Degus (O. degus) used in this study came from two sources (1) a breeding colony at the University of Montana (n = 28) and (2) a colony stocked using outbred (pups from pregnant females caught from the wild) animals at the Institute of Ecology and Biodiversity, University of Chile, Santiago, Chile (n = 4). No apparent differences were observed between degus from the two sources, and they were thus pooled together. All degus were handled and euthanized in accordance with protocols approved by ethics and Institutional Animal Care and Use Committees as part of retirement from breeding and experimental protocols (AUPs 036-18NIPSYC-061918, 033-18NIPSYC-060618, and 001-19NIPSYC-031919). Only degus without prior exposure to pharmacological or invasive procedures were used, and no animals had been subject to significant social or other stressors. All degus were aged in captivity since birth, kept on a 12:12 h light/dark cycle, housed in groups of 2–4, provided with enrichment (chew blocks, nylon bones, shelters, regular dust baths), and given ad libitum water and food.

The age of the 32 degus (13 males, 19 females) used in the study ranged from 1 to 40 months. Animals were assigned to 4 age groups as follows: 8 juveniles (1–3 months old), 11 adolescents (5–8 months old), 7 younger adults (12–19 months old), and 6 older adults (23–40 months old; Table 1). The age choice for juvenile and adolescent groups was based on observations of puberty onset within the colony (appearance of penile spikes and vaginal openings) as well as prior research suggesting that adolescence typically begins around 3–4 months of age (Hummer et al., 2007; Mahoney et al., 2011; Suckow et al., 2012).

An exclusion criterion was applied to ensure tissues with very poor staining were not included in data analysis. This criterion identified outliers as brain regions where both PNN (WFA) and Iba relative intensity levels were two standard deviations from the corresponding degu's age group mean. Only one data point fit this criteria, a 19-month-old degu's prelimbic cortex, where both values were close to zero, indicating poor staining in that brain slice. This single point exclusion did not have considerable effects on linear regression analyses (slightly smaller p-values, although both PNN and Iba1 were already significant with the outlier) but did reveal a significant PNN difference between juvenile and younger adult degu prelimbic cortex. Iba1 relative intensity remained without any significant age group differences after outlier removal.

Statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, CA, USA) and custom python scripts. Simple linear regression plots display R² values, corresponding p-values, and show 95% confidence bands flanking the top and bottom of the linear regression line. Comparisons between degu age groups were conducted using Kruskal–Wallis nonparametric one-way ANOVA tests and were followed by post-hoc Dunn's tests of all head-to-head age group comparison permutations: ^trendp < 0.1, ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001.

Sigmoid inflection analysis was performed using the logistic sigmoid function:

Data were fitted using a least squares regression method with 1,000 iterations. Constraints were placed on specific parameters (|k| > 0.5; x₀∈ [1, 40]; L unconstrained) to ensure sigmoid plot inflection points lay between the ages of the degus studied (1 to 40 months). An improved model with additional constrains (|k|> 0.5; x₀ ∈ [1, 40]; L ≤ max(PNN/mm² or Iba1 relative intensity)); b ≤ min(PNN/mm² or Iba1 relative intensity) was used to obtain a better fit.

Principal component analysis (PCA) was conducted using the sklearn.decomposition. PCA package. As not all degus had all anatomical ROIs available, the following regions were included in PCA analysis: PNN—RSC, TRN, CA3a, S1, SUB; Iba1—RSC, TRN, CA3a, S1, CA1. The top two principal components (i.e., greatest eigenvalues) were used to plot PCA data. Centroids were calculated for each age group in PCA space, and Euclidean distances between each data point and age group centroids were calculated to assess intra- and inter-group distance differences. Multivariate analysis of variance (MANOVA) analysis was carried out on principal components 1 and 2 and followed by post-hoc univariate analysis of variance (ANOVA) and Tukey's HSD (honestly significant difference) tests.

Degus were euthanized by overdose of sodium pentobarbital (390 mg) and sodium phenytoin (50 mg; Euthansol) followed either by decapitation or cardiac perfusion. Right hemispheres were fixed in 4% paraformaldehyde (PFA) in 1 × phosphate buffer saline (PBS, pH 7.4) for 24 h at 4°C, then soaked in 30% sucrose in PBS for 3 days prior to sectioning. Brain coronal sections were cut at a 30 μm thickness using a Leica SM2010R or a Leica SM2000R sliding microtome. Serial free-floating coronal sections were harvested in 1 × PBS and transferred to cryoprotective solution for −20°C long term storage.

Free-floating coronal sections were rinsed three times with 1 × phosphate buffered saline (PBS) prior to incubation in 5% normal donkey serum (NDS) blocking buffer (Jackson ImmunoResearch Laboratories, West Grove, PA; #017-000-121) containing 0.075% (v/v) Triton X-100 in 1 × PBS for 2 h. Sections were then incubated with primary antibodies and stained against perineuronal nets and microglia. Slices containing thalamic reticular nucleus were also stained for parvalbumin to facilitate its identification. Biotinylated Wisteria floribunda lectin/agglutinin (WFA, 1:1,000) was used to stain perineuronal nets. This lectin binds to a sulfation motif in the chondroitin sulfate glycosaminoglycan chains found in most PNNs (Fawcett et al., 2019). Microglia were identified using ionized calcium binding adaptor molecule 1 (Iba1), a well-established marker present in most microglial subtypes (Shapiro et al., 2009; Stratoulias et al., 2019). WFA and primary antibodies against parvalbumin (PV, 1:1,000) and microglia (Iba1, 1:1,000) were diluted in 5% NDS using proper dilutions (Table 2) for 72 h at 4°C. Sections were then washed three times in 1 × PBS and incubated with appropriate fluorescent secondary antibodies (1:200) against primary antibodies and Alexa Fluor 488-conjugated streptavidin (1:500) diluted in 1 × PBS (as detailed in Table 2) for 2 h at room temperature. Sections were washed three more times in 1 × PBS and incubated in 1 × PBS containing 4′,6-diamidino2phenylindole (DAPI; ThermoFisher; D1306, 10 μM) for 30 min at room temperature. Finally, sections were mounted and coverslipped with Fluoromount-G (SouthernBiotech, Birmingham, AL; #0100-01) for microscopic imaging.

Coronal hemisphere slice overviews were imaged using a high-throughput Olympus VS120 scanning system. Confocal microscopy was conducted using an Olympus FV3000 microscope to obtain high resolution images. Serial optical sections were captured for each slice using a 10 × or 20 × objective, with 1 × -3.94 × zoom using a 1.5 μm step interval for 12–15 slices (z-stack). The maximum intensity projection 2D images were processed using Olympus FV31S-SW Fluoview software (Version 2.6, Olympus Life Science). Identical imaging conditions were maintained for each anatomical brain region.

Expression of PNNs and microglia varied in an age and brain region dependent manner. Epiflourescent overviews showed PNNs are present in specific cortical and subcortical regions that span the rostrocaudal axis of the degu brain (Figures 1A–D). Further, these general PNN patterns were missing or less pronounced in juvenile degus (Figure 1A) and develop into their mature appearance with age. Microglia, on the other hand, showed an inverse pattern where juvenile degus exhibit enhanced immunoreactivity (Figure 1E) that decreased with age (Figures 1F–H), similar to what is seen in mice and humans (Brust et al., 2015; Lenz and Nelson, 2018; Menassa et al., 2022). To quantify these results, we focused on a selection of individual brain regions and then considered their common patterns and diversity.

Four cortical regions were selected that offered a sampling across cortical lobes and hierarchy: the prelimbic cortex, entorhinal cortex, retrosplenial cortex, and primary somatosensory cortex.

The prelimbic cortex (PrL), part of the rostral cingulate cortex (the medial prefrontal cortex), has been extensively studied for its role in context-dependent expectation and behavior (Granon and Poucet, 2000; Corbit and Balleine, 2003; Euston et al., 2012; Riaz et al., 2019; Green and Bouton, 2021; Kolk and Rakic, 2022). As seen in the confocal micrographs of Figures 2A–C, there was a clear jump from weak PNN staining in the juvenile degu to numerous PNNs in the adolescent, younger adult, and older adult degus. PNN density was significantly different between juvenile and adolescent/adult degus [Figure 2D; H_{(3, 23)} = 16.37, p = 0.001, Kruskal–Wallis (KW) ANOVA; p = 0.003, p = 0.02, p = 0.009, Dunn's test], and linear regression showed a significant increase with age (Figure 2F; R² = 0.22, p = 0.012). Although microglia levels did not show significant differences between age groups (Figure 2E), there was a statistically significant negative linear regression over age (Figure 2H; R² = 0.21 p = 0.016). We further characterized these datasets by fitting sigmoid curves to them to assess the temporal dynamics of these plasticity markers. PNNs showed a considerably younger inflection age (4.8 months) than microglia (18.2 months; Figures 2G, I). Relative levels of PNN and microglia are illustrated in Figure 2J.

The entorhinal cortex (EC) is reciprocally connected with the PrL and plays an instrumental role in supporting temporal-spatial context for memory and cognition (Coutureau and di Scala, 2009; Takehara-Nishiuchi, 2014; Marks et al., 2021). PNNs were primarily present in layer 2–4 of the degu EC, but relatively absent in juvenile degus (Figures 3A–C). EC PNNs significantly increased with age, with juvenile degus having significantly lower PNN levels than adolescent and younger adult degus [H_{(3, 22)} = 11.15, p = 0.011, KW ANOVA; p = 0.039, p = 0.025, Dunn's test; Figure 3D]. No significant age effects were detected for levels of microglia (Figures 3E, H). Similar to PrL, the PNN inflection age (3.2 months) was substantially younger than in microglia (12.7 months; Figures 3G, I). One juvenile degu (yellow arrow in Figure 3J) exhibited robust PNN signal and lower Iba1 intensity, possibly indicating early EC PNN maturation in that degu.

The third cortical region analyzed was the retrosplenial cortex (RSC), a posterior complement to the rostral cingulate that is also connected with entorhinal and hippocampal memory systems and has important roles in learning and navigation (Vann et al., 2009). Juvenile degus exhibited very small levels of PNNs in their RSC while at the same time expressing strong Iba1 microglia signal (Figures 4A–C). Linear regression showed a significant Iba1 decrease with age (R² = 0.28, p = 0.002), while PNN densities had an increasing statistical trend (R² = 0.09, p = 0.09; Figures 4F, H). In line with this, juvenile degu RSC had significantly less PNN levels than adolescent and older adult degus [H_{(3, 26)} = 14.55, p = 0.002, KW ANOVA; p = 0.003, p = 0.01, Dunn's test; Figure 4D]. Younger adults exhibited significantly lower Iba1 relative intensities than their juvenile and adolescent counterparts [H_{(3, 26)} = 17.43, p = 0.0006, KW ANOVA; p = 0.002, p = 0.006, Dunn's test; Figure 4E]. Inflection ages were earlier for PNN (5.0 months) than microglia (11.4 months), indicative of earlier PNN maturation in the degu RSC (Figures 4G, I). When considering both PNN and microglia together, juvenile, adolescent, and younger adult groups appear to distribute across different vectors (Figure 4J).

The fourth and final cortical area examined was the primary somatosensory cortex (S1; Figure 5), chosen as a representative region of primary sensory cortex to balance “higher-level” areas of frontal, temporal, and parietal cortices. Juvenile degus exhibited low-PNN but high levels of microglia, which gradually inverted in older animals (layer 4, Figure 5A). Juvenile degus had significantly lower PNN levels than adolescent and older degu age groups [H_{(3, 28)} = 19.55, p = 0.0002, KW ANOVA; p = 0.009, p = 0.0001, Dunn's test; Figure 5E] and significantly higher Iba intensity levels than all older age groups [H_{(3, 28)} = 17.58, p = 0.0005, KW ANOVA; p = 0.02, p = 0.005, p = 0.0012, Dunn's test; Figure 5H]. This pattern yielded a positive linear relationship for PNNs (R² = 0.28, p = 0.001) and a negative relationship for Iba1 relative intensity with age (R² = 0.30, p = 0.0009; Figures 5F, I). PNNs had a younger sigmoid inflection age (3.4 months) than microglia (4.9 months), although the difference in these inflection ages was smaller than in other cortical regions (Figures 5G, J). When both plasticity markers are considered together, juvenile degus cluster relatively distinctly from a larger group of adolescent and adult animals (Figure 5K).

Closer analysis of degu S1 revealed PNNs were expressed differentially between cortical layers. Deep layer 6 had the greatest PNN density, which gradually decreased in more superficial layers (L6 > L5 > L4 > L2/3 > L1; Figures 5B–D). This pattern of S1 laminar PNN expression differs from what is seen in humans, mice, rats, and Mongolian gerbils (Brückner et al., 1994; Hausen et al., 1996; Köppe et al., 1997; Ueno et al., 2019; Venturino et al., 2021; de Medeiros Brito et al., 2022). Detailed analysis of one of the juvenile degus beginning to express PNNs showed they do so predominantly in layer 5/6, suggesting PNN establishment and maturation in the degu starts in the deeper layers and expands upwards (Figures 5B1, C1, D1). Intergroup layer analysis revealed juvenile degus had reduced PNNs in all layers except layer 1 (which had very low PNN densities across all degu ages) when compared to older counterparts (Figure 5L).

The hippocampus has been historically a target for studies of lifespan plasticity and learning and memory (McGaugh, 2000; Stickgold, 2005; Andersen et al., 2006). Different subdivisions of the hippocampus serve different functions and processing steps, including those of the classical trisynaptic loop (DG, CA3, and CA1), the subiculum (SUB), as well as further subdivisions along the proximo-distal axis, including CA3c (closest to DG), CA3b, and CA3a (closest to CA1) (Lorente de Nó, 1934; Sun et al., 2017; Lin et al., 2021), offering an opportunity to examine more detailed circuitry associated with PNN and microglia. CA3a exhibited the strongest PNN signal (Figures 6A, B), characterized by diffuse PNN structures outlining cell soma and neuropil, which contrasted from the more structured cortical PNNs that enwrap larger portions of the proximal dendrites. Diffuse PNNs were also observed in DG, while PNNs in CA3b and CA3c enwrapped greater portions of the proximal dendrites. Juvenile degus exhibited significantly lower CA3a PNN signals compared to adolescents [H_{(3, 25)} = 9.208, p = 0.026, KW ANOVA; p = 0.029, Dunn's test; Figure 6F], and there was a statistical trend toward a significant positive linear relationship between PNN density and age (R² = 0.1, p = 0.09; Figure 6H). The CA3a 2D plot showed some juveniles pooling closer to post-sexual maturity age degus along the PNN axis, possibly indicating early CA3a PNN development (Figure 6L). Similar to what was seen in cortical regions RSC and S1, microglia Iba1 intensity had a significant negative linear regression (R² = 0.24, p = 0.005; Figure 6I), with juvenile degus exhibiting robust Iba1 relative intensity that was significantly higher than that of younger adults [H_{(3, 25)} = 16.22, p = 0.001, KW ANOVA; p = 0.0005, Dunn's test; Figures 6C, G]. Sigmoid curves in CA3a showed a similar pattern as in cortical regions, with PNNs having an earlier inflection age (3.1 months) than microglia (7.0 months; Figures 6J, K). Although showing minimal-to-no PNNs, juvenile degu CA1 expressed strong Iba1 relative intensities similar to those seen in other brain regions, which were significantly higher than in younger adults [H_{(3, 25)} = 10.75, p = 0.013, KW ANOVA; p = 0.006, Dunn's test; Figures 6E, M, N].

Juvenile degus exhibited significantly lower SUB PNN densities compared to adolescent degus [H_{(3, 27)} = 11.13, p = 0.0095, KW ANOVA; p = 0.0052, Dunn's test; Figures 7A, D], although no significant PNN linear trend was found with age progression (R² = 0.017, p = 0.48; Figure 7F). This lack of age-related PNN changes was reflected in the fraction of juvenile degus mixed with post-sexual maturity degus in the 2D plot (Figure 7J). Robust Iba1 signal was seen in juveniles that was significantly higher than adolescent and younger adult age groups [H_{(3, 27)} = 18.32, p = 0.0004, KW ANOVA; p = 0.016, p = 0.0002, Dunn's test; Figures 7B, E] with a statistically trending negative linear relation with age (R² = 0.09, p = 0.09; Figure 7H). Sigmoid analysis showed PNN and microglia have, in contrast to previous regions, similar inflection ages (3.8 and 4.1 months, respectively; Figures 7G, I). The SUB 2D plot showed some level of intermixing between all degu age groups, although juveniles mostly populated the upper-left portion of the plot (Figure 7J).

We next examined the developmental profile of the basolateral amygdala (BLA), in part because of its known role in social-emotional information processing (Phelps and LeDoux, 2005), combined with the importance of social behavior to degu behavioral ecology (Ebensperger et al., 2004). We found differences in BLA microglia, where younger adults had a significantly lower Iba1 intensity than juvenile and adolescent degus [H_{(3, 27)} = 16.59, p = 0.0009, KW ANOVA; p = 0.0005, p = 0.016, Dunn's test; Figures 8B, E], and an overall significantly decreasing linear regression (R² = 0.12, p = 0.048; Figure 8H). However, no significant changes in BLA PNN densities were observed (Figures 8A, D, F), which is illustrated in the BLA 2D plot where all 4 groups had some level of mixing (Figure 8J). The inflection age was much smaller in PNN (1 month) than in microglia (11.4 months; Figures 8G, I), suggesting differing PNN and microglia neuroplasticity timecourses in the BLA.

As the thalamus exhibits mechanisms of plasticity that differ from those seen in cortical regions (Kaas, 1999), we next looked at the thalamic reticular nucleus (TRN). The TRN possesses strong PNN signal and has roles in sensory processing, arousal, and cognition (Li et al., 2020). Further, the TRN is composed of GABAergic neurons that highly express the calcium-binding protein parvalbumin (PV), which are the type of neurons most likely to be enwrapped by PNNs (Zikopoulos and Barbas, 2006; Liu et al., 2017; Fawcett et al., 2019). In addition to WFA and Iba1, we immunostained against PV to clearly identify the degu TRN. We found that adolescent degus express robust PNN densities, which were significantly higher than in the juvenile degu [H_{(3, 27)} = 9.017, p = 0.029, KW ANOVA; p = 0.0209, Dunn's test; Figures 9A, D]. However, linear regression did not show any significant PNN changes with age (R² = 0.0002, p = 0.93; Figure 9F). TRN microglia showed a significant negative linear regression (R² = 0.18, p = 0.016; Figure 9H), with juvenile degus exhibiting a significantly higher Iba1 relative intensities than younger adult degus [H_{(3, 27)} = 12.61, p = 0.0056, KW ANOVA; p = 0.0031, Dunn's test; Figures 9B, E]. The TRN sigmoid inflection age was smaller in PNN (1 month) than microglia (7.8 months; Figures 9G, I), similar to what was seen in most of the examined brain regions. The 2D plot shows some mixing between all groups, with adolescent degus showing the broadest territorial coverage (Figure 9H).

The principle goal of the present study was to assess changes in plasticity markers over brain maturation. We found a general pattern of high microglia but low PNN in early age that gradually inverted with brain maturation, with the majority of statistically significant differences involving comparisons with the juvenile age group (Figures 10A, B). However, both qualitative observation across the brain (Figure 1) and detailed analysis of specific regions (Figures 2–9) suggested that the two markers chosen, PNN and microglia, do not follow the same developmental timecourse. To quantify the transitions in PNN and microglia levels from juvenile to adulthood, logistic sigmoid functions were fitted to age datasets and analyzed for curve inflection points in all brain regions.

Data were initially fitted using minimal constraints to ensure sigmoid plot shapes with inflection points/ages occurring in the analyzed degu ages (1–40 months). Results showed PNN inflection points (median = 3.3 months) were significantly lower (p = 0.0003, Mann–Whitney test) than those seen in Iba1 relative intensities (median = 12.6 months). However, variance in older age groups resulted in some brain regions exhibiting poor curve fittings, which motivated us to further constrain the model to ensure a more accurate fitting capable of identifying the youth-to-adult transition in the data (panels G, I in Figures 2–4, 7–9; panels G, J in Figure 5; panels J, K in Figure 6). Using these settings, we continued to observe a statistically significant difference (p = 0.0011, Mann–Whitney test), with PNNs (median = 3.3 months) showing younger inflection point ages than microglia (median = 9.6 months; Figure 10C, latest microglia transition brain regions in squares, earliest in triangles). Two regions known to process abstract and context information, PrL and EC, showed the latest microglia transitions (in squares) and the primary sensory area, S1, was among the earliest to transition (in triangles). Microglia inflection ages also had a significantly higher variance (σ² = 21.7) than PNN inflection ages (σ² = 2.3; ^**p = 0.0087, F-test).

Principal component analysis (PCA) was conducted to further analyze the different age degu populations. Juvenile degus isolated themselves on the right side of the PCA plot, while post-puberty age degus intermix on the left (Figure 10D1). Juvenile, adolescent, and younger adult degus seem to occupy separate territories in the multi-region, PCA space, while older adult degus are more dispersed. PCA data was plotted using the 2 eigenvectors with largest eigenvalues, which accounted for >70% of the variance (Figures 10D2, D3). Euclidean distances from age group centroids were calculated for all degu data points. All age groups, except older age degus (p = 0.13, Mann–Whitney test), had significantly smaller Euclidean distances from their assigned age group than from others (juvenile: p = 0.0001, adolescent: p = 0.0003, younger adult: p = 0.01, Mann–Whitney test), highlighting the distinction between clusters of juvenile, adolescent, and younger (but not older) adult age groups (Figure 10D4). MANOVA analysis found a significant association between degu age groups and their corresponding PCA datapoints [V = 1.0536, F_{(6, 48)} = 8.91, p = 1.54 × 10⁻⁶, Pillai's trace]. Post-hoc univariate ANOVA analysis was significant for principal component 1 [PC1, which had positive loadings for brain regions' PNN densities and negative loading values for Iba1 intensities; F_{(3, 24)} = 32.709; p = 1.209 × 10⁻⁸, ANOVA], and trended toward significance in PC2 [which had negative loadings for both PNN and Iba1 levels; F_{(3, 24)} = 2.668; p = 0.0704, ANOVA]. Subsequent Tukey HSD tests found significant differences between juvenile and all older age groups' PC1 values (p = 0.001 for all 3, Table 3). Adolescent degus had a statistically trending difference in PC2 values when compared to younger adult degus (p = 0.0701, Table 3).