Anatomical location of injected microglia in different activation states and time course of injury determines survival of retinal ganglion cells after optic nerve crush

Abstract Background: Activated microglia release harmful substances to retinal ganglion cells (RGCs), but may also benefit by removing cellular debris and secreting neurotrophic factors. These paradoxical roles remain controversial because the nature and time-course of the injury that defines their role is unknown. The aim of this study was to determine if pharmacological manipulation of microglia to acquire a pro-inflammatory or pro-survival phenotype will exacerbate or enhance neuronal survival after injury.Material and methods: Treated HAP I (highly aggressively proliferating immortalized) microglia were injected into the vitreous or tail vein (T V) of female Sprague-Dawley rats. Retinas were examined at 4-14 days following optic nerve crush (ONC) and the number of surviving RGCs was determined.Results: Injection of untreated HAP I cells resulted in the greater loss of RGCs early after ONC when injected into the vitreous and later after ONC when injected into the T V. LP S activated HAP I cells injected into the vitreous resulted in greater RGC loss with and without injury. When injected into the T V with ONC there was no loss of RGCs 4 days after ONC but greater loss afterwards. Minocycline treated HAP I cells injected into the vitreous resulted in greater RGC survival than untreated HAP I cells. However, when injected into the T V with ONC there was greater loss of RGCs. These results suggest that optic nerve signals attract extrinsic microglia to the retina, resulting in a proinflammatory response.Conclusion: Neuroprotection or cytotoxicity of microglia depends on the type of activation, time course of the injury, and if they act on the axon or cell body. GRAPHICAL ABSTRACT HAPI microglia migrate to the retina or optic nerve following optic nerve injury when injected into the vitreous or tail vein, respectively. Pretreatment with LPS or minocycline differentially effects retinal ganglion cell survival. In most cases, the result late in the injury process is greater retinal ganglion cell loss. HIGHLIGHTS We show here that neuroprotection is not solely determined by the microglial activation state but factors such as the environment and time-course of the injury. Culture microglia can be treated in vitro and then injected in vivo. The cells migrate to the site of injury, cell body of retinal ganglion cells if in the vitreous or to the optic nerve if injected in the tail vein. Retinal ganglion cell death is dependent on the location the microglia act, time-course of injury, and activation state. Proinflammatory microglia can be neuroprotective early in the injury when the primary site of action is on the axons whereas hypoactivated microglia are neuroprotective early in injury when they act on the soma. Later in the injury, both become detrimental.

HAPI microglia migrate to the retina or optic nerve following optic nerve injury when injected into the vitreous or tail vein, respectively.Pretreatment with LPS or minocycline differentially effects retinal ganglion cell survival.In most cases, the result late in the injury process is greater retinal ganglion cell loss.

HIGHLIGHTS
• We show here that neuroprotection is not solely determined by the microglial activation state but factors such as the environment and time-course of the injury.• Culture microglia can be treated in vitro and then injected in vivo .
• The cells migrate to the site of injury, cell body of retinal ganglion cells if in the vitreous or to the optic nerve if injected in the tail vein.• Retinal ganglion cell death is dependent on the location the microglia act, time-course of injury, and activation state.• Proinflammatory microglia can be neuroprotective early in the injury when the primary site of action is on the axons whereas hypoactivated microglia are neuroprotective early in injury when they act on the soma.Later in the injury, both become detrimental.
Glossary : HAPI (highly aggressively proliferating immortalized) rat microglial cells : a spontaneously immortalized cell line derived from primary microglial cell cultures of newly born rat brains.; Intravitreal (IV) injection : administration of substances (i.e.drugs, cells) into the vitreous humor of the eye.; Lipopolysaccharide (LPS) : lipid and polysaccharide components found on the outer membrane of gram-negative bacteria.; Minocycline : a tetracycline antibiotic.; Optic nerve crush (OPC) : optic nerve injury model used to model traumatic optic neuropathies and glaucoma.Created by applying a compression pressure of the optic nerve that results in gradual apoptosis of the retinal ganglion cells.; Retinal Ganglion Cells (RGCs) : a neuronal cell population located in the inner surface of the retina from which the axons make up the optic nerve.The RGCs receive visual information from the photoreceptors via the bipolar and amacrine cells and transmit this information to the brain through the optic nerve.; Retinal Muller Cell line-1 (rMC-1) : an immortalized cell line of rat Muller cells stably transformed from SV40 antigen.Muller cells are glial cells in the retina and support function of retinal neurons and regulate metabolism.; Tail Vein (TV) injection : administration of a substance (i.e.drugs, cells) into the venous circulation through the veins that run the lateral aspects of the tail.

Introduction
Microglia are the innate immune cells of the central nervous system, including the retina, that help maintain normal function as well as respond to immunological stress [ 1 , 2 ].Microglia are of an erythromyeloid precursor cells lineage and their cell bodies can be found in the ganglion cell layer, inner plexiform layer, nerve fiber layer, and the outer plexiform layer [ 3 , 4 ].Under normal conditions, microglia monitor the retina, are involved in scavenging toxins, synaptic pruning, supplying metabolites and growth factors and can regulate neurogenesis [5][6][7].However, in response to injury microglia will proliferate and migrate to the affected areas [ 8 , 9 ].Activated microglial cells also exhibit immunophenotypic changes and release products with neurotrophic or neurotoxic action, however, the exact role of microglia remains controversial [3 , 10-12 ].Microglia are highly flexible cells that can alter their functional state through phagocytosis, antigen presentation, production of cytokines and growth factors [ 13 , 14 ].In early injury, microglia may attempt to modulate the inflammatory environment to promote neuroprotection and regeneration [ 15 , 16 ].However, as the injury progresses, microglial response may become exaggerated leading to tissue damage and disease progression [ 17 , 18 ].Traditionally it was proposed that there exists 2 extreme activation phenotypes that microglia can acquire, the classically activated (M1) phenotype or the alternatively activated (M2) phenotype [ 19 ].Most commonly, M1 microglial are defined by an increase in pro-inflammatory factors such as tumor necrosis factor (TNFα), interleukin 1β (IL-1β), and nitric oxide synthase (NOS) [20][21][22].This can lead to increase in oxidative stress and increase of leukostatsis through the breakdown of the blood retina barrier [21][22][23].However, M2 microglia are characterized by a change of phenotype to an anti-inflammatory state through upregulation of arginase-1, CD2o6, transforming growth factor-β (TGF-β), IL-10 and downregulation of nitric oxide production [24][25][26].M2 microglia help to promote neuroprotection and regeneration, therefore therapeutic agents that shift the polarity to this state have been the target for treatment of neurological diseases [ 27 , 28 ].
The progression of neurodegenerative diseases may be dependent on the activation state that predominates.Microglial activation does not have to exist exclusively as one type or another and can often be a mix of the 2 phenotypes on a spectrum [ 29 , 30 ].The effect of microglial cells can be beneficial or detrimental, depending on if the response is too aggressive or passive.What may determine if microglia have neurotoxic or neuroprotective effect after injury may depend on the time course and nature of the injury, as well as, the anatomical location that the microglia act on.early in neurodegeneration, microglia may try to initiate repair mechanisms and remodel the tissue, however, after prolonged activation there is chronic inflammation that leads to neuronal loss [11 , 15-17 , 31-33 ].Studying the phenotypic, morphological, and functional changes in microglia after injury is important because they can help expose signaling events that determine their activation [2].Due to the diverse phenotypes of microglia and macrophages, there may be many possible targets for therapeutic interventions.
In this study we investigate the effect of systemic infusion or intravitreal injection of exogenously activated microglia on RGC survival after optic nerve crush (ONC).experimentally, LPS can be used to promote an M1 phenotype, while Minocycline can selectively inhibit microglia from taking on the M1 phenotype but not the M2 phenotype [34][35][36].The balance between these 2 cell types may determine if microglia are more beneficial or detrimental after injury and may play a role in neurodegenerative diseases [23,29].It is hard to determine the role of microglia, particularly after injury, since they are difficult to distinguish from macrophages and treatments to elicit phenotypic changes often effect other cell types [25 , 37 ].HAPI (highly aggressive proliferating immortalized) cells are a spontaneously immortalized rat microglial cell line isolated from 3 day old rat brains.Due to being spontaneously immortalized, HAPI microglia do not require fresh isolation each culture reducing animal to animal variations.They have similar characteristics to primary microglia in phagocytic activity, pro-inflammatory cytokine expression, and marker expression [ 38 ].In addition, since HAPI cells are rat in origin, unlike BV-2 cells which are derived from mice, we minimize the risk of immune rejection when transplanted into rats.HAPI cells are a well characterized line that many studies have used to study topics such as immune modulatory therapies, toxicology, and oxidative stress responses [39][40][41][42][43][44][45][46][47].The response of HAPI cells to LPS and minocycline has also been extensively investigated previously [38 , 39 , 48-55 ].The objective of this study was to determine the effect of cultured HAPI cells in quiescent and over-activated states on the survival of RGCs.We sought to explore if pharmacological manipulation of microglia to acquire either a pro-inflammatory or pro-survival phenotype will exacerbate neuronal cell death or enhance neuronal survival after injury, respectively.
When HAPI or rMC-1 cells reached 70% confluency, serum free media was added.The cells were grown until they were 90% confluent and were labelled with wheat germ agglutinin conjugated Texas Red fluorescence (WGA-TR; W831; Molecular Probes, eugene, OR) by washing the cells in WGA-TR diluted in PBS for 10 min (Supplementary Figure 1).The WGA was then removed and the cells were washed 3 times with PBS.Then the cells were prepared for injection, imaging, or cell counting.

LPS activation of HAPI cells
The cells were grown (as described above, except serum free media was added at least 24 h before experiments were to take place) until they were 70% confluent and then were hyper-activated by adding 1 µg/mL LPS (escherichia coli 055:B5; L2637; Sigma Aldrich; St. Louis, MO) in fresh media for 24 h.HAPI cells were washed 5 times in PBS before being labelled with WGA-TR (W831; Molecular Probes, eugene, OR) in PBS for 10 min.The WGA was then removed and the cells were washed 3 times with PBS.

Minocycline treatment of HAPI cells
The cells were grown (as described above, except serum free media was added at least 24 h before experiments were to take place) until they were 70% confluent and then were treated with 10 µg/mL minocycline (Sigma Aldrich; St. Louis, MO) in fresh media for 1 h.HAPI cells were washed 5 times in PBS before being labelled with WGA-TR in PBS for 10 min.The WGA was then removed and the cells were washed 3 times with PBS.Then the cells were prepared for injection, imaging, or cell counting.

Animals
Adult, female Sprague-Dawley rats (n = 112; 225-250 g; Charles River, Wilmington, MA) that were free of common pathogens were used in all experiments.The groups included animals with uninjured (n = 10), optic nerve crush (n = 6 per time point), with intravitreal injection of HAPI cells with and without optic nerve crush (n = 6 per time point and condition), and tail vein injection of HAPI cells with and without optic nerve crush (n = 6 per time point and condition) followed for 4, 7, and 14 days post injury (summary of animal groups and results can be found in Table 1 ).An additional control was introduced using rMC-1 Muller cells injected into the vitreous or tail vein and followed for 4, 7, or 14 days (n = 2 per condition and timepoint).Animals were cared for according all appropriate guidelines including those of the Canadian Council on Animal Care and all experiments were approved by the McMaster University Animal Research ethics Board.The animals were kept on a 12 h light cycle and had access to food and water ad libitum.In all experiments the rats were anesthetized with an intraperitoneal injection of 7% chloral hydrate (Thermo Fisher Scientific, Ottawa, ON; 0.42 g/kg of body weight) during experimental procedures.An ophthalmic eye lubricant (Lacri-Lube; Allergan, Markham, ON, Canada) was applied to the eyes before surgery.Animals were kept warm on a heating blanket (38 °C).Animals were given subcutaneous injections of Anafen (Merial Canada, Baie D'Urfé, QC; 5 mg/kg) to minimize discomfort following surgery.They were also given 5 mL of saline subcutaneously and allowed to recover on a heating blanket.Six animals per group were used and the groups consisted of control, injured control, intravitreal cell injection without injury, tail vein cell injection without injury, intravitreal cell injection with ONC, and tail vein cell injection with ONC.

Retrograde labeling of retinal ganglion cells
Retinal ganglion cells were retrogradely labeled before the optic nerve crush by injecting Fluorogold (FG; Fluorochrome LLC, Denver, CO) bilaterally into the superior colliculus as described by Koeberle and Ball [ 56 ].The rat was placed on a stereotaxic frame (David Kopf M900, Tujunga, CA) and an incision was made on top of the head approximately 1-2 mm from the eyes and 1-2 mm before the ears.Once the incision was made, the transverse suture was located.A 1 mm hole was drilled bilaterally 2.5 mm rostral to lamda and 1.2 mm lateral to the sagittal sutures.Four microliters of FG was injected 3 mm deep into the parenchyma of the brain using a 10 μl Hamilton syringe (Hamilton M701, Reno, NV) with the aid of a microinjector (WPI UltraMicroPump, Sarasota, FL).

Intraocular injections and tail vein injections of microglia
The microglial cells were trypsinized and centrifuged at 19 g (325 RPM; eppendorf 5811 F) for 5 min.The visible pellet of cells on the bottom of the vial were resuspended in PBS.Immediately before ONC, animals received cell injection into the vitreous chamber of the eye by a posterior route or a tail vein injection by the lateral vein route.
The sclera was punctured within 3 mm of the optic nerve with a 30 gauge needle.Then, 5 μL of microglia or rMC-1 cells solution (approximately 30,000 cells) was injected into the vitreous cavity with a 33 gauge Flexifil beveled needle (NF33FBV; World Precision Instruments, Sarasota, FL) on a WPI Nanofil Syringe (300329; World Precision Instruments, Sarasota, FL).Care was taken not to damage the lens or the anterior structures of the eye, which are known to secrete growth factors [ 57 ].Following intraocular injections, the needle was left in place for 2 min and then withdrawn slowly.In the tail vein condition, a 25 gauge needle was used to inject 1 mL of microglia or rMC-1 cell solution (approximately 5 million cells) into the lateral tail vein.

Optic nerve injury
Optic nerve crush (ONC) injury was done one week after FG injection and immediately after injection.The animals were prepared for surgery as previously described.An incision was made in the skin around the rim of the orbital bone to access the optic nerve.The orbital contents were retracted away and the rectus muscle was moved laterally.The eye was rotated temporally to expose the optic nerve.The optic nerve was either crushed by compressing the intradural optic nerve 2 mm from the eye for 3 s (described in Kalesnykas et al. 2012) using Dumont SS fine forceps (Fine Science Tools 11203-23; North Vancouver, BC).

Tissue preparation
The rats were euthanized 4, 7, and 14 days after injury and microglial cell injection by injecting a lethal dose of 7% chloral hydrate.The eyes were enucleated and the cornea and lens were dissected away.The eye cup was fixed in 4% paraformaldehyde, 2% sucrose in Sorensen's phosphate buffer (0.1 M, pH 7.3) for 2 h.The eye cups and optic nerve were rinsed in sodium phosphate buffered saline (PBS; 0.1 M, pH 7.3, 0.9% NaCl) 3 times for 30 min each.The eyecups were placed in 30% sucrose in PBS at 4°C overnight.The eyecups were embedded in OCT compound (Tissue-Tek, Sakura Finetek, Torrence, CA) and sectioned in a cryostat microtome (Leica Microsystems CM1900, Concord, ON) at −20 °C.Transverse sections of 12 μm thickness were picked up on Superfrost Plus slides (Thermo Fisher Scientific, Ottawa, ON).The slides were covered with a glass coverslip with Vectashield (H1000; Vector Labs, Burlington, ON) for visualization under an epifluorescence microscope (Zeiss Axioplan 2, Carl Zeiss, Toronto, ON).

Labeling of retinal ganglion cells with Brn3a antibody
The retinal sections were washed in PBS 3 times for 5 min each.The retinas were incubated in a blocking solution overnight at room temperature.The blocking solution contained PBS with 10% normal donkey serum, 3% bovine serum albumin, 0.1% triton X-100, and 0.1% DMSO.Retinal ganglion cells were labelled with a Brn3a antibody [ 58 ] which is s known to label over 90% of RGCs [ 59 ].The retinal sections were incubated in the Brn3a antibody (α goat; 1:200; Santa Cruz SC-31984 (C-20); immunizing protein human class IV POU domain protein; Santa Cruz Biotechnology Inc., Santa Cruz, CA) in blocking solution over night at room temperature.To detect the primary antibody, donkey α goat Alexa 488 (1:200; Molecular Probes A-11055, Life Technologies, Burlington, ON) in blocking solution was placed on the retinal sections for 4 h at room temperature.The sections were washed in PBS 3 times for 5 min each after each incubation period.The slides were coverslipped in Vectashield (H1000; Vector Labs, Burlington, ON) for visualization under an epifluorescence microscope (Zeiss Axioplan 2, Carl Zeiss, Toronto, ON).

Statistical analysis
Statistical analysis was done using GraphPad Prism (La Jolla, CA).Significance was determined by one way ANOVA analysis followed by Tukey's post-hoc test.All ± values are reported as 95% confidence intervals.exact p values were used to represent significance of the data, however significance was represented as p < 0.001 when the exact p value was lower than 0.001.The RGCs were quantified by manual counting of somas from transverse sections of retinas as previously described [ 60 ].FG is also known to increase in microglia that uptake it as debris from dying RGCs, this FG labelled microglia were excluded as previously described [60].We have previously demonstrated that data from flatmounts and sections were the same and that the measurements were consistent with other reports [61][62][63].The cell counts were made in the mid-peripheral retina approximately 1 mm from the optic nerve and 1 mm from the ora serrata.

Migration of HAPI microglial cells
At 4, 7, and 14 days after HAPI microglial cell injection, cells were only visible around the retina and vitreous after intravitreal injection accompanied by an optic nerve crush ( Figure 2 and supplementary Figure 2 ).However, there were no HAPI cells observed in the optic nerve after intravitreal injection regardless of injury (Figure 2G-l).Similarly, there were no HAPI cells in the retina after tail vein injection regardless of whether the optic nerve was crushed or not ( Figure 3A-F ).HAPI microglia were present in the optic nerve at 4, 7, and 14 days after HAPI microglial cells were injected into the tail vein and the optic nerve was crushed (Figure 3G-I).In the non-injured condition, regardless of the method of injection, there were no HAPI microglial cells seen in the retina or optic nerve (Figures 2D-F, 2J-L, 3D-F, and 3J-L) similar to retinas and optic nerves with no cells injected (Figure 1).

RGC loss after intravitreal microglial cell injection
Four days after HAPI microglia were injected into the vitreous without any injury to the optic nerve, the RGC density was 56.22 ± 2.70 RGCs/mm (n = 6; Figure 2M).This was not significantly different from the control (ANOVA/Tukey's post-hoc test; p = 0.972).Similarly, at 7 days, the RGC density was 54.54 ± 2.15 RGCs/mm (n = 6; Figure 2M).This also was not significantly different from the control (ANOVA/Tukey's post-hoc test; p = 0.999).However, at 14 days after microglial cell injection into the vitreous there was an RGC density of 45.86 ± 1.52 RGCs/mm (n = 6).This was significantly different than control retinas (ANOVA/Tukey's post-hoc test, p ≤ 0.001; Figure 2M).In addition, when rMC-1 cells were injected into the vitreous or tail vein, there was no significant loss of RGCs over the 14 day period.
Four days after microglia were injected into the vitreous and ONC injury, there was a RGC density of 37.45 ± 2.83 RGCs/mm (n = 6; Figure 2M).This resulted in 16.6% more loss of RGCs 4 days after intravitreal injections of microglia and ONC than the loss that would be expected from ONC alone (ANOVA/Tukey's post-hoc test, p = 0.004; Figure 2M).Seven days after intravitreal injection of microglia and ONC, there was greater loss of RCGs.These retinas had a RGC density of 20.39 ± 2.44 RGCs/mm (n = 6; Figure 2M).This is a loss of 40.24% more RGCs 7 days after intravitreal injection of microglia and ONC than would be expected after ONC alone (ANOVA/Tukey's post-hoc test, p ≤ 0.001; Figure 2M).This effect was lost after 14 days of injury.RGC density of retinas 14 days after intravitreal injection of microglia and ONC was 18.26 ± 3.31 RGCs/mm (n = 6; Figure 2M).This was not significantly different than the RGC densities for retinas 14 days after ONC (ANOVA/Tukey's post-hoc test; p = 0.996).
It is interesting to note that the loss of RGCs after intravitreal injection of HAPI cells with ONC was accelerated when compared to ONC alone.The RGC density for the retinas 4 days after intravitreal injection and ONC was similar to that of retinas 7 days after ONC alone (ANOVA/Tukey's post-hoc test; p = 0.737; Figure 2M).Also, the RGC densities for the retinas 7 days (ANOVA/Tukey's post-hoc test; p > 0.05) and 14 days (ANOVA/Tukey's post-hoc test; p = 0.999) after intravitreal injection and ONC was similar to the RGC dentisies 14 days after ONC alone.(G) there is a gradual loss of retinal ganglion cells over time after optic nerve crush.control retinas had 54.74 rGcs/mm, 4 days after onc there were 44.90 rGc/mm, 7 days after onc there were 34.12 rGcs/mm, and 14 days of onc there were 18.9 rGcs/mm remaining.

RGC loss after tail vein microglial cell injection
Four days after injection of HAPI microglia into the tail vein without ONC resulted in a RGC density of 54.29 ± 3.25 RGCs/mm (n = 6; Figure 3A ).This was not statistically different from controls (ANOVA/Tukey's post-hoc test; p = 0.999).The RGC density 7 days after tail vein injection of microglia without injury was 55.31 ± 2.10 RGCs/mm (n = 6; Figure 3A ).This was not statistically different from the control (ANOVA/Tukey's post-hoc test; p = 0.998).However, 14 days after the injection of HAPI microglia in the tail vein the RGC density was 47.95 ± 2.02 RGCs/mm (n = 6; Figure 3A ).Similar to when HAPI microglial cells were injected into the vitreous for 14 days, there was a statistically significant loss of RGCs (ANOVA/Tukey's post-hoc test, p < 0.001) 14 days after microglia were injected into the tail vein in the absence of an injury.
The density of surviving RGCs 4 days after the injection of HAPI microglia in the tail vein and accompanied by an ONC was 44.65 ± 2.62 RGCs/mm (n = 6; Figure 3A ).This was not statistically different from the RGC density 4 days after ONC (ANOVA/Tukey's post-hoc test; p = 0.999).However, the RGC density 7 days after tail vein injection and ONC was 27.36 ± 1.80 RGCs/mm (n = 6; Figure 3A ).This represented a significant loss of 19.81% more RGCs than are lost 7 days 12 µm thick frozen transverse sections of retinas and optic nerve labelled with fluorogold (labelling rGc somas in gold) that had received intravitreal injections of HaPI microglial cells (labelled with WGa-tr; red).(a) 4 days after injection of microglia and onc.transplanted microglia were seen near the vitreal surface of the retina (white arrows).there was a loss of retinal ganglion cells comparable to 4 day onc retina (1e).(B) 7 days after microglia injection and onc.transplanted microglia were seen near the vitreal surface of the retina (white arrows).there was a loss of retinal ganglion cells comparable to (1f).(c) 14 days after microglia injection and onc.fewer microglial cells were consistently seen in the retina (white arrows) at this time point.there was a loss of retinal ganglion cells comparable to (1 f).(D) 4 days after microglia were injected.(e) 7 days after microglia were injected.(f) 14 days after microglia were injected.only at 14 days there was a statistically significant (anoVa followed by tukey's post hoc test) loss of retinal ganglion cells compared to the control.(G) 4 days after microglia injection and onc.(H) 7 days after microglia injection and onc.(I) 14 days after injection of microglia and onc.(J) 4 days after injection of microglia.(K) 7 days after microglial cell injection.(l) 14 days after microglial cell injection.there were no transplanted microglia present in, or near, the optic nerve after intravitreal injections at any time point.(M) rGc survival 14 days after intravitreal injection of microglial cells.Graph showing rGc survival after intravitreal injection of microglia.there was no significant change from control (black line) when microglia were injected without onc (red) until 14 days after injection.the rGc density was lower after microglia were injected with onc (black triangles) when compared to onc alone during the first 7 days.14 days after injection and onc there was no significant difference from onc alone.error bars (or dotted lines): 95% confidence interval.anoVa statistically analysis and tukey's post-hoc test were performed.* compared to onc; & compared to control; # comparison between intravitreal and tail vein injections. Figure 3. 12 µm thick frozen transverse sections of retinas and optic nerve labelled with fluorogold (labelling rGc somas in gold) that had received tail vein injections of microglial cells (red).(a) 4 days after microglial cell injection and onc.(B) 7 days after microglial cell injection and onc.there was a significant loss of rGcs as compared to its control (1e).(c) 14 days after microglia were injected with no onc (D) 4 days after microglia were injected with no onc.(e) 7 days after microglia were injected with no onc.(f) 14 days after microglia were injected with no onc.a statistically significant (anoVa following tuckey's post hoc test) loss of rGcs between the non-injured microglia injected treatment (c) compared to control was at 14 days.there were no transplanted microglia (labelled in WGa-tr; red) visible in the retinal sections after tail vein injections.(G) optic nerve collected 4 days after microglial cells were injected and onc.few transplanted microglia were often seen along the crush site (arrows).(H) optic nerve collected 7 days after microglia were injected and onc.transplanted microglial cells were seen along the crush site and edges (arrows).(I) optic nerve collected 14 days after microglia were injected and onc.few microglial cells were seen throughout the optic nerve (arrows).(J) optic nerve collected 4 days after microglial cells were injected.(K) optic nerve collected 7 days after microglial cells were injected.(l) optic nerve collected 14 days after microglial cells were injected.no transplanted microglia were seen in optic nerves from any of the uninjured time-points.(M) rGc survival 14 days after tail vein injection of microglial cells.there was no significant change from control (black line) when microglia were injected without onc (red) until 14 days after injection.the rGc density was lower after microglia were injected with onc (black circle) when compared to onc alone 7 days after the injury.early in the injury, tail vein injections resulted in no additional loss of rGc than expected with onc alone.(n) Graph comparing rGc survival after intravitrel injection of HaPI microglial cell (HaPI IV; black) and tail vein injection of HaPI microglial cells (HaPI tV; red) injections compared to retinas that did not receive HaPI microglia injections (grey).the rGc density 7 days after tail vein injection and onc (red) was intermediate to that of onc alone (black circle) and after microglial cell injection in the vitreous followed by onc (grey line).error bars (or dotted lines): 95% confidence interval.anoVa statistically analysis and tukey's post-hoc test were performed.* compared to onc; & compared to control; # comparison between intravitreal and tail vein injections.
after ONC alone (ANOVA/Tukey's post-hoc test, p ≤ 0.001).The RGC density 14 days after tail vein injection and ONC was 14.13 ± 1.58 RGCs/mm (n = 6; Figure 3A).This represented a significant loss of 25.40% more RGCs than that would be lost after ONC alone (ANOVA/Tukey's post-hoc test, p = 0.007).Tail vein injections of HAPI microglia accompanied by an ONC resulted in a greater loss of RGCs compared to what would be expected after ONC alone over a 14 day period.

Migration of LPS treated HAPI microglial cells
There were no HAPI cells 4 days after injection into the vitreous without ( Figure 4A ) or with (Figure 4D ) ONC in the retina.At 7 days (Figure 4B and 4e) and 14 days (Figure 4C and 4F ), HAPI cells were observed in the vitreous above the ganglion cell layer (GCL) or in the GCL itself.In the absence of injury, there were no LPS activated HAPI cells seen in the optic nerve (Figure 4 ) after intravitreal injection.However, when When HaPI cells are activated with lPs without onc there was a statistically greater survival of rGcs 14 days after injection as compared to rGc survival after untreated HaPI cell injection.(n) rGc density 4 days after injection of lPs activated HaPI cells with onc (red circle) was not different from control (solid grey line).However, 7 days after onc and injection of lPs activated HaPI cells, the rGc density was lower than the rGc density 7 days after just onc.therefore, there was greater survival of rGcs early in the injury and less late in the injury.(o) 7 days after injection of lPs activated HaPI cells into the tail vein with onc (blue square) there was greater cell death than when lPs activated HaPI cells were injected without onc (red circle).error bars = 95% confidence interval.anoVa/tukey's post-hoc test were performed.# compared to onc; & compared to control; * comparison between lPs treated and untreated.
injected into the vitreous with optic nerve crush, LPS activated HAPI cells and/or other auto-florescent immune cells were seen in the optic nerve (Figure 4).HAPI cells were not present in the retina or vitreous after LPS activated HAPI cells were injected into the tail vein ( Figure 5 ).LPS activated HAPI cells were also absent from the optic nerve when injected into the tail vein and there was no ONC (Figure 5A-F).However, LPS activated HAPI cells were seen along the crush site and the edges of the optic nerve after tail vein injection and ONC injury (Figure 5J-L).
The RGC densities after injection of LPS activated HAPI cell into the vitreous with ONC over a 14 day period was lower than the RGC densities after ONC (Figure 4N).The RGC density 4 days after LPS activated HAPI cells were injected with an ONC was 13.74 ± 5.48 RGCs/mm (n = 6; Figure 4N).This was 69.4% lower than the RGC density 4 days after ONC (n = 6, ANOVA/Tukey's post-hoc test, p ≤ 0.001) and 63.3% lower than when HAPI cells were injected into the vitreous accompanied by an ONC (n = 6, ANOVA/Tukey's post-hoc test, p ≤ 0.001).When LPS activated HAPI cell were injected 7 days earlier accompanied by an ONC there was a RGC density of 19.71 ± 2.76 RGCs/mm (n = 6; Figure 4N).This represented a loss of 42.2% of cells when compared to the RGC density 7 days after ONC (n = 6, ANOVA/Tukey's post-hoc test, p ≤ 0.001).However, the RGC loss was statistically not different from when HAPI cells were injected with ONC for 7 days (n = 6, ANOVA/ Tukey's post-hoc test).The RGC density 14 days after LPS activated HAPI cells were injected accompanied by an ONC was 9.15 ± 4.47 RGCs/mm (n = 6; Figure 4N).This was 51.7% lower than the RGC density 14 days after ONC (n = 6, ANOVA/Tukey's post-hoc test, p ≤ 0.001) and 49.9% lower than when HAPI cells were injected into the vitreous with ONC (n = 6, ANOVA/Tukey's post-hoc test, p ≤ 0.001).The RGC densities were not different from each other over the 14 day period that the LPS activated HAPI cells were injected into the vitreous with ONC (n = 6; Figure 4N; ANOVA/Tukey's post-hoc test; p > 0.05).

RGC loss after tail vein injection of LPS treated HAPI cells
The RGC density 4 days after the injection of LPS activated HAPI cells in the absence of injury was 56.63 ± 4.00 RGCs/mm (n = 6; Figure 5M).This was not significantly different from the control (n = 10; ANOVA/ Tukey's post-hoc test; p > 0.05) and from when HAPI cells were injected into the TV without ONC after 4 days (n = 6; ANOVA/Tukey's post-hoc test; p > 0.05).After 7 days following injecting LPS activated HAPI cells without ONC there were 54.74 ± 6.05 RGCs/mm (n = 6; Figure 5M).This was not significantly different from the control (n = 10; ANOVA/Tukey's post-hoc test; p > 0.05) or from when HAPI cells were injected into the TV without activation 7 days later (n = 6; ANOVA/ Tukey's post-hoc test; p > 0.05).Similarly, there was no difference in the RGC densities when LPS activated HAPI cells were injected14 days previously without injury (55.99 ± 4.43 RGCs/mm; n = 6; Figure 5M) and from the control (n = 10; ANOVA/Tukey's post-hoc test; p > 0.05).However, the RGC density was 16.8% greater than the RGC density 14 days after the injection of non-activated HAPI cells into the TV (n = 6; ANOVA/ Tukey's post-hoc test, p ≤ 0.001).The RGC densities across the 14 day period after injection of LPS activated HAPI cells into the TV were not statistically different from each other and similar to the control (Figure 5M).

Migration of minocycline treated HAPI microglial cells
HAPI cells were not seen in the retina after minocycline treated HAPI cells were injected into the vitreous in the absence of an injury at any of the timepoints after injury ( Figure 6A-C ).However, HAPI cells were first seen in the retina 7 days after injury.It was not until 14 days after injection that many more HAPI cells were observed in the retina.The presence of numerous HAPI cells was often associated with inflammation, to the point that the retinal layers were not distinguishable (Figure 6F).HAPI cells were not seen in the optic nerve after minocycline treated HAPI cells were injected into the vitreous whether there was an injury or not (Figure 6G-L).After minocycline treated HAPI cells were injected into the tail vein, no HAPI cells were present in the retina ( Figure 7A-F ) 4-14 days after injection, regardless of there being an injury or not.However, if there was an optic nerve crush injury (ONC) and the minocycline treated HAPI cells were injected into the tail vein, there were HAPI cells radiating from the crush site 4-14 days after injury (Figure 7J-L).

RGC loss after intravitreal injection of minocycline treated HAPI cells
When minocycline treated HAPI cells were injected into the vitreous without ONC 4 days previously there was an RGC density of 61.38 ± 5.75 RGCs/mm (n = 6; Figure 6M).This was not statistically different from controls (n = 7; ANOVA/Tukey's post-hoc test; p > 0.05) or 7 days after untreated HAPI cells were injected into the vitreous without ONC (n = 6; ANOVA/Tukey's post-hoc test; p > 0.05).There were 62.14 ± 5.76 RGCs/ mm (n = 6) 7 days after minocycline treated HAPI cells were injected into the vitreous without injury.This was not statistically different from controls (n = 7; ANOVA/ Tukey's post-hoc test; p > 0.05) or when untreated HAPI cells were injected into the vitreous (n = 6; ANOVA/ Tukey's post-hoc test).Similarly, 14 days after minocycline treated HAPI cells were injected into the vitreous there was an RGC density of 56.51 ± 4.73 RGCs/mm (n = 6).This was statistically similar to controls (n = 7; ANOVA/Tukey's post-hoc test; p > 0.05) but there was 23.2% more RGCs than when untreated HAPI cells were injected into the vitreous without injury (n = 6; ANOVA/ Tukey's post-hoc test, p ≤ 0.001).
The RGC density 4 days after the injection of minocycline treated HAPI cells into the vitreous accompanied by an ONC was 39.55 ± 4.85 RGCs/mm (n = 4; Figure 6N).Similarly, 7 days after minocycline treated HAPI cells were injected into the vitreous with ONC, there were 34.30 ± 3.92 RGCs/mm (n = 4).These values were not statistically different from densities expected after ONC alone.After 14 days of injecting minocycline treated HAPI cells into the vitreous with ONC there was greater loss of RGCs with 2.67 ± 3.2 RGCs/mm remaining 14 days after HAPI cell injection (n = 5; ANOVA/Tukey's post-hoc test, p ≤ 0.001).

RGC loss after tail vein injection of minocycline treated HAPI cells
There was no significant loss of RGCs when minocycline treated HAPI cells were injected into the tail vein without injury (Figure 7M).The RGC density 4 days after injection of minocycline treated HAPI cells without ONC was 58.77 ± 6.79 RGCs/mm (n = 6; Figure 6M).This was not significantly different from the control (n = 7; ANOVA/ Tukey's post-hoc test; p > 0.05) or from HAPI cells injected into the tail vein with no treatment (n = 6; ANOVA/Tukey's post-hoc test; p > 0.05).Similarly, the RGC density 7 days after injection of minocycline treated HAPI cells into the tail vein was 59.19 ± 5.25 RGCs/mm (n = 6) and 52.16 ± 6.31 RGCs/mm (n = 6) 14 days after injection of minocycline treated HAPI cells.This was not significantly different from the control (n = 7; ANOVA/ Tukey's post-hoc test; p > 0.05) or when untreated HAPI cells were injected into the tail vein without injury (n = 6; ANOVA/Tukey's post-hoc test; p > 0.05).
Injection of minocycline treated HAPI cells into the tail vein accompanied by an ONC resulted in greater RGC death than expected from ONC alone or when untreated HAPI cells were injected into the tail vein with ONC (Figure 7N).After 4 days of injecting minocycline treated HAPI cells into the tail vein with ONC, there were 18.69 ± 7.27 RGCs/mm remaining (n = 6).This was a loss of 58.4% of the RGCs compared to the loss expected from ONC alone (n = 6; ANOVA/ Tukey's post-hoc test, p ≤ 0.001) and 58.1% less RGCs than when untreated HAPI cells were injected into the tail vein with ONC (n = 6; ANOVA/Tukey's post-hoc test, p ≤ 0.001).There were 14.16 ± 5.58 RGCs/mm (n = 6) remaining 7 days after minocycline treated HAPI cells were injected into the tail vein with ONC.This was 58.5% less RGCs than expected after ONC alone (n = 6; ANOVA/Tukey's post-hoc test, p ≤ 0.001) and 48.3% less than the RGC density 7 days after untreated HAPI cells were injected into the tail vein (e) 7 days after injection and onc there were a small number of HaPI cells present in the retina (white arrows).(f) 14 days after injection with onc the retina was very inflamed that even the retina layers were not distinguishable.there were many HaPI cells and may be other auto-fluorescent immune cells present in the retina.the retinal ganglion cells were labeled with Brn3a (green).(G-I) there were no HaPI cells in the optic nerve 4-14 days after the injection of minocycline treated HaPI cells whether there was no injury or (J-l) optic nerve crush injury.(M) the rGc survival 4-14 days after intravitreal injection of minocycline treated HaPI cells.the rGc density was not different from controls (grey line) when minocycline treated HaPI cells were injected into the vitreous without onc (red squares).the rGc density did not differ from when untreated HaPI cells were injected into the vitreous without onc (blue circles) at 4 and 7 days, however, the rGc density was greater than the rGc density 14 days after untreated HaPI cells were injected into the tail vein without onc.(n) Minocycline treated HaPI cells may be neuroprotective early in the injury after intravitreal injection and cytotoxic later in the injury.since the indigenous microglia were not inhibited, we would still expect the loss that is seen after onc. the rGc density was not different from the density 4 days after untreated HaPI cells were injected into the vitreous with onc. the rGc survival was greater at 7 days after injury but there was greater rGc loss 14 days after injury.(o) the rGc survival trend after injection of minocycline treated HaPI cells into the vitreous.the rGc density did not change from values for the controls (grey solid line) when minocycline treated HaPI cells were injected without onc (red square).the rGc density 4 and 7 days after injection were not different than the values expected after onc alone (grey dotted line).However, 14 days after there were much fewer surviving rGcs than there would be after onc alone at the same time point.error bars = 95% confidence interval.anoVa/tukey's post-hoc test were performed.# compared to onc; & compared to control; * comparison between minocycline treated and untreated.

Discussion
This novel method of grafting HAPI cells into the vitreous or injecting them into the tail vein provides an excellent way to study the effect of differentially activated microglia on RGC survival.A problem has been that substances, such as LPS, have been found to activate other cell types in the retina [64][65][66][67] making it difficult to identify the cellular mechanisms involved in LPS mediated cell death.Previously, microglial activation has been studied in vitro, however, this does not help us gain an understanding of how microglial activation effects RGC survival taking into account the complex interplay of different cell types in vivo.The method used in the study permitted us to have the strengths of both in vitro and in vivo approaches.We were able to activate the microglia cells in vitro and introduce them in vivo, without having the side effects of the drug in vivo.This also provides an avenue to investigate future applications of microglia for cell therapy approaches, particularly using different agents to pre-treat the microglia before implantation.

Migration of untreated HAPI cells
Normally, it is difficult to distinguish macrophages from resident microglia because they both have similar morphologies and phenotypes [23 , 68 , 69 ].However, by injecting exogenously labelled cells into the systemic system, it was possible to observe their effect.When HAPI microglia were injected into the vitreous chamber they only migrated to the retina if there was some injury signal for the cells to follow.If the cells were injected into an uninjured eye, there were no cells visible in the retina.Injected HAPI microglia were also not visible in the retina after tail vein injection.However, they were visible at the optic nerve when injected into the tail vein when accompanied by an optic nerve injury.
HAPI microglia 4 days after injection in the tail vein and ONC were located at the crush site but were fewer in number than in the 7 day treatment.At 7 days after HAPI cell injection in the tail vein accompanied by an ONC there were transplanted microglia at the crush site and they were observed near the periphery of the nerve.The most transplanted HAPI microglia were seen at 7 days after injury.By 14 days, the microglia were more scattered throughout the optic nerve.The pattern of microglial cell migration and number coincides with the observation that the greatest amount of death of RGCs was seen 7 days after tail vein injection accompanied by an ONC.Most RGC death also occurred at 7 days following injection of the microglia into the vitreous when there was an ONC.This suggests that transplanted HAPI microglia are activated by day 4 but between day 4 and 7 arrive at the injury site and start to exert their effect.

Loss of RGCs after injection of untreated HAPI cells
The only significant loss of RGCs when HAPI microglia were injected into the vitreous or tail vein in the absence of injury was seen at the 14 day time point (Figure 6).There were no HAPI cells physically located in the retina or at the optic nerve at 14 days without injury.It is possible that this death was due to the cytotoxic effect of dying HAPI cells.However, there was no loss of RGCs when a different glial cell line (rMC-1) was injected into the tail vein or vitreous.This suggests that the injury may be due to secreted factors released from the microglial cells instead of cell death or introduction of exogenous cells.Microglia may produce factors that are capable of inducing cell death, such as reactive oxygen species and pro-inflammatory cytokines [20][21][22][23].Inhibition of these factors may contribute to the survival of neurons, which is often seen with the beneficial effects of NSAIDs on neurodegenerative diseases [23 , 70 , 71 ].
The role of microglia after injury or neurodegeneration is debated.Some argue that microglia do not play a major role in the degeneration or regeneration of axons after injury and that ablation of these cells did not result in changes in RGC death after injury [ 72 ].Others show that ablation of microglia can worsen the progression of injury and that microglia are required for some beneficial role [ 73 ].They can be neuroprotective and have reparative effects or be pro-inflammatory and execute the death of neuronal cells [23 , 74 ].Microglia play a role in killing dopaminergic cells in Parkinson's disease and may contribute to the neurodegeneration seen in mouse models of Amyotrophic lateral sclerosis [ 75 , 76 ].When HAPI microglia were injected into the vitreous of an eye and then the optic nerve was crushed there was a loss of 17% more RGCs at 4 days and 40% more RGCs at 7 days than that would be expected after ONC alone at those time points.At 14 days, there was no additional loss of RGCs.The majority of the death of RGCs was seen between 4 and 7 days.In addition, the density of RGCs at 4 days after intravitreal injection of microglia and ONC was comparable to the density of cells at 7 days after ONC (Figures 3 and 6).The RGC density at 7 and 14 days after intravitreal injection of microglia and ONC was comparable to the density of cells at 14 days (Figures 3 and 6).Initially, HAPI microglial cells injected into the vitreous accelerated the death of RGCs but it leveled off after 14 days.Previous study of endogenous phagocytic microglial dynamics following ONC and ONT has demonstrated that phagocytic microglia begin to clear RGCs 3 days after death and that microglial populations peak at 14 days post injury, partly due to mass migration of microglia from the inner plexiform layer [ 77 ].It is therefore possible that the comparable RGC numbers 14 days post ONC in the treated and non-treated group could be due to the migration and proliferation of endogenous microglia and that injected microglia contribute to the removal of dead and dying cells but do not have a direct or indirect effect on RGCs not destined to die after 14 days.
When HAPI microglia were injected into the tail vein and there was an ONC injury, there was no additional loss of RGCs at 4 days.However, there was a loss of 20% more RGCs at 7 days and a loss of 25% more RGCs at 14 days than would be expected after ONC alone at those time points.This suggests that it takes longer for the loss of RGCs to accelerate when the cells are injected into the systemic system.It takes 7 days for the additional loss of RGCs to begin and continues to gradually accelerate over a 14 day period.The density of RGCs may level off at time points beyond 14 days similar to that seen in the intravitreal injections.This timecourse corresponds well to what is observed for peripheral immune cell migration (such as macrophages) from the blood to the spinal cord following spinal cord injury, which peaks at 1-4 weeks following SCI [ 78 ].However, the death of RGCs was less severe in the tail vein condition than the intravitreal condition (Figure 6C).This may suggest that secreted factors and phagocytosis together mediate the role of RGC death by microglia after injury.This biphasic rate of RGC loss was also observed in other models, such as in laser-induced chronic ocular hypertension.It was noted that there is initially a rapid loss of RGCs of 12% per week for the first 3 weeks followed by a slower loss of 2% per week [ 79 ].Recently, optic nerve axotomy and optic nerve hypertention models have shown to result in biphasic RGC death with an initial fast phase followed by a prolonged secondary phase that could last over 400 days [ 80 , 81 ].It is possible that there is more local microglial involvement during the fast phase of loss and then more involvement of systemic macrophages during the slow phase of loss.
The migration of microglia and immune cells to the injury site was correlated with the loss of RGCs in the retina.At 4 days, after LPS activated HAPI cells were injected into the tail vein accompanied by an ONC, there was greater cell survival than when there was ONC alone or HAPI cells were injected with ONC but without LPS activation.However, 7 days after injury, the RGC density returned to expected values.This suggests that when hyper-activated HAPI cells were injected into the tail vein, they were beneficial to neuronal survival early in the injury.This finding agrees with some of the results from studies done with spinal cord injury models.It has been suggested that the microglial response after injury is required for repair early in the injury, however, the beneficial effect of microglia was lost later in the injury [ 86 ].Also, the recruitment of monocyte derived cells was necessary to terminate the microglial response [86].Perhaps the hyper-activated state is less prone to modulation by other cell types, resulting in a more deleterious effect later in injury.
The amount of RGC that survived after hyper-activated HAPI cells were grafted into the vitreous contrasted the survival of RGCs after tail vein injection.LPS activated HAPI cells were seen in the vitreous and the ganglion cell layer with or without injury after 7 days.In both cases, there was increased RGC loss.When LPS activated HAPI cells were injected into the vitreous without injury, there was a 37-42% loss in RGC compared to controls.This was different from non-activated cells that demonstrated no loss of RGCs after HAPI cells were injected into the vitreous within the first 14 days.This suggests that LPS activation resulted in a pro-inflammatory response, even in the absence of neuronal injury, resulting in the loss of RGCs when hyper-activated microglia are present in the vitreous.The anatomical location (cell body/ optic nerve or local/systemic) where the microglia act may affect neuronal survival.A similar finding was observed over several weeks after middle cerebral artery occlusion where there was a different pattern of microglial activation in the subventricular zone (SVZ) compared to the striatum [ 87 ].The microglia in the SVZ were more ramified and in a resting state, while in the striatum they were more amoeboid and in an activated state [87].Therefore, the properties of the different anatomical areas may modulate the response of the microglial cells.
We also observed increased RGC loss after injection of LPS activated HAPI cells into the vitreous accompanied by an ONC.The RGC density was 42-69% lower than that observed after ONC alone over the 14 day period.When compared to the RGC density after HAPI cells were injected into the vitreous, there were 63% less RGCs at day 4 and 50% less at day 14.Therefore, LPS induced hyper-activity in HAPI cells resulted in a pro-inflammatory response that is detrimental to RGC survival when injected into the vitreous.This was similar to other findings that showed that LPS activation of microglial cells led to a strong pro-inflammatory response due to the release of cytokines and superoxides [39 , 88-91 ].However, our results from the tail vein injections show that LPS activated HAPI cells are not always cytotoxic to RGCs and may help promote survival early in the injury.It has been shown that repeated low-dose exposures to LPS can lead to neuroprotection after CNS injury and resulted in less inflammation [88 ].It has been very difficult to determine what the best microglial response is after injury [ 92 ].The M1 activation state is thought to be neurotoxic but may help promote axonal regeneration, while the M2 state is hard to maintain in an injured environment [92].However, it seems clear that attempting to promote the M2 phenotype and controlling the time and location of action by microglia would help tip the balance to a neuroprotective outcome.
Minocycline may also protect nigral neurons from toxicity by MPTP and 6-OHDA, which used to induce Parkinson's in mice, by modulating microgliosis [93][94][95].However, the protective effect of minocycline has been questioned with reports that minocycline failed to be protective against MPTP in primate models with greater loss of striatal dopaminergic terminals and worse motor score [ 96 , 97 ].Clinical trials using minocycline have been mixed for neurodegenerative disorders such as multiple sclerosis and Alzheimer's Disease [ 98 , 99 ].We here show that the protective effects of minocycline modulated microglia depends on the localization they are injected and where they respond to the injury.There were 56-59% less RGCs than expected from ONC alone when minocycline treated HAPI cells were injected into the tail vein followed by an ONC injury.When the HAPI cells are injected into the tail vein with ONC, the cells migrate to the crush site in the optic nerve.This means that inhibiting M1 microglia from acting at the optic nerve results in a neurotoxic environment that leads to RGC death.This is similar to what is seen in glial scarring in spinal cord injury where the inhibitory extracellular matrix prevents microglia from acting on the axon [ 100 , 101 ].Axonal growth and sprouting is facilitated by using Chondroitinase ABC to digest the eCM and allow microglia to have access to the axon [100,101].Therefore, microglia activation at the axon may be an important in the survival and recovery of injured neurons.

Limitations and future directions
Although HAPI cells have similar characteristics to primary microglia in phagocytic activity, pro-inflammatory cytokine expression, and marker expression, differences between primary cells and different cell lines still exist [38,48].Some benefits for using HAPI cells instead of primary microglia were that they do not require fresh isolation for each procedure which also helps to reduce animal to animal variations.Cells can be thawed, cultured for pre-treatment, and injected in a short period of time.They are also well characterized and been used in many studies to investigate immune modulatory therapies, toxicology, and oxidative stress responses [39][40][41][42][43][44][45][46][47].In addition, compared to BV-2 cells which are immortalized by retroviruses, HAPI cells are spontaneously occurring and isolated from primary microglial cultures [38 , 102 ].The recently development of microglia from induced pluripotent stem cells has opened up a new avenue into investigating human microglia, future studies can explore human microglia behaviors to pretreatment to better understand human disease progressing and treatment [103][104][105].This study provides the base observations and methods that would be necessary for future investigations into microglial interactions, therapeutic time points, cell therapies, and microglial activation modifications.

Conclusion
These experiments describe a novel method of investigation for the role of microglial activation in neurodegenerative disease.We demonstrate that injected HAPI microglia are detrimental after injury by accelerating the loss of RGCs and that paradigms are need to modulate the activity of these cells.HAPI cells treated with minocycline were neuroprotective when acting on the cell body of the RGC early in the injury but not when acting on the axon.The opposite is true for hyper activated (M1) HAPI cells where they are neuroprotective early in the injury by acting at the axon but not when at acting on the cell body.One thing is clear in these studies: long term activation or inhibition of microglia ultimately results in a pro-inflammatory response that exacerbates the loss of neurons.This reveals that any therapeutic intervention targeting microglia to treat neurodegeneration or CNS injury should be early in the injury and targeted to a specific location.This study demonstrated that microglia can be neuroprotective but this is dependent of the time course and location of the injury.These finding needs to be taken into consideration when developing a therapeutic intervention that modulates microglia.

Figure 1 .
Figure 1. 12 µm thick frozen transverse sections of retinas and optic nerve labelled with fluorogold (labelling rGcs in gold).(a) retina with no optic nerve crush (onc).(B) optic nerve with no onc.(c) optic nerve 4 days after optic nerve crush showing the injury 2 mm from the sclera.(D) retina 4 days after onc, (e) retina 7 days after onc, (f) retina 14 days after onc.red auto-fluorescent cells (arrow heads) are negligible in the absence of cell injection.(G)there is a gradual loss of retinal ganglion cells over time after optic nerve crush.control retinas had 54.74 rGcs/mm, 4 days after onc there were 44.90 rGc/mm, 7 days after onc there were 34.12 rGcs/mm, and 14 days of onc there were 18.9 rGcs/mm remaining.

Figure 2 .
Figure 2.12 µm thick frozen transverse sections of retinas and optic nerve labelled with fluorogold (labelling rGc somas in gold) that had received intravitreal injections of HaPI microglial cells (labelled with WGa-tr; red).(a) 4 days after injection of microglia and onc.transplanted microglia were seen near the vitreal surface of the retina (white arrows).there was a loss of retinal ganglion cells comparable to 4 day onc retina (1e).(B) 7 days after microglia injection and onc.transplanted microglia were seen near the vitreal surface of the retina (white arrows).there was a loss of retinal ganglion cells comparable to (1f).(c) 14 days after microglia injection and onc.fewer microglial cells were consistently seen in the retina (white arrows) at this time point.there was a loss of retinal ganglion cells comparable to (1 f).(D) 4 days after microglia were injected.(e) 7 days after microglia were injected.(f) 14 days after microglia were injected.only at 14 days there was a statistically significant (anoVa followed by tukey's post hoc test) loss of retinal ganglion cells compared to the control.(G) 4 days after microglia injection and onc.(H) 7 days after microglia injection and onc.(I) 14 days after injection of microglia and onc.(J) 4 days after injection of microglia.(K) 7 days after microglial cell injection.(l) 14 days after microglial cell injection.there were no transplanted microglia present in, or near, the optic nerve after intravitreal injections at any time point.(M) rGc survival 14 days after intravitreal injection of microglial cells.Graph showing rGc survival after intravitreal injection of microglia.there was no significant change from control (black line) when microglia were injected without onc (red) until 14 days after injection.the rGc density was lower after microglia were injected with onc (black triangles) when compared to onc alone during the first 7 days.14 days after injection and onc there was no significant difference from onc alone.error bars (or dotted lines): 95% confidence interval.anoVa statistically analysis and tukey's post-hoc test were performed.* compared to onc; & compared to control; # comparison between intravitreal and tail vein injections.

Figure 4 .
Figure 4. Brn3a immunoreactive rGcs survival after intravitreal injection of lPs activated HaPI cells in 12 μm thick frozen transverse sections of rat retinas and optic nerve.(a) retinas collected 4 days after lPs activated HaPI cells were injected.(B) retinas collected 7 days after lPs activated HaPI cells were injected.(c) retinas collected 14 days after lPs activated HaPI cells were injected.(D) retinas collected 4 days after lPs activated HaPI cells were injected and onc.(e) retinas collected 7 days after lPs activated HaPI cells were injected and onc.(f) retinas collected 14 days after lPs activated HaPI cells were injected.HaPI cells (arrows) can be seen in both crushed and not crushed conditions 7 days after injection and onward.(G) 4 days after injection but in the absence of onc there were no HaPI cells present in the optic nerve.(H) 7 days after injection but in the absence of onc there were no HaPI cells present in the optic nerve.(I) 14 days after injection but in the absence of onc there were no HaPI cells present in the optic nerve.(J) 4 days after injection and with onc there were HaPI cells and auto-fluorescent immune cells present in the optic nerve at the crush site.(K) 7 days after injection and with onc there were HaPI cells and auto-fluorescent immune cells present in the optic nerve at the crush site radiating outward.(l) 14 days after injection and with onc there were HaPI cells and auto-fluorescent immune cells present in the optic nerve at the crush site radiating outward.(M) HaPI cells injected into the vitreous with or without lPs activation.rGc density after injection of lPs activated HaPI cells (red circles) were not statistically different from the rGc density 7 days after onc (grey dotted line).When HaPI cells were injected without lPs activation (blue squares), the rGc density was not statistically different than control except at 14 days.Injection of hyper-activated HaPI cells leads to greater cell death.(n) HaPI cells injected into the vitreous with or without lPs activation after onc.rGc density after injection of lPs activated HaPI cells and onc red circles) were lower than the rGc density after onc alone (grey dotted line).When HaPI cells were injected into the vitreous without lPs activation and with onc (blue squares), there was greater rGc survival than when lPs activated HaPI cells are injected with onc (except for at 7 days after injection).(o) rGc density after injection of lPs activated HaPI cells and onc was lower than when lPs HaPI cells were injected without onc.Injection of lPs activated HaPI cells into the vitreous led to increase rGc death.error bars = 95% confidence interval.anoVa/tukey's post-hoc test were performed.# compared to onc; & compared to control; * comparison between lPs treated and untreated.

Figure 5 .
Figure 5. Brn3a immunoreactive rGcs (green) survival after tail vein injection of lPs activated HaPI cells labelled with WGa-tr (red) in 12 μm thick frozen transverse sections of rat retinas and optic nerve.(a) retinas collected 4 days after lPs activated HaPI cells were injected.(B) retinas collected 7 days after lPs activated HaPI cells were injected.(c) retinas collected 14 days after lPs activated HaPI cells were injected.(D) retinas collected 4 days after lPs activated HaPI cells were injected and onc.(e) retinas collected 7 days after lPs activated HaPI cells were injected and onc.(f) retinas collected 14 days after lPs activated HaPI cells were injected.no HaPI cells were seen in retinal sections after tail vein injection.(G) 4 days after injection but in the absence of onc there were no HaPI cells present in the optic nerve.(H) 7 days after injection but in the absence of onc there were no HaPI cells present in the optic nerve.(I)14 days after injection but in the absence of onc there were no HaPI cells present in the optic nerve.(J) 4 days after injection and with onc there were HaPI cells present in the optic nerve at the crush site.(K) 7 days after injection and with onc there were HaPI cells present in the optic nerve at the crush site radiating outward.(l) 14 days after injection and with onc there were HaPI cells present in the optic nerve at the crush site radiating outward.(M) rGc survival after injection of untreated HaPI cells into the tail vein (blue square) was not statistically different from the control except at 14 days.rGc survival after injection of lPs activated HaPI cells into the vitreous (red circles) was not statistically different from the control.the rGc density was not different across the 14 day period.When HaPI cells are activated with lPs without onc there was a statistically greater survival of rGcs 14 days after injection as compared to rGc survival after untreated HaPI cell injection.(n) rGc density 4 days after injection of lPs activated HaPI cells with onc (red circle) was not different from control (solid grey line).However, 7 days after onc and injection of lPs activated HaPI cells, the rGc density was lower than the rGc density 7 days after just onc.therefore, there was greater survival of rGcs early in the injury and less late in the injury.(o) 7 days after injection of lPs activated HaPI cells into the tail vein with onc (blue square) there was greater cell death than when lPs activated HaPI cells were injected without onc (red circle).error bars = 95% confidence interval.anoVa/tukey's post-hoc test were performed.# compared to onc; & compared to control; * comparison between lPs treated and untreated.

Figure 6 .
Figure 6. 12 μm thick frozen transverse section of rat retina and optic nerve after intravitreal injection of minocycline treated HaPI cells for 4-14 days.(a) 4 days after injection of the cells there were no HaPI cells (red) present in the retina or vitreous.(B) 7 days after injection of the cells there were no HaPI cells present in the retina or vitreous.(c) 4 days after injection of the cells there were no HaPI cells present in the retina or vitreous.(D) similarly, 4 days after injection and onc there were no HaPI cells present in the retina.(e)7 days after injection and onc there were a small number of HaPI cells present in the retina (white arrows).(f) 14 days after injection with onc the retina was very inflamed that even the retina layers were not distinguishable.there were many HaPI cells and may be other auto-fluorescent immune cells present in the retina.the retinal ganglion cells were labeled with Brn3a (green).(G-I) there were no HaPI cells in the optic nerve 4-14 days after the injection of minocycline treated HaPI cells whether there was no injury or (J-l) optic nerve crush injury.(M) the rGc survival 4-14 days after intravitreal injection of minocycline treated HaPI cells.the rGc density was not different from controls (grey line) when minocycline treated HaPI cells were injected into the vitreous without onc (red squares).the rGc density did not differ from when untreated HaPI cells were injected into the vitreous without onc (blue circles) at 4 and 7 days, however, the rGc density was greater than the rGc density 14 days after untreated HaPI cells were injected into the tail vein without onc.(n) Minocycline treated HaPI cells may be neuroprotective early in the injury after intravitreal injection and cytotoxic later in the injury.since the indigenous microglia were not inhibited, we would still expect the loss that is seen after onc. the rGc density was not different from the density 4 days after untreated HaPI cells were injected into the vitreous with onc. the rGc survival was greater at 7 days after injury but there was greater rGc loss 14 days after injury.(o) the rGc survival trend after injection of minocycline treated HaPI cells into the vitreous.the rGc density did not change from values for the controls (grey solid line) when minocycline treated HaPI cells were injected without onc (red square).the rGc density 4 and 7 days after injection were not different than the values expected after onc alone (grey dotted line).However, 14 days after there were much fewer surviving rGcs than there would be after onc alone at the same time point.error bars = 95% confidence interval.anoVa/tukey's post-hoc test were performed.# compared to onc; & compared to control; * comparison between minocycline treated and untreated.

Figure 7 .
Figure 7. 12 μm thick frozen transverse section of rat retina and optic nerve after tail vein injection of minocycline treated HaPI cells for 4-14 days.(a) 4 days after injection of the cells into the tail vein, there were no HaPI cells (red) present in the retina or vitreous.(B) 7 days after injection of the cells into the tail vein, there were no HaPI cells present in the retina or vitreous.(c) similar to the other time points, there were no HaPI cells in the retina 14 days after injection without onc.the retinal ganglion cells were labeled with Brn3a (green).(D) 4 days after injection and onc of cells into the tail vein, there were no HaPI cells present in the retina.(e) this trend continues 7 days after injection of cells into the tail vein and onc where there were no HaPI cells present in the retina.(f) 14 days after injection with onc.(G) 4 days after injection but without onc there were no HaPI cells present in the optic nerve.(H) 7 days after injection but without onc there were no HaPI cells present in the optic nerve.(I) 14 days after injection but in the absence of onc there were no HaPI cells in the optic nerve.(J) 4 days after injection and with onc there were HaPI cells present in the optic nerve at the crush site.(K) 7 days after injection and with onc there were HaPI cells in the optic nerve at the crush site radiating outward.(l) 14 days after injection and with onc there were HaPI cells in the optic nerve at the crush site.(M) the rGc density 4-14 days after injection of minocycline treated HaPI cells into the rat tail vein.the rGc density was not different from the control (grey solid line) or from when untreated HaPI cells were injected into the tail vein without injury (HaPI tV; blue circle) when minocycline treated HaPI cells were injected into the tail vein without onc (Mino HaPI tV; red square).(n) When minocycline treated HaPI cells were injected into the tail vein with onc (Mino HaPI tV onc; red square), the rGc loss was greater than what was expected after onc alone (grey dotted line) or when untreated HaPI cells were injected into the tail vein with onc (HaPI tV onc; blue circles).(o) Injection of minocycline treated HaPI cells into the tail vein with onc (blue circle) was cytotoxic to rGc, however, when the cells were injected without onc (red square) there was no effect on rGc survival.error bars (or dotted lines) = 95% confidence interval.anoVa/tukey's post-hoc test were performed.Gcl: ganglion cell layer; Inl: inner nuclear layer; onl: outer nuclear layer.# compared to onc; & compared to control; * comparison between minocycline treated and untreated.

Figure 8 .
Figure 8. rGc survival 14 days after intravitreal or tail vein injection of rMc-1 cells.there was no significant rGc loss when rMc-1 cells were injected into the vitreous (blue) or tail vein (red).error bars = 95% confidence interval.the mean number of rGcs/mm in uninjured, uninjected retinas is shown for reference (grey mean line; +/− 95% confidence interval hatched line).anoVa statistically analysis and tukey's post-hoc test were performed.(a) retinal micrographs were similar to control after rMc-1 injection into the vitreous.(B) retinal micrographs were similar to control after rMc-1 injection into the tail vein.(c) retinal micrograph from control retinas (uninjured, uninjected).rGcs are labeled with Brn3b (green).

Table 1 .
overview of the animal groups and summary of the results.