Background
Understanding the conditions that promote or suppress the recruitment of lymphocytes and myeloid cells into central nervous system (CNS) tissue and whether it results in an effector or a protective/regulatory immune response is an active area of investigation with practical consequences for treating inflammatory diseases and neurodegenerative conditions of the CNS. The vast majority of studies concerning immune cell recruitment to the CNS have concentrated on the brain and spinal cord. However, determining the origin, activation/recruitment conditions, and functional output of immune cells, particularly myeloid cells, within or into these tissues can be problematic due to a substantial presence of circulation-derived macrophages in the perivascular spaces and meninges of the brain and spinal cord that are distinct from resident myeloid cells (microglia, MG) in the parenchyma [
1].
The retina is uniquely suited to study CNS myeloid cell maintenance, replacement, and function in that it can be easily removed and cleaned of adjacent tissue, lacks meninges, and is thus a truer representation of neural parenchyma. We have found that recruitment of new myeloid cells to the retina is highly dependent on the stimulus. Using CD11c
DTR/GFP reporter mice [
2], we found an expansion of GFP
hi myeloid cells (GFP
hi MG) within the retina in response to cone photoreceptor cell degeneration [
3] as well as to the limited injury of retinal cells induced by either optic nerve crush, intense light exposure, or partial optic nerve transection [
4,
5]. Subsequent fate mapping, depletion, and parabiosis experiments revealed that the nascent GFP
hi cells found after these injuries were not derived from circulating progenitors but rather were derived from resident MG in either the retina or the optic nerve [
4]. In contrast, we found that ablation of resident myeloid cells and MG from the retina by radiation followed by CD11c
DTR/GFP bone marrow transplant revealed a slow replacement by circulating donor myeloid cells that were primarily GFP
lo [
6]. Furthermore, optic nerve crush injury to these radiation bone marrow chimeras greatly stimulated the replacement of host with donor myeloid cells within the retina and the development of GFP
hi cells from the donor myeloid precursors [
6].
Studies from other laboratories also support the idea that myeloid cell recruitment to the retina is dependent on the stimulus. While the renewal rate of MG in the quiescent retina is unknown, it was initially speculated that they were continually replaced by circulating myeloid progenitors [
7,
8]. However, these studies were done with irradiated bone marrow chimeric mice. It was subsequently demonstrated that under normal, healthy conditions the entry of circulating monocytes into the CNS is highly limited [
9‐
11] and that retinal MG in particular are sustained as a closed, self-renewing population [
12,
13]. Studies involving genetic and chemical depletion of retinal MG showed that replacement MG were either derived solely from residual retinal MG [
14] or from MG in the optic nerve and ciliary body/iris [
15]. Conversely, in certain severe retinal injury models, newly recruited retinal MG were shown to be derived from circulating monocytes [
12,
16‐
18]. Regardless of their origin, the retinal microenvironment compels newly arrived myeloid cells to be the functional equivalent of the endogenous MG [
6,
12], although those derived from circulating monocytes do retain some distinguishing phenotypic signatures [
13].
Although retinal autoimmunity is a fundamentally different process than either the injury or neurodegeneration responses discussed above, determining the origin and role of both the endogenous MG as well as recruited myeloid cells will be crucial to understanding the immunopathogenesis of neuroinflammation. Early studies demonstrated the critical role of myeloid cells in both the induction and resolution of experimental autoimmune uveoretinitis (EAU), a rodent model of retinal autoimmunity [
19‐
21], but could not distinguish the function of resident MG and infiltrating monocytes. More recent studies with immunization-induced EAU suggested that resident MG are crucial for the retinal infiltration of immune cells from the circulation [
22] and that their population expands along the course of the disease [
23]. However, these studies were limited by several factors. First is the reliance on adoptive transfer of activated T cells or self-peptide immunization with strong adjuvants to induce autoimmunity [
24]. This might not be entirely reflective of spontaneous retinal autoimmunity given that the retina lacks lymphatic drainage [
25‐
27], thus the immune cells involved with T cell priming and the effect of the local microenvironment on the priming might not be equivalent. Second is the growing realization that retinal MG are not a homogeneous population, but rather are likely to be composed of distinct subsets [
28,
29]. This is evidenced by our findings of a retinal MG subset, identified as GFP
hi MG in the CD11c
DTR/GFP mouse that, as opposed to GFP
lo MG, expanded in numbers in response to injury and could act as conventional dendritic cells [
5,
30,
31]. Thus, a change in the phenotype and number of retinal MG in response to a stimulus might not represent a change to all MG but rather a change in the balance and function between MG subsets.
In this study we sought to avoid the confounding factors associated with EAU induction by either adoptive transfer of in vitro activated T cells or immunization with adjuvants at remote sites in developing a better understanding of the origin and role of myeloid cells in retinal autoimmunity, particularly those that could be antigen presenting cells (APC), and the resulting T cell response. To do this, we employed R161H transgenic mice [
32‐
34], which spontaneously develop autoimmunity in the retina and uveal tract we term spontaneous autoimmune uveoretinitis (SAU) to distinguish it from the uveoretinitis induced by immunization or adoptive transfer of T cells (EAU), in conjunction with CD11c
DTR/GFP mice. Using R161H
+/− × CD11c
DTR/GFP F
1 mice and parabiosis, we report that SAU was delayed in the F
1 mice relative to normal R161H mice, revealing a distinct prophase of the disease characterized by the expansion of retinal GFP
hi cells prior to observable autoimmunity, and that the induction of SAU strongly correlated with the recruitment of APC (GFP
hi cells) from the circulation into the retina.
Discussion
It has been well established that T cell-mediated autoimmunity in non-immune privileged tissue requires circulating APC. However, in immune privileged sites, particularly CNS tissue, demonstrating a requirement of circulating APC for induction of autoimmunity has been a significant challenge [
49‐
51]. CNS tissues exhibit a wide range of responses to injuries and inflammatory events that can involve the expansion of resident myeloid cells as well as the recruitment of significant numbers of lymphoid and myeloid cells from the circulation. In certain models of neuroinflammation such as experimental autoimmune encephalomyelitis (EAE), a murine model of human multiple sclerosis, circulating APC are thought to be necessary for the induction and progression of the disease [
52‐
55]. Although the retina is a part of the CNS, there is growing evidence that it is unique in its ability to locally control and regulate immune responses compared to other CNS tissue [
6,
56,
57]. The retina lacks meninges, dura mater, subarachnoid space, and choroid plexus, all of which are known to have a substantial population of circulation-derived macrophages [
1], thus making it a truer representation of neural parenchyma. In retinal autoimmune conditions, such as uveoretinitis in humans and EAU in rodents, the blood retinal barrier, the presence of resident macrophages and MG, as well as the difficulty in discriminating between circulating and resident myeloid cells has left in question the origin and nature of the APC that are responsible for initiating immunopathology compared with cells whose presence is simply a consequence of tissue damage. The retina is particularly well suited to study the issue of expansion and/or recruitment of immune cells to CNS tissue since, in its quiescent state, it has a small and easily quantifiable number of immune cells thus allowing changes in response to immunological challenges to be readily detected. Further, the immune cell response within a retina can be visualized and analyzed over the course of the event, especially in mice where the immune cell(s) of interest express a fluorescent label.
In studying retinal immune responses, we have employed a variety of strategies looking for evidence that the presence of specific recruited immune cells versus the expansion of resident immune cells reflects the nature of the challenge and the inherent goal of preserving vision. We have demonstrated there is a subset of retinal MG that dynamically responds to non-autoimmune inflammatory events [
3‐
5]. Further, we and others have shown that, in the absence of severe damage to normal physiological barriers induced by radiation or traumatic injury, retinal MG subsets are replenished by and/or expand from resident myeloid cells within the retina or adjacent CNS tissue [
4,
6,
12‐
15]. Although these and other similar studies represent a significant body of research into the response of retinal innate immune cells, they did not elucidate the nature and origin of APC associated with conditions leading to the onset of retinal autoimmunity.
Several early studies led us to propose that APC recruited from the circulation could be critical for EAU pathogenesis. First was our observation that CD45
+ cells from quiescent murine retina were poor in their ability to present antigen (act as conventional DC) and stimulate an effector response from naive T cells compared to splenic and even brain-derived CD45
+ cells [
56]. Second was our study using radiation bone marrow chimeras between EAU susceptible (Lewis) and non-susceptible (Brown Norway and Buffalo) rat strains showing that EAU induction, following adoptive transfer of activated retinal S-Ag specific T cells, depended on the chimeric rats having circulating monocytes derived from engrafted Lewis bone marrow or splenocytes [
58]. While these results suggested that circulating APC were crucial for EAU induction, we could not discount the possibility that proinflammatory cytokines produced by the activated T cells and/or their intrusion into the retina led to the activation of retinal cells capable of acting as APC. This would remain an issue in studies attempting to determine the origin of APC associated with EAU induction as long as the disease was induced by adoptive transfer of activated T cells or immunization at distal sites.
Our experience with a trackable myeloid cell capable of expansion within or recruitment to the retina that could act as a conventional DC (retinal GFP
hi cells in CD11c
DTR/GFP mice), along with the development of the R161H mouse model of SAU [
32‐
34], allowed us to reexamine whether the induction of retinal autoimmunity was associated with the recruitment of circulating APC to the retina. For several reasons, we found it both useful and necessary to analyze F
1 mice resulting from crossing R161H
+/− (B10.RIII) mice to B6/J mice, a strain permissive but less susceptible for EAU [
24,
59]. First, as most transgenic mice with immune cell markers are on the B6 background, an F
1 was required for use of our CD11c
DTR/GFP mice. Second, as SAU occurs at a young age and progresses rapidly to severe retinal destruction and resolution in R161H
+/− (B10.RIII) mice, we reasoned that R161H heterozygous mice on a partial B6/J background could have delayed onset and slower disease progression, providing the opportunity to study the early events and immunopathogenesis of retinal immune cell responses using parabiosis. Finally, F
1 mice were necessary to ensure MHC compatibility for parabiosis. Although there was a risk of R161H
+/− × B6/J F
1 mice having a low incidence of SAU, our observation that most R161H
+/− × B6/J F
1 eventually developed retinal autoimmunity that was delayed and reduced in severity compared to R161H
+/− (B10.RIII) mice made the F
1 model useful in the present studies. Further, the F
1 model provides a platform unencumbered by the limitations of distal immunization and adoptive T cell transfer for further studies on the origin, role, and response of immune cells in retinal autoimmunity.
Using R161H+/− × B6/J F1 mice, we characterized the development of SAU as having three distinct phases based on immune cell numbers within the retina. Although the prophase, active disease, resolution pattern is likely typical of tissue-specific autoimmune responses, the slow progression of SAU in F1 mice allowed us to closely examine the nature of retinal myeloid cells early in the disease process. In analyzing prophase retinas from R161H+/− × CD11cDTR/GFP F1 mice, we observed a preferential increase in GFPhi myeloid cell numbers compared to GFPlo myeloid cells suggesting an important role for the GFPhi MG subset either in the induction or progression of SAU. We also observed that a significant percentage of the retinal GFPhi cells were also CD45hi indicating they were derived from the circulation. In addition, the more modest increase in GFPlo myeloid cell numbers from prophase retinas was also characterized by those cells also being primarily CD45hi, again suggesting a recruitment of myeloid cells from the circulation. While these results provide circumstantial evidence that circulating APC are likely necessary for induction of SAU or EAU, it could not determine whether circulating GFPhi cells (or precursors thereof) or GFPlo cells were the crucial APC, nor could it eliminate the possibility the signals provided by infiltrating myeloid and lymphoid cells up-regulated CD45 and GFP expression in resident myeloid cells of the retina.
Parabiosis has become a useful tool for studying the origin and contributions of myeloid cells in CNS and peripheral lymphoid tissues under normal and pathological conditions [
1,
9,
55,
60,
61]. For our purposes we deemed it necessary that both the activated, autoreactive T cells (R161H) and the labeled, putative circulating APC (GFP
hi cells) be in one mouse (the donor) while the recipient mouse remain wild-type. While our system is not designed to answer every question about what is needed for autoimmunity it does answer a particular question—is the recruitment of cells from the circulation known to be APC (GFP
hi CD11c
+ cells) associated with retinal autoimmunity. As such, it remains a formal possibility that other circulating hematopoietic cells could also be involved with the induction of SAU. Investigating this would require the use of CD45 congenic markers in combination with our system. However, we believe the limited variability of our system is actually advantageous in that it investigates a specific function of a defined set of cells. It must be noted that our system does not provide a gain of function for the recipient mice but rather simply provides a reporter for circulating CD11c
+ APC that would be present in any mouse.
In analyzing our parabiotic mice, we found that donor age at joining, duration of joining, and the activity of SAU in the donor mouse at time of joining were important factors in the appearance of uveoretinitis in the recipient. Parabiotic mice joined for less than 38 days exhibit a low incidence of disease in recipient mice. Although we did not analyze for chimerism of the circulation in parabionts joined less than 30 days, we believe the low incidence of SAU in these recipients was due to factors associated with its normal course of pathology rather than poor chimerism as full chimerism of circulating hematopoietic cells occurs by 7 to 14 days in parabiotic mice [
36,
62]. Active SAU in the R161H donor mouse led to efficient transfer of the disease to recipient mice. Conversely, R161H donor mice with either sub-clinical or resolving SAU poorly transferred disease to recipient mice. The generation of regulatory T cells (Tregs) is crucial for the control and resolution of EAU [
31,
43‐
45] and we have observed increased regulatory/effector T cell ratios in the retinas of R161H
+/− × B6/J F
1 mice with resolving SAU (unpublished observation). Thus, it is possible that the poor transfer of SAU associated with convalescent donors is due to increased numbers and activity of Tregs from the donor mouse.
However, the most important observation from our experiments with parabiotic mice was that development of SAU in recipient mice was highly associated with the appearance of GFPhi APC in recipient retinas. We found GFPhi cells in the retinas of 19 of 29 recipients with 12 of those developing SAU. Further, we observed that the percentage of recipient retina GFPhi cells that were CD45hi was equivalent to unpaired R161H+/− × CD11cDTR/GFP F1 mice at prophase of SAU. This indicated that there was a faithful replication of the disease in the parabiotic recipients and that the GFPhi cells originated in the circulation. As the onset of clinically or histologically observable SAU is variable, the lack of SAU in any recipient positive for retinal GFPhi cells was likely due to insufficient time between GFPhi cells entering recipient retinas and the time of final analysis. However, we could not discount other factors including Treg activity. Conversely, in recipients that lacked retinal GFPhi cells we did not observe SAU. Since donor and recipient mice were joined long enough to establish chimerism, the circulation within every parabiont contains labeled APC (donor GFPhi cells), the equivalent but unlabeled APC from the recipient, and the R161H autoreactive T cells. As such, and regardless of the disease state of the donor, wild-type recipient parabionts are prime for developing SAU. The successful induction of retinal autoimmunity likely depends on a cascade of signals and events. Although the signal(s) that begin the process of recruiting circulating APC to the retina remain unresolved, our data is clear that a migration of APC from the circulation into the retina appears to be required for the induction of SAU or EAU. Just as important, if the resident retinal microglial APC could support an effector T cell response, we would expect to see pathology in our wild-type parabiotic recipients even in the absence of donor GFPhi cells; this was not observed.
In summary, our results highlight a unique facet of retinal immunology in that circulating, but not resident, APC are likely required for retinal autoimmune pathology. This contrasts with our findings on optic nerve crush injury and cone degeneration associated with RPE65 deficiency showing that recruitment of circulating myeloid cells to the retina is not part of the response to those conditions [
3,
4]. A key difference between these retinal insults is the involvement of T cells. Retinal T cells remained at background levels in our injury and non-inflammatory degeneration models while there was a large influx of T cells into the retinas of R161H
+/− × B6/J F
1 mice even with only minimal levels of clinical or histopathological inflammation. As it has been demonstrated that activated, retinal antigen-specific T cells readily cross the blood retina barrier [
40,
63,
64] we postulate that their arrival in the retina signals for recruitment of circulating myeloid cells capable of acting as APC. In the case of R161H mice the initial T cell activation occurs by recognition of cross-reactive epitopes present on commensal microbiota [
40]. We have studied other retinal-antigen specific T cell receptor transgenic systems and found other T cell stimulating factors such as lymphopenia and Treg depletion were required for T cells activation and subsequent autoimmunity [
31,
65]. Although we have demonstrated there is a subset of retinal MG that can act as conventional DC, it is highly dependent on there being a stimulatory event such as an injury [
5,
30]. In the absence of such an event, infiltrating T cells encounter myeloid cells in an environment that is not conducive for the initiation or progression of autoimmunity. Thus, the induction of retinal autoimmunity likely requires the recruitment of circulating APC which have not been altered by the immunosuppressive environment of the quiescent retina.
Although principally explored in neural tissue, a dynamic between embryonically derived tissue resident macrophages and circulating myeloid cells in homeostasis and disease is emerging in other tissues [
51,
66]. Neuroinflammation and autoimmunity in general are often treated with broad immunosuppressive therapies such as corticosteroids or modulators of T cell activation and effector functions that can interfere with the protective and homeostatic functions of the immune system. It has been proposed that therapies specifically targeting antigen presentation could be effective against autoimmune disorders while limiting the adverse effects associated with broader systemic approaches (
67). Our work suggests that limiting recruitment of circulating APC into tissues could be an effective strategy in limiting autoimmune conditions particularly those involving CNS tissue.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.