View all hidden authors and organizations

Edited by Jeremy Nathans, Molecular Biology and Genetics, Johns Hopkins University School of Medicine, Baltimore, Maryland; received July 8, 2021; accepted October 14, 2021

This work determined the role of microglia (innate immune cells of the central nervous system) in the local control of the retinal vasculature and identified the defects in the early stages of diabetes. Microglia contact neurons and vasculature and express several vasoactive agents. The activation of fractalkine-Cx3cr1 signaling in microglia leads to capillary contraction, and blocking the renin-angiotensin system (RAS) with candesartan can eliminate microglia-mediated vasoconstriction in the retina. In early diabetes, decreased retinal blood flow is consistent with capillary contraction, increased microglia-vascular binding, loss of microglia-capillary regulation, and changes in microglia expression in the RAS pathway. Although candesartan restores the diameter of retinal capillaries in the early stage of diabetes, it is necessary to target microglia-vascular regulation to prevent large retinal blood vessels from dilating and reducing retinal blood flow.

Local blood flow control in the central nervous system (CNS) is essential for normal function and depends on the coordination between neurons, glia, and blood vessels. Macroglia, such as astrocytes and Muller cells, contribute to this neurovascular unit in the brain and retina, respectively. This study explored the role of microglia (innate immune cells of the central nervous system) in the regulation of retinal blood vessels and emphasized the changes during early diabetes. Structurally, microglia are found to be in synaptic contact with retinal capillaries and neurons. In brain and retina explants, the addition of fractalkine (the only ligand for the monocyte receptor Cx3cr1) causes the capillaries in the contact area of ​​the microglia to constrict. This vascular regulation depends on the participation of microglia Cx3cr1, because the genetic and pharmacological inhibition of Cx3cr1 eliminates fractalkine-induced contraction. Analysis of the microglia transcriptome identified several vasoactive genes, including angiotensinogen, which is part of the renin-angiotensin system (RAS). Subsequent functional analysis showed that RAS blockade by candesartan eliminated the capillary constriction caused by microglia. Microglial regulation was explored in the rat streptozotocin (STZ) model of diabetic retinopathy. Due to the decrease in capillary diameter, retinal blood flow decreases after 4 weeks, which is consistent with the increase in microglia. Functional evaluation showed loss of microglia-capillary response in STZ-treated animals, and transcriptome analysis showed evidence of RAS pathway dysregulation in microglia. Although candesartan treatment reversed the capillary constriction in STZ-treated animals, blood flow still decreased due to larger vasodilatation. This work shows that microglia actively participate in neurovascular units, and abnormal microglia-vascular function may lead to early vascular damage during diabetic retinopathy.

The retina is one of the most metabolically active organs in the body and is supplied by the external (choroid) and internal (retinal) vascular network in most mammals (1). Although the choroid provides support for the light detection photoreceptors in the outer retina, the retinal blood supply supports the ganglion cells and numerous neurons and glia in the inner nuclear layer (2). The small arteries supplied by the retinal blood enter the optic disc and branches, which in turn form smaller blood vessels, including retinal capillaries, and establish superficial vascular plexuses. These capillaries penetrate the inner layer of the retina, forming a relatively sparse middle vascular plexus, and deeper into the outer retina, forming a highly anastomosed deep vascular plexus. After completing the vascular circuit, blood returns through the small veins on the surface of the retina, which leave with the optic nerve (3, 4).

The blood flow of the entire retina depends to a large extent on the caliber of blood vessels, which is strictly regulated to meet the metabolic demands of neuronal activity (5). An example of this is the well-defined hyperemia response, whereby increased neuronal activity (via flashing light) leads to dilation of arterioles and increased blood flow inside the retina (6). Unlike peripheral blood vessels, the retina and cerebrovascular system do not have direct neuronal input to regulate vascular tension; on the contrary, large glial cells (Müller cells and astrocytes) are thought to actively adjust the diameter of blood vessels in response to neural activity. Change (7, 8). This type of coupling gave rise to the idea of ​​neurovascular units, including neurons, glia, and blood vessels (7). Although studies in the retina have determined that neuron-dependent calcium increases in Müller cells mediate blood vessel diameter changes (9), recent data indicate that the regulation of the internal retinal vasculature is more complex (10). The evidence for this comes from the fact that the same light stimulation can induce vasoconstriction or relaxation, while Müller cell-dependent calcium signal only controls the capillaries in the middle vascular plexus (11, 12). This indicates that there are multiple regulatory pathways in the retina.

Recently, it has been suggested that microglia, the innate immune cells of the retina, may also play a role in the neurovascular unit, although there is a lack of direct functional evidence (13). The traditional view of microglia is that they cause disease by releasing pro-inflammatory and neurotoxic cytokines (14⇓ –16). However, it is now recognized that microglia play several important non-inflammatory roles in the normal brain and retina, such as dynamic synaptic monitoring and synaptic pruning (17⇓ –19). In the retina, microglia are known to be in close contact with the vasculature and affect blood vessel development (20). Nevertheless, the non-inflammatory response of microglia to neuronal signals and its role in regulating vascular tone has not been confirmed.

Although regulation of retinal blood flow is essential for function (21), vascular dysfunction is known to occur in a variety of pathologies, including diabetic retinopathy (DR). In the early stages of DR progression, vascular lesions such as decreased retinal blood flow, microaneurysms, and non-perfusion areas of blood vessels may occur (22). Decreased retinal blood flow especially appears in the early stages of diabetic patients (23⇓-25) and diabetic animal models (25). Changes in vascular regulation in the retina are considered to be a possible precursor to severe vascular disease in DR (26).

This study investigated whether retinal microglia form functional components of neurovascular units, and whether signal transduction through the fractalkine-Cx3cr1 pathway regulates blood vessel diameter. In addition, this work also explored whether changes in the involvement of microglia in the vasculature of the retina can help explain the decreased retinal blood flow that occurs in the early stages of diabetes. Exploring the mechanisms responsible for the tight regulation between neuronal activity and local blood supply is critical to understanding retinal function in health and disease, and may provide an empirical framework for future treatments against vascular pathogenesis.

Microglia in the central nervous system (CNS) are closely related to the vasculature, especially during development, injury, and disease (20, 27). However, little is known about microglia-vascular interactions in normal tissues. In the retina, microglial cell bodies are usually located in the plexiform layer, and their processes extend to the entire retina (SI appendix, Figure S1). Examination of the superficial vascular plexus revealed that microglia cover the entire tissue (Figure 1A, enhanced green fluorescent protein [EGFP], green) and are closely related to the retinal vasculature (Figure 1A, inset; isolectin B4 [IB4], Red). When quantifying the contact of microglial processes with retinal vessels of different diameters relative to the corresponding area of ​​each vessel diameter category, you can see microglial cells and smaller retinal vessels (≤15 µm), especially The smallest retinal capillary (<10 µm) interaction, when compared to the larger container (Figure 1B) (one-way analysis of variance, for 15 to 20 µm and >20 µm, P <0.05, 0.001, respectively). At the ultrastructural level (Figure 1C), a microglia process (stained as EGFP) is next to a pericyte, which is located on the endothelial cells lining the capillary lumen. An immunohistochemical study of this microglia-pericyte contact was also performed using the NG2-DsRed report mouse, which was labeled with pericyte cell bodies and protrusions (Figure 1D, red). It was observed that the microglia (Figure 1D, Iba-1, green) were in contact with two pericyte cell bodies (Figure 1D, red), and the nuclei were immunolabeled with DAPI (Figure 1D, blue). Orthogonal projections of the boxed area (Figure 1D, top and right) show direct contact between the two cell types. Contact (Figure 1D, asterisk) for further imaging at higher resolution to show the microglia process (Figure 1D, green) and pericyte (red; asterisk in Figure 1E; see also SI appendix, Figure S2 ) Direct contact and movie S1). In the rat retina, the contact degree of microglia with pericyte cell bodies, protrusions (NG2 labeling) and capillary areas without pericyte contact (NG2-/IB4+ areas, possibly endothelial cells) was quantified, and no contradictions were observed. Microglia-Pericytes or microglia prefer vascular contact (Figure 1F).

Retinal microglia are related to the vasculature and neuronal synapses. (A) Whole-mounted mouse retina (Cx3cr1GFP/+) is labeled with anti-EGFP (microglia, green) and G. simplicifolia IB4 (vascular, red). The highlighted area shows that microglia are associated with blood vessels in the superficial vascular plexus (inset). (Scale bar, 500 µm; 50 µm, inset.) (B) Relative to the vascular area of ​​each blood vessel size, quantify the correlation between microglia processes and blood vessels of different diameters, and show that microglia preferentially interact with capillaries Correlation, *P <0.05, ***P <0.001. (C) The ultrastructure of Cx3cr1GFP/+ microglia in the retina in contact with blood vessels shows the process of microglia (for EGFP immunolabeling, black dots) adjacent to pericytes and in contact with endothelial cells in the capillary lumen. (Scale bar, 0.5 µm.) (D) Whole retina from NG2-DsRed pericyte reporter mice (pericytic soma, process, red), using Iba-1 (microglia, green) and DAPI (nucleus, blue) (Color) staining shows the process of microglia in contact with the pericyte cell body. The boxed area is shown in orthographic projection (above and right). (Scale bar, 10 µm.) (E) High-resolution rendered image of microglia-pericyte contact obtained from the asterisk in D. (Scale bar, 5 µm.) (F) Further detection of microglia-pericyte interaction and pericyte cell bodies, protrusions (NG2-), and capillary regions lacking pericyte contact (NG2-/) in the rat retina IB4+) to quantify the degree of exposure. (G) A vertical section from the Cx3cr1GFP/+ retina, marked with blood vessels (IB4, magenta), microglia (EGFP, green), neuronal synapses (VGLUT1, red) and nuclei (DAPI, blue), Shows microglia contacting retinal blood vessels (asterisks) and neuronal synapses (arrows). The boxed area is imaged at higher resolution and rendered to highlight microglia-synaptic interactions (inset). (Scale bar, 50 µm; 5 µm, inset.) (H) In human retina (microglia, Iba-1, green; blood vessels, vitronectin, magenta, asterisk; neuronal synapses, VGLUT1, red , Arrow; nucleus, DAPI, blue). (Scale bar, 50 µm.) (I) When quantifying neurons in the inner retina of Cx3cr1+/GFP mice (vascular, IB4, red; microglia, EGFP, green; neuron synapse VGLUT1, blue)- When microglia contact, most microglia contact neuronal synapses and blood vessels. (Scale bar, 20 µm.) Data are expressed as mean ± SEM, n = 5 (B and F), n = 3 (I, inset). GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; MC, microglia; PC, pericytes; EC, endothelial cells; CL, capillary lumen; ONL, outer nuclear layer; OPL, Outer plexiform layer.

In addition to contacting retinal blood vessels (IB4, magenta, asterisk in Figure 1G), microglia (EGFP, green) were also observed to extend the process to the inner plexiform layer where neuronal synapses are located (Figure 1G, VGLUT1 red , Arrow; DAPI blue). Figure 1G, inset shows rendering of microglia-neuron interaction at higher magnification. This is also observed in the human retina (Figure 1H, DAPI, blue), where microglia (Iba-1, green) contact retinal blood vessels (Figure 1H, vitronectin, magenta, asterisk) and neuronal processes Touch (Figure 1H, VGLUT1), red. arrow). When quantified in the Cx3cr1+/GFP mouse retina, most of the microglia in the internal retina (Figure 1I, EGFP, green) interacted with neuronal synapses (Figure 1I, VGLUT1, blue) and retinal blood vessels (IB4, Red; Figure 1 I, inset) (73 ± 13%, rat retina). The single channel used for immunolocalization is shown in the SI appendix, Figure S3.

In the brain and retina, the contact of macroglia (astrocytes and Müller cells) with neuronal synapses and vasculature is essential for the control of local blood supply in response to neuronal activity (7, 8). To determine whether microglia played a similar role, Cx3cr1GFP/+ retina was isolated and maintained in vitro. Microglia are visualized by their EGFP expression (Figure 2A, green), and blood vessels are labeled with Rhodamine B (Figure 2A, red). Since the fractalkine-Cx3cr1 axis is thought to mediate neuron-microglia communication, blood vessels and microglia are imaged when fractalkine (200 ng/mL) or PBS is perfused into the chamber (movie S2). The blood vessel diameter change is monitored and expressed relative to the baseline value of the same area of ​​the blood vessel.

Microglia contract retinal capillaries and express vasoactive agent genes through fractalkine-Cx3cr1 signaling. (A) The isolated Cx3cr1GFP/+ retina (EGFP; microglia, green) was labeled with rhodamine B (vascular, red) and imaged under a live cell microscope. (Scale bar, 50 μm.) (B) The addition of fractalkine (200 ng/mL) induces vasoconstriction (m+, n = 4 PBS, n = 6 FKN) at the contact site of microglia, but it does not occur significantly in the area The vascular changes lack microglia processes (m-, n = 5 PBS, n = 6 FKN). When performed on the Cx3cr1GFP/GFP retina, there was no significant contraction (n = 5). To support the Cx3cr1 mediated effect, add a small molecule Cx3cr1 inhibitor, AZD8797, to prevent vasoconstriction (inset, n = 4 FKN, n = 3 FKN+AZD8797). (C) The cerebrovascular system's response to fractalkine was tested in the rat thin scull preparation, and it was significantly contracted 120 seconds after injection (n = 3 PBS, FKN). The inset shows the baseline and representative images after adding fractalkine. (Scale bar, 250 μm.) (D) Retinal microglia (EGFP, green), neuronal synapses (VGLUT1, red) and blood vessels (IB4, light blue) in Cx3cr1GFP/+ and Cx3cr1GFP/GFP animals The number of blood vessel and neuron contacts was imaged and the range quantified relative to the microglia volume (see isolated microglia: red, neuron contact; blue, blood vessel contact). (Scale bar, 15 μm.) (E) Grouped data shows that compared with Cx3cr1GFP/+ retina (n = 5), Cx3cr1GFP/GFP retina has less blood vessel contact, but there is no difference in neuronal contact. Cx3cr1GFP/GFP microglia showed reduced process branching (n = 5). (F) Using in vivo OCTA, compared with Cx3cr1GFP/+ retina (n = 4 C57Bl6, n = 6 Cx3cr1GFP/+, Cx3cr1GFP/GFP), the diameter of retinal capillary vessels in Cx3cr1GFP/GFP animals increased, but the diameter of arteries did not change or decreased Veins (A/V ratio shown in the table, n = 4 C57Bl6, n = 6 Cx3cr1GFP/+, n = 5 Cx3cr1GFP/GFP). (G) RNA-seq was performed on rat retinal microglia isolated by FACS, and 268 genes were identified as angiogenesis genes (GO:0001525), 39 genes were involved in vasoconstriction, and 41 genes were involved in vasodilation (vascular For diameter adjustment, go to: 0097746). (H) Quantify blood vessel diameter in rat retinal explants pre-incubated for 10 minutes in Ames (black trace) or Ames + 230 nM candesartan (red trace), and then add fractalkine (FKN, 200 ng/ mL) (shaded area, representing data from 1 retina, n = 5 blood vessels). (I) When analyzing grouped data 10 minutes after adding fractalkine, when pre-incubating with candesartan (n = 7 fractalkine, n = 5 fractalkine + candesartan), the contraction is cancelled. To further support the role of RAS, in vitro incubation with fractalkine up-regulated the expression of microglia Agt, which was not obvious in microglia isolated from the Cx3cr1GFP/GFP retina (inset, n = 6). Data are expressed as mean ± SEM, *P <0.05, **P <0.01, ***P <0.001, ****P <0.0001.

In response to fractalkine, the vascular regions associated with microglia protuberances (m+) contract (Figure 2B, m+) (two-way analysis of variance; PBS and fractalkine, P <0.001), while those away from microglia process (m- ) Showed no significant change in capillary diameter (Figure 2B, m-) (two-way analysis of variance; PBS and fractalkine, P = 0.26). These ex vivo preparations showed minimal microglia process movement at the blood vessel level throughout the imaging process, including during fractalkine exposure (SI Appendix, Figure S4 and Movie S2). When explants lacking Cx3cr1 (Cx3cr1GFP/GFP mice) were exposed to fractalkine, compared with the PBS control, it was at (m+; 105.7 ± 2.7% vs. 94.7 ± 2.3%, two-way analysis of variance P = 0.52) or not ( m-; 98.8 ± 1.2% vs. 97 ± 1.3%, two-way analysis of variance (P = 0.999) contact with microglia (Figure 2B). To further support the Cx3cr1-dependent mechanism, pre-incubation with Cx3cr1 inhibitor AZD8797 (28) inhibits fractalkine (FKN)-induced contraction (Figure 2B, inset) (FKN 82 ± 2%, FKN+AZD8797 96 ± 2%, t test P = 0.015). Finally, in order to explore whether this vascular regulation function of fractalkine is specific to the retina, a thin skull preparation was used to image the superficial blood vessels in the rat brain. These preliminary data indicate that although vehicle delivery caused no change in blood vessel diameter, subdural addition of fractalkine resulted in significant contraction of smaller blood vessels (Figure 2C) (repeated measurement two-way analysis of variance, blood vessels ≤ 15 µm, P <0.05). Although both tissues show fractalkine-induced contraction, the difference in vascular dynamic response may reflect different systems (in vitro and in vivo, respectively) used to explore microglial vascular regulation.

Since the Cx3cr1GFP/GFP retina did not show fractalkine-induced vasoconstriction, the contact of microglia with retinal blood vessels and neurons was explored. High-resolution immunocytochemical analysis of microglia (Figure 2D, EGFP, green) in contact with neuronal synapses (Figure 2D, VGLUT1, red) and blood vessels (Figure 2D, IB4, light blue) allows the quantification of specific Contact area. When calculating the contact volume of each microglia, the blood vessel contact of Cx3cr1GFP/GFP animals is less than that of animals with a functional copy of Cx3cr1 (Figure 2E) (Cx3cr1GFP/+ 7.5 ± 0.4% and Cx3cr1GFP/GFP 5.5 ± 0.3%, t test P = 0.004). Although there is no difference in neuron contact between the two genotypes, the microglia of Cx3cr1GFP/GFP animals have fewer branching processes (Figure 2E) (Cx3cr1GFP/+ 111.5 ± 7.2 and Cx3cr1GFP/GFP 92.2 ± 2.1, t test P = 0.03), reflecting the literature showing that Cx3cr1GFP/GFP has more active inflammation characteristics (29). When the retinal capillary diameter was compared with C57bl6 control animals, Cx3cr1GFP/+ capillaries were similar to the control (Figure 2F) (C57bl6 11.3 ± 0.3 µm and Cx3cr1GFP/+ 10.9 ± 0.2 µm, one-way analysis of variance P = 0.66), and Cx3cr1GFP/GFP showed increased capillary diameter (Figure 2F) (Cx3cr1GFP/+ 10.9 ± 0.2 µm and Cx3cr1GFP/GFP 12 ± 0.4 µm, one-way analysis of variance P = 0.047). There was no difference in the diameter of larger blood vessels for any genotype (Figure 2F, inset) (Cx3cr1GFP/+ and Cx3cr1GFP/GFP P = 0.87 and 0.94, respectively).

RNA sequencing (RNA-seq) was performed on microglia isolated from FACS collected from a 12-week-old dark agouti to determine whether the microglia express vascular regulatory factors. To confirm the purity of the sample, the mapped genes were compared with the published list of microglia markers (30), 23 of 29 markers were identified in our gene group, including microglia specific markers Tmem119 (SI appendix, Table S1) (31). The negative microglial cell fraction (CD11b-) was also queried for characteristic genes of microglial cells, in which key genes such as Cx3cr1, TMEM119 and Slc2a5 were almost not expressed (<3 copies, >1 × 107 transcripts) (SI appendix , Table S1). The microglia transcriptome was also compared with microglia-rich genes reported in several studies, and significant overlap was observed, while the known neuron genes were almost free of contamination (SI appendix, Figure S5). The gene group was compared with genes known to be involved in angiogenesis (Gene Ontology [GO]: 0001525,407 genes) and blood vessel diameter regulation (GO: 0097746, 310 genes). In total, 268 genes expressed in the microglia population were determined to play a role in the angiogenesis pathway (Figure 2G and SI appendix, Table S2), such as hypoxia-inducible factor 1α (Hif1a) and vascular endothelial growth Factors A and B (Vegf A /B). When exploring the regulation of blood vessel diameter, 41 genes were found to play a role in vasodilation, such as phospholipase A2 (Pla2g6) and Sirtuin 1 (Sirt1), and 39 genes were confirmed to have vasoconstrictor effects, including endothelin 1, -3 (Edn1,-3) and arachidonic acid 5-lipoxygenase (Alox5) and angiotensinogen (Agt) (Figure 2G and SI appendix, tables S3 and S4, respectively).

Since angiotensinogen is a component of the renin-angiotensin system (RAS), which participates in the regulation of retinal blood vessels through the type 1 angiotensin II receptor (AT1R) (32, 33), the AT1R antagonist is used. In vitro experiments were performed on desartan. In rat retinal explants exposed to Ames (black trace) and Ames + candesartan (230 nM) (Figure 2H, red trace), the baseline capillary diameter exceeded 10 minutes on average, Then add fractalkine (shaded area in Figure 2H). Similar to what was observed in Cx3cr1GFP/+ mice (Figure 2A and B), the rat retina was exposed to fractalkine-induced capillary contraction, while candesartan prevented Any fractalkine-induced contraction (Figure 2H). When analyzing grouped data, candesartan eliminated fractalkine-induced vasoconstriction (Figure 2I) (80.4 ± 2.0% vs. 97.5 ± 2.3%, t test, P <0.01).

To further support the role of RAS in microglial cell-mediated vascular regulation, the control C57bl6 and Cx3cr1GFP/GFP were exposed to fractalkine in vitro for 2 hours to isolate microglia and quantify the expression of Agt (Figure 2I, inset) . Although exposure to fractalkine increased Agt expression in the control retina, the Cx3cr1GFP/GFP retina, which did not previously exhibit microglia-mediated contraction (Figure 2B), did not show expression changes (Figure 2I, inset) (+FKN, C57bl6 21.8 ± 3.5 copies/1,000 copies of Hprt and Cx3cr1GFP/+ 7.7 ± 0.6 copies/1,000 copies of Hprt, two-way analysis of variance (P = 0.017). Current data indicate that microglia can regulate vasoconstriction in the retina and the wider central nervous system through the fractalkine-Cx3cr1 pathway. Although microglia express several gene transcripts of known vasoactive agents, fractalkine-induced retinal vascular microglia regulation occurs through AT1R.

The regulation of retinal blood supply is essential for normal function. Retinopathy, such as DR, exhibits early retinal blood flow defects and abnormal neurovascular coupling (23, 25, 34). In order to investigate whether microglia vascular regulation changes during early diabetes, adult dark agouti pigs became diabetic through streptozotocin (STZ) and were significantly hyperglycemia during the entire 4-week experiment (SI appendix, Table S5).

Since decreased retinal blood flow is a consistent and early change in diabetic patients and animal models (24, 25), quantitative vascular-dependent dynamic analysis using sodium fluorescein (35) is used to confirm vascular dysfunction. Over time, calculate the average normalized fluorescence intensity of each pixel in the fundus image (Figure 3 AC, inset), group by blood vessel type, and generate a facial heat map (Figure 3 AC, fill time), warmer The color indicates that the filling takes longer (slower blood flow). Vascular-dependent dynamics analysis showed that the small arteries of STZ-treated animals took longer to fill (Figure 3D) (median regression analysis, P <0.05), reflecting reduced blood flow. Due to the continuity of the retinal vasculature, this increase in filling time was also observed in retinal capillaries and venules (Figure 3D) (median regression analysis, P <0.05), and no vessel-specific defects were found (median value Regression analysis, P> 0.05). The drainage time of all retinal vessels was also longer (Figure 3E) (median regression analysis, P <0.05), and its effect was significantly greater than the effect observed with filling time (median regression analysis, P <0.05). The use of velocimetry (SI Appendix, Figure S6) verified that the arteriovenous blood flow of STZ-treated animals was reduced, and the clinically relevant arteriovenous transit time also showed a decrease in blood flow (increased transit time) (SI Appendix, Figure S6). S6D). The decrease in retinal blood flow was not related to systemic changes, and systolic blood pressure, hematocrit, and intraocular pressure remained unchanged (SI appendix, Figure S7).

After 4 weeks of diabetes, retinal blood flow decreases and capillaries shrink. VFA is used to quantify retinal blood flow in control and STZ treated animals. (A–C) The En face heat map depicts the filling time of arterioles, capillaries, and venules. The inset shows the representative average normalized fluorescence intensity trajectories of control (black line) and STZ-treated (red line) animals. (Scale bar, 500 μm in A.) (D and E) Quantify the time required to reach half of the maximum intensity (D, filling time) and half of the maximum intensity (E, drainage time), showing filling and drainage in STZ treatment In animals, the time for all vessel types increased significantly (unfilled bars, control n = 23; filled bars STZ, n = 21). (F) Fundus images of fluorescein sodium were quantified as large vessel curvature (n = 13) and arteriovenous ratio (inset, n = 13), which were not between STZ-treated (stuffed bars) and control (unfilled bars) animals Observe the difference. (G) Immunohistochemistry was used to quantify blood vessel density in control and STZ-treated (unfilled and filled bars) eyes, and no difference was observed between the two groups (n = 11). The rendered image shows the segmented blood vessel types (inset, yellow capillaries, blue arterioles and cyan venules). (Scale bar, 1 mm.) (H) Perform OCTA in vivo to measure the capillary diameter of control and STZ-treated (inset) animals. The measured blood vessels are shown in green. (Scale bar, 50 μm.) (I) Compared with the control in the upper vascular plexus (n = 10), a decrease in capillary diameter was observed in STZ-treated animals (n = 12). No changes were observed in the middle/deep vascular plexus. Group data are expressed as mean ± SEM, *P <0.05.

Since vascular changes affect the blood flow in DR (36, 37), the morphology of large-diameter blood vessels is evaluated based on the fluorescein image of the fluorescence peak intensity. Large vessel tortuosity (Figure 3F) (two-way analysis of variance, small artery P = 0.52, small vein P = 0.98) or small artery/vein diameter (arteriovenous ratio) (Figure 3F, inset) (t-test, P) No change = 0.48) Observed between the two cohorts. Similarly, when using retinal whole immunohistochemistry to quantify the density of arterioles, capillaries, and venules (Figure 3G, the inset shows a rendered image of arterioles, dark blue; venules, cyan; capillaries, yellow), blood vessels No change in density. Different results were observed between control and STZ-treated animals (Figure 3G) (two-way analysis of variance, arterioles P = 0.98, venules P = 0.99, capillaries P = 0.94). Because fluorescein image analysis and immunohistochemistry lack the resolution to assess capillary diameter, optical coherence tomography angiography (OCTA) is used to quantify it in vivo. Images of the superficial retinal capillary network of the control (Figure 3H) and STZ-treated (Figure 3H, inset) animals were obtained. Quantification (Figure 3H, green overlay shows the measured capillaries) shows that the diameter of the capillaries in the STZ is reduced. Treatment group (Figure 3I) (two-way analysis of variance, P <0.05). When similar analysis was performed on the middle and deep capillary plexus, no changes were detected (Figure 3I) (two-way analysis of variance, P = 0.72).

In conclusion, retinal blood flow in diabetic patients was significantly slowed down. In vivo OCTA showed that 4 weeks after STZ-induced diabetes, retinal capillaries in the superficial vascular plexus contracted. These diameter changes are limited to the capillary network, because the larger blood vessels remain unchanged and the retinal vessel coverage does not change.

For control and STZ-treated animals, quantify the degree of contact of microglia (Figure 4A, inset, green, Iba-1) with arterioles, capillaries, and venules (Figure 4A, inset, red, IB4) to determine Whether the contraction of retinal capillaries in diabetic patients is accompanied by changes in the association of microglia. Although microglia showed similar associations with large-diameter arterioles and venules (Figure 4A) (two-way analysis of variance, P> 0.99 and P> 0.66, respectively), in STZ-treated animals, microglia -Increased capillary association (Figure 4A)) (two-way analysis of variance, P <0.05). In addition, the microglia-pericyte association (Figure 4B, inset, microglia green, Iba-1; pericytes light blue, NG2, blood vessel red, IB4) increased in the central retina of STZ-treated animals (Figure 4B, inset, microglia green, Iba-1; 4B) (Two-way analysis of variance, P <0.05). In this early stage of diabetes, there is no loss of blood vessels (Figure 3G) and no loss of retinal pericytes (SI appendix, Figure S8).

After 4 weeks of STZ-induced diabetes, microglia increased their contact with retinal capillaries. (A) The whole retina of the control (inset) and STZ-treated animals were labeled as Iba-1 (microglia, green) and IB4-FITC (blood vessels, red), and the microglia and microglia of each blood vessel were quantified. Type of vascular contact. Although there was no difference in macrovascular contact, the microglia-capillary contact in the central retina of STZ-treated animals increased (solid bars, n = 11). (Scale bar, 50 μm.) (B) Control (inset) and STZ-treated animals are labeled Iba-1 (microglia, green), NG2 (pericytes, light blue) and IB4-FITC (vascular, red) ) And quantify the degree of microglia-pericyte contact of each blood vessel type. The microglia-pericyte association increased in the central retina of STZ-treated animals (solid bars, n = 11). (Scale bar, 50 μm.) (C) Quantify the association of microglia with pericyte cell bodies, protrusions, and capillary regions lacking pericyte contact using immunological markers similar to those in B. Image analysis rendering (inset) highlights pericyte cell bodies (red), pericyte protrusions (green), and pericyte-free blood vessels (blue), while microglia contacting these areas are skeletonized and color-coded for quantification. Although there was no preferential association, all exposures to the STZ treatment (filled bars, n = 5) increased compared to the control (unfilled bars, n = 5) retina (STZ treatment P = 0.0015). (Scale bar, 50 μm.) (D and E) Evaluation of macroglia changes and astrocyte coverage in the control (unfilled bars, n = 11) and STZ-treated (filled bars, n = 11) retina There was no change (D), and no Muller cell gliosis (E) was observed (n = 6). (F) VFA is used to quantify fluorescein shift as a measure of BRB integrity. Although arterioles and venules did not change, capillary deviation increased in STZ-treated animals (unfilled bars, control n = 23; filled bars STZ, n = 21). (G and H) Morphologically assess the inflammatory state of microglia, the number of monocytes/microglia in the central and peripheral retina (G, n = 11), cell body size, average process length or process branches No difference was found at points (H, n = 5 control, n = 8 STZ). (I) Screen the RNA-seq data of retinal microglia taken from control and STZ-treated rats to determine the positive (GO:0050729) and negative (GO:0050728) regulated genes involved in inflammation. Although some inflammation genes were changed, the key inflammation genes remained unchanged after 4 weeks of diabetes. Data are expressed as mean ± SEM, *P <0.05.

Quantitative image analysis was used to further explore the association of microglia with pericytes and capillary regions lacking pericyte contact in control and STZ-treated animals (Figure 4C, inset, rendered image showing red pericyte cell bodies; pericyte protrusions) Green; blue and skeletonized microglia without pericytes). Although microglia did not particularly prefer to contact pericyte cell bodies, protrusions, or capillary areas lacking pericytes (Figure 4C) (two-way analysis of variance, P = 0.16), at 4 weeks of diabetes, microglia had the same effect as all three. The association increased (Figure 4C). 4C) (two-way analysis of variance, P <0.01). In order to determine whether this microglia effect is specific or the result of a broader macroglia response as shown in the later stages of diabetes (38, 39), the effect of astrocyte density and Müller gliosis Quantified. Vessel-specific astrocyte coverage (Figure 4D) and Muller cell gliosis (Figure 4E) did not change after 4 weeks of STZ treatment (two-way analysis of variance, P> 0.92 and 0.99, respectively).

Previous work has shown that the integrity of the blood-retinal barrier (BRB) is compromised in the early stages of diabetes (40). Using vessel-dependent blood flow analysis (Figure 3 AE), we use the fluorescein peak (fluorescein shift) to return to baseline as a measure of BRB integrity. Although no shift changes in larger vessels were observed, retinal capillaries showed a significant increase, indicating a decrease in fluorescein leakage/BRB integrity (Figure 4F) (median regression analysis, P <0.05). The breakdown of BRB can lead to immune cell infiltration and microglia activation. Microglia migration and morphological changes indicate that classical activation is observed 1 month after STZ (41). In order to evaluate whether the altered microglia-vascular association occurred in the presence of monocyte involvement/microglia activation, the whole was co-labeled with IB4 and Iba-1, and the microglia in the central and peripheral retinas were quantified The number and morphology of the cells. Although capillary fluorescein shift increased, the number of monocytes/retinal microglia (Figure 4G) (two-way analysis of variance, center P = 0.4, peripheral P = 0.9) or microglia after 4 weeks of hyperglycemia There was no difference in cell morphology (Figure 4H) (two-way analysis of variance, cell body area P> 0.99, process length/cell P = 0.15, branch point/cell P> 0.99). Nevertheless, the expression of Cx3cr1 and fractalkine in the diabetic retina increased (SI appendix, Figure S9). RNA-seq analysis of microglia isolates from 4-week control and STZ-treated animals showed that among 254 differentially expressed genes, 22 inflammatory response genes were identified, 15 of which were positive regulators (GO:0050729), And 12 are negative regulators of inflammation (GO:0050728) (Figure 4I and SI appendix, Tables S6 and S7). Importantly, the chemokines and cytokines normally associated with microglia activation—including Tlr2, Il-1β, Cxcl10, TNF-a, IL-1a, C1q—have not changed, and there is no infiltration in our RNA Expression of sex monocyte marker gene Ccr2-seq dataset (31, 42). Therefore, in the early stages of diabetes (4 weeks), when retinal capillaries contract, microglia-capillary interactions increase, which is related to monocyte recruitment, classical microglia activation, and wider macroglia Cell response is irrelevant.

To determine whether there is a loss of retinal vasomotor control during early diabetes, breathable oxygen is used to induce hyperoxia challenges, and OCTA is used to quantify the diameter of capillaries in the superficial vascular plexus (Figure 5A, top panel). Although control animals showed significant vasoconstriction in response to 100% oxygen, no contraction was observed in STZ-treated animals (Figure 5A) (two-way analysis of variance, P <0.05). To explore whether this dysfunction is also evident in microglia-mediated vasoconstriction, isolated retinal explants from control and STZ-treated animals were exposed to fractalkine and quantified capillary diameter. Although the contraction was obvious in the control cohort, this response was not present in STZ-treated animals (Figure 5B) (two-way analysis of variance, control P <0.05, STZ-treated P = 0.99). When microglia were isolated from 4 weeks of STZ treatment and control retina and RNA-seq, the expression of angiotensinogen (Agt) increased by 2.4 times (false discovery rate [FDR] = 0.009), while the arene receptor gene Expression (Ahr), a negative regulator of RAS (43), also increased (3.6 times, FDR = 0.002) (Figure 5C). Importantly, these genes (FDR of Agt and Ahr = 0.91) and any other transcripts (FDR> 0.12) did not change in the CD-11b- population.

Four weeks after STZ-induced diabetes, the expression of vasoactive genes in retinal microglia and fractalkine-induced vasoconstriction changed. (A) OCTA was used to explore the responsiveness of retinal blood vessels to hyperoxia challenges in vivo (the inset shows the baseline and OCTA images after exposure to O2). (Scale bar, 200 μm.) Although the hyperoxia challenge (solid bars) caused contraction in the control group (n = 10 normoxia, n = 6 100% O2), no contraction was observed in the STZ cohort (n = 12 normoxia, n = 7 100% oxygen). (B) To study microglia vascular regulation during diabetes, 4 weeks of STZ treatment and control retinas were exposed to fractalkine (representative control and STZ image in inset) (scale bar, 50 μm.) and blood vessels from control retina in vitro Showing fractalkine-induced vasoconstriction (solid bars), STZ retina showed no changes (n = 5 animals). (C) The differential microglia gene expression data from 4-week control and STZ-treated animals and the vascular regulatory gene list (vasoconstriction, GO: 0097746; angiogenesis, GO: 0001525; vasodilation, GO: 0097746) and RAS The positive regulator angiotensinogen (Agt) was compared with the negative regulator (Ahr) and significantly changed (FDR adjustment, citrate control n = 5, STZ n = 4). (D) OCTA was used to quantify the diameter of capillaries on the retinal surface of 4-week control and STZ-treated animals (unfilled and filled strips, respectively) exposed to candesartan or vehicle. In STZ-treated animals, the capillary diameter of the candesartan treatment group returned to baseline (n = 7 controls, n = 8, 5 STZ carriers and candesartan, respectively). (E) The arteriovenous transit time was used to quantify the retinal blood flow and showed increased transit time (slower blood flow) in STZ-treated animals not related to candesartan treatment (n = 8 control, n = 11, respectively) And 8 STZ carrier and Candesartan)). (F) Quantification of the arteriovenous ratio shows that compared to control and vehicle-treated tissues (n = 8 controls, n = 11 and 8 STZ vehicle and candesartan), candesartan treatment increases STZ treatment of the retina The diameter of the larger blood vessel in the middle. Data are expressed as mean ± SEM, * P <0.05, ** P <0.01, *** P <0.001.

Based on the loss of diabetic retinal vasomotor control and the imbalance of the microglia RAS pathway, the animals suffer from diabetes and use candesartan medoxomil or excipients in their drinking water for treatment. 4 weeks after STZ, the capillary diameter and retinal blood flow were quantified. OCTA analysis of superficial retinal capillaries showed a decrease in the internal diameter of the vehicle control group, similar to that observed in Figure 3I (Figure 5D) (91.8 ± 2%, two-way analysis of variance, P <0.05). This capillary contraction was insignificant in STZ-treated animals exposed to candesartan, and the diameter returned to the control level (Figure 5D) (99.9 ± 1.8%, two-way analysis of variance, P> 0.99). However, despite this, retinal blood flow was still slower, and vehicle and candesartan STZ-treated animals had an increased arteriovenous transit time (Figure 5E) (median regression analysis P <0.05 and P <0.001, respectively). The quantification of larger retinal blood vessels (arterioles and venules) showed that systemic administration of candesartan resulted in an increase in the arteriovenous ratio of STZ-treated animals compared with candesartan-treated controls (Figure 5F) (STZ 0.94) ± 0.01, control 0.84 ± 0.01, two-way ANOVA P <0.05) and vehicle-treated control and STZ animals (Figure 5F) (control 0.798 ± 0.03, STZ 0.86 ± 0.02, 2-way ANOVA P <0.001 and 0.05, respectively).

Collectively, these data indicate that in early diabetes, retinal vascular regulation is abnormal, microglia-mediated loss of vasoconstriction, and specific disorders of RAS. However, despite the expansion of retinal capillaries, treatment with the AT1R inhibitor candesartan did not restore retinal blood flow.

This study examined the role of microglia in the local control of the blood supply inside the retina. Microglia preferentially contact the retinal capillaries located in the superficial vascular plexus, as well as the neuronal synapses in the inner retina. The role of microglia in the regulation of blood vessels in the retina and brain is determined, and the addition of fractalkine induces capillary contraction. Subsequent characterization in the retina indicated that this vascular regulation depends on microglial cell contact and Cx3cr1 signaling. The microglial transcriptome contains gene transcripts of known vasoactive agents, and the AT1R inhibitor candesartan blocks capillary contraction, indicating that microglial vascular regulation may occur through the regulation of local RAS. This is supported by data showing fractakine-Cx3cr1 mediated upregulation of angiotensinogen. Under the background of vascular dysfunction during early diabetes, microglia vascular regulation was further explored. After 4 weeks of experimental diabetes, retinal blood flow decreased, consistent with the contraction of retinal capillaries in the superficial nerve plexus and the increase in microglia-capillary association. However, there is no indication of classic microglia activation, nor a more general macroglia response. RNA-seq data showed that the expression of microglia of the RAS component changed, and the function of microglia-mediated capillary contraction was lost during diabetes. Finally, candesartan treatment restored the retinal capillary diameter in STZ-treated animals; however, retinal blood flow was still reduced.

Current data show that microglia are closely related to the retinal vasculature, directly facing the pericytes and capillary areas without pericytes, but do not show a special preference for direct contact. Emphasizing the functional significance of this interaction, the microglial cell-specific receptor Cx3cr1 is stimulated by its sole ligand fractalkine to induce vasoconstriction, not only in the retina of mice and rats, but also in the brain. Although the role of fractalkine-induced vasoconstriction in the brain requires more work to confirm that microglia/Cx3cr1 is involved in the areas that exhibit contraction, in the retina, this effect is spatially discrete and only occurs with Areas related to microglial processes and dependent on Cx3cr1 signal, Cx3cr1GFP/GFP retina did not show contraction, altered microglial cell-vascular contact and capillary diameter. Although previously reported changes in the retina of Cx3cr1GFP/GFP mice may affect these vascular results (44), the data of the Cx3cr1 inhibitor AZD8797 shows the lack of fractalkine-induced contraction, which is consistent with the results of Cx3cr1GFP/GFP, directly implying that microglia are in The capillary reaction of fractalkine. Although previous work has identified microglia as a component of the blood-brain barrier (45) and is involved in the development of the retina and cerebrovascular (20, 46), this article on microglia-mediated vascular regulation The report is unique. In addition, our data and those of others indicate that microglia also monitor and regulate neuronal synapses during development, throughout adulthood, and in response to activity (17, 47, 48), which increases the size of Glial cells may contribute to the possibility of neurovascular coupling, through which local blood flow is regulated by neuronal activity. Since previous work in the retina has shown the existence of vascular regulation mechanisms that do not depend on Müller cells (11, 12), microglial vascular regulation may constitute an alternative approach, especially in the superficial nerve plexus. It is necessary to further explore the structure of microglia-neuron contact, its time response to changes in neuronal activity, and its fractalkine dose-response curve to correctly characterize the role of microglia in neurovascular units.

In order for microglia to directly mediate vasoconstriction, they must express vasoactive factors. RNA-seq data from isolated retinal microglia highlight several genes for vasoactive agents, including endothelin (Edn1, 3), angiotensinogen (Agt), and arachidonic acid 5-lipoxygenase (Alox5), all these genes are known to regulate retinal capillary tension (49). Although retinal neuron/glial cell contamination may confuse the genes identified in the microglia isolates, the low levels of neuronal characteristic genes (SI appendix, Figure S5) indicate that any effects are insignificant. Importantly, pre-incubation with the AT1R antagonist candesartan can inhibit microglia-mediated vasoconstriction, and incubation with fractalkine induces the up-regulation of microglia Agt expression, which was not observed during the ablation of Cx3cr1 gene. These data, together with the dysregulated microglia genes (Agt and Ahr) found during diabetes, indicate that RAS is involved in microglia-mediated vascular regulation. All components of RAS have been observed in the retina, and angiotensin II (AngII) participates in the vasoconstriction of all retinal blood vessels (arterioles, capillaries and venules) through AT1R (32, 33). Although this microglia-mediated vascular regulation by RAS is novel, it is known that microglia express components of this pathway, including angiotensin-converting enzyme, AT1R, AT2R (50). In addition to vasoconstriction by microglia RAS, the activation of microglia and the production of inflammatory cytokines after exposure to AngII in the brain and retina have also been described (51, 52). Therefore, normal vascular control may need to regulate the microglia RAS in normal tissues, and in the pathological process, there may be a positive feedback loop involving AngII, which promotes microglia activation and inflammation.

In view of the ultrastructural and immunocytochemical data indicating that microglia contact pericyte cells and processes, it is possible that microglia directly communicate with pericytes and use their vascular regulation ability (5) to contract capillaries in the retina. Supporting communication between the two cell types, pericytes can modulate the microglia phenotype during inflammation (53), while AT1R is expressed by pericytes to achieve AngII-mediated contraction (33). In addition, in vitro evidence indicates that pericytes express fractalkine, and both fractalkine and Cx3cr1 gene transcripts have been identified in the pericyte transcriptome (54, 55). In addition to pericytes, our data also suggests that microglia can trigger responses by directly communicating with endothelial cells (capillary regions without pericytes), which also express vascular regulators (56). Finally, microglia may communicate with blood vessels indirectly through other retinal glial cells, such as Müller cells, which express components of RAS (57) and have previously been shown to regulate the internal retinal vasculature (9, 10) . Although the proposed mechanism is shown in Figure 6, more work needs to be done to explain how microglia send signals to other members of the neurovascular unit to induce capillary contraction.

Schematic diagram of microglia regulation of retinal capillary contraction. The data from this study indicate that microglia can participate in neurovascular units both in structure and function. Microglia contact neuronal synapses and retinal capillaries (including pericytes) and activation of fractalkine-Cx3cr1 signals causes capillary contraction, which is through an AT1R-dependent mechanism. Ultimately, capillary regulation may occur through direct microglia mechanisms, or may involve the contribution of pericytes and Müller cells. EC, endothelial cells; PC, pericytes.

We found that the retinal blood flow of all retinal vessel types is reduced due to short-term hyperglycemia. This finding is supported by studies on diabetic human and disease animal models (25). Compared with the larger retinal vessels that did not show changes, the diameter of capillaries in the superficial nerve plexus was significantly reduced (~-9%). As far as we know, this is a unique discovery. Although the change in capillary diameter is small, it has a great impact on blood flow, because capillaries make up most of the retinal vasculature (58). One estimate suggests that a 6% expansion of the capillary diameter (~0.32 μm) produces most of the increase in blood flow caused by neuronal activity (5). In addition to static vascular changes, retinal capillaries from STZ-treated animals failed to contract after hyperoxia challenge. This report on the measurement of retinal capillary diameters in the body during a vascular challenge is unique; however, previous human studies reported on hyperoxic retinal vascular response (blood flow) in type 1 (59) and type 2 (60) diabetic patients Has changed.

Since changes in the capillary network are considered to be the basis of the pathophysiology of early and late DR (25, 61, 62), it is tempting to speculate that the microglia of these blood vessels contribute to the vascular dysfunction of early diabetes. Data showing an increase in the number of microglia protrusions associated with capillaries, an increase in microglia Agt expression, and capillary diameter recovery after candesartan cilexetil treatment support this hypothesis. Even the increase in microglia expression of the aryl hydrocarbon receptor (Ahr), a negative regulator of vasoconstriction (43), can be included in this theory because recent work has shown that it contributes to vascular stiffness (63). Therefore, the increased expression of Ahr and Agt may lead to a phenotype of smaller retinal blood vessels and weaker response in early diabetes. Additional support for the specific effect of microglia on the retinal vasculature during diabetes comes from studies of STZ-treated Cx3cr1GFP/GFP animals, which showed an increase in acellular capillaries after 4 months of hyperglycemia (64). It is necessary to use the STZ-processed Cx3cr1GFP/GFP model for further work to specifically explore the capillary contraction confirmed in the early stage of diabetes.

The dysregulation of microglia in RAS suggests that this pathway is altered in diabetes. These data are supported by our supplementary data (SI appendix, Figure S9) and previous studies, which showed increased intravitreal angiotensinogen in individuals with proliferative DR (65), increased AngII concentration in the vitreous of rodents, and retinal AngII Diabetes models with elevated levels of, AT1R and AT2R (66, 67). In addition to causing vasoconstriction, AngII is known to separate pericytes from endothelium, thereby changing vascular permeability and promoting the development of microaneurysms, which are a key clinical determinant of DR (33). To verify the positive effect of candesartan on capillary diameter and further support the role of RAS in DR, early clinical trials have shown that candesartan blockers can successfully prevent clinical-grade DR in non-DR diabetic patients (68) Since these beneficial effects have not been extended to prevent the progression of DR patients, this suggests that RAS disorders are related to the early preclinical stages of DR.

Therefore, when the current data and RAS disorders are considered together, the expression of fractalkine in the retina increases during diabetes [SI Appendix, Figure S9 and other studies (69, 70)], and a hypothesis can be proposed that fractalkine expression is also increased in early diabetes With the enhancement of microglia process-capillary interaction and changes in microglia RAS, capillary contraction increases. Although this potential role of microglia vascular regulation in DR is unique and different from its inflammatory role in the later stages of the disease (41), further research is needed to fully understand this early dysfunction and how it leads to later pathology For example, the hyperemia observed in patients attenuates diabetes (71, 72) and retinal hypoxia leading to advanced DR (38, 73).

Although the candesartan blocker restores the retinal capillary diameter to the control level in this study, the retinal blood flow is still reduced. This is surprising because considering the importance of the microvascular system, reversing capillary contraction is expected to increase retinal blood flow (5, 58) and previous work has shown that candesartan cilexetil can restore blood flow in diabetic rats. Although it was 2 weeks after STZ. 74). However, quantification of the arteriovenous ratio of candesartan-treated STZ animals showed an increase in the diameter of these larger blood vessels. These data, combined with previous work showing AngII-dependent contraction of arterioles and venules (75), indicate that the expansion of larger retinal blood vessels in candesartan-treated STZ animals may lead to a decrease in retinal blood flow velocity, which masks the dilatation. Capillaries. More targeted microglial RAS blockade may overcome these confusions and provide clearer images of telangiectasia and retinal blood flow.

In conclusion, this study established the role of microglia in regulating CNS capillaries, especially the retina. It emphasizes the involvement of the fractalkine-Cx3cr1 signal axis and implies that RAS is involved in the regulation of capillaries mediated by microglia in the early stages of DR in normal tissues. Although inhibiting the RAS pathway changes capillary constriction, it does not change the overall retinal blood flow in early-stage diabetic patients. Further research on the cellular mechanism of microglia-induced vasoconstriction and the intercellular signal conduction between microglia and other components of the neurovascular unit will provide valuable information for retinal vascular response in health and disease.

The animal procedure was approved by the Ethics Committee of the University of Melbourne (#1613867) and complies with the guidelines of the Australian National Health and Medical Research Council and the guidelines for the care and use of laboratory animals (76). To explore the role of microglia in the regulation of retinal blood vessels, Cx3cr1GFP/+ and Cx3cr1GFP/GFP mice were used. One or two alleles of their monocyte-specific receptor Cx3cr1 were replaced by EGFP (77) . In order to show that Cx3cr1 labels microglia in healthy retinas rather than infiltrating monocytes, immunohistochemistry was performed using selective markers (SI appendix, Figure S1). NG2-DsRed pericyte report mice are used to explore pericyte-microglia contact, and provided by BTS through a single intraperitoneal injection of STZ (55 mg/kg) in male adults (6 to 8 weeks old). Hyperglycemia was induced in agouti, trisodium citrate buffer, pH 4.5; Sigma-Aldrich), and control animals received the same amount of vehicle. A single group of animals was drinking candesartan cilexetil (10 µg/mL; Sigma-Aldrich, #SML0245) or vehicle (PEG400/ethanol/Kolliphor EL/water, 10:5:2:83; Sigma-Aldrich) water , 24 hours after diabetes induction. In vivo imaging and tissue separation were performed 4 weeks after diabetes.

Anesthetized Cx3cr1GFP/+ and Cx3cr1GFP/GFP animals (n = 5 and 6, respectively) were injected intraperitoneally with rhodamine B (Sigma-Aldrich) to label the blood vessels, because the IB4 marker on the live cell explants showed microglia crossover Reactivity (SI appendix, Figure S4). After 5 minutes, the animal was overdose (pentobarbital phosphate, 120 mg/kg), and the retina was dissected in frozen Ames medium (Sigma-Aldrich), which was prebubbled with carbon source (95% O2, 5% CO2). The retina was imaged on an inverted confocal microscope (Leica SP5) and perfused with carbonized Ames at 37 °C at a rate of 1 mL/min. Introduce recombinant rat fractalkine (200 ng/mL; R&D Systems, #537-FT-025/CF) or carrier (PBS) after 10 minutes of baseline recording, and image for another 10 minutes. At the end of this culture, measure the diameter of the blood vessel (2 to 4 individual capillary sites per retina, n = 4 to 6 retinas) where there is and no contact with microglia, and the measured value is expressed as the same blood vessel The percentage of the area's baseline diameter (taken as the average vessel diameter at the initial 10-minute baseline). For quantification, the vascular imaging channel was separated from the microglia imaging channel, and the scorer was unaware of microglia contact and treatment (PBS and FKN). In a separate experiment, the retina was pre-incubated with Cx3cr1 inhibitor AZD8797 (10 µM; MedChemExpress, #HY-13848) or carrier for 10 minutes before adding fractalkine (200 ng/mL). The ex vivo preparation was imaged for a total of 30 minutes to limit the variability of vessel caliber. Although this ex vivo preparation may limit retinal blood flow, all explants undergo the same treatment, and all effects are related to the initial baseline. The above protocol was used to measure the vascular response of STZ-induced diabetes to fractalkine after 4 weeks. At the same time, in order to evaluate the role of RAS in fractalkine-induced contraction, the isolated retina was exposed to Ames or Ames + 230 nM candesartan medoxomil (Sigma-Austria). Derich) 10 minutes. Then add Fractalkine (200 ng/mL) and image for 10 minutes (n = 5 fractalkine + candesartan; n = 7 fractalkine), at which time the vessel diameter is quantified relative to the pre-culture line. Although candesartan cilexetil is a prodrug and is usually activated during gastrointestinal absorption, carboxyl esterase is present in the retina (78). Our previous work has shown that candesartan cilexetil is directly delivered to the eye It sometimes blocks angiotensin-induced vascular effects (52).

For the hemodynamic analysis of diabetic animals, the Micron III rodent imaging system (Phoenix Research Labs) (35) was used to perform in vivo video fluorescein angiography (VFA) (n = 21 per group) as described previously. This technique uses sodium fluorescein (1%, 100 µL/kg, fluorescein 10%; Alcon Laboratories) to reliably quantify hemodynamics. Additional details are provided in the SI appendix. Record the time it takes for fluorescein to enter the retina to half of the maximum intensity (filling time), and the time it takes for 30 seconds after imaging to drop from maximum intensity to the midpoint between maximum intensity and final intensity (drainage time).

In order to evaluate the capillary diameter and capillary hyperoxia response in vivo, OCTA (OCT2 Spectralis, Heidelberg Engineering) was performed. OCTA uses motion contrast imaging to generate real-time angiograms of the retinal vasculature (79). The volume scan (15 × 15° area of ​​interest) is performed from two to three disc diameters of the optic nerve. Each area is composed of 512 B scans, and each B scan is composed of 512 A scans. Scan the upper and lower retina with both eyes. The animals were exposed to 100% oxygen through a nose cone (3 L/min), and the vascular response to hyperoxic conditions was measured in 4 weeks of STZ treatment and control animals. After taking the baseline image, after breathing oxygen for 2 minutes, use the follow-up mode to obtain a second capillary image at the same retinal position.

As mentioned above (80, 81), rat, mouse and human retinas were treated with indirect immunofluorescence in whole or in cross-section. Retinal microglia were labeled with rabbit anti-ionic calcium binding adapter molecule 1 (Iba-1, 1:1,000; Wako) or expressed EGFP (Cx3cr1GFP/+, Cx3cr1GFP/GFP), while blood vessels were labeled with Griffonia simplicifolia IB4 (FITC 1 :75; Sigma-Aldrich; 647 fluorophore 1:100; Thermo Fisher Scientific). Although IB4 has shown cross-reactivity with brain microglia and activated retinal microglia (82, 83), we have not observed cross-reactivity in fixed retinal tissues. Compared with the endothelial marker CD-31, we also showed better vessel coverage using IB4 (SI appendix, Figure S10). More details on immunolabeling are in the SI appendix. All imaging was performed on Zeiss META/LSM800 confocal (Carl Zeiss) or Leica SP5 (Wetzlar) at 20 times, while the high-resolution imaging of microglia-pericyte contact and EGFP-expressing microglia was performed at 63 Times to proceed. For the subsequent analysis of the overall installation of the retina, a flat scan was performed in the superficial vascular plexus, and the z-stack (15.6 µm) was used to adapt to changes in the installation of the retina. All subsequent image analysis is performed on the maximum intensity projection.

The open source plug-in ARIA (84) was used to analyze the arteriolar/vein width and tortuosity of fundus images (n = 13 per group) in MATLAB (Mathworks), with an eccentricity of 1.5 and 2 optic disc diameters. AngioTool (85) was used to measure the capillary width (<15 μm) in the superficial vascular plexus (OCTA). The confocal overall image (n = 11 animals in each group) is divided into arterioles, venules and capillaries according to their corresponding VFA contours and vascular masks, which are used in the Metamorph (molecular device) using the angiogenesis tube formation application Subsequent analysis is carried out in segments. The total blood vessel area of ​​each blood vessel type is quantified in NIH ImageJ (86), and the blood vessel density is expressed as a percentage of the blood vessel area covering the total area of ​​the retina. For all subjective measurements, the individual is unaware of the treatment group.

Microglia, pericytes, and astrocytes from STZ-treated and control tissues were analyzed in Metamorph using the axon growth application. Iba-1+ microglia are segmented, counted and masked. The microglia mask is overlaid on the blood vessel/pericyte mask. Cells that overlap the blood vessel by at least 0.82 µm are considered to be in contact and calculated as a percentage of the total cells. For Cx3cr1GFP/+ and Cx3cr1GFP/GFP microglial blood vessels and neuron contacts in the retina, the co-localization area between a single microglia cell and blood vessels and synapses is rendered as a three-dimensional volume and expressed as the total microglia Percentage of volume (Imaris, Bitplane; three microglia per retina per quadrant, n = 5 animals per genotype). To further characterize the microglia-pericyte interaction, use a custom Metamorph script to quantify the contact between microglia (Iba-1+) and pericyte cell bodies and processes (NG2+) (within 0.41 µm), and Areas of capillaries that are not in contact with pericytes (NG2-, IB4+, possibly endothelial cells). Previous work used EGFP to assess the contact of microglia with neurons in the retina and brain (17, 47, 87). The automated neurite outgrowth application (Metamorph) was also used to quantify microglia morphology, while microglia-neuronal synapses and microglia-pericyte images were processed in Imaris. The density of astrocytes in the ganglion cell layer is quantified as the total retinal area and the overlap with each blood vessel type. The quantification of Müller's gliosis is as previously described (44) (3 sections per animal, n = 6 animals).

Retinas from control and 4-week STZ-treated rats (n = 5 control, n = 4 STZ, 12 weeks old) were isolated, papain digested (Worthington Biochemical), and labeled with CD11b-FITC conjugate (Miltenyi Biotec) Used for microglia separation (FACSAria III, BD Bioscience). Collect CD11b-FITC+ (microglia positive) and CD11b-FITC- (microglia negative) parts. Isolate RNA and perform RNA-seq in accordance with the SI appendix. To explore the fractalkine regulation of microglial RAS, retinas from C57bl6 and Cx3cr1GFP/GFP animals (n = 6) were incubated with fractalkine (200 ng/mL; R&D Systems) or PBS for 2 hours at 37°C. Isolation of retinal microglia by FACS using CD11b and EGFP tags. Isolate RNA and perform Smart-seq. 2 Performed 13 cycles of pre-amplification, followed by quantitative PCR (SI appendix).

Statistical significance is determined by a two-tailed unpaired Student's t-test, two-way analysis of variance, or repeated measures analysis of variance, depending on the experiment (Prism 6.0, GraphPad). When needed, a Tukey post-mortem analysis was performed. Median regression analysis (STATA, StataCorp) was used for blood flow analysis. The Alpha level is set to 0.05. Unless otherwise stated, all values ​​are expressed as SEM.

The data reported in this article has been stored in the Gene Expression Omnibus (GEO) database, https://www.ncbi.nlm.nih.gov/geo (accession number GSE139276) (88).

We thank Dr. Leonid Churilov for his help in statistical analysis; Dr. Christine Nguyen and Mr. Darren Zhao for his help in blood gas analysis; Ms. Satya Gunnam for his help in immunohistochemical staining and analysis; and the University of Melbourne's bio-optical microscope and Melbourne cells Technology (MBC node) platform. This work was funded by the National Health and Medical Research Council (APP-2000669, APP-1138509) and the Australian Research Council (DP160102642, DP200102001, FT130100338).

↵1S.AM, AIJ and MAD made the same contribution to this work.

Author contributions: SAM, AIJ, MAD, BVB and ELF design research; SAM, AIJ, MAD, BVB, KAV, JAP, UG, GV, VHYW, CHYW, ZH, FH, JCY, JT, EI, and BTS conducted research; SAM, AIJ, MAD, KAV, UG, GV, VHYW, CHYW, ZH, FH, JCY, JT, EI, BTS and ELF analysis data; SAM, AIJ, BVB and ELF wrote this paper.

The author declares no competing interests.

This article is directly contributed by PNAS.

This article contains online support information at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2112561118/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.