Activated microglia in human retinitis pigmentosa, late-onset retinal degeneration, and age-related macular degeneration

Abstract

Many gaps exist in our knowledge of human retinal microglia in health and disease. We address the hypothesis that primary death of rod photoreceptors leads to activation of resident microglia in human retinas with retinitis pigmentosa (RP), late-onset retinal degeneration (L-ORD), or age-related macular degeneration (AMD). Regions of ongoing photoreceptor cell death were studied by immunocytochemistry with microglia- and other retinal cell-specific markers. In normal human retinas, quiescent microglia were small, stellate cells associated with inner retinal blood vessels. In retinas with RP, L-ORD, or AMD, numerous activated microglia were present in the outer nuclear layer in regions of ongoing rod cell death. These microglia were enlarged, amoeboid cells that contained rhodopsin-positive cytoplasmic inclusions. We conclude that activated microglia migrate to the outer nuclear layer and remove rod cell debris. In other central nervous system diseases such as stroke, activated microglia phagocytose debris from the primary injury and also secrete molecules that kill nearby normal neurons. By analogy with these diseases, we suggest that microglia activated by primary rod cell death may kill adjacent photoreceptors. Activated microglia may be a missing link in understanding why initial rod cell death in the human diseases RP, L-ORD, and AMD leads to death of the cones that are critical for high acuity daytime vision.

Introduction

Retinitis pigmentosa (RP) is a heterogeneous family of inherited retinal degenerations. Loss of vision in RP results from primary death of rod photoreceptors, usually due to mutant rod-specific genes (Pierce, 2001), followed by death of the cones (Milam et al., 1998). The earliest visual abnormality in late-onset retinal degeneration (L-ORD), another inherited disease, is loss of night vision (Kuntz et al., 1996, Milam et al., 2000, Jacobson et al., 2001), consistent with a primary defect in the rods. This is followed by loss of all photoreceptors, as demonstrated by histopathologic and visual function studies (Duvall et al., 1986, Brosnahan et al., 1994, Kuntz et al., 1996, Milam et al., 2000, Jacobson et al., 2001).

Microscopic and psychophysical data indicate that rods in the parafovea are also the first photoreceptors to die in normal aging (Curcio et al., 1993) and in age-related macular degeneration (AMD) (Curcio et al., 1996, Curcio et al., 2000, Jackson et al., 2002). An important unanswered question is why loss of rods leads to death of the cones in RP, L-ORD and AMD. The cones are essential for high acuity, daytime color vision, and loss of cone function leads to poor quality of life for RP, L-ORD and AMD patients.

Several mechanisms have been proposed to explain cone cell degeneration, including toxic products released by dying rods (Bird, 1995), lack of contact-mediated interactions with rods, retinal pigment epithelium (RPE) and Müller cells (Adler et al., 1999), or lack of a rod trophic factor essential for cone cell survival (Sahel et al., 2001). Previous studies indicated that photoreceptors in human RP retinas (Li and Milam, 1995) and AMD maculas (Xu et al., 1997, Dunaief et al., 2002) die by apoptosis, as occurs in several murine models of retinal degeneration (Chang et al., 1993).

Photoreceptor apoptosis was also documented in Royal College of Surgeons (RCS) rats (Tso et al., 1994, Papermaster, 1997). The RPE of these rats is defective, lacking Mertk (Vollrath et al., 2001) and unable to phagocytose shed outer segment tips. Human patients with mutant MERTK develop a severe form of RP (Gal et al., 2000). Photoreceptor death in RCS rats was initially attributed to accumulated outer segment debris that physically removes the photoreceptors from their source of nourishment, the choriocapillaris. However, RCS photoreceptors can be rescued by various growth and survival factors (bFGF, CNTF, etc.), even when a thick layer of debris is present (Steinberg, 1994), suggesting that photoreceptor death is not due solely to lack of photoreceptor metabolic support from the choriocapillaris.

A second, quite different mechanism of photoreceptor death was postulated, involving retinal microglia (MG), which are derived from bone marrow precursor cells that enter the central nervous system (CNS) along blood vessels during embryogenesis (Grossmann et al., 2002). When activated by cell death in the brain, MG migrate to the damaged area and phagocytose cellular debris. In addition to phagocytosis of degenerate cells, activated brain MG also secrete molecules that kill normal neurons around the area of primary degeneration, producing a ‘bystander’ or ‘penumbra’ effect (Streit, 2001, Williams and Hickey, 2002). These molecules include nitric oxide, reactive oxygen species, excitatory amino acids, proinflammatory cytokines, proteases, tumor necrosis factor α, and arachidonic acid derivatives (Lee et al., 2002). In brain and retina, MG cells can also express markers of antigen-presenting cells (Penfold et al., 1993, Penfold et al., 1997, Aloisi, 2001, Yang et al., 2002).

In normal retina, quiescent MG are inconspicuous stellate cells that surround inner retinal blood vessels and function as resident macrophages (Provis et al., 1995). Activated retinal MG are enlarged, amoeboid cells that migrate to the outer retina of RCS rats early in photoreceptor degeneration (Thanos et al., 1994, Roque et al., 1996). MG appear to actively kill the photoreceptors because when MG migration to the outer retina is blocked with MG inhibiting factor, the photoreceptors do not die (Thanos et al., 1996). Activated MG also migrate to the subretinal space in light damaged mouse retinas (Ng and Streilein, 2001) and conditioned media from activated retinal MG kill photoreceptors in vitro (Roque et al., 1999).

In concentric RP (Milam et al., 2001), L-ORD (Kuntz et al., 1996, Milam et al., 2000, Jacobson et al., 2001), and the geographic (GA) form of AMD (Sunness, 1999), a sharp leading edge of photoreceptor apoptosis is tailed by degeneration of adjacent photoreceptors. A similar bystander effect in brain involves activated MG-mediated destruction of normal neurons around a primary area of degeneration, as in stroke (Wood, 1995). We used specific MG markers on retinas with RP, L-ORD or AMD to determine if activated MG migrate to regions of primary rod cell death and phagocytose photoreceptor debris. By analogy with activated brain MG, our results suggest that MG cells may be a missing link in understanding why rod cell death leads to demise of adjacent photoreceptors, including the cones. This mechanism may explain the bystander effect in concentric RP and L-ORD and why a leading edge of photoreceptor apoptosis is associated with degeneration of adjacent photoreceptors in the GA form of AMD.