Research Article

Immune modulation by MANF promotes tissue repair and regenerative success in the retina

See allHide authors and affiliations

Science  01 Jul 2016:
Vol. 353, Issue 6294, aaf3646
DOI: 10.1126/science.aaf3646

Structured Abstract


Regenerative therapies based on cell replacement hold promise for the treatment of a range of age-related degenerative diseases but are limited by unfavorable microenvironments in degenerating tissues. A promising strategy to improve success is to harness endogenous repair mechanisms that promote tissue integrity and function. Innate immune cells are central to such repair mechanisms because they coordinate local and systemic responses to tissue injury by secreting inflammatory and anti-inflammatory signals in a context-dependent manner. A proper balance between these opposing phenotypes of innate immune cells is essential for efficient tissue repair, and immune modulation may be an effective way to promote repair and enhance regenerative therapies. Here, we identified a new evolutionarily conserved immune modulatory function for mesencephalic astrocyte-derived neurotrophic factor (MANF) that biases immune cells toward an anti-inflammatory phenotype, thereby promoting tissue repair in both vertebrates and invertebrates and enhancing retinal regenerative therapy.


In Drosophila, interactions between damaged tissues and hemocytes are essential for tissue repair. We used this model to identify immune cell–derived factors with immune modulatory activity that promote tissue repair after retinal injury. The identification of MANF as such a factor prompted us to test its role in mammalian retinal repair and ask whether its immune modulatory activity helped cell replacement therapies in degenerating retinas.


Using a combination of transcriptome analysis and genetic studies, we identified MANF as a hemocyte-derived factor that is induced by platelet-derived growth factor (PDGF)– and vascular endothelial growth factor (VEGF)–related factor 1 (Pvf-1)/PDGF- and VEGF-receptor related (PvR) signaling. MANF was necessary and sufficient to promote retinal repair after ultraviolet-light–induced retinal injury in Drosophila. MANF also had an autocrine immune-modulatory function in fly hemocytes, which was necessary for its tissue repair–promoting activity. This regulation and function of MANF was evolutionarily conserved: Mouse photoreceptors expressed PDGF-A (a Pvf-1 homolog) in response to damage signals, which promoted MANF expression in innate immune cells. This PDGF-A/MANF signaling cascade was required to limit photoreceptor apoptosis in the retina. Exogenously supplied recombinant MANF protected photoreceptors in several paradigms of retinal injury and degeneration. As in flies, this prorepair function was associated with alternative activation of macrophages and microglia in the retina. Ablation of CD11b+ immune cells and deletion of Cx3Cr1, a chemokine receptor required for MANF-induced alternative activation, prevented MANF-induced repair. Thus, the protective effects of MANF in retinal injury rely on its immune modulatory activity. Finally, MANF supplementation to photoreceptors transplanted into congenitally blind mice increased integration efficiency and accelerated and improved visual function recovery.


Combining genetic studies in invertebrates and vertebrates has rapidly identified factors with promising therapeutic potential. Immune modulation is a promising strategy to optimize regenerative therapies. With its conserved immune modulatory function, MANF is a particularly promising molecule that is likely to be useful for the treatment of inflammatory conditions in many different disease contexts.

MANF in retinal repair.

In Drosophila (left) or mouse (right), the damaged retina secretes Pvf-1/PDGF-A, which acts on innate immune cells. MANF derived from innate immune cells (and other sources) promotes phenotypic changes in immune cells as part of a mechanism required for tissue repair. Therapeutically, MANF supplementation can delay retinal degeneration and improve the success of cell-replacement regenerative therapies in the retina.


Regenerative therapies are limited by unfavorable environments in aging and diseased tissues. A promising strategy to improve success is to balance inflammatory and anti-inflammatory signals and enhance endogenous tissue repair mechanisms. Here, we identified a conserved immune modulatory mechanism that governs the interaction between damaged retinal cells and immune cells to promote tissue repair. In damaged retina of flies and mice, platelet-derived growth factor (PDGF)–like signaling induced mesencephalic astrocyte-derived neurotrophic factor (MANF) in innate immune cells. MANF promoted alternative activation of innate immune cells, enhanced neuroprotection and tissue repair, and improved the success of photoreceptor replacement therapies. Thus, immune modulation is required during tissue repair and regeneration. This approach may improve the efficacy of stem-cell–based regenerative therapies.

Regenerative therapies based on cell replacement hold promise for the treatment of a range of age-related degenerative diseases (1, 2). Moreover, aged and diseased tissues provide a poor microenvironment for integration (3). A case in point is attempting to regenerate the vertebrate retina, a tissue where endogenous repair mechanisms are inefficient and that is subject to a variety of irreversible age-related degenerative pathologies. Human pluripotent stem cells can provide a virtually unlimited source of photoreceptors and retinal pigment epithelial (RPE) cells for replacement and restoration of vision (4), yet the poor integration efficiency of transplanted cells into the host retina has limited clinical applications. Retinal diseases targeted by this therapeutic approach, such as age-related macular degeneration or retinitis pigmentosa, are characterized by microglial activation and proinflammatory microenvironments (59) that will negatively affect integration and repair (3, 10).

Microglia, monocyte-derived macrophages, and other innate immune cell types can both promote and resolve inflammation. Managing these inflammatory responses is essential for tissue repair and regeneration (11). In the central nervous system, resident (microglia) and invading innate immune cells orchestrate a complex response to damage aimed at restoring tissue integrity but can also promote damaging neuroinflammation (1215). This antagonism is at least in part a consequence of different states of immune cell activation. Classical or M1 activation is associated with proinflammatory conditions that can cause tissue damage, whereas alternative or M2 activation is associated with resolution of inflammation and tissue repair (16, 17). This M1/M2 paradigm has been used to describe outcomes of in vitro perturbation of macrophages, yet there is evidence that macrophages in vivo can adopt similar phenotypes and functions (18, 19). Because of these opposing effects of different immune cell phenotypes, immune modulation rather than immune suppression may be an effective way to promote tissue repair and improve regenerative therapies.

Studies in Drosophila have substantially advanced our understanding of tissue repair and regeneration in metazoans (2022). This work has highlighted the critical role of the interaction between hemocytes (Drosophila blood cells with macrophage-like activities) and damaged epithelia in the repair process. Hemocytes are activated in response to tissue damage and coordinate localized and systemic repair responses (2326) but have also been implicated in inflammatory processes in flies (27). A productive model for the genetic dissection of tissue and hemocyte interactions in repair processes is the pupal retina, which responds to ultraviolet (UV) damage by inducing photoreceptor apoptosis in a dose-dependent manner (28, 29). A paracrine interaction between UV-damaged photoreceptors and hemocytes through the platelet-derived growth factor (PDGF)– and vascular endothelial growth factor (VEGF)–related factor 1 (Pvf-1) and PDGF- and VEGF-receptor related (PvR) pathway governs repair of the damaged retina: Damaged photoreceptors secrete Pvf-1 and activate PvR in hemocytes, promoting repair of UV-induced tissue damage (Fig. 1A) (26).

Fig. 1 MANF is a hemocyte-derived damage response factor and promotes retinal repair in Drosophila.

(A) Experimental design and current model for hemocyte-mediated retinal repair in Drosophila. (B) (Left) Representative image of hemocyte smears from third-instar larvae (HmlΔ::Gal4; UAS::GFP) detecting MANF (red) in Hml>GFP+ cells. GFP, green; 4′,6-diamidino-2-phenylindole (DAPI), blue. Scale bar, 5 μm. (Right) Western blot analysis of MANF and GFP proteins in cellular and plasma fractions from hemolymph of third-instar larvae (HmlΔ::Gal4; UAS::GFP). (C) Relative mRNA levels of MANF detected by RT-qPCR in hemocyte samples collected from third-instar larvae of the designated genotypes and treatments (n ≥ 5 for all conditions). For UV treatments, larvae were exposed to 50 mJ of UV at second-instar stage and hemocytes collected 24 hours after. (D) (Left) Relative mRNA levels of MANF detected by RT-qPCR in hemocyte samples collected from third-instar larvae overexpressing Pvf-1 in the retina (n ≥ 5 for all conditions). (Right) Western blot analysis of MANF (intracellular in hemocytes and secreted into the hemolymph) and actin (intracellular in hemocytes) proteins in whole hemolymph collected from third-instar larvae overexpressing Pvf-1 in the retina. (Bottom) Average relative levels of MANF in whole-hemolymph samples normalized to actin. (E and F) (Left) Representative images of adult eye phenotypes from flies with the designated genotypes, after exposure of the right eye of P24 pupae to 17.5 mJ of UV light. Right, average relative size of the UV-treated eye when compared to the untreated eye of the same fly (6 < n < 17 for each genotype; each dot represents one fly). For all quantifications, error bars represent SEM and P values are from Student’s t test.

We performed RNA sequencing (RNA-seq) on isolated hemocytes to identify PvR-dependent genes encoding secreted proteins that were induced after epithelial damage (fig. S1 and table S1). Mesencephalic astrocyte-derived neurotrophic factor (MANF) was found in this screen and, based on its evolutionarily conserved neurotrophic activity (3032), we decided to explore its potential as a retinal repair factor.

Hemocyte-derived MANF is activated downstream of Pvf-1/PvR paracrine signaling to promote retinal repair in Drosophila

We confirmed that MANF is expressed in fly innate immune cells (hemocytes) using immunohistochemistry of hemolymph smears from late second-instar larvae (Fig. 1B, left). In these smears, hemocytes were identified by green fluorescent protein (GFP) expression driven by the hemocyte-specific driver hemolectin:Gal4 (HmlΔ:Gal4) (33). MANF was also detected by immunoblot in the plasma fraction of the hemolymph, confirming its secretion (Fig. 1B, right). Consistent with the RNA-seq data, reverse transcription and real-time quantitative polymerase chain reaction (RT-qPCR) analysis revealed that MANF mRNA levels were significantly higher in hemocytes from UV-treated larvae compared with untreated controls (Fig. 1C, left) and that this induction was PvR dependent (Fig. 1C, right, and figs. S1C and S2A). Overexpression of Pvf-1 in the retina [using GMR:Gal4 (glass multimer reporter) (34) as a driver] was sufficient to induce MANF mRNA specifically in hemocytes, in the absence of damage (Fig. 1D, left), and was accompanied by a significant increase in MANF protein in the hemolymph (Fig. 1D, right, and fig. S2B).

Flies overexpressing MANF in hemocytes (fig. S2C, left) showed significant tissue preservation after UV exposure, even after PvR knockdown in hemocytes (26) (Fig. 1E, left and middle), without affecting PvRRNAi knockdown efficiency (fig. S2A). This protective activity of hemocyte-derived MANF was further confirmed in two genetic models of retinal damage, in which degeneration is induced by retinal (GMR driven) overexpression of the pro-apoptotic gene grim or of mutant Rhodopsin (Rh1G69D) (35, 36) (fig. S2, D and E).

Null mutations in the manf gene (manfmut96 and manfmut112) (31) are homozygous lethal at early first-instar larval stages, yet MANF heterozygotes [which express significantly lower levels of MANF in hemocytes compared with wild type (fig. S2F)] had a significantly increased tissue degeneration response to UV (Fig. 1F, gray dots, and fig. S2G). This increase in tissue loss could be rescued by MANF overexpression in hemocytes (Fig. 1F, black dots) and was recapitulated by hemocyte-specific knockdown of MANF (Fig. 1E, right, and fig. S2C, right).

MANF has immune modulatory properties that are required for retinal repair in Drosophila

The protective effect of hemocyte-derived MANF could be caused by direct neuroprotective activity of MANF on retinal cells or could reflect an indirect effect of MANF on the microenvironment of the damaged retina. To distinguish between these possibilities, we investigated whether MANF could influence hemocyte phenotypes. Hemocytes can acquire lamellocyte phenotypes, characterized by down-regulation of plasmatocyte markers (hemolectin and hemese) and expression of Atilla protein (37), during sterile wound healing (38). These phenotypes correlate with hemocyte activation and may influence tissue repair capabilities, and we recapitulated them in our UV damage paradigm (Fig. 2A). Overexpression of MANF in hemocytes in vivo or treatment of hemocytes in culture with human recombinant MANF (hrMANF) significantly increased the proportion of lamellocytes in hemocyte smears, as detected by Atilla expression (Fig. 2A). This correlated with a decrease in the proportion of cells expressing GFP driven by HmlΔ:Gal4 and a decrease in hml transcripts (fig. S3A). Furthermore, MANF was necessary and sufficient to induce the Drosophila homolog of the mammalian M2 marker arginase1 (arg) (39) in hemocytes (Fig. 2B and fig. S3B), suggesting that these cells may be able to acquire phenotypes similar to alternative activation (16, 17). Most MANF-expressing hemocytes also expressed Arg, suggesting that there is an association between MANF expression and M2-like activation of hemocytes.

Fig. 2 MANF-dependent hemocyte activation is required for neuroprotection in Drosophila.

(A and C) Representative IHC images of hemocyte smears from third-instar larvae of the designated genotypes and treatments, detecting Atilla protein in red. Hml+ cells are identified by GFP expression, green; DAPI, blue. Scale bar, 5 μm. For UV treatments, larvae were exposed to 50 mJ of UV at the second-instar stage, and hemocytes were collected 24 hours later. In (A), all analyses were performed after 24-hour culture in control media [wild-type (WT), UV 50 mJ, and UAS:MANF] or media supplemented with hrMANF protein. In (C), hemocytes were assayed directly after collection and were not cultured (images and left graph) or assayed, as in (A) (right graphs). Percentage of Atilla+ cells in the hemocyte population collected from third-instar larvae of the designated genotypes and treatments is shown (n ≥ 3 for each genotype/treatment). (B and D) Relative mRNA levels of Arg detected by RT-qPCR in hemocyte samples collected from third-instar larvae of the designated genotypes (n ≥ 3 for all conditions). (E) Representative images of adult eyes from flies with the designated genotypes, after exposure of the right eye of P24 pupae to 17.5 mJ of UV light. Right, average relative size of the UV-treated eye when compared to the untreated eye of the same fly (5 < n < 20 for each genotype; each dot represents one fly). For all quantifications, error bars represent SEM and P values are from Student’s t test. RNAi(35) and RNAi(36) correspond to two independent double-stranded RNA interference–expressing lines targeting KdelR transcripts.

To determine whether MANF’s immune modulatory function is required for retinal repair, we assessed retinal tissue preservation in conditions in which hemocytes express and secrete high levels of MANF but are unable to be activated in response to this signal. We generated such a condition by overexpressing MANF in the absence of Kdel receptors (KdelRs). In human cells, KdelRs modulate MANF secretion and cell surface binding. Intracellular KdelR prevents MANF secretion, whereas cell-surface-bound KdelR promotes binding of extracellular MANF (40). Knockdown of the one Drosophila KdelR homolog (41) in hemocytes resulted in a significant induction of MANF transcripts and the detection of MANF protein in the hemolymph (fig. S3, C and D), suggesting that KdelR-depleted hemocytes secrete high levels of MANF. In these hemocytes, MANF-induced lamellocyte formation and Arg expression were significantly decreased (Fig. 2, C and D). Hemocyte activation by extracellular MANF is thus impaired after KdelR knockdown. This genetic perturbation also resulted in a significant enhancement of UV-induced tissue loss, which could not be rescued by MANF overexpression (Fig. 2E). Thus, immune modulation by MANF is critical for tissue repair.

Damage response–associated PDGF-A/MANF paracrine signaling is conserved in mammals

MANF is an evolutionarily conserved protein (31), and we sought to explore its regulation and its potential to allay retinal degeneration and improve retinal repair in vertebrates. We used focal exposure of the central retina of C57BL/6 mice to 8000 lux of bright light for 1.5 hours to induce a retinal innate immune response without generalized photoreceptor apoptosis [C57BL/6 mice carry a protective variant of the Rpe65 gene, preventing excessive retinal damage in response to light (42)]. This protocol resulted in a moderate and transient increase in the presence of innate immune cells in the retina (Fig. 3 and fig. S4, A and B).

Fig. 3 PDGF-A/MANF damage–associated paracrine signaling is conserved in mammals.

(A) Cellular layers in the mouse eye. (B) to (H) are from C57BL/6 mice. (B and D) IHC showing expression of PDGF-A, CD11b, and MANF after light exposure or in controls. See also fig. S4, A and B. (C) Retinal mRNA levels of MANF (RT-qPCR) relative to controls (n = 3). (E and F) IHC showing expression of CD11b [(E) and (F)] and MANF (F), 1 day after intravitreal injection of mrPDGF-AA or vehicle (PBS). Details in (F) highlight CD11b+ cells detected in the vitreous (right) and choroid (left) blood vessels and MANF coexpression. (E) Average number of CD11b+ cells in the vitreous (mrPDGF-AA, n = 5; PBS, n = 6; three sections per eye for each animal; each dot represents one animal). (G and H) TUNEL staining, 2 days after light exposure. (G) After intravitreal injection of antibody to PDGFRα or vehicle [goat immunoglobulin G (IgG)]. (H) In Manf+/− and Manf+/+ littermates. Average number of TUNEL+ nuclei is quantified [(G) No light: antibody to PDGFRα, n = 5; IgG, n = 5; light exposure: antibody to PDGFRα, n = 6; IgG, n = 5. (H) No light: Manf+/+, n = 5; Manf+/−, n = 3; light exposure: Manf+/+, n = 5; Manf+/−, n = 5. Twelve sections per eye for each animal; each dot represents one animal). (I) Retina of BALB/cJ mice, stained with TUNEL, 2 days after intravitreal injection of hrMANF or vehicle (PBS) and exposure to 5000 lux of bright light for 1 hour. Average number of TUNEL+ nuclei per retinal field is shown (hrMANF, n = 8; PBS, n = 8; each dot represents one retinal field). (J) Retina of P28 Crxtvrm65 mice, stained with DAPI, 14 days after intravitreal injection of hfib-MANF or hfib-Cntrl. Red dashed lines indicate the thickness of the ONL after hfib-MANF delivery for comparison. Quantification of photoreceptor preservation as a percentage of nuclei rows in ONL relative to untreated controls (hfib-MANF, n = 8; hfib-Cntrl, n = 8, five sections per eye, untreated controls for relative quantifications; n = 4, five sections per eye; each dot represents one animal). For all quantifications, error bars represent SEM and P values are from Student’s t test. Scale bars, 20 μm.

PDGF-family and VEGF-family proteins are the mammalian homologs of Drosophila Pvf-type ligands (43, 44), and we detected PDGF-A–expressing cells in the neural retina 6 hours after light exposure (Fig. 3B). The induction of PDGF-A was followed by a significant increase in MANF transcripts (Fig. 3C) and the detection of MANF+ innate immune cells, identified by CD11b expression (45, 46), in the vitreous (Fig. 3D and fig. S4A, 12 hours). Resting microglia, localized to the plexiform layers [inner plexiform layer (IPL) and outer plexiform layer (OPL)] in control retinas, did not express MANF (Fig. 3D, no light exposure). Thirty-six hours later, MANF+ innate immune cells were found within the outer nuclear layer (ONL) (Fig. 3D and fig. S4A, 36 hours). This innate immune cell activation and/or recruitment was also accompanied by a redistribution of MANF protein from the cell bodies of Müller glia [where it is detected in control conditions in the inner nuclear layer (INL)] to glial processes (identified by staining against glial fibrillary acidic protein) (fig. S4C). Microglia and/or macrophages recruited and/or activated after light exposure expressed reduced levels of MANF when PDGF signaling was inhibited using neutralizing antibodies against PDGFRα (47) (fig. S4, D and E). Conversely, intravitreal injection of mouse recombinant PDGF-AA (mrPDGF-AA) significantly increased CD11b+ innate immune cells in eyes in the absence of light exposure (Fig. 3E). These CD11b+ cells also expressed MANF and were found in the vitreous (Fig. 3F, right, arrowheads) and the choroidal blood vessels (Fig. 3F, left, arrowheads).

Reduction of PDGFRα signaling or MANF levels [in heterozygotes for a null allele (48)] significantly enhanced photoreceptor apoptosis [detected by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL)] in response to light exposure (Fig. 3, G and H). Homozygotes for this MANF allele are embryonic or perinatal lethal (48). Reduction of MANF expression in bone marrow–derived macrophages (BMDMs) of these mice was confirmed by RT-qPCR (fig. S4F).

Damage signals from retinal cells thus engage a conserved retinal repair response in both flies and mice that involves the Pvf/PDGF-mediated recruitment/activation of MANF-expressing innate immune cells and that is essential to prevent excessive apoptosis in response to light.

MANF has a conserved neuroprotective function in the mammalian retina

To determine whether MANF protein supplementation would be sufficient to ameliorate retinal degeneration, we used the light-sensitive BALB/cJ strain, which lacks the protective variant of the Rpe65 allele, rendering them susceptible to light-induced retinal damage (42). This phenotype is accompanied by activation of proinflammatory microglia and by chemokine production that modulates photoreceptor degeneration (49). Exposure of these mice to 5000 lux of bright light for 1 hour resulted in photoreceptor apoptosis (fig. S4G). In addition, we used two genetic models of retinal degeneration (Crxtvrm65, a slow model of genetically induced retinal degeneration, and Pde6bRd1, a fast model of genetically induced retinal degeneration), whose dynamics of photoreceptor loss have been previously described (5052).

We injected hrMANF protein or vehicle [phosphate-buffered saline (PBS)] into the vitreous immediately before light exposure or at the onset of retinal degeneration [postnatal day 14 (P14) for Crxtvrm65 mice and P7 for Pde6bRd1 mice] and evaluated photoreceptor apoptosis by TUNEL. MANF injection significantly reduced apoptosis in all three models of retinal degeneration (Fig. 3I and fig. S5, A and B). As photoreceptors degenerate, the number of nuclei in the ONL is reduced, and in Crxtvrm65 mice there are on average five to six rows left at P21. In hrMANF-treated eyes, there was a significant preservation of photoreceptors in the ONL (fig. S5C), suggesting that inhibition of apoptosis effectively slows retinal degeneration in this model. Similar results were observed in the Pde6bRd1 mouse model analyzed 5 days after intravitreal delivery of hrMANF (fig. S5, D and E).

Finally, we asked whether a persistent source of MANF could further delay retinal degeneration in Crxtvrm65 mice. We infected human fibroblasts (hfib) with a lentivirus driving the expression of a functional MANF-GFP fusion protein (40). MANF-GFP expression could readily be detected in these fibroblasts and in the media supernatant, confirming that the fusion protein was efficiently secreted (fig. S5, F and G). When MANF-secreting fibroblasts were injected into the vitreous of P14 Crxtvrm65 mice, their retinas degenerated more slowly than control fibroblast-injected retinas, and a significant amount of photoreceptors were preserved in the ONL (Fig. 3J). Survival of injected fibroblasts was confirmed at the time of analysis (2 weeks after injection) by detecting the presence of GFP-expressing cellular aggregates within the vitreous.

MANF can thus prevent photoreceptor apoptosis broadly and delay retinal degeneration, independently of the damaging stimulus.

MANF-dependent modulation of immune cell phenotypes mediates retinal protection

After intravitreal injection of MANF-secreting fibroblasts, MANF+ innate immune cells (CD11b+) could be detected in the vitreous of Crxtvrm65 mice (Fig. 4A, top panel). These CD11b+ cells with round morphology also expressed markers of alternative activation (16, 17) (Fig. 4A, bottom panel, and Fig. 4B, left; fibroblasts injected into the vitreous were detected by GFP expression at the time of dissection and were completely removed along with the lens). Intravitreal delivery of hrMANF had similar effects on innate immune cell phenotypes in Crxtvrm65 (Fig. 4B, right) and light-damaged retinas (Fig. 4C and fig. S6A), supporting an immune modulatory function for MANF. Accordingly, the recruitment of MANF+ innate immune cells in response to PDGF-AA treatment (Fig. 3, E and F) was accompanied by a significant increase in the number of CD11b+ cells colabeled with Ym1+ (Fig. 4D and fig. S6B).

Fig. 4 MANF-dependent immune modulation mediates retinal neuroprotection.

(A) IHC showing expression of CD11b, MANF, and Ym1 in P28 Crxtvrm65 mice 14 days after intravitreal injection of hfib-MANF or hfib-Cntrl. Arrowheads indicate coexpression. (B) Average number of Ym1+ or Arg1+ cells, per eye cryosection, in P28 or P21 Crxtvrm65 mice, 14 or 7 days after intravitreal injection of hfib (hfib-MANF, n = 6; hfib-Cntrl, n = 6; five sections per eye; each dot represents one animal) or recombinant protein (hrMANF, n = 6; PBS, n = 6; each dot represents one section). (C) (Left) Average number of Arg1+ cells per eye cryosection, in BALB/cJ mice, 2 days after intravitreal injection of hrMANF or vehicle (PBS) and light exposure (hrMANF, n = 7; PBS, n = 7; three sections per eye; each dot represents one animal). (Right) Percentage of CD11b+/MANF+ cells in the retina of BALB/cJ mice after the same treatment (hrMANF, n = 13 sections; PBS, n = 5 sections; each dot represents one section). See also fig. S6A. (D) Average number of Ym1+ cells, per eye cryosection, in C57BL/6 mice, 1 day after intravitreal injection of mrPDGF-AA or vehicle (mrPDGF-AA, n = 5; PBS, n = 6; three sections per eye; each dot represents one animal). See also fig. S6B. (E, F, and H) Relative mRNA levels (RT-qPCR) in BMDMs from WT [(E), n = 3] or Cx3Cr1-deficient [(H), n = 3] mice, stimulated with hrMANF or vehicle (PBS) or raw macrophages transfected with MANF targeting siRNA pool or a nontargeting siRNA pool [(F), n = 5]. See also fig. S9. (G and I) TUNEL staining, 2 days after intravitreal injection of hrMANF or vehicle (PBS) and light exposure of CD11b:DTR (G) or Cx3Cr1tg(YFP-CRE-ER) (Cx3Cr1−/−) mice (I). Average number of TUNEL+ nuclei is shown [(G) No light: PBS, n = 3; DT, n = 4. Light: PBS, n = 3; DT, n = 4; DT+hrMANF, n = 5; four sections per eye. (I) No light: Cx3Cr1+/−, n = 4; Cx3Cr1−/−, n = 8. Light: Cx3Cr1+/−, n = 6; Cx3Cr1–/–: PBS, n = 7; hrMANF, n = 6; 12 sections per eye; each dot represents one animal). For all quantifications, error bars represent SEM and P values are from Student’s t test. Scale bars, 20 μm.

The innate immune cell population recruited after MANF delivery was mostly composed of monocytes and monocyte-derived macrophages (60 to 80%, identified by F4/80 or CD68 expression). Ly6-G+ (Gr-1high) neutrophils represented about 15% of the population. The majority of both macrophages and neutrophils (80%) expressed MANF and Arg1, suggesting that MANF expression is associated with markers of alternative activation (fig. S6C), similar to what we observed in fly hemocytes (fig. S3B).

In vitro stimulation with hrMANF for 3 hours was also sufficient to induce markers of alternative activation (Arg1 and Ym1) (16, 17) and Il-13, an anti-inflammatory cytokine (53), in BMDMs (Fig. 4E) and in a macrophage cell line [RAW264.7 (54)] (fig. S7, A and B). Silencing of MANF with a targeting small interfering RNA (siRNA) pool in this cell line resulted in the repression of the same set of genes (Fig. 4F). This suggests that MANF has a direct immune modulatory function in macrophages and that at least part of the mechanism is autocrine.

To determine whether immune modulation by MANF is required for its neuroprotective activity, we assessed retinal damage after light exposure and after MANF supplementation in mice with impaired immune cell function. We depleted macrophages and microglia using diphtheria toxin (DT) administration in CD11b:DTR mice (55, 56). DT, but not sham (PBS), injection resulted in a significant reduction in the number of innate immune cells in the retina (fig. S7C) and induction of photoreceptor apoptosis in response to light exposure (Fig. 4G). Intravitreal supplementation of hrMANF protein did not significantly reduce photoreceptor apoptosis in these mice (Fig. 4G), supporting an essential role for immune cells in mediating the protective effects of MANF.

We further used mice deficient in Cx3Cr1 (57, 58) to test the requirement of immune modulation for the protective effects of MANF. Here, we aimed at generating a condition in which immune cells were present but failed to induce alternative activation in response to MANF signaling, similar to KdelR deficiency in flies. Cx3Cr1 is a chemokine receptor expressed in different immune cell populations, including retinal microglia and peripheral monocytes (58). High Cx3Cr1 expression has been associated with a functionally distinct class of monocytes with immune patrolling activity and with a molecular profile of macrophage differentiation resembling alternative activation (18, 19). Loss of Cx3Cr1 results in retinal degeneration in response to several stimuli and is associated with proinflammatory activation of immune cells (9, 14, 59). Thus, we hypothesized that loss of Cx3Cr1 could be an effective way to impair MANF-induced alternative activation. Indeed, BMDMs derived from Cx3Cr1-deficient mice failed to induce genes associated with alternative activation upon MANF stimulation (Fig. 4H), despite expressing normal levels of MANF (fig. S7, D and E). Light-induced photoreceptor apoptosis in Cx3Cr1-deficient mice could not be rescued by intravitreal delivery of hrMANF (Fig. 4I), suggesting that it is not only MANF derived from macrophages that mediates the protective effects but rather a more complex mechanism that depends on MANF immune modulatory activity. We cannot exclude, however, that Cx3Cr1 deficiency may also result in other alterations that contribute to the loss of protective effects of MANF observed in these conditions, which may be independent of macrophage functions.

MANF promotes cell integration and restoration of visual function in the mammalian retina

Retinal repair by transplantation of mouse and human photoreceptor precursors can restore vision in mouse models of retinal degeneration (60, 61). Integration efficiency depends on the ontogenetic stage of donor cells (60) and on the status of the degenerative microenvironment (62) and negatively correlates with the presence of classically activated macrophages within the retinal tissue (63). We injected photoreceptors derived from Nrl-GFP mice subretinally into wild-type retinas and found that microglia and/or macrophages located at sites of integration expressed MANF, suggesting a possible role for MANF-mediated immune modulation in promoting integration (Fig. 5, A to C). Supporting this hypothesis, integration efficiency was significantly reduced in Cx3Cr1 mice (Fig. 5D).

Fig. 5 MANF enhances the efficiency of retinal regenerative therapies.

(A) Cartoon representing the transcorneal subretinal injection method. (B) IHC showing expression of CD11b, MANF, and GFP at an integration site of Nrl-GFP donor photoreceptors 1 week after transplantation. (C) Average number of MANF+CD11b+ cells/field in integration sites versus sites of no integration (10 fields per condition; all fields contained cells in the subretinal space; each dot represents one field). (D) Quantification of integration into WT (n = 8) or Cx3Cr1−/− (n = 6) mice, analyzed by IHC for GFP expression, 7 days after subretinal injection of P7 Nrl-GFP donor photoreceptors (PhR). Each dot represents one animal. (E) Quantification of integration in C57BL/6 mice, analyzed by IHC for GFP expression, 7 days after subretinal injection of Nrl-GFP donor photoreceptors (PhR) supplemented with hrMANF protein (n = 10, P14; n = 7, P21) or vehicle (PBS) (n = 8, P7 and P14; n = 9, P21). Each dot represents one animal. See also fig. S8A for representative images of P21 transplants. (F) Representative images and quantification of integration in WT (n = 8, same as in Fig. 5D) or Crxtvrm65 mice, analyzed by IHC for GFP expression, 7 days after subretinal injection of P7 Nrl-GFP donor photoreceptors (PhR) supplemented with hrMANF protein (hrMANF, n = 4) or vehicle (PBS, n = 4). Each dot represents one animal. (G) Examples of ERG waves obtained in MANF-supplemented (blue) and PBS-supplemented (black) transplants of P7 Nrl-GFP PhRs in Crxtvrm65 mice. (H) Maximal b-wave amplitudes measured 1 to 4 weeks after subretinal injections of P7 Nrl-GFP PhRs supplemented with MANF (n = 3 to 7 at each time point), PBS (n = 4 to 6 at each time point) and of eyes that did not receive a transplant (n = 6 to 10), all in Crxtvrm65 host. Each dot represents one animal. See also fig. S8B for b waves after hrMANF or PBS injection without cells and fig. S8C for b waves of WT eyes. P values are from a two-way ANOVA analysis. For all quantifications, error bars represent SEM. P values in (C) to (F) are from Student’s t test. Scale bars, 20 μm.

To further test this hypothesis, we asked whether MANF supplementation would increase integration of subretinal-delivered photoreceptors derived from Nrl-GFP mice into a wild-type host. Integration efficiency declines with increased maturity of injected photoreceptors (60, 64). Accordingly, we observed a strong decline in integration efficiency (assessed 1 week after injection) when using P21 rather than P7 or P14 photoreceptors in a wild-type host (Fig. 5E). hrMANF supplementation rescues this decline while having no effect on P14 cells (Fig. 5E and fig. S8A), suggesting that MANF may act either directly on refractory photoreceptors to improve their integration capabilities or indirectly by inducing a more supportive environment for such cells.

To distinguish between these possibilities, we tested the effects of MANF on integration efficiency in degenerating retinas. The inflammatory microenvironment in degenerating retinas is a likely cause for poor integration efficiency (62) and thus a critical limitation in clinical settings. We used Crxtvrm65 retinas to model a degenerating environment and found that integration of even young (P7) Nrl-GFP photoreceptors, which efficiently integrated into wild-type retinas (Fig. 5E), was significantly reduced in Crxtvrm65 retinas (Fig. 5F). MANF supplementation significantly improved integration in this context (Fig. 5F), supporting the notion that MANF improves the environment for integration even in a disease context.

Importantly, MANF accelerated and improved restoration of visual function, as evaluated by maximal b-wave amplitudes measured in sequential electroretinogram (ERG) testing over the course of 4 weeks (Fig. 5, G and H). Eyes that received MANF-supplemented transplants showed signs of light responsiveness based on a detectable b-wave as early as 1 week after transplantation, whereas eyes that received control transplants had the earliest detectable b-wave only at 3 weeks. Comparing ERG b-wave amplitudes of untreated Crxtvrm65 mice (no transplant) to treated mice at 1 to 4 weeks confirmed a functional improvement in vision in the MANF-supplemented cohort only, whereas the PBS-supplemented group did not significantly differ from untreated controls. The ERG changes reflected cell integration and not an effect of MANF supplementation alone (fig. S8B) and represented a recovery of about 60% of visual function when compared with normal ERG b-wave amplitudes of wild-type mice (Fig. 5H and fig. S8C). This is a significant improvement over nonsupplemented transplants, which yield about 20% of visual function recovery (Fig. 5H and fig. S8C) [see also (61)].


Our results identify MANF as an evolutionarily conserved immune modulator that plays a critical role in the regulatory network mediating tissue repair in the retina (Fig. 6A). The ability of MANF to increase regenerative success in the mouse retina highlights the promise of modulating the immune environment as a strategy to improve regenerative therapies (Fig. 6B).

Fig. 6 Model for the evolutionarily conserved immune modulatory function of MANF and its implication in tissue repair and regeneration.

(A) In Drosophila (left) or mouse (right), the damaged retina secretes Pvf-1/PDGF-A, which acts on innate immune cells—hemocytes in Drosophila or microglia/macrophages in mice. MANF derived from innate immune cells (or other sources) promotes phenotypic changes—atilla and arginase expression in hemocytes or alternative activation of microglia/macrophages—which are part of the mechanism involved in tissue protection. (B) MANF supplementation is an enhancer of retinal regenerative therapies by increasing the integration efficiency of exogenously supplied photoreceptors for retinal repair.

The usefulness of immune modulation for regenerative medicine has been anticipated based on studies of tissues where regeneration is sustained endogenously by resident stem cells (3, 11, 6569). Our study provides strong support for this hypothesis.

MANF has previously been described as a neurotrophic factor (30, 70, 71), and it may also exert a direct neuroprotective effect in the retina, yet our data suggest a more expansive role: Because MANF cannot promote tissue repair in flies in which the hemocyte response to MANF is selectively ablated, or in mammalian retinas depleted of innate immune cells or containing macrophages that are unresponsive to MANF, we propose that MANF’s role in promoting alternative activation of innate immune cells is central to its function in tissue repair. Further studies will be required to determine the specific contribution of alternative-activated macrophages in mediating these effects. Although our data point to an important role of macrophages in mediating the effects, it does not exclude the possibility that other cell types are involved in the process or that macrophages’ functions other than polarization may influence the outcome of MANF’s protective effects.

Clinically, MANF may thus have a distinct advantage over previously described neurotrophic factors both in improving survival of transplanted cells directly and in promoting a microenvironment supportive of local repair and integration. Because integration efficiency correlates with the extent of vision restoration (61), it can be anticipated that MANF supplementation will have an important effect in clinical settings.

Further studies involving tissue-specific knockdown of MANF in mammals will be required to evaluate the relative contribution of different cellular and tissue sources for MANF in homeostatic and damage conditions. Although we found that MANF is strongly expressed in immune cells, we also observed MANF expression in other cell types, in agreement with previous reports (72).

Similarly, the molecular mechanism involved in MANF signaling remains elusive. To date, a signal-transducing receptor for MANF has not been identified, although protein kinase C signaling has been described to be activated downstream of MANF (73). MANF can further negatively regulate nuclear factor κB (NF-κB) signaling in mammalian cells (74), and loss of MANF in Drosophila results in the infiltration of pupal brains with cells resembling hemocytes with high Rel/NF-κB activity, potentially representing proinflammatory, M1-like phenotypes (75). The identification of immune cells as a target for MANF in our study may accelerate the discovery of putative MANF receptors and downstream signaling pathways.

Because neurotoxic inflammation has been implicated in Parkinson’s disease (76), it is possible that the protective effects of MANF in this context (71) are also mediated by immune modulation, as we show here for retinal disease. Indeed, recent reports suggest that the MANF paralog, cerebral dopamine neurotrophic factor, has an anti-inflammatory function in murine models of Parkinson’s disease (77) and in nerve regeneration after spinal cord injury (78). A recent study has further shown that loss of MANF leads to beta cell loss in the pancreas (48). Beta cell loss is commonly associated with chronic inflammation, and it is thus tempting to speculate that MANF is broadly required in various contexts to aid conversion of proinflammatory macrophages into prorepair anti-inflammatory macrophages. Future studies will clarify the role of MANF in resolving inflammation and promoting tissue repair not only in the retina and brain but also in other tissues. A deeper understanding of MANF-mediated immune modulation and its effect on stem cell function, wound repair, and tissue maintenance is thus expected to help in the development of effective regenerative therapies.

Materials and methods


All mice used in the described studies were housed and bred at the Association for Assessment and Accreditation of Laboratory Animal Care International accredited vivarium of the Buck Institute for Research on Aging, in a specific-pathogen-free facility, in individually ventilated cages on a standard 12:12 light cycle. All procedures were approved by the Buck Institute Institutional Animal Care and Use Committee. For details on the mouse strains and lines used, see the supplementary materials (SM).

Drosophila stocks and culture

Fly stocks were raised on standard cornmeal- and molasses-based food. All experiments were performed at 25°C. Both sexes gave the same results in all experiments, unless otherwise described. For details on the fly lines used, see SM.

Intraocular injections in mice

For intravitreal injection, recombinant proteins or cells in 1-μl volume were injected into the right eye using a graduated pulled glass pipet and a wire plunger (Wiretrol II, 5-0000-2005, Drummond Scientific Company). For details on the test articles injected and the procedure for intravitreal injections, see SM.

For subretinal injection, dissociated GFP-expressing mouse retinal cells from the Nrl-GFP mice were transplanted into the subretinal space of recipient mice using the transcorneal subretinal injection method. For details on the preparation of Nrl-GFP cells and the procedure for transplantation, see SM.

Light damage in mice

Mice were dark-adapted for 18 hours before the procedure. Test eyes were exposed to 5000 to 20,000 lux of bright light using a 144 light-emitting diode microscope ring light (AmScope) for 1 to 2 hours. After light damage, mice were allowed to recover from anesthesia, returned to their cages, and housed in darkness until analysis. Undamaged control mice were housed in regular conditions throughout the experiment. For details, see SM.

UV damage in Drosophila pupae retina and larvae

Pupae retinas were exposed to 17.5 mJ of UV light as previously described (26, 28). Second-instar larvae were exposed to 50 mJ of UV light as previously described (25). For details on the procedures and quantification methods, see SM.

Histological analysis, imaging, and quantification methods

Retinal sections, macrophages, and hemocyte smears were analyzed by immunohistochemistry (IHC) and other histological methods (see SM for details) and imaged using a LSM 700 confocal laser-scanning microscope; images were used for quantification purposes. For details on staining methods and quantification methods, see SM.


All ERGs were carried out under scotopic conditions using a handheld multispecies electroretinograph (HMsERG, OcuScience) and analyzed using ERGVIEW Version 4.3 (OcuScience). For details on the procedures, see SM.

Cell culture

Raw 264.7 macrophages (ATCC, TIB-71, lot 61524889) and BMDMs were used in hrMANF stimulation experiments and MANF knock-down experiments. BMDMs were differentiated in culture from bone marrow using 20 ng/ml of macrophage colony-stimulating factor (M-CSF, Sigma, M9170) for 7 days. Raw 264.7 and BMDMs were stimulated for 3 hours with 10 μg/ml of hrMANF before analysis. For details on the procedures and on the knockdown experiments, see SM.

Expression analyses

RNA sequencing and RT-qPCR were used to quantify mRNA levels, and Western blot analysis was used to quantify protein levels. For details on the methods employed in each technique, see SM.

Statistical analysis

All counts are presented as average and standard error of mean (SEM). Statistical analysis was carried out using Microsoft Excel or GraphPad Prism, and Student’s t test or two-way analysis of variance (ANOVA) was used to determine statistical significance, assuming normal distribution and equal variance.

Supplementary Materials

Materials and Methods

Figs. S1 to S9

Tables S1 to S5

References (7982)

References and Notes

  1. Acknowledgments: We acknowledge Amarantus Biosciences Inc. for kindly supplying the hrMANF protein used in these studies. Work in H.J.’s laboratory is supported by NIH grant EY018177, and some of the fly work was supported by Amarantus Biosciences Inc. Work in D.A.L.’s laboratory is supported by NIH grant EY025779 and the Foundation for Retinal Research. J.N. is supported by the Glenn Foundation for Medical Research. The full RNA-seq data set is provided as supplementary materials. J.N., H.J., and D.A.L. are the inventors and the Buck Institute for Research on Aging is the applicant for an international patent application for the use of MANF as an enhancer of cell replacement therapies.
View Abstract

Stay Connected to Science

Navigate This Article