Absence of aged Treg leads to retinal neurodegeneration

In the healthy retina, Treg are present in the eye [13] and the retinal parenchyma [14]. Its presence, however, is notably sparse [13, 19]. To determine the number of Treg present in the aged eye, we performed flow cytometry staining of the retina and RPE/choroid of young (2 months) and aged (> 15 months) mice. As previously reported [13, 19], we found very few Treg in the retina of both, young and aged mice (~ 1–2 per mouse retina), which represented around 1% of the CD4+ T cells present (Additional Fig. 1 A, B). This number was increased in the RPE/choroid, were we found ~ 20 Treg cells per mouse and Foxp3+ Treg accounted for about 20% of the CD4+ T cells present (Additional Fig. 1C, D). However, we did not observe any difference in Treg number associated with age. To determine the role of Treg in the adult and aged retina in the absence of apparent disease, we depleted Treg using diphtheria toxin (DT) in young (< 4 months) and aged (> 18months) FoxP3-DTR mice [8, 20] (Fig. 1A). We first confirmed that Treg systemic depletion was successful characterising the presence of Foxp3+ Treg within CD4+ T cells and observed a significant Treg depletion in the lymph node, spleen and blood of both young and aged mice (Additional Fig. 1F-I). As previously described [20], we observed an enhanced systemic inflammation validated by the enhanced proportion of CD25+ CD4+ T cells in the lymph nodes (Additional Fig. 1J). Next, we evaluated changes in the neural retina and retinal pigment epithelium (RPE). Histological analysis at 2.5 weeks post-Treg depletion identified a significant decrease in total retinal thickness in aged but not young mice (Additional Fig. 2A, B). On further analysis, no changes in photoreceptor nuclei were observed in young mice lacking Treg, while aged mice had a significant reduction of total photoreceptors (DAPI+ rows in the ONL) following Treg depletion (Fig. 1B, C). Specifically, photoreceptor loss was associated with a decrease in both, Cone arrestin+ (CA+) cones (Fig. 1B, D) and rods, which were determined by subtracting the count of CA+ cells from the number of DAPI+ nuclei within a specific area (Rods, DAPI+CA−) (Fig. 1B, E). To further investigate Treg depletion-associated neurodegeneration, we examined the second-order neurons in the retina. In normal retinas, the bodies of rod bipolar cells reside in the outer region of the inner nuclear layer (INL). These cells exhibit a cluster of dendrites that extended into the outer plexiform layer (OPL). Treg depletion did not affect overall rod or cone bipolar cell number in young or aged Treg-depleted mice (Fig. 1F-J). However, in PKCα+ rod-bipolar cells and secretagogin+ cone-bipolar cells, Treg depletion in aged mice altered cell body laminarity (Fig. 1F, I). This was more pronounced in PKCα+ rod-bipolar cells; in which despite unaltered total number of somas, there was a significant increase of ectopic PKCα+ cell bodies located below the OPL due to shorter axons (Fig. 1F-H). Upon Treg depletion, we also found bipolar dendrites and synaptophysin sprouts expanding from the OPL towards the ONL (Fig. 1F). This cellular remodelling (consisting of the retraction of the bipolar cells, sprouting of dendrites and formation of ectopic synapses) has been previously observed in diseases associated with photoreceptor loss [21]. We did not observe any alteration in retinal ganglion cells across experimental groups (Additional Fig. 2C-E) suggesting that neurodegeneration following Treg depletion in aged mice predominantly impacted the outermost layers of the retina. Despite the overall systemic inflammation detected (Additional Fig. 1J), we did not observe an enhanced T cell infiltration in aged mice upon Treg depletion at the time point examined (Additional Fig. 2F-H), suggesting that the neurodegeneration observed in the aged Treg depleted mice is not linked to non-specific T cell activation and retinal infiltration due to the absence of Treg.

Fig. 1

Treg depletion in aged but not young mice accelerates retinal neurodegeneration. A Diagram of the treatment regime and research groups. B-E Immunostaining and quantitative analysis of photoreceptors. Representative images showing photoreceptors in the ONL (B) (Cones, CA+; Rods DAPI+ CA−) in young and aged Foxp3-DTR (scale bar = 50 µm). Quantitative analysis of the number of rows of DAPI+ nuclei in the ONL (C). Quantification of the number of cone photoreceptor cells (D) (CA+ cells) and the number of rod photoreceptors (E) (CA−DAPI+ cells in ONL). F-J Immunostaining and analysis of rod bipolar cells (F–H) and cone bipolar cells (I, J). Representative image of PKC-α (F) (red) and synaptophysin (blue, lower panels). Arrows indicate abnormal location of some PKC-α+ cell soma lower within the INL. Arrowheads indicate abnormal rod bipolar dendrite and synaptic vesicles sprouts into the ONL (scale bar = 25 µm). Quantitative analysis of the number of PKC-α+ cells (G) (n = 3–6 mice) and (H), percentage of PKC- α+ somas found in a lower location out of the total of PKC-α+ cells (n = 3–6 mice). Representative image (I) (scale bar = 25 µm) and quantification of rod bipolar cells in the INL (J) using secretagogin staining. K-N Analysis of IBA-1+ microglial cells and GFAP+ Müller cells determined by immunohistochemistry. Representative image of microglial cells (K) (IBA-1+, purple) and gliotic Müller cells (GFAP+ fibers, green) (scale bars = 50 µm). Quantitative analysis of the percentage of total GFAP+ area (L), the number of GFAP+ fibers in the ONL (M) and total number of IBA-1+ cells in all layers of the retina (N). Data information: B-N, n = 3-6 mice, data presented as mean ± s.e.m. *P < 0.05; ** P < 0.01; ***P < 0.001; 1-way ANOVA followed by Bonferroni's multiple comparisons test. Statistical comparisons that were not significant are not indicated in the graphs

Fig. 2

Adoptive transfer of young but not aged Treg rescues aged retinal neurodegeneration. A Diagram showing the experimental design. B-E Immunostaining and analysis of the effect of Treg adoptive transfer on photoreceptors. Representative images (B) (scale bar = 50 µm) and quantification photoreceptors (C-E). Total photoreceptors (C), CA+ cones (D) and CA− rods (E) in the ONL of control, Treg depleted and Treg reconstituted aged retinas. F-J Alterations in bipolar cells after Treg adoptive transfer in aged animals. Representative images (F) (scale bar = 25 µm) and quantification of total rod bipolar cells (H) and rod bipolar cells with a shorter axon and displaced nuclei (I) identified by PKC-α (red) and synaptophysin (cyan) immunostaining (n = 4-6 mice). Representative images (G) (scale bar = 25 µm) and quantification of cone bipolar cells (J) identified by secretagogin immunostaining. K-N Changes in Müller cell gliosis and microglia in aged animals after the adoptive transfer of Treg. Representative image showing microglia (IBA-1+, purple) and Müller cell gliosis (GFAP+ fibers, green) (K) (Scale bar = 50 µm). Quantification of GFAP+ area (L), the number of GFAP+ fibers crossing the ONL (M) and total IBA-1+ microglia (N) in the aged retina. Data information: B-N, n = 3-7 mice, data presented as mean ± s.e.m. ns: not significant, *P < 0.05; ** P < 0.01; ***P < 0.001; 1-way ANOVA followed by Bonferroni's multiple comparisons test

Retinal gliosis is associated with the lack of Treg in the aged retina

To determine if Treg depletion affected retinal glia, we first examined Müller glial cell responses. In the absence of Treg we observed a significant increase in GFAP+ staining in the neuroretina in aged but not young Treg-depleted mice (Fig. 1K), suggesting an intense Müller cell gliosis. In addition to overall increased GFAP expression (GFAP+ area of retina) (Fig. 1K, L) we also observed an expansion in reactive Müller cell number in the ONL (Fig. 1K, M). Müller cell hypertrophied branches could be a compensatory response growing towards the outer limiting membrane to compensate for photoreceptor loss in an attempt to maintain retinal cytoarchitecture, as described in other retinal diseases characterised by photoreceptor loss, such as retinitis pigmentosa, in which Müller cell hyperreactivity has also been observed upon photoreceptor loss [22].

We next examined whether Treg depletion modified microglial activity in the neuroretina. Microglia are the localized immune cells in the retina, and they show an increased activation in the aged retina [1, 3]. While Treg depletion in young mice did not affect retinal microglia, a significant increase in the density of IBA-1+ microglia was observed in the neuroretina of aged mice (Fig. 1K, N). This increase was detected in all retinal layers, but most significantly in the ONL which is normally devoid of microglia (Additional Fig. 3A). IBA-1+ cell increase was mostly related to ameboid-shaped microglia (Additional Fig. 3B), a morphological and phenotypic shift usually associated with pro-inflammatory responses [23]. This morphological change was further confirmed by measuring microglial branching as a function of the distance to cell soma. Treg depleted aged mice show a significantly decreased branching when compared to control aged mice (Additional Fig. 3C). To validate this pro-inflammatory phenotype, we quantified the number of MHCII+ pro-inflammatory microglia [24, 25] in aged Treg-depleted mice and observed a significant increase of these IBA-1+MHCII+ microglia in the neuroretina (Additional Fig. 3D, E).

Fig. 3

Treg depletion leads to blood-retinal barrier damage and innate immune cell infiltration in the subretinal space of young and aged mice. A-B Analysis of RPE cell changes. Representative image of phalloidin (A) indicating the geometry of RPE cells. Stars indicate enlarged cells and arrowheads fragmented cytoskeleton (scale bar = 50 µm). Quantification indicating the number of RPE cells with an enlarged morphology per area of RPE/choroid flatmount (B). C-D Determination of the number of CD68+ cells in subretinal space. Representative image of CD68 and phalloidin immunostaining in the RPE/choroid flatmount (C) (scale bar = 50 µm) and (D) quantitative analysis of the number of CD68+ cells in the subretinal space. E–F Determination of CD68+MHCII+ cells in the subretinal space. Representative image (E) (scale bar = 50 µm) and quantification (F) of CD68+MHCII+ innate immune cells in the subretinal space upon Treg depletion and reconstitution in aged mice (scale bar = 50 µm). G Confirmation of the CD68+Tmem119− nature of the phagocytes observed in the subretinal space. Arrowheads indicate the scarce presence of CD68+Tmem119+ cells (scale bar = 100 µm). Data information: A-F, n = 2–5 mice, data presented as mean ± s.e.m. *P < 0.05; ** P < 0.01; ***P < 0.0051; 1-way ANOVA followed by Bonferroni's multiple comparisons test

These results indicate that Treg are necessary to prevent photoreceptor death and distorted lamination of bipolar cells in aged retinas. This appears to be linked with Müller cell hypertrophy as well as the accumulation of pro-inflammatory microglia in the nuclear layers of the retina.

Treg adoptive transfer prevents retinal degeneration and rescues gliosis

We next sought to determine if Treg reconstitution could rescue age-associated retinal neurodegeneration. To do so, we adoptively transferred purified young and aged Treg (Additional Fig. 2I) into Treg-depleted aged mice to determine whether ageing affects the capacity of Treg to maintain retinal normal immunoregulation and limit the negative effects of an uncontrolled inflammation in the aged neuroretina (Fig. 2A). We first examined if reconstitution with young and aged Treg diminished photoreceptor loss. Surprisingly, transfer of young but not aged Treg rescued photoreceptor (cone and rod) loss in the neuroretina back to control levels (Fig. 2B-E). Although no differences were detected in the number of rod and cone bipolar cells upon Treg depletion (Fig. 2F, G, H, J), adoptive transfer of young but not aged Treg improved the bipolar lamination pattern, showing a decrease in the number of ectopic PKCα+ somas, as shown by the diminished number of PKCα+ cell bodies with a lower nuclear location (Fig. 2F, I). Additionally, young Treg also succeed in reducing the bipolar dendrite sprouting into the ONL (Fig. 2 F, I). Overall, young Treg adoptive transfer rescued retinal neurodegeneration features back to control levels, while aged Treg adoptive transfer was not significantly different to aged Treg depleted mice. We next evaluated if adoptive transfer of Treg in aged mice inhibited Müller cell gliosis. In line with what we observed for photoreceptor loss, only young but not aged Treg transfer prevented Müller cell gliosis in Treg-deficient aged retinas (Fig. 2K, L, M), supporting the concept of Müller cell gliosis being a compensatory response to retinal degeneration as described previously [26, 27]. Lastly, we studied the effect of Treg adoptive transfer on microglia density in the neuroretina. In contrast to what we observed with Müller cells, both young and aged Treg reduced the number of total IBA-1+ microglia as well as the number of IBA-1+MHCII+ microglia present in the neuroretina, however aged Treg were not capable of restoring microglial branching and complex morphology (Fig. 2K, N, Additional Fig. 3A-E). Therefore, since adoptive transfer of aged Treg reduces the number of microglia in the neuroretina, but not photoreceptor loss, these data indicate that the increased microglia in the neuroretina is not triggering photoreceptor loss and neurodegeneration at the outermost layers of the retina. The adoptive transfer of young Treg cells successfully rescued Treg depletion phenotype, as illustrated in Fig. 2. When the efficiency of adoptive transfer was assessed (in a separate study of the same animals aged Treg were not identified in blood, spleen or lymph nodes, in contrast to adoptively transferred young Treg [28]. Despite this observation, the fact that young Treg rescued retinal neurodegeneration phenotype (Fig. 2) and both, young and aged Treg rescued neuroretina microglial phenotype (Fig. 2K, N, Additional Fig. 2A, B, D, E) is evidence for the presence of adoptively transferred Treg cells in the organism. Despite adoptive transfer of young but not aged mice rescuing Treg depletion phenotype, we are cautious with the interpretation of these result, as not being able to detect adoptively transferred Treg in the classic lymphoid organs prevents us from determining whether the reconstitution levels were equivalent for both, young and aged Treg.

Treg maintain RPE integrity and limit phagocyte accumulation in the subretinal space

Since the retinal neurodegeneration and gliosis observed in the absence of Treg in aged mice appears to be associated primarily with the outer retinal layers, which are in closer contact with the retinal pigment epithelium (RPE), we next evaluated the effect of Treg depletion in RPE integrity and immune cell infiltration in the subretinal space. It was previously reported that a range of RPE defects occur in the ageing retina. With age, RPE cells show an altered morphology, which includes discontinuity of the cytoskeletal bands between adjacent cells and altered cytoarchitecture (appearance of enlarged and irregular cells), contributing to altered retinal immune regulation and enhanced chronic low-grade inflammation described in the aged retina [29]. In agreement with this, we also observed significantly more enlarged RPE cells in aged animals (Fig. 3A, B). Next, we sought to investigate the effect of Treg depletion in age-related RPE alterations. Surprisingly, Treg depletion exacerbated age-related RPE pathology not only in aged, but also in young Treg depleted mice (Fig. 3A, B). In addition, CD68+ phagocytes were significantly increased in the subretinal space of both, young and aged mice upon Treg depletion (Fig. 3C, D). As per our observations in the neuroretina, phagocyte infiltration in the subretinal space mainly consisted of CD68+MHCII+ pro-inflammatory macrophages (Fig. 3E, F). Considering that RPE dysmorphology can lead to alterations in the outer blood retinal barrier, we next studied the nature of the phagocytes present in the subretinal space to determine whether they were mainly macrophages accumulated in response to systemic and/or local inflammatory cues, or alternatively, microglia that had migrated to the subretinal space because of the damaged RPE. Unlike what we described for the neuroretina, where the majority of IBA-1+ cells were Tmem119+ microglia (Additional Fig. 3F, G), in the subretinal space the vast majority of CD68+ cells were negative for Tmem119, indicating that CD68+ cells in the subretinal space are peripheral monocyte-derived macrophages, with few Tmem119+ microglia (Fig. 3G). Thus, Treg depletion led to accumulation of innate immune cells in the subretinal space and RPE alterations in both young and aged mice. However, these alterations were associated with neurodegeneration and gliosis only in the aged retina. Taking into account that neurodegeneration is present mostly in the outermost part of the retina, these data suggest that aged retinas are more susceptible to the neurotoxic effects of accumulated pro-inflammatory innate immune cells in the subretinal space. Hence, aged retinas could be more vulnerable and thus, have a higher dependency of Treg-mediated immunoregulation and maintenance of tissue homeostasis. Interestingly, transfer of young Treg diminished the RPE cell morphological changes, while transferred aged Treg failed to do so efficiently (Fig. 3A, B). Similarly, transfer of aged Treg did not limit CD68+ and CD68+MHCII+ cell infiltration in the subretinal space, while transfer of young Treg was highly efficient at doing so (Fig. 3C-G). In line with what was described in the retina, adoptive transfer of aged Treg did partially rescue the accumulation of pro-inflammatory MHCII+ macrophages in the subretinal space although to a lesser extent than young Treg, even though no statistically significant differences were detected between the two adoptive transferred groups (Fig. 3E, F). Therefore, Treg are essential to limit age-related RPE dysmorphology and innate cell accumulation in the subretinal space. However, older age Treg have a limited capacity to control an already established retinal inflammatory environment.

Treg secretome rescues photoreceptor cell death upon inflammation ex vivo

The differential susceptibility of young and aged retina to Treg depletion might be related to a) a higher vulnerability of aged photoreceptors and RPE to systemic pro-inflammatory cues in the absence of Treg or b) differences in the local function of young and aged Treg in the retina. Considering that upon adoptive transfer young but not aged Treg can efficiently restore retinal neurodegeneration, we next investigated what changes in Treg with age.

To do so, we took advantage of two available datasets investigating the transcriptomic changes associated with Treg ageing (Guo et al. 2020 (GSE130419); de la Fuente et al. 2023 (GSE218804)). Guo et al. found 1204 genes differentially expressed between young and aged Treg, with 638 genes upregulated and 566 genes downregulated with age (Fig. 4A), while in our dataset 1758 genes were found differentially expressed (DEG) with 1456 being upregulated and 302 downregulated in aged Treg (Fig. 4B). We focused on those genes that followed common traits for subsequent Gene Ontology (GO) analysis and combined both datasets, identifying 216 genes commonly upregulated (Fig. 4C) and 26 genes commonly downregulated (Fig. 4D). We next investigated the GO Biological processes associated to these genes and found that within the genes upregulated in aged Treg, GO Biological processes associated with inflammation (e. g. inflammatory response, regulation of cytokine production,) and the nervous system (e.g. axon guidance, neuron projection development) were enriched, suggesting that although Treg maintain most immune functions with age [30], aged Treg also acquire some pro-inflammatory phenotypes as shown by the increased expression of IFNγ and IL-1β or increased expression of other T cell effector genes [28, 31, 32] (Fig. 4E). The GO biological processes enriched in those genes that are downregulated with age are associated with sphingolipid metabolism, cell chemotaxis or acetyl choline receptor pathway (Fig. 4F).

Fig. 4

Identification of the biological processes enriched in aged Treg and effect of the secretome in retinal explants. A-B Differentially expressed genes (DEG) in aged Treg when compared to young Treg in Guo et al., 2020 (GSE130419) (A) and de la Fuente et al. 2023 (GSE218804) (B). C, D Venn diagram the genes that are commonly differentially expressed between both datasets. (C) Venn diagram showing the genes commonly upregulated in aged Treg across both datasets and (D) Venn diagram showing the genes downregulated in aged Treg in both datasets. E Quantification showing the fold enrichment of the GO Biological processes enriched amongst the genes upregulated in aged Treg and that are associated with inflammatory processes or nervous system processes. F Quantification showing the GO Biological processes fold enrichment of the genes downregulated in aged Treg. G, H Determination of the effect of aged Treg secretome in retina explants. (G) Representative images of cone-arrestin+ cones (CA+, red) in retinal explants 6h after treatment with LPS, LPS and Treg secretome, control Treg media and Treg secretome in vitro (Scale bar = 50µm) and (H) quantification of the cone survival shown as percentage of the control condition. Data information: E-F n = 4 mice, data presented as mean ± s.e.m. *P < 0.05; 1-way ANOVA followed by Bonferroni's multiple comparisons test

Due to the systemic Treg depletion used in our diphtheria toxin model, it is not possible to determine whether Treg exert a local function within the retina or act indirectly through the maintenance of systemic homeostasis. To build on this and taking into account the enhanced inflammatory and nervous system related genes expressed by aged Treg, we next investigated local Treg effects in a retinal explant model [18]. We exposed retinal explants to Treg secretome in the presence or absence of a pro-inflammatory stimulus in the form of lipopolysaccharide (LPS) and investigated cone survival by staining for CA. LPS is a widely used pro-inflammatory activator of microglia and macrophages [33], which at least in part, will mimic the phenotype described in the aged retina when Treg are depleted. As described previously [34], LPS increased CA+ photoreceptor cell death, which was prevented by treatment with Treg-conditioned media (Fig. 4G, H). However, Treg conditioned media alone did not seem to enhance photoreceptor survival at least ex vivo, suggesting that Treg products limit immune-mediated retinal neurodegeneration. Nevertheless, the identification of the Treg secretome components that mediate this effect in the retina is outside the scope of this current work.