Reactive microglia extend into the pyramidal layer of the CA1 region of the hippocampus. Among known mouse-adapted prion strains, in C57BL/6J mice, SSLOW caused disease with the shortest incubation time (mean ± SD 123 ± 4 days post inoculation [dpi] to terminal disease via intracerebral inoculation [i.c.] route). In the SSLOW-infected C57BL/6J host, PrPSc accumulated in multiple brain regions, including the cortex, thalamus, caudate putamen, and hippocampus (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI181169DS1). These same brain regions developed spongiform vacuolation and profound neuroinflammation, as seen by immunostaining for reactive microglia and astrocytes (Supplemental Figure 1 and Supplemental Figure 2, A and B).

By the clinical onset of the disease, marked thinning of the pyramidal layer in the CA1 area of hippocampus was observed in SSLOW-infected C57BL/6J mice (Figure 1, A and B). Neuronal loss in the pyramidal layer coincided with an increase in density of reactive microglia (Figure 1, A–C). The microglia extended processes around neuronal somas or intruded as whole cells into the neuronal layer (Figure 1, B and C). Extensive microglia-neuronal contacts in the CA1 area prompted us to examine other brain regions affected by prions.

Reactive microglia in prion-infected mice extend into pyramidal layer of CA1 area of hippocampus. The CA1 area of hippocampus in noninfected age-matched control C57BL/6J mice (Ctrl) and C57BL/6J mice infected with SSLOW via i.c. examined at clinical onset and terminal stage using staining with anti-IBA1 (a marker of microglia, A and C) and anti-NeuN (a nuclear neuronal marker, A and C). (B) Quantification of the mean intensity of IBA1 and NeuN signals in pyramidal layer of hippocampus. Colors represent different brains. Dots represent mean intensity values in individual fields of view. Average values for each brain are shown as circles. Black lines mark means. n = 3 animals per group. *P < 0.05; **P < 0.01; ***P < 0.001, ordinary 1-way ANOVA followed by Tukey’s multiple-comparison tests. (C) 3D reconstruction of confocal microscopy images of reactive microglia invading pyramidal layer in terminal SSLOW-infected mice. Scale bars: 50 μm (A); 100 μm (C).

In SSLOW-infected mice, reactive microglia envelop neurons. Coimmunostaining of clinically ill C57BL/6J mice infected with SSLOW via intraperitoneal (i.p., 159 ± 13 dpi to terminal disease, mean ± SD) or intracranial (i.c., 123 ± 4 dpi to terminal disease) routes revealed a substantial population of cortical neurons partially or fully enveloped by the soma of reactive microglia (Figure 2, A–C, and Supplemental Figure 3, A–D). In noninfected cortices, such contacts between neurons and microglial soma were very rare or absent (Figure 2D and Supplemental Figure 3A). Three-dimensional reconstruction of confocal imaging further illustrates partial neuronal envelopment by the soma of reactive microglia in prion-infected mice (Figure 2B and Supplemental Figure 4, A and B, Supplemental Video 1, and Supplemental Video 2).

Reactive microglia in cortexes of prion-infected mice envelop neuronal soma. Terminally ill C57BL/6J mice infected with SSLOW via i.p. (A and B) or i.c. routes (C) or noninfected age-matched controls (D), stained with anti-IBA1 (A–D) and anti-MAP2 (A and B) or anti-NeuN (C and D) antibodies and examined by epifluorescence microscopy (A) or confocal microscopy following by 3D reconstruction (B–D). Smaller panels in A and B show enlarged images of individual microglial cells that partially or fully envelopes neuronal soma. In A, dashed circles mark nuclei, arrows and arrowheads point at neuronal and microglial nuclei, respectively. Scale bars: 20 μm (A); 10 μm (B); 50 μm (C and D).

Microglia typically execute phagocytosis by forming pseudopodia with a phagocytic cup (14, 37, 38). This cup engulfs the target, which may include synapses, apoptotic cells, or protein aggregates. Full engulfment of a target by the phagocytic cup results in the formation of phagosomes within pseudopodia. Consistent with this mechanism, pseudopodia with phagocytic cups were observed in adult 5XFAD mice, where microglia partially engulfed Aβ aggregates or neurons (Supplemental Figure 3, E and F). In contrast to the traditional phagocytic mechanism involving pseudopodia and a phagocytic cup, in prion-infected mice, neurons were surrounded by microglial soma (Figure 2, A–C, and Supplemental Figure 3, B–D). The proximity of microglial and neuronal nuclei supports the notion that neurons were encircled by microglial somas, which often exhibited a characteristic cup-shaped appearance (Figure 2A and Supplemental Figure 3D) rather than by extended pseudopodia. Envelopment of individual neurons by microglia in prion-infected brains occurred at a one-to-one ratio (Figure 2, A–C, and Supplemental Figure 3D).

Neuronal envelopment is a common phenomenon observed across prion strains. C57BL/6J mice, infected i.p. with 4 prion strains (SSLOW 159 ± 13 dpi, RML 209 ± 10 dpi, 22L 216 ± 24 dpi, and ME7 293 ± 32 dpi to terminal disease), showed partial envelopment of neurons by reactive microglia (Figure 3, A–D and F). Notably, by the terminal stage of the disease, SSLOW mice showed the highest percentage of partially encircled neurons. Among the 4 strains, SSLOW was characterized by the shortest incubation time and displayed the most pronounced contacts between microglial soma and neurons (Figure 3, A and F). In the mock-inoculated control group, close contacts between microglial soma and neurons were very rare (Figure 3, E and F).

Envelopment of neurons is a common property among prion strains. (A–D) Representative images of neuronal envelopment (arrows) by reactive microglia in cortex of C57BL/6J mice infected with SSLOW (A), RML (B), 22L (C), and ME7 (D) via i.p. and mock-inoculated age-matched mice (E) stained using anti-IBA1 (red) and anti-MAP2 (green) antibodies. The dataset from the 22L-infected and mock-inoculated control groups were also used to report microglia-neuronal colocalization in the Figure 5 of the manuscript by Sinha et al. (49). (F) Quantification of the percentage of neurons that are undergoing envelopment in cortexes of mice infected with SSLOW, RML, 22L or ME7, and mock-infected control mice (Ctrl). Colors represent different brains. Dots represent individual fields of view. Average values for each brain are shown as circles. Black lines mark strain means. n = 4 animals per group. **** P < 0.0001; **P < 0.01, statistical significance versus control; #P < 0.05, statistical significance versus SSLOW by ordinary 1-way ANOVA followed by Dunnett’s multiple-comparison tests. Scale bars: 50 μm.

Mouse-adapted prion strains employed here display distinctive strain-specific cell tropism. For instance, 22L PrPSc was predominantly associated with astrocytes, while ME7 PrPSc exhibited a prevalent localization with neurons (39). In the case of RML, the preference of PrPSc for astrocytes versus neurons is brain region dependent (39). Despite these strain-specific cell tropisms, the observation of neuronal envelopment across all strains, coupled with comparable percentages of envelopment among ME7 (mean 17.9%), 22L (mean 18.1%), and RML (mean 17.1%) strains, irrespective of their cell tropism, implies that the envelopment is not solely driven by PrPSc association with neurons (Figure 3F).

The majority of neurons exhibit only partial envelopment. Envelopment of neurons was observed across all brain regions affected by prions, including the cerebral cortex, caudate/putamen (or striatum), hippocampus, and thalamus (Figure 4, A–D). The same brain regions showed spongiform change, PrPSc accumulation, and reactive gliosis as judged from ionized calcium-binding adaptor molecule 1 (IBA1) and glial fibrillary acidic protein (GFAP) staining (Supplemental Figures 1 and 2), suggesting that neuronal envelopment is linked to prion pathology. Interestingly, fully enveloped neurons were rare, whereas most of the enveloping events were partial.

The vast majority of neurons are only partially enveloped. (A–D) Envelopment of neurons (arrows) in cerebral cortex (Ctx) (A), caudate/putamen (CP) (B), hippocampus (Hp) (C), and thalamus (Th) (D) of C57BL/6J mice infected with SSLOW via i.c. route at the terminal stage. Staining was performed using anti-IBA1 (red) and anti-MAP2 (green) antibodies. (E–I) Quantification of the neuronal area enveloped by microglia. With epifluorescence microscopy, light penetrates the full depth of a cell; thus an overlap of signal from neuronal (NeuN, E) and microglial (IBA1, F) markers is observed when a neuronal body is undergoing envelopment by a microglial cell (G). The percentage of neuronal area enveloped by microglia is estimated for individual neurons as a fraction of NeuN signal overlapped with IBA1 signal. (H) Merged thresholds of images from IBA1 and NeuN channels. Arrow points to a rare event of a complete envelopment. (I) Frequency distribution of the enveloped areas of the individual neurons quantified for the cortex of SSLOW-infected mice. n = 3 animals, n = 1,169, 1,422, and 1,551 envelopment events for individual animals. (J) A gallery of confocal microscopy images of neurons partially or fully enveloped by microglia. Scale bars: 100 μm (A–D); 20 μm (E–H); 5 μm (J).

To estimate the degree of envelopment for individual neuronal cells, we calculated the area of NeuN signal covered by IBA1 signal in the cortex (Figure 4, E–I). Rigorous examination of confocal microscopy Z-stack images confirmed that we indeed observed genuine encircling events, rather than mere vertical adjacency of NeuN+ and IBA1+ cells (Figure 4J and Supplemental Figure 5, A and B, Supplemental Video 3, and Supplemental Video 4). The lack of envelopment of astrocytes argues that envelopment is specific to neurons and is not simply due to microglia proliferation and overcrowding (Supplemental Figure 2B). In the cerebral cortex of SSLOW-infected mice analyzed at the terminal stage, approximately 36% of neurons showed no contact with microglia, while the remaining neurons exhibited varying degrees of encircling by the IBA1+ cells (Figure 4I). Intriguingly, the complete encircling of the entire neuronal signal with the IBA1 signal was observed in only a small percentage of neurons (less than 1%) (Figure 4I). Consequently, the majority of neurons were found to be only partially encircled.

The envelopment of neurons by microglia in a cortex does not lead to a reduction in the total number of neurons. Previous studies have shown that the phagocytic clearance of a cell by microglia in a mouse brain takes 25 minutes to 2 hours (15, 40). In the terminal stage of prion disease, as few as 15.1% of cortical neurons (for RML) were observed to undergo envelopment (Figure 3F). If 15% or more of the neuronal population are phagocytosed every 2 hours, the entire population would be cleared in less than 12 hours. Consequently, we aimed to establish a timeline for neuronal loss in prion-infected cortex.

In C57BL/6J mice challenged with SSLOW via i.p. route, clinical onset occurred around 120 dpi (Figure 5A). Mice were euthanized at 157–166 dpi, when they showed 20% weight loss along with severe motor impairment and behavior deficits (Figure 5A). Neuronal quantification and the percentage of neurons under envelopment were assessed in cortices at regular time points, starting at 64 dpi. At 64 dpi, 78 dpi, and 92 dpi, the percentage of neurons with extensive body-to-body neuron-microglia cell contacts was comparable to the control group, matched in age to the 64 dpi group (Figure 5, B–D). However, a statistically significant increase in the percentage of neurons undergoing envelopment was noted at 106 dpi (mean 12.4% versus 3.0% in control), 2 weeks prior to clinical onset (Figure 5, B and C). The percentage of neurons undergoing envelopment rose with disease progression, peaking at 146 dpi (mean 44.5%) (Figure 5, B and C). However, despite a substantial number of microtubule associated protein 2–positive (MAP2+) neurons being partially enveloped at the clinical stage of the disease, the total number of MAP2+ cells in the cortex did not decrease with the disease progression, suggesting that envelopment does not result in neuronal death (Figure 5B). Since mice must be euthanized at 20% weight loss, defined as the endpoint per requirements of the animal care committee, we do not know whether any neuronal loss would occur in cortices after this point and upon reaching the actual terminal stage of the disease.

Time course of the envelopment. (A) The clinical onset of the disease in C57BL/6J mice infected with SSLOW via i.p. established using the EPM test. Mice were subjected to EPM sessions once per week starting from the preclinical stage. After the first training sessions (not shown), mice naturally acquired a strong preference for the closed arms. The clinical onset was defined as a time point, when the time on open arms consistently increased in comparison with the noninfected age-matched control group (shown by arrow). n = 5 for control; n = 14 for SSLOW until 119 dpi; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001, by Tukey’s multiple-comparison test. (B) Change in the total number of MAP2+ neurons (upper plot) and the percentage of MAP2+ neurons undergoing envelopment (lower plot) during disease progression. Ctrl1 and Ctrl2 are age-matched controls for 64 dpi and terminal, respectively. Aged 740-day-old mice. Colors represent different brains. Dots represent individual fields of view. Average values for each brain are shown as circles. n = 3–5 animals per time point. Means are marked by black lines. Comparison of means with Ctrl1 was performed by nonparametric Mann-Whitney U test. *P < 0.05; **P < 0.01. (C) Representative images of neuronal envelopment in the cerebral cortex of SSLOW-infected C57BL/6J mice collected at 92 dpi, 106 dpi, 122 dpi (disease onset), and 146 dpi. (D) Representative images of aged brains (740 days old). Staining with anti-IBA1 (red) and anti-MAP2 (green) antibodies. Arrows point at neurons undergoing envelopment. Scale bars: 50 μm.

In a cohort of aged C57BL/6J mice (604–704 days old), the number of neurons undergoing envelopment was comparable to that in 2 control groups, matched in age either to the 64 dpi or 157–166 dpi groups (Figure 5, B and D). This observation suggests that the phenomenon of envelopment is not prominently associated with normal aging.

The analysis of gene expression in SSLOW-inoculated mice revealed that, despite the absence of a decline in the cortical neuronal population, neuron-specific genes were downregulated. Notably, genes associated with ligand-gated ionic channels related to memory and learning (Gabrg1, Grin1, Grin2b, Grm2) and synaptic neurotransmission (Snap25, Syn2, Syp) exhibited reduced expression levels (Supplemental Figure 6).

In summary, the above findings establish that neuronal envelopment events were on the rise prior to the clinical onset of the disease. Neuronal envelopment was not accompanied by neuronal loss. These findings are consistent with the fact that, in the majority of cases, the encircling of neuronal soma was only partial.

The envelopment is not selective toward the subpopulation of the most vulnerable PV+ neurons. In previous studies, GABAergic parvalbumin-positive (PV+) neurons were identified as the most vulnerable in prion diseases (41–44). In both humans affected by CJDs and mice infected with mouse-adapted strains, a decline in PV+ neurons was observed at the subclinical stage of the disease (41–44). If microglia are responsible for the loss by selectively targeting PV+ neurons, we would expect PV+ neurons to be enveloped early, i.e., prior to clinical onset. Contrary to this hypothesis, statistically significant envelopment of PV+ neurons was observed only at terminal time points (Supplemental Figure 7, A and B). Moreover, both PV+ and PV– neurons were subject to envelopment, arguing that this process is not selective toward PV+ neurons (Supplemental Figure 7A).

Neurons under envelopment lack apoptotic markers. To investigate whether neurons that undergo envelopment remain viable, we assessed the presence of apoptotic markers. In previous studies on adult neurogenesis through apoptosis-coupled phagocytosis, the neuronal cells at the early stages of phagocytosis by microglia, i.e., the stage of partial engulfment, displayed the strongest activated (cleaved) caspase-3 signal (cCasp3) (15). Staining terminal SSLOW-infected C57BL/6J mice with an antibody to cCasp3, an early apoptosis marker, revealed that the majority of neurons under envelopment were cCasp3 negative (Figure 6, A, D, and E). In fact, the number of cCasp3+ neurons was low and comparable to that of normally aged mice (Figure 6E). On the contrary, in positive controls, C57BL/6J mice subjected to ischemia showed extensive neuronal cCasp3 staining (Figure 6C).

Neurons undergoing envelopment lack apoptotic markers. (A–C) Coimmunostaining of SSLOW-infected C57BL/6J mice at the terminal stage (A and B) or C57BL/6J mice subjected to MCAO and analyzed 5 days after insult (C) using antibody to activated cCasp3 and anti-IBA1 antibody. (B) Confocal microscopy imaging illustrates the intracellular localization of cCasp3 (pointed by arrows) in microglia of SSLOW-infected mice. (D) Coimmunostaining of SSLOW-infected C57BL/6J mice using anti-cCasp3 and anti-NeuN antibodies. (E) Percentage of cCasp3+ microglia (IBA1) and neurons (NeuN) in cortexes of SSLOW-infected mice and in neurons of aged 607- to 740-day-old C57BL/6J mice. ***P < 0.001, Brown-Forsythe and Welch’s ANOVA followed by Dunnett’s T3 multiple-comparison tests. n = 6–7 fields of view. (F) TUNEL staining of SSLOW-infected C57BL/6J mouse at the terminal stage, and sections from the same mouse pretreated with DNase and used as positive controls. Dashed circular lines represent examples of neuronal envelopment. Scale bars: 50 μm (A, C, and D); 5 μm (B); 100 μm (F).

Interestingly, in prion-affected brains, reactive microglia exhibited cCasp3-positive puncta (Figure 6, A, B, and E). Confocal microscopy imaging confirmed that cCasp3 immunoreactivity was associated with microglia and displayed intracellular localization (Figure 6B). Caspase-3 activation in microglia has been recognized as a switch between proinflammatory activation and cell death, as observed in neurodegenerative diseases including Alzheimer’s and Parkinson’s diseases (45, 46).

As an alternative approach for identifying apoptotic cells, we employed the TUNEL assay. TUNEL stains apoptotic cells by detecting DNA fragmentation. In the cortices of terminal SSLOW-infected mice, neurons undergoing engulfment were negative for TUNEL staining (Figure 6F). However, pretreatment of brain slices from the same mice with DNase revealed DNA fragmentation detected by the TUNEL staining (Figure 6F). In summary, these results indicate that neurons undergoing envelopment by microglia lack apoptotic markers.

Changes in neuronal functional state precede their envelopment. Transcriptome analysis identified several neuron-specific genes that were markedly downregulated at the terminal stage. Among these was Grin1, which encodes a critical subunit of the N-methyl-d-aspartate (NMDA) receptor, essential for synaptic plasticity, memory, and learning. To test whether neuronal changes precede their envelopment by microglia, Grin1 levels were quantified in individual cortical neurons at regular time points using confocal microscopy (Figure 7A). A modest yet statistically significant drop in Grin1 signal was observed at 78 dpi, followed by a steeper decline at 92 dpi (Figure 7B). This decline suggests alterations in the neuronal functional state, indicative of neuronal stress. Notably, these changes occurred before the onset of neuronal envelopment by microglia, which begins at 106 dpi (Figure 5B). Confocal microscopy confirmed that microglia envelop neurons with reduced Grin1 signal (Figure 7C).

A decline in neuronal levels of Grin1 with disease progression. (A) Confocal microscopy images of C57BL/6J mice infected with SSLOW via i.p. examined at 92 dpi and terminal stage (157–166 dpi) using anti-Grin1 (red), anti-NeuN (gray), and anti-IBA1(red) antibodies. (B) Quantification of Grin1 levels during disease progression. Colors represent different brains. Grin1 mean intensity values in individual neurons are shown as dots. Average values for each brain are shown as circles. Black lines mark time-point means. The time-point data were compared with Ctrl1 (age-matched controls for 64 dpi mice) by ordinary 1-way ANOVA followed by Dunnett’s multiple-comparison tests. *P < 0.05; ****P < 0.0001. n = 3–4 brains per time point. Age-matched control for terminal mice (Ctrl2) is provided as a reference. (C) Confocal microscopy 3D reconstruction images of individual neurons in age-matched control and SSLOW-infected C57BL/6J mice illustrating low Grin1 signal in neurons enveloped by microglia. Scale bars: 50 μm (A); 10 μm (C).

Neurons are enveloped by PrPSc-positive microglia. Reactive microglia may target viable neurons because they sense the accumulation of prion aggregates on neuronal surfaces. To test this, brains of SSLOW-infected C57BL/6J mice were coimmunostained using anti-PrP antibody 3D17 in combination with anti-MAP2, anti-IBA1, or anti-GFAP antibodies. To avoid confusion between diffuse PrPSc and PrPC, fluorescence microscopy imaging was optimized to preferentially detect granular PrPSc deposits, characterized by a bright emission. In SSLOW-infected animals, the majority of granular PrP immunoreactivity was associated with reactive microglia, including the microglia that enveloped neurons (Figure 8, A and C). Noninfected, age-matched controls did not display any granular PrP immunoreactivity, arguing that in prion-infected animals, the granular PrP signal is attributed to PrPSc (Figure 8, B and F). In SSLOW-infected mice, only a subtle granular PrP signal was detected in neurons (Figure 8, A and D). Very minimal, if any, PrP immunoreactivity was detected in astrocytes (Figure 8, A and E). A similar pattern of granular PrP immunoreactivity, the majority of which colocalized with microglia, was observed in a parallel staining using different anti-PrP antibody SAF-84 (Supplemental Figure 8). Unlike neurons or astrocytes, microglia do not replicate PrPSc but can acquire PrPSc positivity via phagocytic uptake (26, 27).

PrPSc colocalizes with reactive microglia. (A) Quantification of PrPSc colocalization with MAP2+, GFAP+, and IBA1+ cells in SSLOW-infected C57BL/6J mice analyzed at the terminal stage. *P < 0.05; **P < 0.01 by Brown-Forsythe and Welch’s ANOVA followed by Dunnett’s T3 multiple-comparison tests.(B) Quantification of PrP immunoreactivity in SSLOW-infected and age-matched control C57BL/6J mice. Colors represent different brains. Dots represent individual values for field of view. Average values for each brain are shown as circles. Black lines mark group means. n = 3 animals per group. ***P < 0.001, unpaired t test with Welch’s correction. (C–F) Representative images of SSLOW-infected C57BL/6J at the terminal stage (C–E) and age-matched control mice (F) coimmunostained using anti-PrP (3D17) and anti-IBA1 (C and F), anti-MAP2 (D), or anti-GFAP antibodies (E). In C, dashed circles show neuronal envelopment; a gallery of images on the right shows PrPSc+ microglia that envelop neurons. (G and H) 3D reconstruction of confocal microscopy images of PrPSc+ microglia enveloping neurons in SSLOW-infected C57BL/6J mice at terminal stage. Staining using anti-PrP (SAF-84) (G and H), anti-IBA1 (G and H), and anti-MAP2 antibodies (H). Scale bars: 50 μm (C–F); 5 μm (G and H).

Three-dimensional reconstruction of confocal microscopy imaging confirmed the intracellular localization of SSLOW PrPSc aggregates in reactive microglia (Figure 8G). Furthermore, triple coimmunostaining using anti-MAP2 and anti-IBA1 antibodies revealed that PrPSc aggregates in microglia enveloping neurons localized to perinuclear sites (Figure 8H, Supplemental Figure 9, and Supplemental Video 5). Confocal microscopy Z-stack video demonstrated that PrPSc deposits occupy substantial cell volume in microglia enveloping neurons (Supplemental Figure 9 and Supplemental Video 5).

In mice infected with 22L, granular PrPSc was also found in association with microglia (Supplemental Figure 10A), though it appeared to be less abundant. In microglial cells enveloping neurons, 22L and SSLOW PrPSc showed similar patterns of intracellular localization, as observed through 3D reconstruction of confocal microscopy images (Supplemental Figure 10B). The lower abundance and smaller size of granular 22L PrPSc compared with SSLOW PrPSc could be attributed to reduced proteolytic stability and a higher rate of clearance of 22L in lysosomes.

Phagocytic uptake of PrPSc by microglia precedes neuronal envelopment. The observation of PrPSc deposits in reactive microglia raises the possibility that microglia phagocytically uptake PrPSc prior to engaging in neuronal envelopment. To test this hypothesis, we quantified the percentage of PrPSc+ microglial cells and the percentage of neurons undergoing envelopment in the cortices of C57BL/6J mice at regular time points starting at 64 dpi. A significant increase in the percentage of PrPSc+ microglia was observed at 78 dpi (Figure 9A and Supplemental Figure 11), whereas envelopment became noticeable only from 106 dpi onward (Figure 9B). At 106 dpi, nearly 50% of microglial cells showed well-detectable PrPSc deposits (Figure 9A). Remarkably, despite a substantial increase in the percentage of PrPSc+ microglia, the total amount of PrPSc in a crude brain, as quantified by Western blot, remained at very low levels until 106 dpi (Figure 9D). These findings suggest that microglia manage to control prion replication through the removal of PrPSc via phagocytic uptake. Notably, a substantial increase in the number of IBA1+ cells was observed after 106 dpi (Figure 9C), tripling their phagocytic capacity. While infiltration of myeloid cells from the periphery cannot be entirely excluded, previous studies have established that the expansion of microglia in prion diseases is primarily due to the proliferation of resident myeloid cells, with minimal recruitment of peripheral myeloid cells (34, 47). Consistent with a substantial boost in phagocytic capacity of IBA1+ cells was an increase in CD11b and Gal3 levels observed during the clinical stage (Figure 9, F and G). CD11b and Gal3 are involved in 2 different phagocytic pathways. Notably, the percentage of PrPSc+ microglia continued to rise even after 106 dpi (Figure 9A) alongside their substantial proliferation (Figure 9C), indicating that proliferated cells were actively engaged in the phagocytic uptake of PrPSc (Figure 9D). However, despite the profound boost in phagocytic capacity attributed to proliferation, prion replication spun out of control after 106 dpi (Figure 9, D and F). This time point coincides with the rise of microglial envelopment of neurons, which only intensifies with disease progression (Figure 9B). To summarize, phagocytic uptake of PrPSc by microglia preceded neuronal envelopment by 4 weeks, whereas clinical onset followed neuronal envelopment.

Time course of PrPSc uptake, neuronal envelopment, and microglia proliferation. Changes in the percentage of PrPSc+ microglia (A), the percentage of MAP2+ neurons under envelopment (B), the total number of IBA1+ microglial cells per field of view (C), the amounts of total PrP (no PK treatment) and PrPSc (after PK treatment) as estimated by Western blot (D), and the total number of PrPSc+ microglia per field of view (E) in C57BL/6J mice infected with SSLOW via i.p. route with disease progression. Data are represented as mean ± SEM. Comparisons with Ctrl (combined age-matched controls for 64 dpi and terminal points) were done using Kruskal-Wallis test followed by Dunn’s multiple-comparisons test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. n = 18–63 fields of view (3–5 brains) per time point for B, and n = 17–60 (4–5 brains) for A, C, and E. Term, terminal animals collected at 157–166 dpi. Dashed line shows clinical onset. (F) Representative Western blots of total PrP (no PK treatment, -PK), PrPSc (after PK treatment, +PK), Tubb3, CD11b, and Gal3 in brains of SSLOW-infected C57BL/6J mice. PrP is detected by 3D17 antibody. (G) Quantification of Western blots of Tubb3, CD11b, and Gal3; signal intensities were normalized per intensities of actin for each individual Western blot. In F and G, Ctrl1 and Ctrl2 are age-matched controls for 64 dpi and terminal time points, respectively. Data are represented as means ± SD. n = 3–5 animals per group, *P < 0.05; ***P < 0.001; **** P < 0.0001; each time point was compared with the combined control group (Ctrl1 + Ctrl2) by Brown-Forsythe and Welch’s ANOVA followed by Dunnett’s T3 multiple-comparison tests.

Microglia engaged in envelopment are characterized by activated hypertrophic lysosomes. The observation of PrPSc-positive microglia aligns well with previous studies that document profound upregulation of phagocytic activity in microglia during prion diseases (48–50). Given that sustained phagocytic activity necessitates the upregulation of lysosomal degradation, we quantified cathepsin D, a proteolytic enzyme used to assess lysosomal activity, during disease progression (36). In noninfected mice, the majority of cathepsin D immunoreactivity was not associated with IBA1+ cells (Figure 10A). However, in SSLOW-infected mice, granular cathepsin D deposits were observed in microglia (Figure 10A). The upregulation of microglia-associated cathepsin D coincided with a rise in neuronal envelopment (Figure 10B), with both processes progressing in parallel with the disease advancement. Confocal microscopy revealed colocalization of cathepsin D with PrPSc deposit in reactive microglia (Figure 10C). Careful examination of individual envelopment events revealed cathepsin D–positive microglia engaged in encircling neurons (Figure 10D). Moreover, cathepsin D–positive vacuoles were observed on the periphery of neuronal material that was fully engulfed by microglia (Figure 10D).

Microglia engaged in envelopment have activated hypertrophic lysosomes. (A) Epifluorescence microscopy images of cortices of noninfected age-matched control and C57BL/6J mice infected with SSLOW via i.p route and examined at the terminal stage using anti–cathepsin D (red) and anti-IBA1 (green) antibodies. (B) Changes in the integrated density of cathepsin D associated with microglia (black) and the percentage of MAP2+ neurons undergoing envelopment (red) with disease progression. Data are represented as mean ± SEM. n = 11–63 fields of view (3–6 brains) per time point. Comparisons to Ctrl (age-matched control for 64 dpi and terminal points) were done using Kruskal-Wallis test followed by Dunn’s multiple-comparisons test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Terminal animals collected at 157–166 dpi. (C and D) 3D reconstruction of confocal microscopy imaging of SSLOW-infected C57BL/6J mice illustrating colocalization of PrPSc (SAF-84, gray) with cathepsin D (red) in microglia (IBA1, green) (C) and envelopment of neurons (NeuN, gray) by cathepsin D–positive (red) microglia (IBA1, green) (D). (E) Maximum intensity projection confocal images of LAMP1+ compartments (green) in microglia (IBA1) in cortices of SSLOW-infected C57BL/6J mice analyzed at the terminal stage along with age-matched control mice. (F) Quantification of LAMP1 integrated density in individual microglial cells engaged or not engaged in neuronal envelopment in cortices of SSLOW-infected mice, and age-matched controls. n = 21–39 individual cells. **P < 0.01; ****P < 0.001, by Kruskal-Wallis test followed by Dunn’s multiple-comparisons test. (G) Confocal microscopy image of LAMP1+ compartments in cortices of SSLOW-infected C57BL/6J mice. Scale bars: 50 μm (A and C); 20 μm (G); 10 μm (D and G).

Next, we quantified lysosomal compartments in individual microglial cells by confocal microscopy imaging of lysosome associated membrane protein 1 (LAMP1), which is highly expressed in lysosomal membranes. In contrast to age-matched controls, in SSLOW-infected animals, we observed substantial LAMP1 immunoreactivity associated with microglia (Figure 10E). Remarkably, within the SSLOW-infected brains, LAMP1 immunoreactivity was significantly higher in microglial cells engaged in neuronal envelopment compared with microglia not involved in envelopment (Figure 10, F and G). Collectively, these results support the idea that microglial cells engaged in neuronal envelopment have activated, hypertrophic lysosomes.

Neuronal envelopment is independent of the CD11b pathway. Among several microglial phagocytic pathways, the CD11b-dependent pathway was previously found to be responsible for phagocytosis of newborn cells, including neurons during development (37, 51). CD11b was also shown to be involved in the phagocytosis of neurons in glia-neuronal cocultures (52). To test whether the same pathway is responsible for envelopment in prion-infected mice, we analyzed the time course of the disease and neuronal envelopment in CD11b-knockout mice infected with the SSLOW strain (Figure 11). There were no differences with respect to incubation time to the terminal disease or the amount of PrPSc between CD11b–/– and control C57BL/6J (WT) groups (Figure 11, A–C). Neurons were undergoing envelopment in both CD11b–/– and control groups (Figure 11D), while the percentage of enveloped neurons was the same in the 2 groups (Figure 11E). Moreover, no differences between CD11b–/– and the control WT group were found with respect to the density of MAP2+ neurons (Figure 11E), the level of expression of neuron-specific protein Tubb3 (Figure 11, B and C), or microglia activation, as judged from IBA1 immunoreactivity (Figure 11E). Modest upregulation of galectin 3 (Gal3) was seen in the CD11b–/– versus the control group (Figure 11, B and C). Gal3, which is involved in alternative phagocytic pathways, is released by activated myeloid cells and acts as an opsonin by binding galactose residues on the cell surface.

Analysis of neuronal envelopment in CD11b–/– mice. (A) Incubation time to terminal disease in CD11b–/– and C57BL/6J control mice (WT) inoculated with SSLOW via i.c. route. n = 10 animals per group. Mantel-Cox test of survival curves indicated no significant difference between the groups. (B) Densitometric quantification of Western blots for PrPSc, Tubb3, and Gal3 in SSLOW-infected CD11b–/– and WT mice. Data are represented as means ± SD. n = 6–10 per group. *P < 0.05, by 2-tailed, unpaired Student’s t test. (C) Representative Western blots of selected markers in CD11b–/– and WT mice at the terminal stage. For analysis of PrPSc, BHs were digested with PK and stained with 3D17 antibody. (D) Envelopment of neurons by microglia in the cortex of SSLOW-infected CD11b–/– and WT mice at the terminal stage stained using anti-IBA1 (red) and anti-MAP2 (green) antibodies. (E) Percentage of MAP2+ neurons undergoing envelopment, the total number of MAP2+ neurons, and IBA1 immunoreactivity in SSLOW-infected CD11b–/– and WT mice at terminal stages. n = 4 animals per group. Colors represent different brains. Dots represent individual values. Average values for each brain are shown as circles. Means are marked by black lines. Scale bar: 20 μm.

Neuronal envelopment in sCJD individuals. To check whether the partial envelopment phenomenon occurs in human prion diseases, we examined sCJD brains using staining for the β-chain of human HLA-DR, a marker of reactive microglia. In all examined sCJD subtypes, including MM1, MM2C, VV1, VV2, and MV2K (53), we found reactive microglia engaged in partial envelopment (Figure 12). Akin to prion-infected mouse brains (Figure 12A), such partial envelopment was detected across different human brain regions affected by prions, including frontal and temporal cortex, thalamus, striatum, midbrain, and hippocampus (Figure 12, B–F).

Partial envelopment by reactive microglia in sCJD. (A) Representative image of reactive microglia stained with anti-IBA1 antibody in the cortex (ctx) of 22L-infected C57BL/6J mice provided as a reference. (B–F) Representative images of reactive microglia strained with anti-HLA-DR+DP+DQ (CR3/43) antibody in the following subtypes of sCJD: MV2K (B), MM1 (C), MM2C (D), VV1 (E), and VV2 (F). Fc, frontal cortex; thal, thalamus; str, striatum; mdb, midbrain; occ, occipital cortex; hipp, hippocampus; tc, temporal cortex. Arrows point at microglia engaged in envelopment. Scale bar: 20 μm.