Retinal And Brain Microglia in Multiple Sclerosis And Neurodegeneration Part 2

3.3. Ocular Manifestations of Multiple Sclerosis

Post-mortem and retinal imaging studies have revealed MS-related pathologies such as optic nerve damage by demyelination and atrophy and GCL, IPL, and RNFL thinning (Figure 2) where the RGCs and microglia can be found [79,80]. It is theorized that retinal thinning is a downstream effect of the optic nerve and RGC degeneration following microglia-mediated immune responses [79,81]. 

In modern society, optic nerve damage is a common and worrying problem. The optic nerve is the nerve that connects the eyes to the brain, and if damaged, we can face a range of difficulties and challenges, such as loss of vision and memory loss.

However, we should not lose heart in the face of challenges. Optic nerve damage did not directly affect memory. Memory is a function of the brain, and the optic nerve is just a channel for transmitting signals. If we actively keep our brains healthy, we can maintain good memory.

There are many ways to promote brain health. For example, by exercising, we can increase blood circulation to the brain, thus improving the health of the brain. In addition, we can also maintain a proper diet and increase social activities, these methods can effectively maintain the healthy state of the brain.

In addition, in the face of optic nerve damage, we can also take some positive coping methods. For example, we can find appropriate treatments that will help us restore our vision. We can also develop some new hobbies and activities to keep our memory and brain healthy.

In summary, while optic nerve damage can present many challenges, we should not lose faith. By actively maintaining a healthy brain, we can maintain a good memory and better cope with life's challenges. This shows that we need to improve our memory. Cistanche can improve our memory, because Cistanche can also regulate the balance of neurotransmitters, such as increasing the level of acetylcholine and growth factors, which are very important for memory and learning In addition, meat can also improve blood flow and promote oxygen delivery, which can ensure that the brain receives sufficient nutrition and energy, thereby improving the vitality and endurance of the brain.

Click know ways to improve brain function

Interestingly, these hallmarks are symptomatically present in more than 20% of pMS and asymptomatically present in more than 68% of pMS, making almost a staggering 90% of pMS who are estimated to have such retinal pathologies. The typical symptomatic features consist of photopsia, monocular loss of visual acuity, and color vision with mild eye pain, whilst some atypical features consist of photophobia, binocular prolonged vision loss, and severe or absence of pain [62,67,82]. These symptoms may develop from a couple of hours to days, peaking at roughly 2–3 weeks from the initial signs followed by recovery [83].

Clinicians have been encouraged to use neuroimaging techniques such as magnetic resonance imaging (MRI) to confirm MS-associated diagnoses [63,79]. Such neuroimaging methods can be used to image the entire CNS, including the visual pathway. 

For example, an ON-patient MRI showed ventricular brain enlargement and atypical thickening of the optic nerve, which were indicative of brain atrophy and optic nerve demyelination [84]. Optical coherence tomography (OCT) and motion perception assessments, both of which examine the visual system, may also be used as adjuncts to MRIs [85]. The majority of neuroimaging techniques may be complex, low reproducibility, uncomfortable, and extremely costly (e.g., MRI) [63,79,80]. In contrast, OCT retinal imaging offers a noninvasive, simpler, and less expensive technique to provide high-resolution images with detailed analysis of distinct retinal layers [63,79,80,83]. 

Whilst sophisticated MRI techniques can be used to identify dynamic inflammation, typically seen in earlier stages of MS, standard MRI parameters can be correlated to OCT evidence indicative of irreversible neurodegeneration and, therefore, significant reductions to retinal thickness [79]. Despite the shortcomings of both methods, each provides unique information relating to distinct aspects of MS pathogenesis. This emphasizes that both MRI and OCT should be used in conjunction to maximize vital information extraction, enabling in-depth monitoring of MS-associated atypical pathologies and making MS trajectory predictions.

3.4. Microglia and Multiple Sclerosis

Like other neurodegenerative disorders, changes in microglial activation have also been observed both in models of MS and in pMS. Experimental autoimmune encephalomyelitis (EAE) is often used in scientific investigations to model MS and ON [48]. Although all experimental models of diseases are not fully representative, studies have shown similarities between human and animal investigations. For instance, compared to healthy, EAE-induced mice were found to have more “de-ramified” amoeboid microglia in the optic nerve and spinal cord [86]. Horstmann et al. found in MS-modeled mice by injection of MOG35-55 (encephalitogenic peptide which induces EAE) that retinal microglia were significantly more abundant compared to that of healthy mice injected with PBS [87]. The same investigation also found that, at 60 days post-immunization, the MS-induced mice had non-significantly but slightly more amoeboid microglia but significantly more ramified microglia compared to the control mice [87]. Another investigation also used MOG35-55 to induce EAE in mice [88]. '

Within the first week of immunization, retinal microglia density had increased significantly [88]. In this investigation, amoeboid microglia were also found to be more abundant at 7 and 28 days post-MOG35-55-immunisation; however, no statistical analyses were carried out. These variations of microglial morphotypes indicate that retinal microglia may behave dynamically, becoming “activated” and “de-activated” throughout the progress of EAE. This could correlate to changes in microglial activation, which may occur during fluctuations of demyelination and remyelination seen in MS. Additionally, more sophisticated retinal layer-specific analyses were performed on EAE-induced and healthy mice to find layer-specific microglial morphotypes [86]. 

Whilst both EAE and healthy mice consisted of ramified microglia in the IPL and OPL, amoeboid microglia were only found in the GCL layer of EAE-induced mice [86]. This interestingly correlates to the GCL and IPL thinning hallmarks of MS, although it has not yet been investigated if this thinning pathology is directly linked to microglial activation. Ultimately, this suggests that retinal microglia may respond and activate in a temporally specific and layer-specific manner [32,86].

Similar implications can also be seen in human studies whereby there was a significantly reduced expression of P2RY12 (homeostatic microglia marker) and significantly increased expression of CD68 in both normal-appearing white matter and areas of active lesions of MS brains compared to healthy brains [89]. Since these markers have been associated with specific functions and microglia morphotypes, the results from Zrzavy et al. suggest that MS brains have fewer surveillant ramified microglia whilst having more phagocytic amoeboid [32,36,89].

Myelin phagocytosis is an additional pathological feature of MS [51]. When Hendrickx et al. investigated the phagocytic capabilities of myelin by microglia through flow cytometry, microglia derived from normal-appearing white matter of post-mortem brains seemed to engulf more myelin from MS donors compared to that from healthy donors [51]. Furthermore, myelin debris processing has been correlated to microglia taking on other “dystrophic” morphologies [51,90]. Cuprizone (CPZ) is a toxic and systemic copper chelator that can be administered intravenously and orally to model MS and demyelinating diseases [90,91]. 

Cantoni et al. investigated demyelination and microglia characteristics in healthy or CPZ administered [90]. Histological observations revealed that healthy animals were unaffected by their myelin integrity whilst most microglia appeared ramified. Conversely, CPZ-administered animals exhibited hippocampally and corpus callosum (CC) neuron demyelination that became more “profoundly” evident with longer exposure to CPZ [90]. Interestingly, amoeboid microglia were abundant up to 4 weeks of CPZ administration, but as this increased to 6 or 12 weeks, more ramified microglia were observed [90]. This suggests that microglial activation plays a more active role in the earlier stages of demyelination or demyelinating diseases such as MS. This is further supported by clinical evidence showing that younger MS patients in earlier stages of the disease are more responsive to therapeutic interventions such as Rituximab that can reduce microglial activation [92,93]. Conversely, transcriptional investigations in microglia from normal-appearing white (NAWM) and grey matter (NAGM) of MS and healthy volunteers showed some interesting differences [94]. 

Firstly, the expression of surveillant microglia markers such as P2RY12 was found to be similar in NAGM and NAWM [94]. Secondly, compared to the NAGM, the NAWM had higher levels of genes involved in the NF-κB pathway, which is involved in the immunomodulatory responses in MS pathology [94]. Unlike the findings from the CPZ MS models [90], van der Poel et al. [94] suggest that surveillant properties of microglia may remain unaffected in the initial stages of MS whilst the transcriptional profile of human-derived microglia is region-specific. Although evidence from human studies would provide a valuable translational comparison, there are, to date, no reports that investigate the morphological characteristics of retinal microglia in patients with MS.

3.5. Immunological Markers

Retinal microglia are also unsurprisingly involved in uveitis, an infrequent manifestation of MS. Experimental autoimmune uveitis (EAU) is a commonly used model of uveitis, which can be induced by immunization of interphotoreceptor retinoid-binding protein (IRBP) [95]. Whole-mounted retinas of EAU-induced mice were found with P2RY12 positive microglia that underwent morphological activation from ramified to amoeboid morphology within 7 days post-IRBP-immunisation, indicating the early involvement of retinal microglia in EAU pathology [95]. Interestingly, ablation of microglia with the antagonist PLX5622 in IRBP-immunised was found to suppress EAU. This was observed by a decrease in infiltrating inflammatory cells, lack of photoreceptor folds, and retinal granulomas compared to untreated animals [95]. Additionally, three-dimensional images were constructed from images of renovascular leukocytes, including T-cells and microglia, labeled with CD11b, CD4/CD8, MHCII, and P2RY12 [95]. The expression of all markers could be found co-localized with lectin, a marker of blood vessels. Together, these results emphasize the active role of microglia activation in initiating the transport of leukocytes between the blood-retinal barrier [95]. Whether the morphological shift had a direct correlation to EAU pathogenesis was not explored.

3.6. Extracellular Vesicles: Effects on Multiple Sclerosis

The EAE model and MS patients have been investigated in parallel to observe microglia-derived microvesicle (M-MV) EVs in MS/EAE brain pathology [96]. Firstly, M-MVs were found to be neuroinflammatory as evidenced by the upregulation of inflammatory markers such as CD86 (T-cell receptor ligand), iNOs, and IL1ß, by the transfer of the vesicular components from the origin microglial cell to the recipient [96]. Additionally, EAE mice injected with M-MV into the corpus callosum had many amoeboid microglia, which were not observable in control EAE mice (injected with saline, liposomes, or mesenchymal stem cells). Mice with and without acid sphingomyelinase knockout (ASKO) were induced with EAE [96]. ASKO-EAE mice showed significantly reduced M-MVs in the CSF, and CD45lowCD11b+ microglia and inflammatory T-cells in the spinal cord, whilst no ASKO-EAE mice had developed EAE pathologies [96]. In humans, healthy participants were found to have M-MVs in their CSF with increased concentrations in MS: significantly greater in CIS compared to RRMS. These results suggest that M-MVs are differentially present in MS patients and EAE models, which may cause microglial activation and morphotype changes whilst these responses may be reduced by preventing M-MV shedding [96]. This also reveals the potential of targeting M-MVs to monitor disease and its therapeutic strategies. Their role in the retina is still to be established, although if they are found to influence retinal microglia morphology, this could have significant implications for using the eye as a monitoring and treatment efficacy tool.

3.7. Genetic Factors: Effects on Retinal Microglia in Multiple Sclerosis

The MOG35-55-induced EAE model of MS was used to identify the genetic and morphological differences of retinal microglia between 8-week-old C57BL/6 mice and healthy controls [97]. Iba-1 staining of cross-sections of the retina revealed mostly ramified microglia in healthy controls whilst there were rounder and amoeboid-like microglia with higher expression of iNOS in early EAE mice (16 days post-MOG35-55 induction— 16DPI) [97]. This iNOS expression was reduced by 41DPI, although still higher than that in healthy controls. Additionally, there was increased expression of C1q and TNF-alpha in EAE [97]. This could suggest that early transcriptional changes in retinal microglia reflected by morphological changes may be pathological features of optic neuropathies.

Additionally, Bell et al. investigated the retinal microglia transcriptional profile of Cx3cr1CreER: R26-tdTomato C57BL/6 mice that were induced with endotoxin-induced uveitis (EIU) by a single intravitreal injection of LPS [98]. Isolated retinal microglia were analyzed with flow cytometry, Fluorescence-Activated Cell Sorting (FACS), mRNA sequencing, and quantitative PCR to discover a total of 1069 differentially expressed genes (DEGs) [98]. There were 613 DEGs found at 4 h post-EIU, 537 DEGS at 18 h, and none at 2 weeks [98]. Initially, activation markers such as Fas and CD44 were upregulated whilst homeostatic markers such as P2rY12 and Mertk were downregulated in EIU-induced samples whilst healthy controls did not show a similar profile, e.g., high P2RY12 levels throughout [98]. Additionally, confocal microscopy of tdTomato+ microglial cells in naïve was ramified whilst EIU cells underwent morphological changes from ramified at 4 hours to amoeboid at 18 h, which returned to ramified by 2 weeks post-injection [98]. Together these results indicate that uveitis may be characterized by some initial transcriptional changes that may be reflected by retinal microglia morphology changes [98]. Similar investigations in patient cells would be useful to determine the clinical translatability of these findings.

4. Other Neurodegenerative Diseases

4.1. Alzheimer’s Disease

AD is the most common cause of dementia. It is characterized by the progressive loss of neurons as a result of the build-up of extracellular amyloid-beta (Aß) plaques and intracellular neurofibrillary tangles composed of hyperphosphorylated tau proteins [99–101]. By the time the first cognitive symptoms start to appear, roughly 20 years of irreversible neurodegeneration would already have occurred [100]. Although the exact pathogenesis is still unknown, microglial activation in the brain and retina has been associated with AD pathology [4,13,21,99–103]. This highlights the unmet need for earlier detection of AD pathogenesis and that characterization of microglia may be a potential answer to this dilemma.

Brain AD investigations have shown more activated microglia in the cortex and hippocampus compared to age-matched controls [99,103]. Specifically, Cao et al. found a significantly greater degree of microglial activation in these areas of AD-modelled mice at 12 months old compared to other ages and compared to age-matched or older healthy controls [99]. Additionally, 4-month-old 5× familial AD-modeled mice showed that Aß plaques were surrounded by activated microglia in the cortex [103]. These activated microglia were also found to co-localize with Aß aggregates, and as Aß accumulated, microglial cell death was observed [103]. Similarly, Aß plaques were found surrounded by retinal microglia with a large cell body and fewer and thicker ramifications, suggesting activated microglia observations could be seen in the AD-modeled retina [100–102]. 

Some sources have found activated retinal microglia in 3× transgenic AD mice as young as 5 postnatal weeks old, suggesting an initial microglial response in AD pathology [102]. This suggests that some microglial responses to AD may occur in the retina before the brain. Despite these findings, it is important to note the ambiguous definition of activated microglia. For instance, some literature refers to all non-resting microglia as “activated microglia”, and some may use the term interchangeably with “activation” of microglia, whilst more recent literature may use the term about the unique morphological description of the activated microglia.

Rod microglia may have a region-specific involvement in AD pathology [4,21]. For instance, when three brain regions of AD patient autopsies were examined, rod microglia were only seen in the parietal cortex and not in the hippocampus or temporal cortex [4]. Additionally, there were no significant observations seen relating rod microglia in hippocampal and cortical brain regions of patients who had traumatic brain injury [4]. However, rod microglia were found to be significantly abundant in the parietal cortex but not the hippocampus of AD patients compared to healthy controls [4]. Rod microglia have also been observed in those with mutated C9orf72 whilst they were absent in age-matched controls whilst additionally, a high density of rod microglia in the grey matter of those with Down’s Syndrome AD correlated to more severe pathological hallmarks compared to those just with AD [21].

Recently, Grimaldi et al. investigated the retinal pathology in post-mortem AD patient retinas [15]. These samples also had Aß plaques and phosphorylated tau tangles, with signs of retinal-neuronal degeneration evidenced by increased expression of caspase 3, a marker of AD-associated neurodegeneration [15]. Additionally, the AD retinal slices examined had a significantly higher density of iba-1 positive microglial cells which also expressed higher levels of DAM marker, IL-1ß, compared to healthy samples [15]. TREM2 is also a DAM-associated marker, although this was not upregulated in the iba-1-positive AD microglia [15]. Despite this, the AD retinas expressed higher levels of TREM2 mRNA compared to healthy controls [15]. This unexpected trend in AD retinal microglia may be due to historic evidence of post-mortem retinal samples with Aß deposits expressing downregulated TREM2 [15]. Investigations that observe other retinal DAM markers in parallel to morphological parameters may provide useful insight into elucidating AD retinal pathologies.

4.2. Parkinson’s Disease

Parkinson’s disease (PD) is the second most common type of neurodegenerative disease, affecting over 10 million people worldwide [104,105]. Its pathological hallmark consists of Lewy body (LB) aggregates composed of alpha-synuclein (α-syn) that accumulate inside the neuronal cell bodies and axons [104,105]. This is followed by the gradual neurodegeneration of dopaminergic neurons in CNS regions such as the substantial niagra (SN) pars compacta [6,106,107]. By the time the initial clinical symptoms such as tremor, rigidity, akinesia, and postural stability start to appear, 50–80% of dopaminergic neurons have already been lost [107,108]. Much like AD, PD research has revealed many potential mechanisms of pathogenesis, including neuroinflammation as a result of microglial activation [6,105,109]. Recent studies have also revealed that up to 80% of PD patients may experience some visual dysfunction at least once throughout their disease progression [105]. Additionally, some studies have reported these retinal changes to occur before that of the brain [110]. Considering these factors, exploring the retinal microglia characteristics may serve as an ideal starting point in identifying earlier biomarkers of PD.

One study investigated the characteristics of SN and hippocampal microglia in three participant groups: PD patients, matched healthy participants, and prodromal PD (p-PD) participants, with age-matched analyses performed (age range from 56 to 96 years old) [111]. Whilst there were almost no α-syn and LBs found in both brain regions of healthy controls, there were some in that the p-PD patient specimens and significantly more in PD [111]. Additionally, healthy controls had significantly more ramified microglia compared to p-PD and PD groups, p-PD had significantly more amoeboid microglia compared to healthy participants, and PD patients also had significantly more than p-PD [111]. “Primed” or activated microglia were prominently seen in p-PD compared to any other participant group. This suggests that activated microglia may be an early indicator whilst amoeboid microglia are associated with PD pathology.

Similarly, healthy retinas were free of α-syn whilst there were some in 5-month-old retinas of transgenic mice (TgM83) modeled for PD and significantly more in the OPL of 8-month-old TgM83 retinal samples [112]. Only 8-month-old mouse retinas indicated microglial activation due to significant increases in CD68 expression (microglial activation marker) and visual observation of amoeboid microglia, whilst this was not seen in other experimental groups [112]. This suggests that there may be a differential expression of activated microglial markers and morphological changes through the early to intermediate stages of PD pathology. However, other studies show that whilst there was OPL α-syn accumulation in mice modeled for α-syn-dependent diseases (Plp-α-syn), there were mostly ramified microglia found in whole-mounted retinal samples with no sign of microglial activation for all age groups investigated [113]. This may have been due to differences in disease models used, age groups investigated, and methods of analysis.

4.3. Glaucoma

Glaucoma is one of the main causes of blindness worldwide, which is estimated to affect more than 120 million people by 2040 [114]. It is characterized by the gradual loss of RGCs and, like AD and PD, significant neurodegeneration would have already occurred before clinical symptoms start to emerge [114]. Numerous methods for detecting the activity of glaucoma have been developed, including elevated intraocular pressure and an increased number of dying retinal cells, which are measured by detecting apoptotic retinal cells (DARC) [115,116]. Additionally, microglial activation may also play a role in glaucoma pathogenesis.

Microglia activation has been identified as one of the first events in glaucomatous neural damage, which may occur before RGC death [117,118]. For instance, when Swiss albino mice were modeled with the unilateral ocular hypertension (OHT) model of glaucoma, there was microglial activation [108]. This was seen in the form of increased cell body size and retraction of cell processes as early as 1-day post-surgery (PS) and increased cell density by 3 days PS. Ramírez et al. additionally found that retinal microglia were P2RY12+ at 1 day PS [108]. This expression was decreased at 3 and 5 days PS, which then returned to normal levels by 15 days PS [108]. Moreover, the extent of neurodegeneration could also be correlated with early microglial changes, including microgliosis or microglial activation, as evidenced in transgenic Cx3cr1GFP/+ DBA/2J mouse retinas modeled for pigmentary glaucoma [117]. In addition, in another experimental glaucoma model, treatment with minocycline or with a high dose of irradiation reduced microglial activation, resulting in lower RGC death [119,120].

Rod microglia have also been associated with several disease states. In a model of progressive RGC degeneration in rats, it has been shown that optic nerve transection (ONT) induced retinal rod microglia formation [23]. This was observable within the first 7 days, peaking at 14 days and disappearing after 2 months [23]. These periodic trends were interestingly reflected by the trends of ONT-induced RGC death. Additionally, these retinal rod microglia were found to orient themselves along the axon and cell body of RGCs, suggesting their active role in ONT-associated neurodegeneration [23].

4.4. Inherited Eye Disease: Retinitis Pigmentosa

Another major cause of inherited irreversible blindness is retinitis pigmentosa (RP). It affects roughly 1 in 4000 people in the world. Typically, patients may undergo loss of retinal rod cells, whilst more advanced cases may be caused by loss of cone cells. Although the exact mechanism of RP is still unclear, there is evidence of early retinal microglia activation occurring initially in the process of photoreceptor degeneration [121]. For instance, Di Pierdominico et al. observed P23H-1 rats, a model of inherited photoreceptor degeneration [121]. A reduction in rows of photoreceptor nuclei occurred between P10-21 with a greater reduction between P15-21 [121]. This occurred simultaneously with microglial morphology changes to amoeboid microglia, observed through retinal cross-section histology analyses showing that iba-1 positive microglia with shorter processes by P15 whilst the total retinal microglia density was significantly greater in P23H-1 rats compared to controls between P15-45 [121]. The same study investigated Royal College of Surgeons (RCS) rats, another model for inherited photoreceptor degeneration, to find a similar trend. The most significant photoreceptor degeneration occurred between P33-60 whilst retinal microglia took on the amoeboid appearance by P21-33 [121]. These microglia also migrated to the outer retinal layers between P21-60. RCS retinas had a significantly higher density of microglia compared to controls P21-60, whilst there was a significant decline between P45-60 [121].

5. Perivascular Microglia

Investigating microglia that reside in or around the perivascular region (PM) has become a popular field of research because of the associations that have been made between immune responses to neurological conditions and the CNS vascular network [122]. These associations are seen where patients with diabetic retinopathy had a significantly elevated number of hyper-ramified retinal PM microglia cells in comparison to that of normal subjects [34]. Additionally, PM microglia are said to closely track any incoming compounds from outside the blood-brain barrier through the vessels and into the CNS [95]. For instance, drug-induced depletion of PM microglia in mice with experimental AD resulted in cardiovascular accumulation of amyloid plaques [122]. Other investigations also found that glaucoma-modeled mice (ocular hypertension model induced by laser) presented with rod microglia both in the retinal PM and adjacent to the retinal axons [22]. This indicates that microglia morphotypes are not specific to perivascular regions.

6. Imaging Retinal Microglia

The establishment of a robust method to image retinal microglia both non-invasively and in vivo would provide an invaluable avenue for tracking disease-associated retinal microglia pathologies. There are indeed transgenic, e.g., CX3CR-1GFP mice, which are specially bred to have enhanced expression of the green fluorescent protein (GFP) in the chemokine receptor 1 (CX3CR-1) gene [123–125]. As a result, scanning laser ophthalmoscopes (SLOs) and OCT systems may be used to non-invasively image the fluorescing retinal microglia at a cellular resolution, to investigate their shape, spatial distribution, and density in real time [37,124–126]. Each of these studies could use distinctive methods with unique SLO/OCT systems including multi-color confocal SLO to detect multiple fluorescent markers [124], widefield autofluorescence imaging [125], and 488 nm fluorescence imaging using commercial SLO/OCT devices [126]. Although this reveals the imaging versatility of the Cx3CR-1GFP mice, genetic modifications often have undesirable adverse effects and are not currently clinically translatable.

Despite this, there have also been reports of attempts at in vivo and non-invasive imaging methods in humans [127,128]. For example, Liu et al. developed a multimodal adaptive optics (AO) system which combined SLO and OCT systems [127]. Superluminescent diodes were also used to produce the imaging beams from each of the AOSLO and AOOCT with the addition of mirrors positioned precisely for it to be directed into the eye, enabling the retina to be imaged. Although they were able to image microglia in the inner limiting membrane (ILM), these were identified by subjective visual interpretation, and the authors emphasize the need for improvement due to the lengthy process (1 hour to the image) [127]. Although AOOCTs can produce high-resolution and cellular levels of imaging, this is compromised by a ~6-fold reduction in the field of view compared to a commercial machine [128]. For instance, Castanos et al. used spectral-domain en face OCT imaging in the reflectance mode (OCT-R) to obtain larger images of the retina [128]. On examination of the ILM in healthy participants, uniformly distributed ramified macrophage-like cells were observable [128]. In patients with diabetic retinopathy, central retinal vein occlusion, and open-angle glaucoma, these cells appeared amoeboid and were not uniform [128]. Through these morphological and distributional characteristics, the authors interpret these cells as microglia or hyalocytes. Although this reveals the possibility of non-invasive in vivo imaging of the human retina, the authors recognize limitations to their technology, including manual subjective interpretations of cell morphology and the indefinite cell type distinction [128].

7. Contradictory Findings: Reasons and Limitations

Although the aforementioned studies had similar objectives of finding densitometric or morphological features of microglia, there were some which produced contradictory results. This may have been because many of the methods of analysis used were subjective, varied, and even ambiguous. These aspects could have been amended to allow for more consistent, representative, and less biased results. For example, when specific regions of the CNS, such as the striatum, SN, hippocampus, or retina, have been investigated, randomly chosen “sections” or “slices” are analyzed [86,111,129]. Additionally, some investigations took subjective measures to distinguish “ramified” and “amoeboid” cells whilst also using manual counting methods [99]. There are indeed other methods that can certainly be used to reduce the impact of these subjective and non-representative techniques. For instance, more extensive and efficient imaging methods can be achieved by scanning the whole retina rather than sectors and including z-stacks of all the layers and maximum intensity projections [24]. In this way, observations of the whole retina and all its layers can be comprehensively analyzed. Furthermore, algorithms that automate cell counting can be used to quantify densitometric parameters [24]. 

In light of encouraging future investigations, avoidance of subjective categorization of microglia morphotypes has successfully already been performed using changes in 15 parameters defined through the course of microglial inflammation enabling morphometric analyses output by automated software (Fraclac; ImageJ) [130]. However, it is important to note that Fernández-Arjona et al., manually selected the microglia cells to be analyzed based on a set of criteria that consisted of a fully visible cell body and processes which are nonoverlapping with parts of other adjacent cells [130]. Therefore, the resulting data were not fully representative or free from bias, although this demonstrates the principle of using quantitative data to determine the different morphotypes.

Clinical and preclinical trials often have only male subjects [60,91]. This may be to reduce the effects of physiological differences such as the ovarian cycle, confounding the results of the trial or experiment. However, are known sexual dimorphisms for the structural and functional differences in microglial responses to aging and disease, including MS [131]. This may be due to X chromosomes that express the greatest number of immune-related genes in the whole human genome and epigenetic and environmental influences [131,132]. 

Han et al. also suggest that hormonal differences may influence males having more activated microglia in development stages, putting them at higher risk of disease, whilst for females, this occurs during adulthood [131]. Other investigations on mediating hypersensitivities to pain have shown the involvement of microglia in male mice whilst other immune cells such as T-cells are required in female mice [133,134]. Although there are no reports which investigate the sexual differences in the morphological characteristics of retinal microglia in MS, transcriptome microarray analyses of male and female mice of two age groups (3 or 24 months) have shown that age-associated changes of retinal microglia cell-specific gene expression are sexually dimorphic [135]. Designing trial/experimental studies based on one sex may limit the ability to identify sex-associated pathological differences in MS, a disease more common in females [60]. Using stratified analyses in investigations that are inclusive of both sexes instead would benefit understanding pathological differences between different sub-sets of patients within the heterogeneous MS population. This would also enable larger patient group sizes to be investigated in larger multicentre research collaborations with consequently greater impact and generalisability.

There has also been a recent interest in the use of computer vision and machine learning algorithms to enable automated recognition and/or categorization of “cells” or “regions of interest” in neuroimaging data. Commonly used in biomedical research are support vector machines (SVMs), regarded as supervised machine learning techniques that can categorize data and perform regression analyses [136]. Linear SVMs can use the input “training” data for each category to extract features that may be used to define each category [137]. Then, it uses the input “training” data to effectively categorize each data point by examining said defining features [137]. This is achieved by automatically identifying the optimal hyperplane that allows each data point to be distinguished in one category from those of another. These machine learning algorithms have already been implemented in research in efforts to predict dementia in patients using data from high-resolution MRIs and even mental state assessment scores [136]. Although these studies do not yet produce algorithms that have 100% accuracy and precision, the concept of using shareable automated algorithms presents as an excellent starting point for improving biomedical data processing and analysis [136].

8. Concluding remarks

In conclusion, microglia offers an exciting avenue for research in neurodegenerative disorders, including MS. There is also mounting evidence to show some early retinal microglial changes in response to diseases, further supporting the idea of using retinal biomarkers as an earlier method of disease detection. Additionally, there is a growing number of studies that investigate retinal and brain microglia concerning their morphological, genetic, and molecular characteristics. However, the key to our understanding of their role is the ability to perform accurate and consistent quantitative analyses of the dynamic changes in structure, as morphotypes may be closely linked to microglia function. We would encourage the development of standardized experimental designs and validated data analysis algorithms, as these are necessary to comprehensively delineate the role that these important cells have in disease pathogenesis and severity.

Author Contributions:

Writing—planning, S.C., L.G., and M.F.C.; writing—original draft preparation, S.C.; writing—review and editing, S.C., L.G., and M.F.C.; visualization—S.C.; supervision—L.G. and M.F.C. All authors have read and agreed to the published version of the manuscript.

Funding:

This research received no external funding.

Conflicts of Interest:

The authors declare no conflict of interest. The funders had no role in the writing of the manuscript.

References

1.Au, N.P.B.; Ma, C.H.E. Recent Advances in the study of bipolar/rod-shaped microglia and their roles in neurodegeneration. Front. Aging Neurosci. 2017, 9, 128. [CrossRef] 

2. Li, F.; Jiang, D.; Samuel, M.A. Microglia in the developing retina. Neural Dev. 2019, 14, 1–13. [CrossRef] 

3. Silverman, S.M.; Wong, W.T. Microglia in the retina: Roles in development, maturity, and disease. Annu. Rev. Vis. Sci. 2018, 4, 45–77. [CrossRef] 

4. Bachstetter, A.D.; Ighodaro, E.T.; Hassoun, Y.; Aldeiri, D.; Neltner, J.H.; Patel, E.; Abner, E.L.; Nelson, P.T. Rod-shaped microglia morphology is associated with aging in 2 human autopsy series. Neurobiol. Aging 2017, 52, 98–105. [CrossRef] 

5. Holloway, O.G.; Canty, A.J.; King, A.E.; Ziebell, J.M. Rod microglia and their role in neurological diseases. Semin. Cell Dev. Biol. 2019, 94, 96–103. [CrossRef]

6. Ramirez, A.I.; De Hoz, R.; Salobrar-Garcia, E.; Salazar, J.J.; Rojas, B.; Ajoy, D.; López-Cuenca, I.; Rojas, P.; Triviño, A.; Ramírez, J.M. The role of microglia in retinal neurodegeneration: Alzheimer’s disease, Parkinson, and glaucoma. Front. Aging Neurosci. 2017, 9, 214. [CrossRef] [PubMed] 

7. Rashid, K.; Akhtar-Schaefer, I.; Langmann, T. Microglia in Retinal Degeneration. Front. Immunol. 2019, 10, 1975. [CrossRef] [PubMed] 

8. Laprell, L.; Schulze, C.; Brehme, M.L.; Oertner, T.G. The role of microglia membrane potential in chemotaxis. J. Neuroinflammation 2021, 18, 1–10. 

9. Karlstetter, M.; Scholz, R.; Rutar, M.; Wong, W.T.; Provis, J.M.; Langmann, T. Retinal microglia: Just bystander or target for therapy? Prog. Retin. Eye Res. 2015, 45, 30–57. [CrossRef] [PubMed] 

10. Wang, J.; Wang, J.; Wang, J.; Yang, B.; Weng, Q.; He, Q. Targeting Microglia and Macrophages: A potential treatment strategy for multiple sclerosis. Front. Pharmacol. 2019, 10, 286. [CrossRef] 

11. Walker, D.G.; Lue, L.-F. Immune phenotypes of microglia in human neurodegenerative disease: Challenges to detecting microglial polarization in human brains. Alzheimer’s Res. Ther. 2015, 7, 1–9. [CrossRef] 

12. Luo, C.; Jian, C.; Liao, Y.; Huang, Q.; Wu, Y.; Liu, X.; Zou, D.; Wu, Y. The role of microglia in multiple sclerosis. Neuropsychiatr. Dis. Treat. 2017, 13, 1661–1667. [CrossRef] 

13. Yao, K.; Zu, H.-B. Microglial polarization: Novel therapeutic mechanism against Alzheimer’s disease. Inflammopharmacology 2020, 28, 95–110. [CrossRef] 14. Ransohoff, R.M. A polarizing question: Do M1 and M2 microglia exist? Nat. Neurosci. 2016, 19, 987–991. [CrossRef] 

15. Grimaldi, A.; Pediconi, N.; Oieni, F.; Pizzarelli, R.; Rosito, M.; Giubettini, M.; Santini, T.; Limatola, C.; Ruocco, G.; Ragozzino, D.; et al. Neuroinflammatory processes, A1 astrocyte activation, and protein aggregation in the retina of Alzheimer’s disease patients, possible biomarkers for early diagnosis. Front. Neurosci. 2019, 13, 925. [CrossRef]

For more information:1950477648nn@gmail.com