Astrocyte process coverage, morphology, and interactions differ between regions of chronic active lesions

Immunofluorescent (IF) imaging using GFAP labeling showed that astrocyte morphology is greatly altered in and around chronic active lesions. Using Iba1 to label myeloid cells and identify the lesion rim, astrocyte density intensified moving through the perilesion and became increasingly gliotic approaching the lesion core (Fig. 1a). Higher magnification shows significant GFAP density changes within the lesion core compared to NAWM (Fig. 1b, c). Importantly, core-associated astrocyte processes were closely associated with remaining SMI-labeled axons, some of which appear to have swollen retraction bulbs (Fig. 1c, d). These data demonstrate significant morphological heterogeneity between astrocytes and their processes from the perilesion that continues into the core, indicating that the function of astrocytes may differ between regions of the chronic active lesion. Further, the close association of astrocytic processes with axons in the core suggests a vital interaction that could have either protective or detrimental consequences.

Fig. 1

Astrocyte coverage, morphology, and cellular interactions differ spatially across chronic active lesions. Astrocytes, myeloid cells, and axons were identified using antibodies directed against GFAP, Iba1, and SMI31/32, respectively. a The density of Iba1+ cells was used to identify the lesion rim, outlined with white dotted lines. Astrocyte GFAP density was increased in the lesion rim and core compared to NAWM. b NAWM astrocytes appeared less reactive and more diffuse compared to c the lesion core, where d astrocytes maintained close association with axons, as indicated by white arrowheads. Scale bars, 500 μm (a) and 10 μm (b–d)

Distinguishing the regional protein profiles of astrocytes in chronic active lesions

To gain insight into the functions of astrocytes in and around chronic active lesions, we utilized cell-specific spatial protein profiling to examine labeled astrocytes within defined ROIs. We used the GeoMx® Digital Spatial platform as outlined in Fig. 2a. ROIs were identified using tissue maps of characterized chronic active lesions labeled with PLP and MHCII in serial sections (Fig. 2b, c). Since GFAP labeling was ubiquitous throughout the lesion and closely associated with other cell types (Fig. 1), it presented as a poor morphological marker for isolating astrocytes as a spatial identifier. To specifically mask for astrocytes within designated ROIs, we employed the astrocyte-specific marker ALDH1L1, as it primarily labels the cell body and large processes and was relatively evenly distributed between ROIs (Fig. 2d, e). Following tissue processing, the lesions were scanned, and ROIs were identified based upon proximity to the MHCII+ microglial/macrophage line from serial sections and nuclear density (Fig. 2f, g). The protein panels assessed included those related to cell identification and activation status (Neural Cell Subtyping, Glial Cell Subtyping), confirmed and feasible drug targets (IO Drug Target), and those that included indicators of cell survival, proliferation, and inflammation (MAPK Signaling) (Fig. 2h). Together, these panels allowed for comparison of the activation and functional status of astrocytes between various regions of the chronic active lesion and provide insight into potential drug targets.

Fig. 2

Parameters of the NanoString GeoMx® Digital Spatial platform. a Schematic representation of the NanoString GeoMx® Digital Spatial platform protocol. Chronic active lesions were characterized using b PLP to assess myelin loss and c MHCII to identify the lesion rim. ALDH1L1 was used for identification of astrocytes in both d the NAWM and e lesion and demonstrated co-localization with GFAP. f Regions of interest labeled with ALDH1L1 and SYTO13 were collected from g NAWM, the perilesion, lesion rim, and lesion core areas. h Spatial protein profiling analysis of collected ROIs was conducted using antibody panels for neural cell subtyping, glial cell subtyping, IO drug targets, and MAPK signaling. Schematic in a was created using BioRender.com

Astrocytes from the lesion core are the most spatially diverse

Analysis of normalized protein levels per ROI revealed that astrocytes in the lesion core exhibited the most distinct protein expression profile compared to those in other regions (Fig. 3a). This trend was the most prominent comparing the lesion core to the NAWM and perilesion and diminished at the lesion rim (Fig. 3b–d). While astrocytes within the NAWM and perilesion were similar, the lesion rim had a modest amount of differentially expressed proteins compared to both the NAWM and the perilesion (Fig. 3e–g). These data are corroborated by single-nucleus RNA sequencing of chronic active lesions reported by Absinta et al. [21], where the largest astrocytic transcriptional differences similarly occurred in the lesion core (Supp. Figure 1a, b). These data reveal a distinct molecular signature of astrocytes in the chronic active lesion core, setting it apart from other regions, particularly the NAWM. Based on transcriptome analysis, it is known that the chronic active lesion contains a multitude of reactive glial cell phenotypes [21]; however, here we describe for the first time the partial protein profile of spatially distinct astrocytes to begin to assess their functional significance.

Fig. 3

Lesion core astrocytes are spatially diverse. a Normalized protein read count was subdivided by ROI across patient lesions and values were sorted by lesion core expression. Astrocytes in the lesion core had the largest number of differentially expressed proteins compared to the b NAWM and c perilesion but were mostly similar to the d lesion rim. Modest changes in protein expression were seen in the lesion rim compared to the e NAWM and f perilesion, while no significant changes were observed in the g perilesion compared to the NAWM. Significant q values were set at > 1 or < − 1 and were determined using the Benjamini–Krieger–Yekutieli multiple comparison analysis for false discovery

EGFR signaling in lesion core-associated astrocytes

Multiple proteins within the MAPK signaling pathway, including EGFR and p44/42 MAPK ERK1/2 (ERK1/2), were upregulated in the core relative to both the NAWM and perilesion, indicating potential activation of the EGFR signaling axis (Fig. 4a, f). EGFR signaling is critical for maintenance of astrocyte growth and differentiation and is also associated with axonal growth and guidance [24]. Additionally, subsequent protein kinase (ERK1/2) activation plays a role in many cellular processes including survival, proliferation, growth, and inflammation [25, 26]. Further, a downstream effector (Fig. 4j), phospho-p90RSK, was also increased (Fig. 4k) and of the proteins analyzed, is one of the only phosphorylated proteins upregulated in lesion core astrocytes. Phospho-p90RSK maintains a critical role in a multitude of cellular functions including, but not limited to, growth factor signaling and cellular metabolism that may vary depending on cell type [27].

Fig. 4

MAPK pathway proteins are upregulated in lesion core astrocytes. a Quantification of EGFR protein expression across ROIs revealed elevated expression in the lesion core compared to NAWM and the perilesion. b, c IF labeling and d high-resolution 3D reconstruction using Imaris software confirmed EGFR protein expression in lesion core astrocytes. e Reanalysis of transcriptomic data [21] using BioTuring Lens software revealed a corresponding increase in EGFR transcript within astrocytes in the core compared to other regions of chronic active lesions. f p44/42 MAPK ERK1/2 expression was similarly found to be elevated in the lesion core and confirmation of astrocyte-specific expression was demonstrated via g, h IF labeling and i Imaris 3D reconstruction. k Additionally, effector molecules of the MAPK pathway, including phospho-p90RSK, had a corresponding pattern of expression. l, m IF labeling confirmed increased expression in the lesion core compared to NAWM and o 3D reconstruction further demonstrated overlap with astrocytes. j Schematic representation of the EGFR pathway demonstrates the activation order of target proteins. p IF labeling and q high-resolution 3D reconstruction using Imaris software demonstrated the lack of EGFR, p44/42 MAPK ERK1/2, and phospho-p90RSK protein expression in control tissue astrocytes. r IF labeling and s Imaris 3D reconstruction confirmed EGFR, p44/42 MAPK ERK1/2, and phospho-p90RSK protein expression in chronic inactive lesion astrocytes. Differences between groups were tested using a one-way ANOVA with Tukey's test for multiple comparisons. Error bars = mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars, 10 μm. Schematic in j was created using BioRender.com

Validation of EGFR-related proteins via IF staining showed increased expression of all three targets in the lesion core compared to NAWM (Fig. 4b, c, g, h, l, m; Supp. Figure 4a–f), and 3D reconstruction using Imaris software confirmed astrocyte-specific expression (Fig. 4d, i, o). To determine if any of these proteins followed similar transcriptional trends, comparative analysis was conducted using a publicly available spatial transcriptomic dataset [21]. Notably, EGFR was the only transcript found to be significantly increased in astrocytes in the core of chronic active lesions (Fig. 4e). However, additional GWAS analyses revealed significant associations between single nucleotide polymorphisms (SNPs) in the ERK1/2 genes, MAPK1 and MAPK3, in both MS onset [22] and severity [23] (Supp. Figure 2b, c, e, f). To determine if this pattern of protein expression was unique to the chronic active lesion core, we performed IF labeling of control and ALS tissue as well as chronic inactive lesion core tissue. Of note, although ALS is traditionally considered a gray matter disease, demyelination and oligodendrocyte death have been observed [28]. Analysis of labeled tissues revealed that none of the EGFR signaling targets were expressed in control tissue astrocytes (Fig. 4p; Supp. Figure 4 g), only astrocytic ERK1/2 was observable in ALS tissue (Supp. Figure 3a, d), and all three were present in the chronic inactive lesion core (Fig. 4r; Supp. Figure 4 h). We used 3D reconstruction to confirm these expression profiles (Fig. 4q, s). These findings further support the importance of the EGFR and MAPK signaling pathways in MS lesion core astrocytes.

Immune checkpoint proteins are expressed by lesion core astrocytes

Immune checkpoints are proteins of the immune system that modulate the duration and amplitude of an inflammatory response to maintain self-tolerance and minimize tissue damage [29]. We and others have recently described the influence of astrocytic immune checkpoint expression during neuroinflammation, and specifically in chronic active MS lesions [30, 31]. Similarly, here we report that lesion core astrocytes upregulate several proteins involved in the regulation of the immune response by immune checkpoints compared to NAWM and perilesion astrocytes including 4-1BB, Tim-3, and MERTK (Fig. 5a, f, k). 4-1BB, a co-stimulatory receptor and member of the TNFR superfamily, has primarily been studied in T cells and can vary in function depending on the inflammatory context and cell types involved [32, 33]. In murine obesity-induced inflammation, upregulation of astrocytic 4-1BB led to increased inflammation [34]; however, in a murine model of MS, agonism of 4-1BB inhibited autoreactive T cell responses and limited clinical relapse [35]. Tim-3 has been primarily studied in cancers but has been found to be upregulated by astrocytes during CNS injury and to reduce inflammation, acting as an inhibitory immune checkpoint [36, 37]. MERTK is commonly known to facilitate the phagocytosis of apoptotic cells [38], but it also stimulates the PD-1/PD-L1 axis [39], a prominent inhibitory checkpoint pathway known to be expressed in chronic active lesions [30].

Fig. 5

Immune checkpoint proteins are increased in lesion core-associated astrocytes. a Levels of 4-1BB protein were quantified across ROIs and demonstrated highest expression in the lesion core compared to NAWM and perilesion. b, c Imaging of 4-1BB co-localization with ALDH1L1 showed elevated protein expression within the lesion core compared to the NAWM and was further validated using d Imaris 3D reconstruction. e No significant changes in TNFRSF9 (4-1BB) transcript were observed across ROIs. f Tim-3 followed a similar expression pattern across ROIs. g, h IF imaging demonstrated increased protein expression in lesion core astrocytes compared to NAWM, which was corroborated by i Imaris 3D reconstruction. j HAVCR2 (Tim-3) transcript levels were highest in NAWM astrocytes compared to other ROIs. k Protein profiling analysis revealed that MERTK was most highly expressed in lesion core astrocytes compared to NAWM and perilesion. l, m Elevated MERTK expression was confirmed in the lesion core compared to NAWM via IF labeling and n 3D reconstruction further illustrated expression in astrocytes. o A corresponding increase in MERTK transcript within astrocytes was demonstrated in the lesion core compared to other areas. IF labeling combined with 3D Imaris reconstruction of p, q control tissue and r, s chronic inactive lesions confirmed the lack of observable MERTK, Tim-3, and 4-1BB protein expression in ALDH1L1+ astrocytes. Transcript data was sourced from Absinta et al. [21]. Differences between groups were determined using a one-way ANOVA with Tukey's test for multiple comparisons. Error bars = mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars, 10 μm

We visualized these immune checkpoint-associated proteins using IF co-labeling with ALDH1L1 in both the NAWM and lesion core of chronic active lesions, validating that 4-1BB, Tim-3, and MERTK were expressed and localized with astrocytes in the lesion core (Fig. 5b, c, g, h, l, m; Supp. Figure 5a–f). 3D reconstructions of astrocytes in the lesion core were generated to highlight the prominence and cellular localization of these immune checkpoint proteins (Fig. 5d, i, n). Using spatial transcriptomics, Absinta et al. [21] found that while transcript levels of TNFRSF9 (4-1BB) did not differ between regions, MERTK followed a similar expression profile to that of protein, where expression was most pronounced in core astrocytes (Fig. 5e, j). Interestingly, HAVCR2 (Tim-3) transcript levels had the opposite trend compared to spatial protein expression, with HAVCR2 being the highest in NAWM and diminishing approaching the lesion core (Fig. 5o). Of note, the immune checkpoint targets investigated were largely absent in astrocytes from control, chronic inactive MS lesions (Fig. 5p–s; Supp. Figure 5g, h), and ALS tissue (Supp. Figure 3b, e). These findings emphasize the power of combining spatial transcriptomics with protein analyses and suggest that astrocytes may work to modulate the immune profile of the lesion core microenvironment.

Debris clearance in the lesion core

While astrocytes are not classically thought of as highly phagocytic cells, this function of astrocytes has been highlighted more prominently in the last decade. Microglia and astrocytes have several phagocytosis receptors in common that are used for engulfment of dead and dying cells in the CNS, including MERTK [40]. Interestingly, spatial protein profiling analysis revealed astrocytic expression of another critical mediator of ingestion and degradation in the CNS, cathepsin D (CTSD). CTSD is a lysosomal protease critical for the autophagy-lysosomal system in the CNS [41] and was increased in lesion core astrocytes compared to those in NAWM and in the perilesion (Fig. 6a). We visualized the expression of CTSD in the context of ALDH1L1+ astrocytes and found that CTSD co-localization with astrocytes was more robust in the lesion core compared to NAWM (Fig. 6b, c; Supp. Figure 6a, b). 3D reconstructions were generated to highlight the prominence and cellular localization of CTSD in lesion core astrocytes (Fig. 6d).

Fig. 6

Protein expression profiles in lesion core astrocytes indicate that they are phagocytically active. a Spatial protein profiling of astrocytes in chronic active lesions demonstrated increased expression of CTSD in the lesion core compared to NAWM and perilesion. b, c Protein expression was validated via IF imaging and revealed increased expression in astrocytes in the lesion core compared to the NAWM. d 3D reconstruction using Imaris software confirmed expression within ALDH1L1+ astrocytes. e, f TMEM119 was expressed within astrocytes in the lesion core, but not in the NAWM, and g, h 3D reconstruction illustrated direct overlap of TMEM119 within CTSD+ lysosomes in lesion core astrocytes. i IF labeling and j high-resolution 3D reconstruction using Imaris software confirmed TMEM119 and CTSD protein expression in control tissue microglia with an apparent absence in astrocytes. k IF labeling combined with l high-resolution 3D reconstruction revealed TMEM119 and CTSD protein expression in chronic inactive lesion astrocytes. Differences between groups were tested using a one-way ANOVA with Tukey's test for multiple comparisons. Error bars = mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars, 10 μm

TMEM119 is widely used as a microglial marker; however, it was found to be increased in astrocytes of the lesion core relative to the NAWM and perilesion (Fig. 3b, c). We wondered if TMEM119 was either being expressed or engulfed by astrocytes. Using IF imaging, we labeled chronic active lesions for TMEM119, Iba1, ALDH1L1, and CTSD. As expected, in NAWM, TMEM119 preferentially overlapped with Iba1+ microglia/macrophages; however, in the lesion core, TMEM119 appeared clustered within CTSD-labeled lysosomes of ALDH1L1+ astrocytes (Fig. 6e, f; Supp. Figure 6a, b). Using 3D reconstructions of labeled astrocytes, we observed prominent lobular TMEM119+ staining encapsulated by CTSD+ structures (Fig. 6g, h). Notably, TMEM119 often appeared intracellular and granular in both microglia and astrocytes possibly indicating that both cell types were trafficking microglial proteins to either autolysosomes or lysosomes, respectively. To confirm this potential engulfment, additional phagocytic and lysosomal markers were used to label NAWM and the chronic active lesion rim and core. As expected, NAWM and lesion rim Iba1+ microglia/macrophages expressed high amounts of CD68 and LAMP1 (Supp. Figure 7a–d). However, in the lesion core, the phagocytic and lysosomal markers CD68 and LAMP1 were prominent in ALDH1L1+ astrocytes with Iba1+ fragments found encapsulated by LAMP1+/CD68+ structures (Supp. Figure 7e, f). In contrast, TMEM119 overlapped more prominently with Iba1+ microglia/macrophages with little upregulation of CTSD in control tissue, as expected (Fig. 6i, j; Supp. Figure 6c). Similar to this, little upregulation of either CTSD or TMEM119 was observed in ALS tissue (Supp. Figure 3c, f). Interestingly, however, the expression pattern in chronic inactive lesions more closely resembled that seen in the chronic active lesion core where TMEM119 appeared co-localized with lysosomes of ALDH1L1+ astrocytes (Fig. 6k, l; Supp. Figure 6d). Given these data, astrocytes may have a critical role in removing dead and dying cell debris within the MS lesion core as another mechanism of fostering a microenvironment conducive to neuronal survival and maintenance.