. For example, it is known that cortical neurogenesis is unusually extended in primates relative to rodents (Workman et al., 2013). We consider all time points available, and we do not derive specific translations for heterochronic processes though it is possible to derive age translations for specific biological and behavioral processes (Clancy et al., 2001; Workman et al., 2013). These different approaches mean there may be some differences in specific age alignments from past approaches. We chose this approach because we are here concerned with overall age alignments rather than translating specific biological or behavioral processes.
Specimen collection for MR imaging
Mice were maintained under a 14/10 h light/dark photoperiod with Purina Mills rodent diet (Animal Specialties and Provisions) and water available ad libitum. Mouse brains at postnatal days (P)3, 4, 12, 21, and 60 were extracted, immersion fixed in 4% paraformaldehyde, and stored at 4°C. Mouse brains were collected for purposes other than this study, which meant that this research was exempt from IACUC approval. Postmortem mouse brains were sent to the University of Delaware Center for Biomedical and Brain Imaging (CBBI). Details of the samples are in Table 2.
These show details about mouse brain scans from the 9.4 T MR scanner collected at the University of Delaware
MR scanner
We used a 9.4 T Bruker Biospec 94/20 small animal MR system (Bruker BioSpec MRI, Ettlingen Germany) with a 1H receive-only 2 × 2 mouse brain surface array coil along with a 86 mm active detuned 1H transmit–receive coil to scan postmortem mouse brains at the University of Delaware CBBI. Mouse brains were first immersed in phosphate-buffered saline (PBS) doped in gadolinium-diethylenetriamine pentaacetic acid for 24 h, and they were then rinsed in PBS for 6 h and then placed in Fomblin (a chemically inert perfluoropolyether fluorocarbon; Solvay Solexis), before and during brain scanning. The cerebellum of some mice incurred some damage, and this damage shows in some of the scans. We acquired diffusion-tensor images (DTI) using a spin-echo–based echoplanar imaging sequence with the following parameters: an echo time of 45 ms, a repetition time of 500 ms, isotropic 100 μm3 voxel resolution, 65 sampling directions with a b value of 4,000 s/mm2, five b = 0 s/mm2 images, and a slice thickness of 0.1 mm, with fat suppression and field-of-view saturation. The total acquisition time for each dataset is ∼2 h. Details on the individuals scanned (sex, male or female) including their age and spatial resolution are listed in Table 2.
Diffusion metrics
We used the DSI Studio software (https://dsi-studio.labsolver.org/) to extract several diffusivity metrics, which include anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) from P3 to 60 mouse brains (n = 15). We set parameters as those described above to reconstruct these images. We omitted a brain scan from these analyses because one brain had some damage in the caudal cortex. These diffusion metrics capture many dimensions of brain structure. In general, FA is a measure of microstructural integrity, MD is an inverse measure of membrane density (Alexander et al., 2011), AD is a measure of the rate of diffusion along its main axis, and RD is a measure of the magnitude of direction perpendicular to fiber tracts (Song et al., 2002; Sun et al., 2006; Soares et al., 2013). We use these measures to characterize the development of microstructural properties in the cortical white matter.
We relied on the relative brightness from FA images and the direction of diffusion to define the cortex white matter. Some pathways are very small at P3 making it a challenge to quantify individual pathways over the course of postnatal development. Rather, we selected white matter fibers that course tangential to the gray matter of the cortex. That is, we selected fibers that primarily course across the medial to lateral direction within medial regions of the cortex and across the dorsal to ventral direction in lateral areas of the cortex. In one case, many fibers in the rostrolateral cortex projected across the rostral to the caudal axis. Our definition spanned lateral regions of the cortex in caudal regions in other cases. This region of interest (ROI) should include the transcallosal, corticothalamic, and thalamocortical pathways (Fig. 8B). The boundaries of the cortex white matter were bounded laterally by primary and supplementary somatosensory areas in relatively rostral regions and the temporal association area in caudal regions. In most rostral and caudal regions, FA values are relatively low in the white matter, and fiber orientation can be intermingled. We therefore omitted measurements through these most rostral and caudal areas. We fit a polynomial regression to test whether age is a significant predictor of these metrics.
Tractography reconstruction
We used high angular resolution diffusion imaging (HARDI) to reconstruct fiber tracts from postmortem mouse brain diffusion weighted MR scans (Fig. 3). The tractography was reconstructed with the Diffusion Toolkit (trackvis.org) to visualize and quantify pathways. Tracts were set to not exceed a 45° angle between two consecutive orientation vectors (angle threshold; Cottam et al., 2023). We used the fiber assessment by continuous tracking algorithm to reconstruct tracks coursing through the whole brain. No fractional anisotropy threshold was applied to reconstruct tracts, which is consistent with past work (Takahashi et al., 2012; Charvet et al., 2022; Cottam et al., 2023).
Tractography validation
Observations from tract-tracers guided the selection of tractography parameters to reduce the probability of producing inaccurate reconstructions. Indeed, it is well known that the tractography does not systematically align with findings from tract-tracers (Jones et al., 2013; Chen et al., 2015; Johnson et al., 2019; Crater et al., 2022). There are a number of reasons for such inconsistencies (Jones et al., 2013; Oh et al., 2014; Maier-Hein et al., 2017). For example, there is a likelihood that tracts misconnect in zones of crossing fibers. In mouse brains, tracts may be misconnected at the gray to white matter cortex because of the sharp turn these fibers make at the white to gray matter boundary, though they may be less of a problem in gyrencephalic brains, which do not make such sharp turns at the white to gray matter (Charvet et al., 2022). Tractography parameters are set to filter tracts that make such sharp turns because such parameters may lead to misconnections of pathways within the white matter. These uncertainties make validation of results from tractography an important component of diffusion MR tractography studies.
Our past work showed that there is strong concordance between diffusion MR tractography and tract-tracers of commonly studied pathways. This is the case for the fornix as well as the hippocampal commissure (Charvet et al., 2022). Here, we focused on olfactory association pathways, and we used tract-tracers from the Allen Brain Institute to compare tractography reconstructions with tract-tracers in the mouse brain (Fig. 4). We found strong concordance in the location and orientation of white matter pathways in mice and in humans, but it was not clear whether the terminations within the gray matter are accurate. Given these observations, we measured cross-sectional areas of pathways in the white matter, and we did not focus on the precise terminations of pathways within the gray matter.
Pathway maturation quantification
We measured the cross-sectional area of cortical and olfactory association pathways in mice from P3 to 60 with TrackVis (Tables 2, 3). We set ROIs to visualize pathways and to measure cross-sectional pathway areas. Some fibers were measured from coronal and sagittal slices. We set ROIs through coronal slices at the level of the anterior commissure to measure the cingulate bundle and olfactory association pathway. The cingulate bundle was defined as a pathway coursing through the rostral to caudal axis of the cingulate cortex (Bubb et al., 2018). We included fibers coursing across the rostral to caudal axis of the cingulate cortex. In the medial cortex, some fibers course across the rostral to caudal and medial to lateral direction, but these were not included in the definition of the cingulate bundle. The olfactory association pathway was defined as a pathway coursing across the rostral to caudal axis of olfactory cortices. We also measured the area of the corpus callosum and the anterior commissure from midsagittal slices. We used these data to test whether age is a significant predictor of the white matter pathway area as a means to evaluate the timeline of growth during mouse development.
Results
We first discuss how we equate corresponding ages across mice and humans (Figs. 1, 2) and how we compare the timeline of brain pathway maturation in mice relative to those of humans (Fig. 4).
Translating Time across humans and mice
We collected time points (n = 474) across pre- and postnatal ages in humans, mice, and rats. We used observations from rats to increase the sample available for comparison across mice and humans because we have many observations from rats as part of the Translating Time project. In some cases, observations from a given time point are available for both rats and humans, but they are not available in mice. We used observations in rats to predict corresponding ages in mice when data were not available for mice (see Materials and Methods; Figs. 1, 2). We fit a smooth spline through log-transformed time points (expressed in age postconception) of mice and humans. This smooth spline fit to humans and mice accounted for 86.6% of the variance (Fig. 1B), which means that we can with good certainty predict age in humans from equivalent time points in mice. We also fit a polynomial equation to generate an equation that can find corresponding ages across species (Fig. 1) as follows:
An interesting observation from these cross-species age alignments is that the pace at which species proceed through time points varies as a function of age. Humans take several years to proceed through milestones postnatally, but these equivalent time points only span a few months in mice. Accordingly, humans in their late teenage years equate to a mouse at ∼P60 [estimate ± SE (P38; P91); Fig. 1]. We can evaluate how the pace of biological processes in humans compares with those of mice. We fit smooth splines through different biological programs to evaluate whether or multiple biological processes underlie the relative deceleration in the pace of processes in humans (Fig. 2C,D). According to these smooth splines, several biological processes show relative accelerations in humans versus mice and humans versus rats (Fig. 2). Those include life history, bone, and brain development, which show a relative deceleration in humans relative to mice, but such deceleration diminishes as development progresses into aging. These observations suggest the development of human infancy and childhood is relatively extended compared with mice. We next evaluate the development of pathways from diffusion MR tractography, and we use cross-species age alignments to map the timeline of white matter pathway maturation in mice compared with humans.
Diffusion MR tractography
We collected diffusion MR scans of mouse brains (n = 16) from P3 to 60 (Fig. 3). Whole-brain tractography shows white matter pathways, which are color-coded by the average fiber direction. At P60, many white matter pathways course across the medial to lateral axis (e.g., colossal, corticosubcortical pathways), and others course across the rostral to caudal axis. These pathways include a relatively large bundle that spans olfactory structures named the olfactory association pathway as well as the cingulate bundle that spans the cingulate cortex. We found that the diffusion MR tractography from P3 to 60 mouse brains is concordant with tract-tracer information from mice at P56 (Fig. 4). Similar to what is observed in diffusion MR tractography, tract-tracers identify axon bundles coursing across white matter olfactory structures as well as rostral to caudal cingulate fibers coursing across the cingulate cortex (Fig. 4, Tables 3, 4). These consistencies bolster confidence in the white matter pathway reconstructions, and their analyses can be used to capture changes over the course of postnatal development.
We integrate these observations with temporal changes in gene expression (Fig. 7B) and diffusion MR tractography (Fig. 7C,D). At the earliest age examined (P3), many association pathways are evident. Some pathways grow substantially during postnatal development through P60 (Figs. 5, 6). The olfactory association pathways and the cingulate bundle are relatively small at P3–4, but they expand postnatally (Figs. 5, 6). In contrast, transcallosal and corticosubcortical pathways appear relatively invariant with age (Figs. 5, 6). We fit linear models to test whether the pathway area varies significantly with age (log-transformed age expressed in days postbirth; Fig. 7). The olfactory association pathway (y = 0.23x − 0.046; n = 16; adj R2 = 0.42) and the cingulate bundle (y = 0.10x − 0.013; n= 16; adj R2 = 0.63) both significantly increase with age (slope, p = 0.004 and p = 0.00015, respectively). However, the anterior commissure (y = 0.03x + 0.07; slope p = 0.11376; R2 = 0.11) and the corpus callosum (y = 0.08x + 0.79; slope p value = 0.44; R2 = 0.02) do not increase significantly with age. Left and right olfactory association pathways and cingulate bundles also show substantial growth up to P60 (Fig. 7E). We found global increases in white matter FA (Fig. 8B), but MD, AD, and RD were relatively consistent over the first 3 postnatal weeks of development, with a slight decrease starting at P21 (Fig. 8C–E). We fit polynomial regressions to predict whether the log-transformed age accounted for a significant percentage of the variance in these diffusion metrics. Age accounted for a significant percentage of FA (y = 0.42 + 0.1x – 0.01 × 2; R2 = 0.405; F = 4.1; p < 0.05; n = 15), but age accounted for a relatively low percentage of the variance for the other diffusion metrics (AD, R2 = 0.35; F = 3.27; p > 0.05; MD, R2 = 0.2; F = 1.5; p > 0.05; RD, R2 = 0.12; F = 0.79; p > 0.05; n = 15). Together, these results demonstrate that white matter matures extensively at postnatal ages and that some white matter pathways grow through P60 in mice.
Coronal slices show tractography of mouse brains at different age (P3, P12, P21, p60). At P3, many major cortical association fibers are present and many of them grow substantially. Dashed circles draw attention to the cingulate bundle as well as the olfactory association pathways and how they grow postnatally across successive postnatal ages. A one-slice-thick coronal slice captures fibers coursing through these areas. P, postnatal day.
We measured the area of pathways at multiple stages of development (i.e., P3, 12, 21, and 60). We measured the area of the corpus callosum, the cingulate bundle, and the olfactory association pathway. We used ROIs to visualize coronal or sagittal slices and to quantify the area of various pathways over the course of development (Fig. 7). P, postnatal days.
We measured the area of pathways (A), which include the cingulate bundle, the olfactory association pathways, the corpus callosum, and the anterior commissure. B, We also evaluated temporal changes in transcriptional profiles of NEFH and MBP expression in mice. These are markers of long-range projecting neurons and myelination. Accordingly, NEFH and MBP expression increases with age in mouse brains, especially between P4 and 14. The asterisk shows areas of protracted increase in expression in olfactory structures from P28 to 56. C, We fit a nonlinear model to NEFH expression from RNA sequencing data from the frontal cortex of mice and humans, which show plateaus in peak expression occur at ∼50 d in mice and in their teens in humans. D, We fit linear models to test whether age could be used to predict the cross-sectional area of these pathways. E, Left and right olfactory association pathways and cingulate bundles grow. The olfactory association pathway and cingulate bundle increase significantly with age, but this is not the case for the corpus callosum and the anterior commissure, which are relatively invariant with age. We fit linear models on these graphs when the slope of the model is significant. In situ expression data are from the Allen Brain Institute. See Table 3.
A, We used FA images from P3 to 60 mouse brains to quantify developmental changes in the cortex white matter. The color coding reflects the average direction of fibers within each voxel. The direction of diffusivity is shown for select voxels with relatively high FA values. Bright voxels show white matter. B, We also show examples of three-dimensional definitions of the white matter cortex. C–F, Diffusion metrics (i.e., FA, MD, AD, RD) are plotted against the log-transformed age. C, Polynomial fit to these data show that FA significantly increases with age, but this is less the case for other diffusion metrics (D–F). A, Color-coding directions are the same as in Figure 3.
Area measurements from diffusion MR scans
Summary of tract-tracer information from the Allen Brain Institute (Oh et al., 2014)
Transcription to validate pathway growth
We leveraged spatiotemporal patterns in gene expression to confirm quantitative analyses of growth trajectories made with diffusion MR tractography. We considered the spatiotemporal expression of myelin basic protein (MBP) and neurofilament heavy polypeptide (NEFH) with in situ expression because transcripts that will translate to these proteins are expressed by large neurons with long axonal structure and can be used as markers of neurons with long-range projections (Zečević et al., 1998; Khalil et al., 2018). There is a sharp increase in the expression of MBP and NEFH from P4 to 14 (Fig. 7B). The expression of MBP and NEFH increases in some structures up to at least P56, and this includes olfactory structures (Fig. 7B). Such increases are also observed from RNA sequencing data, which show peak expression in NEFH in the frontal cortex in the first postnatal month (P19.5) and teens in humans (12.3 years postbirth). A similar situation is observed with frontal cortex NEFH expression, which increases up to ∼50 d in mice and in their teens in humans (Fig. 7C). These data further confirm that white matter pathways mature well into human teenage years, which equates to the second postnatal month in mice.
Discussion
In this study, we leveraged Translating Time as well as neuroimaging techniques to assess spatiotemporal changes in the growth of murine white matter pathways for comparison with humans. We also generate cross-species age alignments in humans and mice, which can serve as a resource for researchers who need to map findings across species (Clancy et al., 2007). Additionally, this study demonstrates that diffusion MR tractography is a valuable tool to characterize the brain's white matter tracts in model systems, especially once the tractography is integrated with other methods to validate the tractography.
Translating Time
We expand on our long-term research project called Translating Time, which equates corresponding ages across model systems and humans (Clancy et al., 2001; Workman et al., 2013; Charvet and Finlay, 2018; Charvet et al., 2023). We incorporate time points from multiple metrics to find corresponding ages during pre- and postnatal development. In humans, biological processes that span some postnatal ages, including neural circuit maturation and bone ossification, proceed more slowly at postnatal ages than they do during prenatal stages. This relatively lengthened pace of postnatal development supports the notion that humans have a relatively extended duration of childhood and helplessness during which infants can absorb vast amounts of information. Learning without the need for immediate task performance results in a powerful foundation model that supports rapid and versatile learning later on (Cusack et al., 2024). These time translations enrich our understanding of conserved as well as modified biological processes leading to phenotypic variation.
There is much interest in identifying equivalent timelines across mice and humans. We can use these age alignments to predict the timeline of myelination and senescence (Salat et al., 2010; Yeatman et al., 2014; Slater et al., 2019; Conte et al., 2024). For example, the process of cortical myelination in humans extends well through the 30 s (Grydeland et al., 2019), which corresponds to mice in their second year. Our age alignments can be used to predict the timeline of different aspects of myelination (Slater et al., 2019). More generally, these age alignments can predict the timeline of a multitude of biological processes across humans and mice.
The age alignments from this study are valuable for pinpointing the most suitable ages in mice and to investigate human developmental disorders, including autism and schizophrenia. These alignments also provide insights into the utility of mice as models for age-related diseases, including Alzheimer's disease (Westergaard et al., 2019). Our age alignments show that a 65-year-old human maps onto a ∼1.6-year-old mouse. Given that Alzheimer's disease risk substantially increases in the 70 s and beyond, our study suggests that a comparable age in mice would extend beyond 1.6 years of age. Importantly, mice rarely live past 18 months of age (Fahlström et al., 2011) and do not show the full spectrum of age-related human diseases (Marx et al., 2013; Rigby Dames et al., 2023). Accordingly, mice are limited to study human aging due to their shortened lifespan compared with humans. Other model systems such as companion animals (e.g., cats, dogs) show much promise to understand human aging because they spontaneously recapitulate brain plaques and tangles, and they can develop a condition called cognitive dysfunction syndrome, which is a condition akin to human dementia (Landsberg et al., 2012; Chambers et al., 2015; de Sousa et al., 2023). Therefore, broadening our repertoire of model systems, integrating noninvasive lines of investigations, and collaborating with veterinarians are exciting venues to define and remedy long-standing issues in aging.
Diffusion MR tractography captures white matter pathway growth
Diffusion MR can be used to reconstruct brain pathways and should only be paired with other biological approaches such as transcript expression, histology, tract-tracers, or other techniques (Axer et al., 2011; Gutman et al., 2012; Calabrese et al., 2015; Aydogan et al., 2018; Winnubst et al., 2019; Axer and Amunts, 2022; Charvet et al., 2022; Charvet, 2023; Johnson et al., 2023). Here, a comparison of tract-tracers and tractography reconstruction guided the tractography reconstruction parameters (Chen et al., 2015; Martin-Lopez et al., 2018; Charvet et al., 2022; Cottam et al., 2023). White matter association pathways grow for an extended duration in mice. These data show that brain pathways mature for an extended time at postnatal ages in both mice and humans.
There are substantial differences in wiring across species. For example, humans possess several cortical association pathways (e.g., arcuate fasciculus, inferior longitudinal fasciculus), which are linked to higher cognitive capacities and language. Many of these cortical association pathways are found in apes and monkeys, but they become progressively more elusive in increasingly distant species. Indeed, many of the association pathways found in humans are also elusive in mice. Mice also possess association pathways (e.g., olfactory association pathways) that do not have a clear homology to humans. Given these observations, we did not focus on comparing specific pathways across humans and mice. Instead, we focus on overall trajectories in white matter pathway development in mice and how such timelines compare with humans.
We found heterogeneity in white matter pathway maturation in mice, which is a situation akin to what is found in humans (Cohen et al., 2016. In mice, the corpus callosum and the anterior commissure do not grow significantly through postnatal ages whereas the olfactory association pathway and the cingulate bundle grow well through P60. The FA measurements we collected revealed global increases in cortical white matter through studied postnatal ages in mice. Similar increases in FA have also been observed in humans where FA increases through many white matter pathways through childhood (Krogsrud et al., 2016). Therefore, our study shows that the extended development of some white matter pathways in mice follows a timeline that is similar to humans. While the overall timelines are similar across species, our study also highlights interesting cross-species differences. Here, mice show protracted development of limbic structures, including the cingulate bundle and olfactory association pathways. In humans, cortical association pathways mature for a relatively extended duration. In humans, these pathways are linked to higher cognitive processing, and some of our species-specific adaptations, including language, are especially protracted throughout development, and their growth extends well into our 20s. The protracted development of limbic pathways in mice is linked to murine specializations for olfaction. These observations raise the intriguing possibility that protracted development of pathways may be a signature of sensory or cognitive species-specific specializations.
Conclusion
Translating Time is a powerful tool that provides an easily accessible method to correlate studies between animal models and humans. We collected 474 time points related to brain, dental, bone, and locomotor maturations, as well as lifespans from previous studies, and we equated them across humans, rats, and mice. We also use diffusion MR tractography to identify key tract characteristics in white matter pathways such as the cingulate bundle and corpus callosum in P3, 12, 21, and 60 mice and map the developmental timeline to their equivalencies in humans. The growth of the Translating Time tool is providing researchers with important translational context to increase the impact of the utility of model systems and the reliability of cross-species experimentation and comparison.
Data Availability
Diffusion MR scans and cross-species time points are available on Dryad (https://doi.org/10.5061/dryad.8pk0p2nzt). The P60 and 21 mouse diffusion brain scans used in this study are available on a previously published Dryad dataset.
Footnotes
This work was supported by Eunice Kennedy Shriver National Institute of Child Health and Human Development Grants 1R21-HD-101964-01A1 and 7R21-HD101964-02 (to C.J.C.); Institutional Development Award Networks of Biomedical Research Excellence Pilot Grant P20-GM-103446 from the National Institute of General Medical Sciences (NIGMS; to C.J.C.); NIGMS Core Center Access Award P20-GM-103446 (to C.J.C.); startup funds from Auburn University (to C.J.C.); and an Auburn undergraduate research fellowship (to M.B.). The Centers of Biomedical Research Excellence Grant 5P20-GM-103653 was used for research at Delaware State University; Opinions are not necessarily those of the National Institutes of Health.
↵*N.C.C. and K.O. contributed equally to this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Christine J. Charvet at charvetcj{at}gmail.com.
