Infiltrative microgliosis: activation and long-distance migration of subependymal microglia following periventricular insults
Subventricular microglia (SVMs) are positioned at the interface of the cerebrospinal fluid and brain parenchyma and may play a role in periventricular inflammatory reactions. However, SVMs have not been previously investigated in detail due to the lack of a specific methodology for their study exclusive of deeper parenchymal microglia.
We have developed and characterized a novel model for the investigation of subventricular microglial reactions in mice using intracerebroventricular (ICV) injection of high-dose rhodamine dyes. Dynamic studies using timelapse confocal microscopy in situ complemented the histopathological analysis.
We demonstrate that high-dose ICV rhodamine dye injection resulted in selective uptake by the ependyma and ependymal death within hours. Phagocytosis of ependymal debris by activated SVMs was evident by 1d as demonstrated by the appearance of rhodamine-positive SVMs. In the absence of further manipulation, labelled SVMs remained in the subventricular space. However, these cells exhibited the ability to migrate several hundred microns into the parenchyma towards a deafferentation injury of the hippocampus. This "infiltrative microgliosis" was verified in situ using timelapse confocal microscopy. Finally, supporting the disease relevance of this event, the triad of ependymal cell death, SVM activation, and infiltrative microgliosis was recapitulated by a single ICV injection of HIV-1 tat protein.
Subependymal microglia exhibit robust activation and migration in periventricular inflammatory responses. Further study of this population of microglia may provide insight into neurological diseases with tendencies to involve the ventricular system and periventricular tissues.
KeywordsEpendymal Cell Left Lateral Ventricle Parenchymal Microglia Stereotaxic Head Holder Periventricular Tissue
List of abbreviations
bovine serum albumin
stratum oriens of the hippocampus contralateral to stab lesion
green fluorescent protein
modified Gallyas silver degeneration stain
human immune deficiency virus 1 encephalitis
phosphorylated/activated extracellular signal-regulated kinase
rhodamine latex microbeads
standard error of the mean
tumor necrosis factor alpha.
It has become increasingly evident that the central nervous system is an immunocompetent organ . Microglia are the primary immune effector cells of the brain parenchyma and functionally resemble tissue macrophages elsewhere in the body [1, 2]. The brain ventricles are also under immune surveillance by intraventricular macrophages which patrol the cerebrospinal fluid (CSF), choroid plexus, and supraependymal surface . At the interface of the CSF and brain proper is the ciliated neuroepithelial ependymal cell which lines the ventricular system of the brain and spinal canal. The ependyma not only functions as a physical barrier preventing foreign proteins and organisms from entering the brain from the CSF, but also displays immunological effector ability such as phagocytosis of fluorescent beads injected into the CSF  and upregulation of MHC-II in response to interferon gamma challenge in vivo . These diverse cell types may work in concert establishing the basis for the innate immune system of the CNS.
Importantly, a population of resident subventricular microglia (SVMs) are found in the subependymal zone [6, 7] suggesting the ependyma and microglia may cooperate to prevent invasion of the CNS from the ventricular system. Juxtaventricular microglia have been shown to react to both direct periventricular/ependymal damage as well as the mere presence of cytokines in the CSF. For instance, the intracerebroventricular (ICV) injection of lentiviral tat protein in low nanomolar quantities is sufficient to kill ependymal cells and cause a periventricular inflammatory reaction including the characteristic microglial nodules of human HIV-1 encephalopathy (HIVE) . Alternatively, Kong et al , noting the high CSF levels of inflammatory cytokines in multiple sclerosis (MS), demonstrated a vigorous periventricular activation of microglia with ICV injection of IFN-γ alone or in combination with endotoxin or TNF-α. In these cases, microglia were activated in the absence of primary tissue damage, but were thought to contribute secondarily to immune-mediated periventricular damage via potentiation of cytokine release and bystander effect. Of note, both HIVE and MS often present with enigmatic periventricular inflammatory lesions in humans [10, 11, 12, 13].
The reaction of SVMs to periventricular damage has not been described in detail. Specifically, the functional repertoire displayed by activated SVMs including phagocytosis and long-distance migration have not been investigated. Here, we directly examine SVM function by exploiting a novel methodology which selectively activates and labels SVMs in vivo combined with confocal timelapse techniques for dynamic analyses in adult mouse brain tissue. We hypothesized that SVM activation is a general consequence of periventricular insults that can be caused by diverse circulating substances in the CSF known to damage the ependyma. Our work indicates that activated, phagocytic SVMs are capable of infiltrating deep within the parenchyma.
Adult male C57bl/6 mice (6–8 wks) were obtained commercially from Harlan (Indianapolis, IN) and cared for in accordance with Public Health Service and University of Virginia guidelines.
To damage the ependyma 2–3 μl of 0.2% Sp-DiI (D-7777, Molecular Probes) in DMSO; 1:20 rhodamine-conjugated latex microspheres (Lumafluor, Naples, FL) in sterile PBS; 0.25 U Neuraminidase (Sigma); or 2.0 nM recombinant HIV-1 tat protein in 100 mg/ml BSA, 0.1 mM DTT in PBS were stereotaxically injected into the left lateral ventricle at the following coordinates: L, 1.5 mm; P, -0.5 mm; D, 2.0 mm. GFP-expressing adenovirus (109 plaque forming units); 100 mg/ml BSA, 0.1 mM DTT in PBS, or deactivated tat  were used as volume-matched controls. Forebrain stab lesion for deafferenting lesion of the contralateral hippocampus was performed as previously described . Briefly, mice were anesthetized with a Xylazine/Ketamine mixture and placed in a stereotaxic head holder (Benchmark, myneurolab.com). Temperature was maintained with a ventral heat pad. A right parietal craniectomy extending 3–4 mm from midline and spanning Lambda to Bregma was created with a microdrill. Beginning at the level of Bregma, a 3 mm lesion was created in the sagittal plane 1.5 mm from midline at a depth of 3.5 mm with a sterile no. 11 scalpel blade held in the stereotaxic device. After achieving hemostasis, the bone was replaced and sealed with carboxylate cement (Durelon, Fisher Sci). This right sided lesion results in deafferentation injury to primarily stratum oriens of the left hippocampus.
Brains were collected and processed for histopathology as previously described . All antibodies and staining procedures have been described previously  except for rat anti-cd11b/MAC-1 (1:50), mouse anti-foxj1 (1:1000), mouse anti-nestin/rat-401 (1:100), and rat anti-F4/80 (1:10). Histochemistry for biotinylated Griffonia simplicifolia lectin IB4 was performed 1:100 in PBS overnight at 4°C and visualized with either 1:200 FITC- or Alexa Fluor 350-conjugated streptavidin (Sigma and Molecular Probes, respectively).
200–400 μm live slices were prepared from adult C57BL/6 mice as described previously [16, 17]. Briefly, mice were acutely anesthetized in a chamber with halothane and decapitated. The brain was rapidly removed, blocked, and covered with 10% agar at 37°C in a specimen mold (Tissue-Tek 4566, Fisher). Live slices were obtained with a vibratome and placed individually on Millicell-CM inserts (Millipore PICM03050, Fisher). Culture medium consisted of CCM1 (Hyclone, Logan, UT) with 20% heat-inactivated normal horse serum. Vital labelling of microglia in tat experiments was performed with Alexa 488 or 568 IB4 (Molecular Probes) . Laser-scanning confocal images were acquired on a Nikon IX-70 inverted microscope with Fluoview 300 software (Olympus). A z-series stack covering 40 μm of slice thickness was taken every 1.5–4 minutes, creating a three-dimensional timelapse data set. To create timelapse movies from the data set, 4 to 6 z-plane images were collapsed as 2D projections using ImageJ 1.31 u and compiled into quicktime movies with Quicktime Pro 6.3. Movies were analyzed for migration speed and distance as described . Values represent the mean ± SEM. Statistical analysis was performed with ANOVA or student's t-test. Pairwise post-hoc analysis was performed with a t-test and the Bonferroni correction factor. A p < 0.05 was considered statistically significant. Rose plots were created in Kaleidagraph 3.0.
Selective ependymal death is induced by high-dose rhodamine dyes
SVMs phagocytose ependymal debris
SVMs may become RHO+ as a result of phagocytosis of the dye-labeled ependymal debris. To provide direct evidence of this hypothesis we performed confocal time-lapse microscopy in living slices from adult mice given dye injection 24 h prior to sacrifice. Grossly, we observed a dramatic increase in RHO+ periventricular cells over several hours suggesting active clearance of labelled debris (not shown). Further, SVMs displayed dynamic behavior consistent with phagocytosis of ependymal debris (Fig. 2c, Video 1(Additional file 1)). Therefore, SVMs became rhodamine positive after high-dose dye injection due to phagocytosis of labelled ependymal debris.
To determine if SVM activation is a general response to periventricular damage we injected animals with a sublethal dose of RhoB to label ependymal cells without causing damage. 24 h later we injected 0.25 U neuraminidase to damage the ependyma  via an alternate mechanism. Histological examination of sections at 7 d revealed loss of the ependyma and the presence of RHO+ SVMs (Fig. 2d, left panels). Control injection of PBS vehicle alone resulted in no ependymal damage or SVM labelling (Fig. 2d, right panels). Thus, ependymal cell damage of diverse etiologies incites a reactive response by SVMs including phagocytosis of debris.
Long-distance infiltration by SVMs after parenchymal injury
ICV injection of HIV-1 tat protein causes IMG
We have shown that SVMs represent a pool of microglial cells which are highly reactive to periventricular damage and are responsible for clearance of resulting cellular debris. Further, activated SVMs are capable of migrating away from the ventricle towards injury cues from damaged regions in the parenchyma several hundred microns away. We confirmed these findings dynamically in acute slice preparations from adult mice. Both the juxtaventricular origin and the extensive migratory capacity of activated SVMs have important implications for neurobiology and disease.
Periventricular/subependymal microglia have been noted by histologists since microglia were first identified as a distinct cell type (reviewed in ). Little attention has been paid to these cells in the literature until recently due to their intimate arrangement among the subventricular neural progenitors [6, 7]. Possible phenotypic differences between SVMs and parenchymal microglia have not been investigated, however, the location of SVMs among stem cells and within close proximity to ependymal cells and ventricular CSF is unique. A few neurological diseases demonstrate altered CSF constituents and often pathological involvement of the periventricular tissues [8, 9, 10, 11, 12, 13]. Therefore, the specific function of microglia in these specialized regions of the brain under normal and pathological conditions deserve further investigation.
While evaluating rhodamine dyes for selective labeling of ependymal cells  for other studies we discovered that the ependyma of the injected hemisphere became rapidly damaged after dye uptake. Upon death of the ependyma, SVMs phagocytosed the ependymal debris, and thereby became rhodamine-positive. This serendipitous finding results in rapid and selective labelling of SVMs allowing study of this specific cell population in vivo. For instance, in this study we were able to demonstrate phagocytosis, mitosis, and migration of activated SVMs using histopathological techniques alone. We have shown these cellular activities can be confirmed in situ with timelapse confocal microscopy further validating this versatile protocol. Mechanistic investigations are possible by combining our in vivo and in situ protocols with genetic or pharmacological techniques.
We found SVMs only infiltrated the brain after selective ependymal damage with rhodamine dyes if a distant lesion was also present, likely providing a gradient of chemoattractive cues. CC chemokines are known to be upregulated rapidly after deafferenting injury of the hippocampus . The extensive migration of SVMs in response to HIV-tat injection, on the other hand, may be due to a direct effect of tat on microglia or possibly an indirect effect due to upregulation of chemokines by neurons and glia [27, 28]. Further, that ICV injection of recombinant HIV-1 tat protein alone is sufficient to damage the ependyma, activate SVMs, and incite infiltrative microgliosis supports the "cytokine dysregulation hypothesis" [8, 22] of damage in HIV-1 encephalitis whereby overactivation of microglia/monocytes may be more critical than actual CNS viral load [33, 34].
In summary, we have shown that SVMs are a highly reactive pool of cells which, when activated, can infiltrate the parenchyma in response to injury cues from damaged brain regions or exposure to HIV-1 tat. These findings provide new in vivo and in situ models for the study of SVM function, further insight into microglial dynamics after brain injury, and novel hypotheses for the role of microglia in periventricular reactions in neurological diseases.
We thank Stephen Brody (Washington University) for the gift of foxj1 antibodies; Rooshin Dalal and Claire Brown for expert assistance with confocal microscopy; John Stock for expert assistance with electron microscopy, Sean Aeder and Isa Hussaini for the gift of the Ad-GFP; Isa Hussaini, Akira Sakakibara, and Scott Vandenberg for helpful discussion; and Margo Roberts for comments on this manuscript. The following reagents were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: TAK-779 and HIV-1 Tat protein from Dr. John Brady and DAIDS, NIAID. This research was supported by National Institutes of Health Grants NS-047378 (to J.W.M.) and GM-232442 and GM-064346 (to A.F.H.). W.S.C. was supported by the UVA Medical Scientist Training Program, a Raven Society Fellowship, and an award from the National Neurotrauma Society.
- 22.Grondona JM, Perez-Martin M, Cifuentes M, Perez J, Jimenez AJ, Perez-Figares JM, Fernandez-Llebrez P: Ependymal denudation, aqueductal obliteration and hydrocephalus after a single injection of neuraminidase into the lateral ventricle of adult rats. J Neuropath Exp Neurol. 1996, 55: 999-1008.CrossRefPubMedGoogle Scholar
- 27.Albini A, Ferrini S, Benelli R, Sforzini S, Giunciuglio D, Aluigi MG, Proudfoot AE, Alouani S, Wells TN, Mariani G, Rabin RL, Farber JM, Noonan DM: HIV-1 tat protein mimicry of chemokines. Proc Natl Acad Sci USA. 1998, 95: 13153-13158. 10.1073/pnas.95.22.13153.PubMedCentralCrossRefPubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.