Focal Transient CNS Vessel Leak Provides a Tissue Niche for Sequential Immune Cell Accumulation during the Asymptomatic Phase of EAE Induction

Abstract

Peripheral immune cells are critical to the pathogenesis of neurodegenerative diseases including multiple sclerosis (MS) (Hendriks et al., 2005; Kasper and Shoemaker, 2010). However, the precise sequence of tissue events during the early asymptomatic induction phase of experimental autoimmune encephalomyelitis (EAE) pathogenesis remains poorly defined. Due to the spatial-temporal constrains of traditional methods used to study this disease, most studies had been performed in the spine during peak clinical disease; thus the debate continues as to whether tissue changes such as vessel disruption represents a cause or a byproduct of EAE pathophysiology in the cortex. Here, we provide dynamic, high-resolution information on the evolving structural and cellular processes within the grey matter of the mouse cortex during the first 12 asymptomatic days of EAE induction. We observed that transient focal vessel disruptions precede microglia activation, followed by infiltration of and directed interaction between circulating dendritic cells and T cells. Histamine antagonist minimizes but not completely ameliorates blood vessel leak. Histamine H1 receptor blockade prevents early microglia function, resulting in subsequent reduction in immune cell accumulation, disease incidence and clinical severity.

INTRODUCTION

Although disruption of CNS vessel integrity is a hallmark of MS, it is not clear whether such vascular compromise is the initiator or the result of the neuroinflammatory pathophysiology (Lassmann et al., 2007). Gross vascular compromises have been reported as an early onset indicator in the cerebellum and spine (Floris et al., 2004; Koh et al., 1993; Muller et al., 2005; Tonra et al., 2001). Current available data, however, does not adequately describe the nature and dynamics of the vessel leak. Previous studies on rats have shown: (1) via radiolabelling, that vessel leaks can be detected in the lower sacral region one day before clinical disease, (Koh et al., 1993) and; (2) via MRI, vessel leaks in the cerebellum and spine are detected by gadolinium on day 11 preceding macrophage infiltration and onset of clinical disease on day 14 (Floris et al., 2004). These observations are of major breaches in the vessel integrity, generally in the mm to cm size range. On the other hand, early vascular and associated tissue changes in the pre-symptomatic murine cerebral cortex undergoing EAE pathogenesis are less well defined.

Current dogma dictates that during EAE induction the blood-CNS barrier is rendered “leaky” by irradiation or pertussis toxin (PTx) through histamine-mediated effects (Linthicum et al., 1982). As engagement of H1R induces microglia activation (Dong et al., 2014), it is believed that microglia activation occurs during early EAE induction phase (Ponomarev et al., 2005). However, pathogenic T cell infiltration does not require microglial expression of the major histocompatibility antigen class II (MHCII) molecule (Greter et al., 2005), suggesting that another antigen-presenting cell (APC) subset, the dendritic cells (DCs), plays an crucial role in re-activating myelin-specific T cells in the CNS (Bailey et al., 2007). How circulating DC precursors are attracted to EAE lesions is not entirely clear.

Myelin-specific Th17 and Th1 T cells play critical roles in EAE and MS (Korn et al., 2009). Intravital two-photon microscopy (TPM) studies have examined the recruitment and behavior of pathogenic T cells in the mouse brain and spinal cord (Bartholomaus et al., 2009; Flügel et al., 2007; Herz et al., 2011; McGavern and Kang, 2011b; Mues et al., 2013; Siffrin et al., 2010). These investigators focused their observation on T cell behavior during peak clinical symptomatic phase. As such, these studies did not report how such T cells access the brain during the transition from the initiation stage to the infiltration stage of EAE pathogenesis (Lodygin et al., 2013a; Odoardi et al., 2012; Sallusto et al., 2012). However, entry of T cells into the spinal cord in early EAE has been documented previously (Gordon et al., 2001; Kim et al., 2010; Rothhammer et al., 2011).

Here, we applied sequential TPM to examine cellular behavior at the tissue interface between the subarachnoid space (SAS) and cortex during the first 12 days of asymptomatic induction phase of EAE. We observed transient vessel leaks within the cortical grey matter within the first 3 days of disease induction, which initiated microglia activation, leading to sequential accumulation of DCs followed by myelin-specific T cell infiltration. We further showed that inhibiting histamine-induced microglia activation early in EAE induction could reduce disease progression. Specifically, we described early focal, intermittent and transient disruptions of post-capillary junctional venules following PTx administration, resulting in dextran uptake by microglia, and immune cellular accumulation. Transient vessel leaks initiated an early parenchymal inflammatory response first evidenced by microglial uptake of blood contents, followed by an influx of both CD11b⁺CD11c⁺ and CD11b⁺CD11c⁻ APCs. The increasing numbers of activated CNS-resident and blood-derived APCs on days 3–6 facilitated the ingress and retention of activated T cells in a myelin-specific manner into the Virchow-Robin Space (VRS) and the parenchyma on days 6–12 before the clinical onset of disease. In contrast to other intravital imaging studies involving the spine, we focused our imaging effort on the cortex, which is relevant to human MS. Interestingly, a distinct pattern of immune cells congregated in patchy tissue niches that were in close proximity to sites of prior microvessel leaks. Early activation of CNS APCs and antigen recognition by T cells in the cortex subsequently affected the pattern of T cell migration in the cerebellum and spine. In addition, we show that histamine H1 receptor antagonist curtailed the frequency of PTx-induced focal vessel leakage, reduced the degree of microglia activation, diminished subsequent recruitment of circulating immune cells, and protected mice from EAE incidence and disease severity.

MATERIALS AND METHODS

Mice

2D2 (Bettelli et al., 2003) mice expressing the TCR transgene specific for MOG35-55 peptide / I-A^b (Jackson Labs, #006912) were crossed with B6.129 (ICR)-Tg actin-beta CFP mice (Jackson Labs, #004218) to derive mice expressing 2D2-CFP cells for in vivo tracking by TPM. Similarly, OTII mice expressing the TCR transgene specific for OVA323 339 peptide / I-A^b (Taconic, #1896) were crossed with C57BL/6-Tg ubiquitin GFP mice (Jackson Labs, #004353) to derive OTII-GFP cells for in vivo TPM tracking. C57BL/6J, B6.FVB-Tg CD11c-DTR/GFP mice (Jackson Labs, #004509) and B6.129P CX₃CR1^+/GFP mice (Jackson Labs, #005582) were obtained from Jackson Laboratory. All animals were housed and handled in accordance with protocols approved by the CWRU IACUC.

Active EAE induction

EAE was actively induced in 8–12 week old female mice as previously described (Stromnes and Goverman, 2006). 3×10⁶ naive splenic 2D2-CFP CD4 T cells, isolated by negative depletion using Dynal beads (Life Technologies, Grand Island, NY, USA) were injected into recipient mice. 24 hours later, an emulsion with 200μg myelin oligodendrocyte glycoprotein (MOG) peptide 35 55 (MOG35 55; Anaspec, San Jose, CA, USA) with CFA (8 mg/mL H37RA and Incomplete Freund’s Adjuvant (IFA); Difco Laboratories, Detroit, MI, USA) was injected subcutaneously (s.c.) bilaterally on the lower back of the recipient mouse (Hooke Laboratories, Lawrence, MA, USA). Control mice were injected s.c. bilaterally with an emulsion consisting of PBS and CFA. 100 ng Pertussis toxin was administered intraperitoneal (i.p.) on days 0, 1 and 2. Daily clinical assessment of mice was performed. Clinical scores were assigned as previously described (Mi et al., 2007; Stromnes and Goverman, 2006).

Blood brain barrier leakiness analysis

To compare the pattern of vessel leakiness induced by different inflammatory stimulants, 100ng LPS (Sigma-Aldrich, St. Louis, MO, USA), 100ng PTx or 200 μL of PBS were injected i.p. into C57BL/6 mice on days 0, 1 and 2. 700ng of 150kDa TRITC dextran (Sigma-Aldrich, St. Louis, MO, USA) vessel dye was injected i.v. on day 1 and day 2. On day 3, QTracker-655 (Life Technologies, Grand Island, NY, USA) was injected to highlight the vessels. The mice were sacrificed and whole brain tissues were imaged immediately by TPM.

Priming of antigen-specific T cells

To determine whether myelin specific T cells require antigen-specific APCs for recruitment to the CNS, EAE was actively induced using a combination of two antigenic peptides in 8–12 week old female recipient mice. 3×10⁶ naïve 2D2-CFP and 3×10⁶ naïve OTII-GFP CD4⁺ T cells were isolated separately by negative depletion using Dynal beads and co-injected into naive recipient mice. 24 hours later, an emulsion of 200 μg MOG35-55 only, 10 μg OVA323-339 only, or 200 μg MOG35-55 and 10 μg OVA323-339 combined (Anaspec, San Jose, CA, USA) with CFA (8 mg/mL H37RA and IFA; Difco Laboratories, Detroit, MI, USA) was injected s.c. bilaterally on the lower back of the recipient mice. Control mice were injected s.c. bilaterally with an emulsion of PBS and CFA. 100 ng PTx (List Biological Laboratories, Inc, Campbell, CA, USA) was administered i.p. on days 0, 1 and 2 to induce EAE.

Flow cytometry analysis

For single-cell APC and T cell characterization, brain tissue was harvested on days 0 through 12 of EAE induction. Tissues were dissociated by collagenase D (1 mg/mL) and DNAse (250 units/mL) (Sigma-Aldrich, St. Louis, MO, USA) to form a single cell suspension after gentle dissociation with a dounce homogenizer and subjected to a 70%/37%/30% Percoll (Sigma-Aldrich, St. Louis, MO, USA) gradient (Cardona et al., 2006). Isolated cells were stained for CD45, CD11c, CD11b, CD4, CD8, IL-17, IFN-γ and FoxP3. Absolute numbers of cells per brain were determined by multiplying the number of events on FACS per collection volume by the total volume of cell samples. Spleen and LN were isolated from mice at the end of imaging on Day 12. Organs were crushed with the ends of 1ml syringes through a 40-μm cell strainer into single cell suspensions and were stained for Vα2 (OTII TCRα chain), Vβ11 (2D2 TCRβ chain), CD4, CD25, CD44, and CD69. Samples were collected and analyzed using the Accuri flow cytometer (BD Biosciences Co, San Jose, CA, USA).

Tissue histology analysis

After injection of TRITC vessel dye and completion of intravital imaging, CX3CR1^+/GFP mice were sacrificed and subjected to transcardial perfusion with 4% PFA. The brain tissue was harvested and 12-μm cryo-sections were performed and the samples were then subjected to confocal microscopy to obtain fluorescent images.

Intravital TMP imaging

Cranial windows were created as previously described (Mostany and Portera-Cailliau, 2008). The window surgery was performed 4 to 10 days before EAE induction to allow the local surgical site inflammation to subside. On the day of intravital imaging, 150kDa TRITC dextran or QTracker 655 was injected i.v. 10 minutes prior to imaging to highlight the vasculature. Image acquisition was conducted using Leica SP5 microscope system on an upright DM6000 stage (Leica Microsystems, Inc.) fitted with a 20X (N.A. 1.0) water immersion lens. The SP5 microscope was coupled with a Ti/sapphire laser (16W Cameleon XR laser, Coherent, Inc) tuned to 880 nm to illuminate fluorescent elements within the live tissue. An image volume of 750 × 750 × 150 μm was routinely collected every 30 seconds to yield xyzt imaging data set.

Hydroxyzine treatment

Oral administration at 5mg/kg/day was used for treatment with hydroxyzine. For the incidence curve, hydroxyzine was administered via drinking water at a concentration of 0.1mM starting from day −1. Later, for FACS and TPM analyses, 5 mg/kg of hydroxyzine was administered via daily gavage from day −1 to day 12.

Imaging data analysis

Second harmonic generation (SHG) signal produced by high collagen content structures were used to identify the pial layer and to discriminate parenchyma from meninges. Relative fluorescent intensity region of interest (ROI) measurements of dextran leaking into the parenchyma were made using the LAS-AF software (Leica Microsystems Inc, Buffalo Grove, IL, USA). Timing and duration of vessel leaks were manually processed. Cellular dextran uptake and CD11c-GFP were identified by fluorescent intensity. Co-localization of TRITC and GFP signals and volume data were calculated using the Imaris software (BitPlane Inc, Zurich, Switzerland). T cell motility data was also analyzed using the Imaris software, which allows cell identification, as well as 3D tracking over time by producing information on individual positions, time and track length. T cell interactions with APCs were grouped into short contacts (< 2 min) and prolonged interactions (≥ 2 min). Individual T cell migration tracking was further analyzed for track length, velocity, mean fluorescent intensity and cell flux using MATLAB software (Mathworks Inc, Natick, MA, USA). Cell flux was calculated by determining the vector between the beginning of the T cell track and the point at which the T cell came within a 50-μm radius from the center of an APC cluster. Contact time between APC and T cells, and T cell movement to and from the blood vessel were analyzed manually by visual inspection of individual image slices over time using the Imaris software.

Statistical consideration

Statistical analyses were performed using GraphPad Prism 5 software (GraphPad Software Inc, San Diego, CA). For all analyses P < 0.05 was considered significant and degrees of freedom (df) are noted.

RESULTS

Transient focal vessel leaks occurred early in EAE

EAE induction efficiency and clinical presentation can be variable. Therefore, to reliably capture rare cellular events by TPM we introduced fluorescent-labeled naïve 2D2 cells into recipient animals prior to disease induction in combination with a low dose of PTx on the first 3 days of induction without irradiation (Stromnes and Goverman, 2006). This resulted in 100% disease incidence compared to mice without transferred 2D2 cells (+2D2: n=6 from 2 experiments; −2D2: n=20 from 6 experiments), with hind limb paralysis achieved by day 21 (Figure 1a). We focused our intravital observation in the cortex during the first 12 days of asymptomatic EAE induction (Luo et al., 2008). A cranial observation window was surgically implanted 4 to 10 days prior to induction in order to limit the inflammatory changes to structures and cellular response (Barkauskas et al., 2013; Dorand et al., 2014).

Figure 1. PTx administration caused early, transient and focal CNS vessel leak.

(a) Consistent clinical scores were achieved in mice pre-transferred with naïve 2D2 T cells (■; n=6 from 2 experiments) as compared to non-transferred controls (□; n=20 from 6 experiments) prior to EAE induction. (b) Dynamic TPM imaging of pial vessels (red) on day 3 after EAE induction. Scale bar = 50 μm. (c) Mean fluorescent intensity (MFI) measurements of the indicated ROI’s in (b) over time revealed the transient and focal nature of the vessel leaks in ROI 1 and 2. (d) Control mice with cranial window implantation and 7 days of recovery. (e) The duration of the leaks in (d) was short. There were no flares detected after day 3, up to 21 days. N=4 from 1 experiment. (f) Frequency of cerebral vessel leaks was quantified by TPM during the first 12 days of EAE induction. Arrows: PTx injections. N=19 from 8 independent experiments. (g) The duration of individual vessel leaks in (f) was measured dynamically by TPM. Arrows: PTx injections. (h) ROI 5 (a cross-sectional view of ROI 1 in (b)) revealed a slower clearance of vessel dye under the pial surface in the parenchyma (P) as compared to the meninges (M). Scale bar = 25 μm. (i) The presence of vessel dyes (red) can be detected by IHC in the perivascular areas of the CNS parenchyma in fixed CX3CR1^+/GFP brain tissue, both near the superficial vessels and the deep vessels away from the meningeal-parenchyma interface (*). Dash line: Imaging depth achieved by TPM. Scale bar = 100 μm. Data are shown as mean ± SD.

PTx has been shown to disrupt the BBB integrity in EAE (Linthicum et al., 1982), although the exact location, timing and dynamics of the vessel breach have not been carefully characterized to date. Using TPM and fluorescent-labeled vessel dye, we directly observed focal transient leakage of CNS vascular contents hours to days after PTx injections on days 0 to 2. These transient BBB breaches were visualized as blooms of dye leaking from the vessels (Figure 1b; Movie 1). The effect of PTx on CNS vessel disruption was neither uniform nor permanent, as certain post-capillary venule junctions were prone to intermittent vessel leaks while other vessels of similar caliber remained completely intact (Figure 1c). Control mice exhibited little to no vessel leaks (n=4 from 1 experiment), indicating that these leaks were not due to damages associated with TPM or prior surgical procedures (Figures 1d and 1e). Furthermore, we observed vessel leaks that originated from vessels outside of the imaging field, arguing against this phenomenon as an artifact of TPM (Movie 2). When the vessel leaks were serially characterized in the same mice through the first 12 days of EAE induction, we observed the highest frequency of vessel leaks during the first 3 days (n=19 from 8 experiments), coincident with PTx administration (Figure 1f). The duration of individual leaks during this early period was variable, with some lasting ~30 minutes before resolving (Figure 1g). While vessel content quickly dissipated at the pial surface by the bulk CSF flow in the meningeal space, blood solutes released into the parenchyma took a longer time to clear after resolution of the BBB breach (Figure 1h). Evidence of blood persistence can be found deep in the parenchyma, as TRITC-labeled dextran was observed on day 3 in fixed brain tissue (Figure 1i).

Identity of TRITC⁺ cells in early EAE

Next, we examined the focal and intermittent nature of the vessel breach by measuring the location and the dextran uptake activity in the cortex throughout EAE induction. CX₃CR1 is constitutively expressed by microglia and a sub-population of macrophages in the CNS (Kettenmann et al., 2011). To identify the TRITC⁺ cells, we induced EAE in Cx₃cr1^+/GFP mice. Five hours after PTx administration on day 0, the only TRITC⁺ cells in the meninges and parenchyma were 100% CX₃CR1⁺ cells (Figure 2a).

Figure 2. Focal vessel leaks induced localized activation of endogenous CNS immune cells and accumulation of DCs.

(a) Day 0 after EAE induction showed that all TRITC⁺ cells (red) were confined within CX3CR1^+/GFP (green) cells. (b) TPM vessel imaging of cortex on day 3 following PTx administration revealed aggregates of TRITC⁺ cells (red) while other similarly imaged areas were devoid of pinocytic activities. The numbers indicate dextran-laden cell counts within the boxed parenchymal volume. Scale bar = 100 μm. (c) Quantification of TRITC⁺ cells in the meninges and parenchyma following systemic exposure to LPS and PTx. * p<0.01, ** p<0.001, *** p<0.0001. PBS, n=7; PTx, n=6; LPS, n=4 from 2 independent experiments. (d) Sequential CNS imaging of the CD11c-GFP mice over the first 12 days of EAE induction showed progressive changes in the accumulation of CD11c⁺ (green) and TRITC⁺/CD11c⁺ cells (purple) in left panel and TRITC⁺ only cells (red) in right panel over time. Scale bar = 60 μm. (e) Quantification of CD11c⁺ (green), TRITC⁺/CD11c⁺ (purple) and TRITC⁺ cells (red) cells in the meninges and parenchyma over a 12-day sequential imaging experiment as shown in (d). N=5 mice from 2 independent experiments. Data are shown as mean ± SD.

Intermittent vessel leaks resulted in localized uptake of blood vessel contents by microglia

When TRITC-labeled dextran was introduced intravenously on 3 consecutive days during PTx administration, we observed a patchy distribution of TRITC-labeled cells throughout the parenchyma, while other areas within the same imaging field have little to no detectable TRITC-labeled cells (Figure 2b). The distribution of TRITC⁺ cells is consistent with the distribution pattern of transient vessel leaks observed in Figure 1. Serial imaging of the same CNS location also revealed that the vessel leaks coincided with observed increase in the TRITC-dextran uptake and activation of nearby microglia, as judged by morphologic changes (Figure 2a, inset), thereby starting a cellular cascade that results in the infiltration and accumulation of circulating immune cells.

Toxin-dependent phagocytic activation

To assess how the phagocyte distribution may be affected by neuroinflammatory stimuli, we compared the distribution and numbers of TRITC⁺ cells following systemic exposure to PTx and the TLR4 agonist, lipopolysaccharide (LPS). We observed a distinct distribution pattern of parenchymal TRITC⁺ cells between the two stimuli: systemic LPS produced an even distribution of TRITC⁺ cells throughout the entire parenchyma and meninges, whereas PTx treatment resulted in a patchy TRITC⁺ distribution (data not shown) as seen with EAE induction (Figure 2b). Both PTx and LPS administration resulted in a significant increase in the appearance of TRITC⁺ cells in both the meninges (PTx, df=11, p=0.0082; LPS, df=8, p=0.0004) and parenchyma (PTx, df=11, p=0.046; LPS, df=8, p=0.0009) (Figure 2c; two-tailed unpaired t-test). This method allows serial documentation of the vessel leak patterns in the same hosts with finer time and space resolution than previously reported (Findling et al., 1994).

CD11c⁺ DCs infiltrate the CNS following phagocyte activation

CD11c⁺ DCs are required in EAE induction (Bailey et al., 2007; Miller et al., 2007). To determine the time course of infiltrating DCs and their phagocytic activities in the CNS, we imaged CD11c-GFP⁺ mice daily during the first 12 days of active EAE induction (Figure 2d). We quantified the number of CD11c-GFP⁺ cells, TRITC⁺ cells, or both in the meninges and parenchyma (Figure. 2e; 1 way ANOVA with df=41). We observed a significantly increased number of CD11c⁺TRITC⁻ cells in the meninges starting on days 3 through 12, with increasing dextran uptake from days 9 to 12. The entry of CD11c⁺TRITC⁻ DCs into parenchyma was much slower compared to those in the meninges, with DCs increasing in number starting on day 6. Their phagocytic capacity increases over the subsequent 6 days. In contrast, the number of CD11c⁻ TRITC⁺ cells in the meninges decreased between days 3 to 12, even as parenchymal CD11c⁻ TRITC⁺ cell numbers remain constant throughout this period.

Presentation of myelin antigens by CNS APCs enhanced activated myelin-specific T cell entry and retention

Following the characterization of CD11c⁺ and CD11c⁻ APCs in the EAE brain, we investigated the timing and antigen requirement of 2D2 cell recruitment and behavior within the CNS. More 2D2 cells began to infiltrate the brain parenchyma on day 3 and significantly more on day 9 (Figure 3a; n=5 from 2 experiments, df=8, p=0.0082, 1 way ANOVA). However, when mice adoptively transferred with naïve OTII and 2D2 cells were immunized only with OVA323-339, we saw a near complete absence of infiltrating 2D2 or OTII cells in the meninges or the CNS parenchyma (Figure 3b), even though both T cell populations were found in equivalent numbers within the peripheral lymph nodes (Figure 3c). The few OTII cells resided only transiently in the CNS next to blood vessels and migrated with a reduced speed as compared with 2D2 cells (Figure 3d). The divergent behaviors between the two T cell populations were observed in the presence of equivalent leaky CNS vessels, phagocytic APC accumulation and robust T cell activation within the same host. Taken together, our data imply that the cooperation between PTx-induced leaky vessels, resident and infiltrating APCs, the availability of CNS-derived antigens and activated myelin-specific T cells are preconditions for MOG-specific T cell entry into the CNS, which in turn amplifies other cellular infiltration later in the disease progression.

Figure 3. Peripherally activated T cells accumulated and migrated near sites of vessel leak and DC aggregation in an antigen-dependent manner.

(a) Direct TPM observation of naïve OTII and 2D2 co-transferred mice that were immunized with both OVA323-339 and MOG35-55 peptides in the presence of CFA and PTx. N=5 from 2 independent experiments. (b) TPM observation of naïve OTII and 2D2 co-transferred mice that were immunized with only the OVA323-339 peptide. N=3 from 1 experiment. (c) Flow cytometry analysis of OTII (Vα2) and 2D2 (Vβ11) cells in the peripheral draining lymph nodes of mice in (a). (d) Comparison of the average migration speeds of OTII and 2D2 cells found in the brain. (e) Sites of focal vessel leaks early in EAE induction (day 4) resulted in TRITC⁺ (red) and CD11c-GFP⁺ (green) cell accumulation (days 6–9), along with localized 2D2 cell (blue) migration and accumulation (day 9). Colored tracks on day 9 highlight the paths taken by individual 2D2 cells (blue) near areas of CD11c-GFP⁺ cell clusters (green) in the parenchyma (P) compared to the meninges (M). (f) Contact duration of CD11c-GFP⁺ cells by 2D2 T cells on day 9. (g) Average velocity of 2D2 cells based on interaction times. (h) Total number of unique 2D2 cells migrating towards CD11c+ rich regions (GFP^hi) versus low CD11c-GFP (GFP^lo) regions on days 9 and 12. N=5 from 2 independent experiments. (i) 2D2 cells exhibited similar overall contact time with APCs on days 9 and 12. N=5 from 2 independent experiments. (j) More 2D2 cells egressed from the blood vessels into the parenchyma as compared to those entering into the blood vessels from the CNS on days 9 and 12. N=5 from 2 independent experiments. **, p<0.001; n.s., not significant. Data are shown as mean ± SD.

Localized blood vessel leak preceded phagocyte accumulation and the infiltration of DCs and T cells

To further examine the cellular and structural cooperation leading to EAE, we performed multiplex sequential imaging to capture cellular interactions among TRITC⁺ cells, DCs and 2D2 cells during the first 9 days of EAE induction. Following transient vessel leaks (Figure 3e, day 4), the vessel content was cleared with a concomitant increase in the number of TRITC⁺ cells and the arrival of CD11c⁺ DCs by day 6 (Figure 3e, days 6–9). This was followed on day 9 by a large number of infiltrating 2D2 cells that congregated near sites of previous vessel leaks (Figure 3e, day 9), eventually leading to extensive inflammatory changes, with tail and limb weakness beginning on days 10 onward (Figure 1a). The 2D2 cells surveyed a large parenchymal area, while maintaining focal movements surrounding sites of prior vessel leak where clusters of CD11c⁺ cells were also found (Figure 3e, day 9 tracks; Movie 3). The 2D2 cells exhibited both transient (< 2 min) and prolonged (≥ 2 min) contacts with DCs (Figure 3f). Those 2D2 cells that interacted briefly with CD11c⁺ DCs migrated at faster velocities (~ 14.8 ± 4.2 μm/min), while those that exhibited prolonged interaction with DCs traveled with reduced velocities (~ 12.5 ± 3.6 μm/min) (Figure 3g). In all, 83% of 2D2 cells were motile, with 6.1 ± 5.1% of 2D2 cells residing in the perivascular areas with limited motility.

2D2 cells preferentially congregated in DC-rich regions

To assess whether 2D2 cells were preferentially drawn to sites of DC aggregates, we analyzed the number of 2D2 cells entering a 50-μm radius of a CD11c⁺-GFP⁺ cellular aggregate. Our analysis revealed a significant increase in the number of unique 2D2 cells moving towards an area high in GFP signals as compared to areas with low GFP signals within the same imaging field (Figure 3h; n=5 from 2 experiments, df=21, p=0.0069, two-tailed unpaired t-test). Contact duration analysis of 2D2 cells with CD11c⁺ and CD11c⁻ phagocytes showed a trend toward longer duration with CD11c⁺ cells (Figure 3i; n=5 from 2 experiments, df=17, p value not significant, 1 way ANOVA). We also found a trend for a greater number of 2D2 cells egressing from nearby blood vessels to enter the CD11c⁺ rich regions as compared to those entering the vessels from the parenchyma (Figure 3j; Movies 4, 5; n=5 from 2 experiments, df=24, p value not significant, two-tailed unpaired t-test).

Histamine H1 receptor blockade reduced microglia activation, DC recruitment, disease incidence and clinical severity

Histamine released by mast cells is implicated in PTx-mediated vessel leak (Linthicum et al., 1982). We tested whether blockade of the histamine H1 receptor (H1R) could ameliorate observed vessel leak and subsequent cellular accumulation events. We administered hydroxyzine (HXYZ), a H1R antagonist that readily crosses the BBB, in the drinking water of mice undergoing EAE induction. We observed a 60% reduction in EAE incidence among mice receiving HXYZ treatment (Figure 4a; n=10 from 2 experiments). Despite HXYZ treatment via gavage, however, we still observed PTx-induced vessel leaks (Figures 4b, 4c; Movie 6). Nevertheless, HXYZ treatment did result in a dramatic reduction in the number of TRITC⁺ parenchymal cells, with numbers comparable to PBS control (Figure 4d; PBS: PTx + HXYZ, df=7, p value not significant, two-tailed unpaired t-test). These findings were further corroborated by TPM, showing a trend toward reduction of TRITC⁺ cells and DCs as late as day 12 (Figure 4e, day 3; df=18, p=0.0146; day 12 df=22, p=0.0386; two-tailed unpaired t-test). Flow cytometry analysis revealed a reduction in the number of both CD11b⁺CD11c⁺ DCs (Figure 4f; df=7, p=0.006, two-tailed unpaired t-test) and CD11b⁺CD11c⁻ macrophages (Figure 4g; df=12, p=0.0049, two-tailed unpaired t-test), as well as a similar reduction in the number of infiltrating CD8⁺ (Figure 5a; df=8, p=0.0013, two-tailed unpaired t-test) and CD4⁺ T cells (Figure 5b; df=8, p=0.0005, two-tailed unpaired t-test) in HXYZ-treated groups early on day 3. This was accompanied by a reduced number of CNS infiltrating IL-17 and IFN-γ producing CD4⁺ T cells (Figures 5c, day 9; IL-17: df=8, p=0.036; IFN-γ: df=8, p=0.024; two-tailed unpaired t-test), even though the spleens of these mice contained similar numbers of IL-17 and IFN-γ producing CD4⁺ T cells (Figure 5d).

Figure 4. Histamine receptor blockade inhibited early activation of CNS TRITC⁺ cells, diminished DC and 2D2 cell accumulation, and blunted EAE severity.

(a) 60% of mice displayed absent or reduced severity of disease when exposed to hydroxyzine (HXYZ) via drinking water during EAE induction, with 20% disease-free incidence and 40% with a delayed onset and reduced overall clinical severity. N=10 per group from 2 independent experiments. (b) Frequency of CNS vessel leaks were observed by direct TPM observation after HXYZ was administered to mice via gavage followed by EAE induction. No statistical differences in vessel leaks were observed in HXYZ-gavaged mice as compared to control mice during EAE induction. N=6 per group from 2 independent experiments. (c) EAE-induced mice treated with HXYZ showed a dramatic reduction in TRITC⁺ cells. Scale bar = 50 μm. (d) Cell number in both meninges and parenchyma with PTx treatment versus PTx + HXYZ treated as compared to PBS controls. PBS, n=7; PTx, n=6; PTx+HXYZ, n=4 from 2 independent experiments. (e) Sequential TPM imaging data analysis of HXYZ-treated and control mice showed a substantial decrease in parenchymal and meningeal CD11c⁺ and CD11c⁻ TRITC⁺ numbers during EAE induction between days 3 and 12. HXYZ-treated, n=6; control, n=5 from 2 independent experiments. (f) After whole brain percoll isolation of cells from HXYZ-treated and control mice, flow cytometry analysis revealed a significant decrease in the number of CD11b⁺CD11c⁺ cells in the CNS of HXYZ-treated mice on days 3 and 6 following EAE induction. N=8 from 2 independent experiments. (g) The number of CD11b⁺CD11c⁻ cells in the CNS of HXYZ-treated mice was also significantly lower as compared to control on day 6 following EAE induction. N= 8 from 2 independent experiments. Data are shown as mean ± SD. **p<0.001; ***p<0.0001; n.s., not significant.

Figure 5. Effects of HXYZ on T cell numbers during EAE induction.

HXYZ was administered via gavage from day −1 to day 12 while EAE was being induced. Whole-brain percoll isolations of individual cells from HXYZ-treated and control mice were performed. Flow cytometry analysis of the isolated cells showed the following: (a) A significant decrease in the number of CD8⁺ cells in the CNS with HXYZ treatment on Day 3. N=8 from 3 independent experiments. (b) A significant decrease in the number of CD4⁺ cells in the CNS with HXYZ treatment on Day 3. N=8 from 3 independent experiments. (c) Although the total number of CD4⁺ cells in the CNS was similar between treatment and control groups (b) on day 9, there was a significant decrease in the number of IFN-γ and IL-17 producing cells within the CNS. N=5 from 2 independent experiments. (d) The total number of CD4⁺ T cells expressing IFN-γ, IL-17 and FoxP3 were not significantly different with HXYZ treatment on day 9 in the spleen. N=5 from 2 independent experiments. *p<0.01; **p<0.001; ***p<0.0001; n.s., not significant.

DISCUSSION

CNS vessel leak has been implicated as a trigger in neurodegenerative disorders including epilepsy, Alzheimer, and Parkinson’s disease (Hawkins and Davis, 2005; Zlokovic, 2011). Although changes in CNS vessel permeability have been documented in vivo, the timing and kinetics at the level of detail as described in the current study has not been reported previously (Davalos et al., 2012; Findling et al., 1994; Kenne and Lindbom, 2011; McGavern and Kang, 2011a). We provided evidence of the transient and focal nature of altered vascular permeability in early EAE induction as an initiating event, not a byproduct of effector T cell aggregation and tissue destruction near the perivascular pathologic lesions. These leaks occurred during PTx treatment, where blood contents including plasma proteins and fibrinogen (Davalos et al., 2012) are released into the parenchyma and initiate a parenchymal inflammatory response including the activation of the phagocytic capacity of resident CX₃CR1⁺ microglia (Neumann et al., 2009), well in advance of any observable clinical symptoms. Vessel leaks were found in the meninges and throughout the parenchyma. During the early vessel leaks following PTx administration, we did not detect influx of circulating monocytes or T cells in the cortex, suggesting that the opening of the BBB at this stage does not permit a large influx of circulating cells into the imaged region of the brain. At later time points (days 4–6) when the vessel leakage subsides, we began to observe increasing immune cell infiltrate coincident with a second wave of vessel leak (Figures. 1d, 1e; days 7–9), an observation consistent with the two-step paradigm of EAE induction (Sallusto et al., 2012). Interestingly, while PTx administration resulted in focal dextran-uptake activity, systemic administration of LPS, the classical TLR4 agonist, resulted in a diffuse pattern of dextran uptake in the CNS presumably due to the direct effect of LPS on endothelial and immune cells throughout the body (Andonegui et al., 2003; Andonegui et al., 2009). The observed differential effects between PTx and LPS on CNS dextran uptake are interesting, since PTx has been shown to evoke intracellular signaling through TLR4 (Kerfoot et al., 2004).

Mice with functionally deficient microglia exhibited delayed EAE onset and reduced severity, suggesting microglia contribution in EAE pathogenesis (Greter et al., 2005; Heppner et al., 2005). Peritoneal macrophages activated with ConA display increased pinocytosis concomitant with an increase in phagocytosis, and prolonged pinocytosis may be an indicator of macrophage activation (Cohn, 1966). Depending on the mode of activation, however, pinocytosis may also decrease as compared to non-activated macrophages (Montaner et al., 1999; Tsang et al., 2000). Pinocytosis is a highly metabolic processes and requires an ATP source such as dextran (Edelson et al., 1975). In our study, we found that microglia surrounding the vessel leaks were pinocytic. Using the CX₃CR1^+/GFP reporter mouse, we showed that microglia and macrophages constitute the main pinocytic populations that absorb vessel contents in the first 3 days. These cells can internalize vessel contents in response to TLR agonist stimulation by one of two ways. First, they act as scavengers, removing blood solutes following vessel leaks. Second, perivascular CX₃CR1⁺ cells can sample blood contents directly within intact blood vessels through intravascular processes (Barkauskas et al., 2013).

Bone marrow derived myeloid cells are also important in regulating CNS inflammatory response, where infiltrating DCs with functional MHCII are required for EAE induction (Bailey et al., 2007; Greter et al., 2005; Heppner et al., 2005; Juedes and Ruddle, 2001; Miller et al., 2007; Pashenkov et al., 2003; Serafini et al., 2000). Activated OTII cells can access the CNS only in the presence of co-activated 2D2 cells. In contrast, activated OTII cells in the presence of naïve 2D2 cells were incapable of doing so, despite the presence of pinocytic microglia and DCs. These results suggest that T cells do not simply enter CNS passively through disrupted BBB but require activated 2D2 cells recognizing myelin-presenting APCs. Our data support the view that APCs presenting endogenous CNS peptides are critical to license T cell recruitment and retention in the EAE brain. Our TPM imaging and FACS data are in good agreement with reports showing that 5–10% of CD11b⁺ cells present myelin basic protein as early as day 1 after EAE induction, whereas CD11c⁺ DCs are not detected till after day 5 and only present myelin basic protein from day 7 onwards (Sosa et al., 2013).

T cell infiltration into the brain was thought to only occur at the choroid plexus, although recent data suggest an alternate route via VRS and the pial vessels in the SAS (Goverman, 2009; Ransohoff and Engelhardt, 2012). Our imaging data captured T cells migrating directly into the parenchyma from VRS, and migrated near CD11c⁺ DCs in a pattern that were in contrast to published data showing non-directed T cell behavior during the peak of EAE disease (Bartholomaus et al., 2009; Herz et al., 2011; McGavern and Kang, 2011a; Mues et al., 2013; Siffrin et al., 2010). Our finding of the focal T cell accumulation and their repeated surveillance with DCs during the asymptomatic phase of EAE suggests direct communication between newly arrived 2D2 and DCs. These short contacts were similar to other reports, which utilized calcium flux and NFAT reporter mice to show that infiltrating T cells in the spine of passively induced EAE mice were in an activated state while meandering and making short contacts with phagocytic APCs (Lodygin et al., 2013b; Mues et al., 2013). Here, we showed that 2D2 cells are re-activated by DCs in the cortex before these T cells exert their pathogenic effector function, eventually leading to the development of clinical symptoms.

PTx has been shown to increase the permeability of vessels (Linthicum et al., 1982; Locht et al., 2011; Miyata and Morita, 2011) with direct effects on the choroid plexus and cerebral endothelial cells (Brückener et al., 2003; Lapilla et al., 2011). In the brain, PTx is postulated to act through histamine, which causes vasodilation, changes adhesion molecule expression, and alters endothelial tight junctions and astrocyte end feet barrier integrity. Mice deficient in Histamine receptors 1 and 2 have reduced EAE severity and BBB permeability (Saligrama et al., 2012). The CSF of MS patients contains elevated levels of histamine, and H1R antagonists have shown clinical benefits in MS (Jadidi-Niaragh and Mirshafiey, 2010). In a clinical trial, hydroxyzine was shown to stabilize or improve symptoms in 75% of MS patients (Logothetis et al., 2005).

Interestingly, we only observed a trend towards diminished frequency, but not a complete absence of vessel leakage when HXYZ was administered. Our observation suggests that mechanism(s) other than histamine are responsible for vessel permeability changes under neuroinflammatory conditions, such as tachykinins (e.g. substance P), serotonin, and bradykinin (Annunziata et al., 2002; Annunziata et al., 1998; Feng et al., 1996; Kakizoe et al., 1997; Olesen, 1985; Reinke et al., 2006; Saria et al., 1983). Regardless, we showed that HXYZ treatment reduced EAE disease incidence or severity in 60% of the mice. Furthermore, HXYZ initially reduced the extent of dextran uptake by meningeal macrophage as documented by TPM and decreased the myeloid-derived APC infiltration globally in the CNS. Importantly, preventing the early activation of microglia with HXYZ resulted in a reduction of both non-specific T cell infiltration on day 3 and antigen-specific T cell infiltration on day 12. These data provide further evidence supporting the view that microglia play a critical role in early EAE induction, and that blocking microglia activation during the inflammatory phase can lessen myelin and neuronal destruction (Giunti et al., 2013; Juedes and Ruddle, 2001; Saijo and Glass, 2011). Future work remains to fully understand the regulation of focal and intermittent CNS vessel disruption that was only partially blocked by H1R blockade.

The exact kinetics of cellular inflammation in the brain has been hard to study, especially during the very early stages of EAE induction prior to disease manifestation, due to the low number of relevant cells detected using conventional bulk cellular assays. Our present study helps to shed new light on the precise, local and sequential interplay among CNS tissues and cellular factors that coordinately manifest in early EAE pathophysiology involving early intermittent BBB disruption, local activation of microglia, subsequent recruitment of DCs, and the surveillance behavior of pathogenic CD4⁺ T cells in the EAE brain.

Supplementary Material

1. Movie 1: Transient and dynamic vessel leaks with vessel content persistence in the parenchyma.

Region 1: At time 00:31:30, a focal transient leakage of TRITC vessel dye (red) starts and the fluorescence persists in the parenchyma until the end of the movie. Region 2: Starting at time 00:16:00, a focal transient leakage of TRITC vessel dye (red) initiates and resolves, then initiates again at time 00:38:00 and resolves with no residual fluorescence persisting in the parenchyma. Note: other vessels of similar caliber within the same imaging volume remained completely intact. Total time: 52 min 30 sec. Playback speed: 300X. Scale bar = 60 μm. Time stamp: hh:mm:ss.

2. Movie 2: Transient vessel leaks originated from outside the field of dynamic TPM imaging.

Maps of the whole cranial window were used to create a composite image on which a movie was overlaid. Overlaid movie on the static composite image enables an approximation of the vessels that gave rise to vessel leaks seen in the dynamic movie. Total time: 60 min. Playback speed: 300X. Scale bar = 100 μm. Time stamp: hh:mm:ss.

3. Movie 3: Infiltrating 2D2 cells trafficked to locations of CD11c⁺ DC accumulations.

Imaging on day 9 after EAE induction shows 2D2 T cells (blue) surveying the parenchyma concentrating around CD11c-GFP⁺ rich region. TRITC vessel dye also highlights pinocytic cells (red). Colored tracks show the paths taken by 2D2 cells during the duration of the video, revealing high traffic around CD11c-GFP+ rich region. Total time: 90 min. Playback speed: 300X. Scale bar = 100 μm. Time stamp: hh:mm:ss.

4. Movie 4: Egress of a 2D2 cell away from vessel into CNS parenchyma.

Dynamic intravital TPM on day 9 following EAE induction captures a 2D2 T cell leaving the blood vessel en route to the brain parenchyma. The white arrow at the beginning of the video points to a 2D2 cell (blue), with part of its cell body still within the blood vessel lumen, leaves the vessel and migrates away from the vessel in the ensuing 11.5 min. Yellow line denotes the path taken by the 2D2 cells during the imaging period. Total time: 11.5 min. Playback speed: 60X. Scale bar = 10 μm. Time stamp: hh:mm:ss.

5. Movie 5: Perivascular movement of 2D2 cell.

Another example of 2D2 T cell behavior on day 9 following EAE induction, showing 28 min of perivascular patrolling behavior of a 2D2 cells, with part of its cell body still within the blood vessel lumen in the beginning of the video, before departing the perivascular space into CNS parenchyma. Yellow line denotes the path taken by the 2D2 cells during the imaging period. Total time: 30 min. Playback speed: 60X. Scale bar = 15 μm. Time stamp: hh:mm:ss.

6. Movie 6: HXYZ treatment did not prevent vessel leak but prevented vessel content uptake.

Mice were given HXYZ via gavage over the course of EAE induction. The left panel shows the dynamic imaging of a full EAE-induced mouse while the right panel shows the dynamic imaging of an HXYZ-treated EAE-induced mouse on day 3. At time 00:18:00 the EAE mouse exhibited a vessel leak (arrow) and pinocytic cells can be seen throughout the whole field. The HXYZ treatment mouse showed 2 locations of vessel leak (arrow) and no pinocytic cells. From time 00:01:00 00:14:00 one leak was observed. At a second location, the same vessel leaked 3 times during the imaging session, at times 00:03:30, 00:08:00 and 00:28:30. Total time: 59 min. Playback speed: 300X. Scale bar = 100 μm. Time stamp: hh:mm:ss.

Highlights.

• Focal vessel leaks occur in early EAE prior to MOG-specific T cell infiltration.

• Vessel leaks precede microglia activation, followed by DC and T cell infiltration.

• MOG T cells contact DCs in the cortex prior to clinical EAE development.

• Histamine antagonist does not completely ameliorate blood vessel leak.

• H1R blockade reduces microglia activation, immune cell influx and EAE severity.