Ethics Statement

All experiments were approved by the Institutional Animal Care Committee of the University of North Carolina School of Medicine and Public health and performed according to the legislation for the protection of animals (IACUC ID: 12–038.0).

Experimental Animals

Wild type (WT) C57BL/6 mice were purchased from the Jackson lab. Platelet-endothelial cell adhesion molecule–deficient (PECAM-1^−/−) mice on a C57BL/6 background were kindly provided by Dr P. Newman (Blood Research Institute, Blood Center of Wisconsin, Milwaukee). C57BL/6 immorto-mice expressing a SV40 large T antigen were obtained from Charles River Laboratories (Wilmington, MA).

Inflammatory Response in Mouse Cremaster Muscles or Ear Skin and Intravital Microscopy

WT or PECAM-1^−/− male mice (∼25 g) were injected intrascrotally (i.s.) with saline, interleukin-1β (IL-1β) (50 ng/mouse) or tumor necrosis factor-α (TNF-α) (500 ng/mouse) (R&D Systems, Minneapolis, MN). In IL-1β-injected mouse cremaster muscles, the responding leukocytes (>90%) were previously verified to be PMNs [14]. In selected experiments where PMNs need to be depleted, 100 µg of anti-GR1 antibodies (Abs) [15] (RB6–8C5, BD Biosciences, Chicago, IL) were injected intraperitoneally (i.p.) into each mouse 24 hours before i.s. administration of IL-1β. To block PMN extravasation, in some experiments each mouse was intra-splenically pre-injected with 100 µg of Abs against either mouse intercellular adhesion molecule-1 (ICAM-1) (YN1/1.7.4, Biolegend, San Diego, CA) [16] or α₆ integrin (GoH3, Biolegend, San Diego, CA) [17], [18] 15 minutes before IL-1β administration. Control animals received isotype-matched control Abs (Biolegend, San Diego, CA). Two hours after IL-1β injection, the cremaster muscles of fully anaesthetized mice were exteriorized and transferred to the stage of an Olympus FluoView FV1000 upright microscope and the cremaster muscles were constantly superfused with warm buffered solution. The numbers of rolling or static adherent leukocytes were quantified starting from the administration of superfusion buffer as described before [14]. In some experiments, cremaster muscles were clamped using a micro aneurysm clip (Harvard Apparatus, Holliston, MA) and superfused with buffer containing Norepinephrine (NE, 10 µM, Sigma-Aldrich, Saint Louis, MO) or Tolazoline (100 µg/ml, Sigma-Aldrich, Saint Louis, MO). Fifteen min later, the inflammatory response was stopped immediately by perfusion of 4% paraformaldehyde (PF). The cremaster muscles of all experimental mice were finally harvested and fixed in PF for further immunostaining. In some experiments, IL-1β (5 ng/mouse) or TNF-α (50 ng/mouse) in 50 µl saline was subcutaneously injected into mouse ears for two hours to induce an inflammatory response in skin.

Immunostaining of Mouse Cremaster Muscles and Ear Skin

Fixed mouse cremaster muscles and ear skin were immunostained for multiple proteins as described previously [9]. An Alexa 488- or 647-labeled rat anti-mouse PECAM-1 Ab (MEC13.3 BioLegend, San Diego, CA) was applied to label ECs. The vascular BM was displayed by a rabbit anti–mouse collagen IV polyclonal Ab (Abcam, Cambridge, MA) plus appropriate secondary Abs conjugated to Alexa 405 or 488 (Invitrogen, Carlsbad, CA). To mark the PMNs, a goat anti–mouse MRP-14 Ab (Calgranulin B (M-19), Santa Cruz Biotechnology, Santa Cruz, CA) or Allophycocyanin-conjugated rat anti-mouse CD11b Ab (M1/70, eBioscience, San Diego, CA) was used. To stain the pericytes, tissues were further incubated with a Cy3-labeled anti-mouse α-SMA Ab (1A4, Sigma-Aldrich, St. Louis, MO) or a goat IgG against mouse Platelet-derived growth factor receptor-β (PDGFR-β) (R&D Systems, Minneapolis, MN) plus fluorescent secondary Abs. In all studies, appropriate control IgG (BioLegend, San Diego, CA) was used in parallel with the specific primary Abs.

Imaging and Analysis of Immunostained Cremaster Muscles and Ear Skin

Immunostained samples were imaged in three-dimensions (3D) using a Zeiss 510 Meta confocal microscope with lasers excited at 488, 545, 633, and 740 nm respectively, and with a x40 water-dipping objective (numerical aperture = 1.25). The imaged tissues were quantified for transmigrating leukocytes, defined as the number of leukocytes within the immunostained venular walls (per 200 µm vessel segment) and extravasated leukocytes, defined as the number of leukocytes in the interstitium of the whole tissues (per mm² cremaster muscle) or in the interstitial tissues within 100 µm around a 200 µm vessel segment. 3D images of vessels were split in the middle along the longitudinal axis using the LSM software and fluorescence intensity of “hemi-vessels” was analyzed. The area of gaps between adjacent pericytes, the area of Collagen IV LERs, and the protein contents of Collagen IV in LERs (fluorescence intensity, pixels/unit area), were measured as previously described [9].

Isolation of Mouse Peritoneal PMNs

Mouse PMNs were harvested from the thioglycollate-stimulated peritoneal cavity as previously described [19]. Briefly 1 ml of 4% thioglycollate solution (Sigma-Aldrich, St. Louis, MO) was injected into the mouse peritoneal cavity. Four hours later, exudate cells were harvested and purified for PMNs by Percoll density gradient centrifugation, yielding PMN preparations of around 90% purity.

Collection of Mouse Bone Marrow PMNs and Culture of Mouse Retinal Pericytes

Bone marrow neutrophils were harvested from wash-outs of femurs of mice using three-layer Percoll gradients, yielding neutrophil preparations of around 85% purity [20]. Primary pericytes were isolated from retinas of WT Immorto mice (6–7 pups, 4 week old). Twelve to fourteen retinas were collected and digested at 37°C for 45 min in serum-free DMEM containing collagenase type II (1 mg/ml, Worthington, Lakewood, NJ) and 0.1% BSA. After washes, digested tissue was resuspended in DMEM supplied with 10% FBS, 2 mM L-glutamine, 100 g/ml streptomycin, 100 U/ml penicillin, and murine recombinant interferon-γ (at 44 U/ml, R&D Systems, Minneapolis, MN) and cultured at 33°C with 5% CO₂. Cells were progressively passed to larger plates [21] and used for different assays within 20 passages.

Collection of Transmigrated PMNs in vitro

Within a 6-well Boyden transwell, mouse primary retinal endothelial cells were seeded on the top surface of each FN-pre-coated insert (with 3 µm pores) and grown to confluence. Then the endothelial cell layer was incubated with serum-free DMEM containing TNF-α (10 ng/ml) overnight and freshly isolated mouse bone marrow PMNs were added on the top of the endothelial cells. The inserts were put back into the wells containing serum-free DMEM in the presence of 3.0×10⁻⁸ M of interleukin-8 (IL-8, R&D Systems, Minneapolis, MN) and incubated at 37°C for 4 hours. Transmigrated PMNs were harvested and washed for further experiments.

Retroviral Packing and Infection

Retroviral packaging and infection were performed as previously described [22]. Briefly, human 293 FT cells (Invitrogen, Carlsbad, CA) were placed on poly (L-Lysine)-pre-coated dish and transfected with the packing plasmids pRSV-Rev, pMDLg/pRRE, pCMV-VSV-G (Addgene, Cambridge, MA) and the pLentiLox 5.0 shuttle plasmid encoding WT, constitutively active (CA) Q63L, or dominant negative (DN) T19N human RhoA. Virus-containing medium, together with 6 µg/ml of polybrene, was added to pericytes which were plated at 50% confluence. Forty eight hours later, protein expression was checked by using fluorescence microscopy and western blot.

Plasmids and shRNA Construction

To generate a lentiviral construct encoding short-hairpin RNA (shRNA) specific to mouse RhoA, forward hairpin oligos 5′-[Phos] TGTCAAGCATTTCTGTCCAATTCAAGAGATTGGACAGAAATGCTTGACTTTTTTC -3′ and reverse oligos 5′-[Phos] TCGAGAAAAAAGTCAAGCATTTCTGTCCAATCTCTTGAATTGGACAGAAATGCTTGACA -3′ were annealed and ligated into the HpaI and XhoI sites of the pLentiLox5.0 (pLL5.0) vector (Addgene, Cambridge, MA). The dual cassette vector (pLL5.0-shRNA) co-expresses enhanced green fluorescent protein (EGFP) as a reporter gene and shRNA targeting mouse RhoA under the control of CMV and the U6 promoter, respectively. For expression of exogenous RhoA, plasmids encoding human RhoA (GenBank Accession number: NM_001664), were directly constructed into the SacII and BamHI sites of the pLL5.0 vector with an EGFP reporter under the control of CMV. All plasmid constructions were verified by sequencing.

RhoA Activity Assays

RhoA activity assays were performed as described previously [23]. Primary pericytes or cells expressing GFP vector or GFP-tagged WT, CA or DN RhoA were cultured in serum free medium containing TNF-α (15 ng/ml). Next day, cells were stimulated at 37°C for 1 hour by serum-free DMEM containing DMSO, Forskolin (100 µM, Ascent Scientific, Princeton, NJ) or lysophosphatidic acid (LPA, Enzo Life Sciences, Plymouth Meeting, PA) (1 µg/ml)) or mouse PMNs, which were pretreated with DMSO or phorbol 12-myristate 13-acetate (PMA, Sigma-Aldrich, Saint Louis, MO) (100 ng/ml) and washed for four times with PBS. Cells were lysed and incubated with glutathione S-transferase (GST)-Rho binding domain (RBD) beads. Samples were washed and then resolved by SDS-page gels. The endogenous RhoA in the bound fraction (active) and in the whole cell lysates (total) was immune-blotted with an anti-RhoA Ab (26C4, Santa Cruz Biotechnology, CA). The exogenous active or total RhoA was detected with an anti-GFP monoclonal Ab (Roche Applied Science, Indianapolis, IN). The results were quantified using Image J software. The relative amount of active RhoA was determined by taking the ratio of RhoA sedimented by GST-RBD beads to the amount of total RhoA.

Impedance-based Assay of Pericyte Responses to Either Reagents or PMNs

Mouse primary pericytes were grown in serum free medium containing TNF-α (15 ng/ml). Next day, 4.0×10⁴ cells were seeded onto gold electrodes in each well (surface area of each well is equal to 19.2 mm²/well) of a RTCA analyzer (Roche Applied Science, Indianapolis, IN) and electrical impedance was monitored over time. Changes in impedance may be used to measure a number of characteristics of cells in a real-time manner including cell spreading, degree of cell confluence, cell invasion, and permeability of a cell layer, depending on the experimental design (https://www.roche-applied-science.com/sis/xcelligence/index.jsp?&id=xcect_020000). Here we use the RTCA analyzer to monitor the spreading and morphology of pericytes within a confluent culture and their response to a number of factors. To verify our experimental results collected by the RTCA analyzer, in some selected experiments, changes of cell morphology at indicated time-points were analyzed in parallel, by imaging techniques as detailed below. In an equivalent experiment using 96-well plates, we found that 4.0×10⁴ pericytes were able to make a confluent cell monolayer and cover a surface of 19.2 mm² 2 hours after spreading. Our preliminary experiments indicated that the impedance for each well reached a plateau from 2 to 6 hours after cell seeding. Within this time frame, reagents such as DMSO, PBS, Forskolin (100 µM), C3 Transferase (C3) (2 µg/ml, Cytoskeleton Inc, Denver, CO), Y27632 (10 µM, Calbiochem, Billerica, MA), Blebbistatin (10 µM, Sigma-Aldrich, Saint Louis, MO), Tolazoline (100 µg/ml), NE (10 µM), or LPA (1 µg/ml), or 5.0×10⁴ mouse PMNs, were added to the pericyte layers and the impedance was recorded in real-time. In the case of adding PMNs, these cells were pretreated with DMSO, saline, PMA (100 ng/ml), or MnCl₂ (2 mM) and extensively washed using 10 ml PBS at least four times before adding to the pericyte monolayers.

Image-based Analysis of Pericyte Responses to Reagents or PMNs

Coverslips (12 mm in diameter) placed in the wells of a 24 well plate were pre-coated with human Fibronectin (FN, 10 µg/ml). Mouse pericytes (primary cells or cells expressing GFP vector or GFP-tagged WT, CA or DN RhoA) were grown in serum free medium containing TNF-α (15 ng/ml). On the next day cells were seeded onto each well at 37°C for 4 h, allowing cells to fully spread. During this time period, PMNs were freshly isolated from mouse bone marrow and treated with DMSO, saline, PMA (100 ng/ml), MnCl₂ (2 mM) or recombinant murine CXCL1 (KC) (1×10⁻⁷ M) at 37°C for 15 min. After being extensively washed, 5.0×10⁵ pretreated PMNs were added onto the pericyte layer at 37°C for 1 hour and cells were fixed in 4% PF. Alternatively, chemicals such as DMSO, PBS, Forskolin (100 µM), C3 (2 µg/ml), Y27632 (10 µM), Blebbistatin (10 µM), Tolazoline (100 µg/ml), NE (10 µM), or LPA (1 µg/ml) were added at 37°C for 1 hour to the cultures. In some selected experiments, pericytes were co-incubated for 1 hour with mouse peritoneal PMNs or PMNs which had passed across an endothelial cell layer and migrated to IL-8 in the bottom well of a Boyden chamber transwell system. Fixed cells were stained with Alexa-labeled phalloidin (Invitrogen, Carlsbad, CA), Hoechst 33342 (Invitrogen, Carlsbad, CA) and primary Abs against either paxillin (rabbit anti-paxillin Abs, Abcam, Cambridge, MA) or phosphorylated myosin light chain 2 (ser19) (anti-phospho-MLC Abs, Cell Signaling Technology, Danvers, MA) plus appropriate fluorescent second Abs and imaged using a Zeiss 510 Meta multi-photon confocal microscopy with lasers excited at 488, 545, 633 and 740 nm respectively, using a x63 oil objective (numerical aperture = 1.4). To view the behavior of pericytes, DMSO- or PMA-pretreated PMNs were added onto a TNF-α-induced pericyte layer, which had been grown on a FN-coated glass surface of culture dishes (MatTeK, Ashland, MA), and the interaction of pericytes and PMNs was recorded with a 20 x A-Plan ph1 objective and an ORCA ER CCD digital camera within a Zeiss Axiovert 200 M inverted microscope equipped with a heating stage and a CO2 supply and driven by a MetaMorph 7.1.7 software (MDS Analytical Technologies, Exton, PA).

Measurement of the Area of Gaps between Adjacent Pericytes Grown in vitro

3.0×10⁵ mouse pericytes (primary cells or cells expressing GFP vector or GFP-tagged WT, CA or DN RhoA) in serum free medium containing TNF-α were seeded on FN-coated coverslips (12 mm in diameter) placed in the wells of a 24 well plate overnight. At this density, pericytes became confluent next day and were labeled with Calcein green. 5.0×10⁵ DMSO- or PMA-pretreated PMNs were labeled by Calcein blue and added on the top of each pericyte monolayer for 1 hour. Alternatively, compounds such as DMSO, Forskolin (100 µM), Tolazoline (100 µg/ml), NE (10 µM), or LPA (1 µg/ml) were added. Cells were fixed in 4% PF and 15 random fields per sample were imaged using a 20 x A-Plan ph1 objective in a Zeiss Axiovert 200 M microscopy using the setting for phase contrast or fluorescent images. The area of pericyte gaps (the area of negative staining in terms of fluorescence intensity) was measured using Image J software.

Transmigration of PMNs Through a Pericyte Single Layer

Within Boyden transwells, 4.0×10⁴ pericytes (primary cells or cells expressing GFP vector or GFP-tagged WT or CA RhoA), which were grown in serum free medium containing TNF-α overnight, were seeded on the top surface of each FN-pre-coated insert (with 3 µm pores). Three hours later, cells in the inserts were stimulated with DMSO, Forskolin (100 µM), NE (10 µM), or NE (10 µM) plus Tolazoline (100 µg/ml) at 37°C for 1 hour and 2.0×10⁵ of Calcein-green labeled resting PMNs were added into each unit. Alternatively, pericytes were stimulated with 1.0×10⁵ DMSO- or PMA-pretreated PMNs for 1 hour followed by gentle washes with PBS and then 2.0×10⁵ of Calcein-green labeled resting PMNs were loaded into each insert. The inserts were put back into the transwells which were filled with serum-free RP1640 medium containing 3 10⁻⁸ M of IL-8 and incubated at 37°C for 4 h. Fluorescent PMNs which moved into the bottom wells were counted.

Statistics

All results are expressed as means ± s.e.m. Statistical significance was assessed by one-way analysis of variance (ANOVA) plus Neuman-Keuls multiple comparisons. Where two variables were analyzed a Student’s t test was used. P<0.05 was considered to be significant.

Inflamed Venules Exhibit Expanded Pericyte Gaps and Display Larger but Thinner LERs

Previous studies have suggested that the low expression regions (LERs) of matrix proteins in the vascular basement membrane (BM) act as channels for leukocyte transmigration and tend to align with the gaps between pericytes. These LERs are plastic with respect to their size and thickness during acute inflammation [9], [10]. However, little is known about how this phenomenon occurs and whether pericytes contribute to the remodeling of the vascular BM at these sites. To address this question, we injected mouse cremaster muscles or skin with IL-1β or TNF-α to elicit local acute inflammation. Two hours after injection, inflamed tissues were fluorescently immunostained and imaged by confocal microscopy. The area of gaps between adjacent pericytes, indicated by negative staining of α-SMA (or PDGFR-β, data not shown), and the area of LERs in the collagen type IV network were measured. Our data showed that pericyte gaps were enlarged after IL-1β injection (16.98±0.43 µm²), compared to those in untreated venules (9.70±0.30 µm²) (Figure 1A, 1C). PMNs used these gaps for transmigration (Figure 1A, 1B), which primarily occur at post-capillary venules rather than capillaries (Figure S1). In contrast, smooth muscle cells (SMCs), which share a common origin with pericytes in venules [11], revealed compact coil-like cell sheaths regularly surrounding the arterial vessels walls and few gaps between SMCs could be found (Figure S2). Moreover we were almost unable to detect PMN extravasation occurring in arterioles during acute inflammation (Figure 1A and Figure S1). This observation suggests that the occurrence of PMN extravasation at venules relies, at least in part, on the existence of gaps between adjacent pericytes in venular walls. Consistent with previous studies, IL-1β-stimulated LERs, as judged by collagen type IV immunofluorescence, expanded in area (Figure 1A, 1D) and became thinner (Figure 1F). Importantly, our data revealed a positive linear relationship between the area of pericyte gaps and collagen IV LERs but a negative correlation with LER thickness (Figure 1E, 1G). This is consistent with the expansion of pericyte gaps contributing to the remodeling of the vascular BM as LERs enlarge during acute inflammation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Expansion of pericyte gaps and collagen IV LERs in IL-1β-stimulated venular walls.

(A) 3D images of vessel walls in mouse cremaster muscles in the presence and absence of IL-1β revealed pericyte gaps and collagen IV LERs (circled areas) in venules. α-SMA (a marker of venule pericytes and arteriole smooth muscles), collagen IV (a vascular BM marker) and MRP14 (a PMN marker). (B) Cross-sectional images revealed PMNs migrating through pericyte gaps and collagen IV LERs (arrows) in an IL-1β-stimulated venule. (C) Distribution frequency of area of pericyte gaps showed pericyte gap expansion induced by IL-1β (n = 2400). (D) Collagen IV LERs expanded in IL-1β-stimulated venular walls, compared to those in untreated vessels (In each group, n = 1500). (E) The area of pericyte gaps (n = 1500) was plotted against the area of collagen IV LERs (n = 1500). (F) Mean fluorescence intensity (MFI) of collagen IV in LERs decreased after IL-1β injection, indicating thinning of the vascular BM in these areas. (G) MFI of collagen IV in LERs was plotted against the size of pericyte gaps. Bar = 10 µm. t test, ***P<0.001.

https://doi.org/10.1371/journal.pone.0045499.g001

PMN Transmigration Expands Pericyte Gaps and LERs

Since migrating leukocytes prefer to use pericyte gaps and LERs in the vascular BM [9] and the expansion of pericyte gaps contributes to the enlargement of LERs during acute inflammation, we asked whether PMN transmigration might regulate the opening of pericyte gaps and hence the ensuing enlargement of LERs. To examine this, we fluorescently immunostained resting or IL-1β-injected mouse cremaster muscles and analyzed them using confocal microscopy. Normally few PMNs were observed in the lumen of resting venules (left panel in Figure 2A). Analyzing the association of PMNs and vessel walls, we noted two types of venular segments in IL-1β-stimulated tissues. In some segments transmigrating and fully extravasated PMNs were seen (middle panel in Figure 2A), whereas in others neither transmigrating nor extravasated PMNs were observed in the venular walls or in the perivascular tissues within 100 µm of these vessel segments (right panel in Figure 2A). In the former type of vessel segments, we refer to the pericyte gaps being used by transmigrating leukocytes, as leukocyte-associated pericyte gaps (circled in the middle panel in Figure 2A). In the latter, we refer to them as non-leukocyte-associated pericyte gaps (boxed in the right panel in Figure 2A). Accordingly, we refer to LERs as leukocyte-associated or non-leukocyte-associated collagen IV LERs (Figure 1A). We found that the average area of pericyte gaps in untreated venules was less than that of non-leukocyte-associated gaps in IL-1β-stimulated vessels (Figure 2B), implying a general expansion of pericyte gaps occurs during inflammation induced by this cytokine. However, leukocyte-associated pericyte gaps were even larger than those non-leukocyte-associated (Figure 2B), suggesting that the interaction of pericytes with PMNs had expanded pericyte gaps. To examine this possibility, we depleted most (>89%) of circulating leukocytes using anti-Gr-1 antibodies [15]. This decreased the number of both transmigrating and extravasated PMNs in inflamed tissues (Figure 2C–2E). Since not all circulating PMNs can be removed by Gr-1 antibodies, in situations where transmigration occurred in anti-Gr-1 antibody-pretreated mice (right panel in Figure 2C), the leukocyte-associated pericyte gaps still had an average area similar to that found in animals pretreated with control antibodies (Figure 2F). However, when the total area of pericyte gaps and collagen IV LERs was normalized to a 200 µm vessel segment, leukocyte-associated pericyte gaps and collagen IV LERs in Gr-1 antibody-treated mice were much smaller than in control mice (Figure 2G, 2H). These experiments indicate that the presence of transmigrating PMNs further expands pericyte gaps as well as the LERs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Depletion of circulating PMNs decreases the expansion of pericyte gaps and collagen IV LERs during inflammation.

(A) 3D images showing the association of pericyte gaps and PMNs in mouse cremaster muscles that were untreated (left) or treated with IL-1β (middle and right). Arrows indicated pericyte gaps and a circle showed PMNs migrating through pericyte gaps. In IL-1β-stimulated tissues, two types of venular segments were noted. In some vessel segments (middle), both transmigrating and extravasated PMNs were seen. In others, no PMNs were found in the vessel walls or within 100 µm of the perivascular tissues (right). In the first situation (middle), pericyte gaps being used by transmigrating PMNs (circled area) are referred to as “leukocyte-associated pericyte gaps”. In the second situation (right, boxed area), pericyte gaps are referred to as “non-leukocyte-associated pericyte gaps”. Note: we did not classify the pericyte gaps not circled in the middle panel since we were not sure whether these gaps were used by leukocytes or not. However, these were included in the measurements when analyzing the total pericyte gaps. Similarly, collagen IV LERs were described as “leukocyte-associated LERs” or “non-leukocyte-associated LERs”. (B) Average area of pericyte gaps in untreated or IL-1β-stimulated mouse cremaster muscle venules. (C) 2D images along the longitudinal axis of venules treated with IL-1β together with control or anti-Gr-1 antibodies. Transmigrating PMNs are circled. (D) Number of transmigrating PMNs within venular walls treated with saline or IL-1β together with control or anti-Gr-1 antibodies. (E) Number of extravasated PMNs within mouse cremaster muscles treated with saline or IL-1β together with control or anti-Gr-1 antibodies. (F) Average area of pericyte gaps in untreated venules or vessels treated with IL-1β together with control or anti-Gr-1 antibodies. (G) The total area of leukocyte-associated pericyte gaps was normalized relative to a 200 µm length of vessel. (H) The total area of leukocyte-associated collagen IV LERs was normalized relative to a 200 µm length of vessel. Bar = 10 µm. 5–6 mice were used in each treatment. In (B) and (F), ANOVA plus Neuman-Keuls multiple comparisons were performed and in others, t tests were applied. *P<0.05, **P<0.01, and ***P<0.001, compared to untreated groups. ^###P<0.001, compared between gaps as indicated.

https://doi.org/10.1371/journal.pone.0045499.g002

Direct PMN-pericyte Contact Mediates Enlargement of Pericyte Gaps and LERs

We next asked how migrating PMNs regulated the opening of pericyte gaps. The pericyte response to leukocytes could be mediated either by diffusible mediators or direct cell-cell contacts [11], [24]. Here we examined the role of direct contact between PMNs and pericytes. We took advantage of the observation that PMNs from PECAM-1^−/− mice display retarded transit across the IL-1β-stimulated venular BM within the first 4 hour time frame after IL-1β injection due to their failure to express α₆ integrin on their cell surface [17], [20]. Hence in our study, we injected IL-1β into the cremaster muscles of wild type (WT) or PECAM-1^−/− mice to induce inflammation and used them as tools to test whether abolition of PMN-pericyte contact influenced the opening of pericyte gaps. In IL-1β-injected PECAM-1^−/− mice, the number of PMNs completing extravasation was significantly lower than in WT mice (Figure 3A, 3B), despite no difference in the number of PMNs involved in adhesion (observed by intravital microscopy) or attempting transmigration (Figure 3C). Migrating PECAM-1^−/− PMNs might pass across the endothelium but frequently stopped at the vascular BM, being separated from the pericyte sheath by this barrier (bottom right in Figure 3A), whereas WT PMNs broke through the vascular BM and reached the pericyte sheath (bottom left in Figure 3A). This measurement was previously confirmed by transmission electron microscopy [17]. The average area of total pericyte gaps and non-leukocyte-associated pericyte gaps in IL-1β-stimulated PECAM-1^−/− venules was greater than that in resting vessels, suggesting that a general expansion of pericyte gaps also occurs in PECAM-1^−/− venular walls during acute inflammation. Unlike WT PMNs, however, migrating PECAM-1^−/− PMNs did not cause any additional expansion of leukocyte-associated pericyte gaps (Figure 3D). These observations indicate that transmigrating PMNs separated from the pericyte sheath fail to enlarge pericyte gaps beyond the level induced by IL-1β.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Direct cell-cell contact between transmigrating PMNs and pericytes mediates expansion of pericyte gaps and collagen IV

LERs. (A) 2D images of IL-1β-stimulated venules along the longitudinal axis. The areas outlined in the top panels are shown magnified in the bottom panels, and reveal that transmigrating PECAM-1^−/− PMNs become trapped at the vascular BM. (B) Number of extravasated PMNs in WT or PECAM-1^−/− mouse cremaster muscles treated with saline or IL-1β, or in WT tissues injected with IL-1β together with control, anti-ICAM-1 or anti-α₆ integrin antibodies. (C) Number of transmigrating PMNs within WT or PECAM-1^−/− venular walls treated with saline or IL-1β, or within WT vessels treated with IL-1β together with control, anti-ICAM-1 or anti-α6 integrin antibodies. (D) Average area of pericyte gaps in untreated or IL-1β-injected WT or PECAM-1^−/− venules in mouse cremaster muscles. (E) The area of leukocyte-associated pericyte gaps was compared for WT and PECAM-1^−/−cremaster muscles, as well as with those treated with the indicated antibodies. (F) The area of leukocyte-associated collagen IV LERs was compared for WT and PECAM-1^−/−cremaster muscles, as well as with those treated with the indicated antibodies. (G) The normalized MFI of leukocyte-associated collagen IV LERs was compared for WT and PECAM-1^−/− cremaster vessels, as well as with those treated with the indicated antibodies. Bar = 10 µm. 7–8 mice for each treatment. In panel (D), (E) and (F), ANOVA plus multiple comparisons (Comparisons with untreated (D) or control Ab (E, F) groups were displayed by *. Additional comparisons between two different groups were indicated by #), and in others, t tests were performed. * or ^#P<0.05, **P<0.01, and ***P<0.001.

https://doi.org/10.1371/journal.pone.0045499.g003

Using intravital microscopy, we noted that anti-ICAM-1 antibodies dramatically inhibited leukocyte adhesion to the endothelium (data not shown). Immunostaining of whole-mounted tissues revealed that these antibodies reduced both transmigrating and extravasated PMNs in IL-1β-stimulated WT cremaster muscles (Figure 3B, 3C). Anti-α₆ integrin antibodies did not influence the number of transmigrating PMNs in WT mice (Figure 3C) but trapped these leukocytes in the vascular BM (data not shown), therefore decreasing the number of emigrated PMNs (Figure 3B). Importantly, we found that both these antibodies markedly prevented expansion of both the leukocyte-associated pericyte gaps and the LERs (Figure 3E, 3F), similar to what was observed in PECAM-1^−/− animals (Figure 3E–3G). Taken together, these findings suggest that blocking contact between PMNs and pericytes prevents the signaling in pericytes required to expand the gaps between them.

Engagement with Activated PMNs Induces Pericyte Relaxation

To examine the signaling initiated in pericytes by PMNs, we incubated mouse primary pericytes with PMNs which were pre-treated with DMSO- or phorbol 12-myristate 13-acetate (PMA), an elevator of integrin expression as well as an activator of integrins on the PMN surface [25], [26]. Analyzed by time-lapse images, no significant change in pericyte shape was observed after pericytes were stimulated by DMSO-treated PMNs (Figure 4A), whereas exposure to PMA-activated PMNs induced pericytes to extend prominent lamellipodia, elongate and increase their motility (Figure 4B, Movie S1 and Movie S2). In addition, pericytes disassembled stress fibers and focal adhesions and decreased myosin light chain (MLC) phosphorylation upon stimulation of PMA-activated PMNs, indicative of cell relaxation [27] (Figure 4C–4G).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Engagement of activated PMNs induces pericyte relaxation.

(A) Changes in mouse pericyte morphology before and after adding DMSO-treated PMNs or (B) before and after adding PMA-treated PMNs. (C) After incubation of DMSO- or PMA-treated PMNs with mouse primary pericytes, cells were fixed and stained for F-actin (green), nuclei (blue), and paxillin (red) or phosphorylated MLC (red) (D). (E) Change in the level of phosphorylated myosin light chain (phospho-MLC) in pericytes responding to DMSO- or PMA-treated PMNs. Change in the area (F) and the number (G) of focal adhesions in pericytes responding to DMSO- or PMA-treated PMNs. (H) Change in the level of phospho-MLC in pericytes responding to NaCl- or CXCL1-treated PMNs. Change in the area (I) and the number (J) of focal adhesions in pericytes responding to NaCl- or CXCL1-treated PMNs. T test, *P<0.05, **P<0.01 and ***P<0.001. (K) Response of mouse pericytes to PMNs activated by CXCL1. Cells were stained for F-actin (green), paxillin (red) and nuclei (blue). (L) Response of mouse pericyte to PMNs activated by CXCL1. Cells were stained for F-actin (green), phospho-MLC (red) and nuclei (blue). (M) Response of mouse pericytes to PMNs that have passed across an endothelial layer in vitro in response to IL-8 or that have been collected from the mouse peritoneal cavity following thioglycollate intra-peritoneal injection. Cells were stained for F-actin (green), paxillin (red) and nuclei (blue). Bar = 10 µm.

https://doi.org/10.1371/journal.pone.0045499.g004

At concentrations above 1 ng/ml, PMA induces a similar phenotype in pericytes to that caused by PMA-activated PMNs (data not shown). We were concerned that the effects of activated PMNs on pericytes might be due to PMA leaching out of the PMNs. To exclude this possibility, we developed two experimental strategies. In one of these, we cultured pericytes on two sides of a FN-coated coverslip and stimulated them with serum-free medium containing TNF-α. After washes, PMA-activated PMNs were added onto the top side of pericyte layers. Cells were fixed and stained for F-actin. We found that pericytes on the top surface (right image in Figure S3A) exhibited large lamellipodia and lost actin stress fibers, compared to the cells growing on the bottom surface (left image in Figure S3A) where contact with PMNs was not possible. In another strategy, we collected the medium from the tissue culture wells in which mouse primary pericytes had interacted with PMA- or DMSO-treated PMNs and used these conditioned medium to stimulate another batch of pericyte layers growing on FN-coated coverslips. Consistent with the aforementioned results, exposure to PMA-treated (top left image in Figure S3B) rather than DMSO-treated PMNs (top right image in Figure S3B) led to loss of actin stress fibers and focal adhesions in pericytes. However, conditioned medium collected from the wells, in which either PMA-treated or DMSO-treated PMNs had interacted with pericytes, had no influence on the morphology and cytoskeleton in the second batch of pericytes (bottom images in Figure S3B). These observations suggest that the pericyte response was not induced by residual PMA, but rather resulted from direct contact with activated PMNs that induced relaxation of the pericyte actomyosin system.

We also examined pericyte morphology and cytoskeletal organization in response to PMNs activated by more physiological agents. PMNs treated with murine CXCL1 (KC), an inflammatory chemokine that has neutrophil chemoattractant activity [28], induced a loss of stress fibers and focal adhesions, and decreased MLC phosphorylation (Figure 4H–4L). Pericytes also exhibited a relaxation phenotype (decreased stress fibers and focal adhesions) after engagement with PMNs that had been stimulated with IL-8 and collected after passage across an endothelial cell layer in vitro (Figure 4M, middle panel). We also induced an inflammatory response in mouse peritoneal cavities by injection of thioglycollate. PMNs were then isolated from the peritoneum and incubated with pericytes. Again, we observed that these physiologically activated PMNs induced a loss of pericyte stress fibers and focal adhesions (Figure 4M, lower panel). Additionally, a similar phenotype was seen when pericytes were exposed to PMNs that had been activated by MnCl₂, which increases integrin affinity [29], [30] (Figure S4).

Pericyte Relaxation Results from Inhibition of the RhoA/Rho Kinase (ROCK) Signaling Pathway and Suppression of Actomyosin-based Contractility

Because PMN interaction with pericytes induced stress fiber disassembly, we examined the role of RhoA and its effector, ROCK [31] in regulating pericyte contractility. Pericytes in culture usually display prominent stress fibers and focal adhesions, indicative of a high level of contractility. We examined whether RhoA in cultured pericytes can be activated or inactivated after exposure to different stimuli. To this end, a series of compounds was added to primary pericyte cultures individually. Lysophosphatidic acid (LPA) [32] increased RhoA activation in mouse primary pericytes, while Forskolin, an elevator of cAMP [33], inhibited its activation (Figure 5A). This suggested that RhoA activation in pericytes could be either increased or suppressed in our culture conditions, depending on the type of stimulus. Morphological analysis demonstrated that RhoA activators, including LPA and thrombin [34], did not have much effect on pericyte shape. However, RhoA inhibitors such as Forskolin and C3 Transferase (C3) [35] decreased MLC phosphorylation in pericytes (Figure 5B, 5D) and induced loss of stress fibers and focal adhesions (Figure 5C–5F) resulting in a migratory phenotype. These responses are similar to those triggered by interaction with activated PMNs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Inhibition of endogenous RhoA leads to a relaxation phenotype in pericytes.

(A) Mouse primary pericytes were treated with LPA, Forskolin (For) or their vehicles. RhoA activity in cell lysates was detected using a pull-down assay. n = 5, t test and **P<0.01 as indicated. (B) Mouse primary pericytes were untreated or treated with LPA or Forskolin. Cells were fixed and stained for F-actin, phospho-MLC and nuclei (shown in blue). (C) Mouse primary pericytes were treated with different compounds such as Thrombin, C3, LPA and Forskolin. Cells were fixed and stained for F-actin, paxillin and nuclei (shown in blue). Bar = 10 µm. (D) Change in the level of phosphorylated MLC in pericytes responding to different compounds. Change in the area (E) and the number (F) of focal adhesions in pericytes responding to the different compounds. Three independent experiments and ≥10 fields/experiment were analyzed in D, E and F. T test, *P<0.05, **P<0.01 and ***P<0.001, compared to vehicles (i.e. DMSO or PBS).

https://doi.org/10.1371/journal.pone.0045499.g005

In order to examine the effect of activated PMNs on RhoA activity within pericytes, we used viral vectors to express GFP-tagged WT RhoA in pericytes (Figure S5). In addition, we also expressed GFP-tagged constitutively active (CA) Q63L or dominant negative (DN) T19N RhoA or GFP alone in pericytes (Figure S5). This allowed us to distinguish RhoA in pericytes from RhoA in PMNs when investigating the two cell types together. Activated PMNs decreased the activity of exogenously expressed WT RhoA in pericytes when compared with non-activated PMNs (Figure 6A). However, no depression was seen in cells expressing the constitutively active RhoA (Figure S6). As expected, LPA activated WT RhoA, whereas Forskolin inhibited it. Neither treatment influenced the activity of CA or DN RhoA (Figure S6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Engagement of activated PMNs inhibits RhoA activity in pericytes.

(A) Pericytes expressing GFP-tagged WT human RhoA were exposed to DMSO-treated or PMA-treated PMNs. Activity of GFP-tagged RhoA was detected. n = 3, t test. ***P<0.001. (B) Pericytes expressing GFP-vector or GFP-tagged CA human RhoA were exposed to DMSO-treated or PMA-treated PMNs for one hour. Cells were fixed and stained for F-actin (green), paxillin (red), and nuclei (blue). GFP was shown in white. Expression of CA RhoA in pericytes blocked the response induced by PMA-treated PMNs. Bar = 10 µm.

https://doi.org/10.1371/journal.pone.0045499.g006

Compared to cells expressing GFP alone, pericytes expressing CA RhoA maintained prominent focal adhesions and stress fibers after exposure to PMA-activated PMNs (Figure 6B). In addition, knocking down endogenous RhoA by ShRNA led to loss of stress fibers and focal adhesions in pericytes (Figure S5A, S5B and Figure S7). We also examined pericyte response to Y27632 and Blebbistatin, inhibitors for ROCK [36] and myosin ATPase [37] respectively that are downstream of RhoA. We found that both of these compounds decreased the phosphorylation of MLC and induced disassembly of stress fibers and focal adhesions in pericytes (Figure 7A–7E). Together these results support a model in which activated PMNs induce the loss of stress fibers and focal adhesions in pericytes by inhibiting the RhoA-ROCK pathway leading to reduced pericyte actomyosin-based contractility [31], [38].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Inhibition of ROCK or myosin ATPase, downstream of RhoA, induces a relaxation phenotype in pericytes.

(A) Mouse primary pericytes were treated with the inhibitors Y27632 or Blebbistatin. Cells were fixed and stained for F-actin, phospho-MLC and nuclei. (B) Mouse primary pericytes were treated with the inhibitors Y27632 or Blebbistatin. Cells were fixed and stained for F-actin, paxillin and nuclei. Bar = 10 µm. (C) Change in the level of phosphorylated MLC in pericytes responding to different compounds. Change in the area (D) and the number (E) of focal adhesions in pericytes responding to the different compounds. Three independent experiments and ≥10 fields/experiment were analyzed in C, D and E. T test, *P<0.05, **P<0.01 and ***P<0.001.

https://doi.org/10.1371/journal.pone.0045499.g007

Pericyte Relaxation Affects the Morphology of Pericyte Monolayers

Our results suggest that the pericyte sheath serves as a barrier to leukocyte extravasation. We wished to examine the barrier properties of pericytes further and particularly the formation of gaps between pericytes in tissue culture. However, unlike endothelial cells, pericytes do not form monolayers in which adjacent cells are held together by adherent junctions and tight junctions. Consequently, conventional electrical resistance measurements are not a practical approach for monitoring changes in permeability as PMNs migrate across the pericyte monolayer. However, electrical impedance measurements provide a measure of cell area and adhesion to the underlying substratum [39], [40] and we have taken advantage of this approach. Compounds decreasing actomyosin contractility, including Forskolin, C3, Y27632 and Blebbistatin, decreased the impedance of pericyte layers and led to a relaxation phenotype of pericytes (Figure 8A, 8B, Figure 5 and Figure7). However, agents that increased RhoA activity slightly increased or had no effect on impedance (data not shown). Incubating the pericytes with PMA-activated or MnCl₂-activated PMNs reduced impedance of the pericyte layer (Figure 8C and Figure S4). Examining pericyte monolayers revealed that the spaces between pericytes were enlarged by activated PMNs and by the presence of relaxants, such as Forskolin or C3 (Figure 8D, 8E). Agents that stimulate contraction, such as LPA, did not induce or expand the gaps between pericytes (Figure 8D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Relaxation of pericytes expands pericyte gaps in a pericyte monolayer.

(A)–(C) On the left, the relative impedance is shown for pericytes responding to different treatments. Representative real-time read-outs are shown on the right. The relative impedance at 1 hour after adding a chemical (indicated by the red vertical line) was normalized to the impedance when reagents or cells were added (indicated by the black vertical line). In (A), the effects of Forskolin or C3 were analyzed, and in (B), the effects of Y27632 or Blebbistatin were compared. (C) the effect of DMSO- or PMA-treated PMNs on the impedance of primary pericyte monolayers. (D) Change in area of pericyte gaps in culture. Calcein green-labeled pericytes were grown to confluence on FN-coated coverslips in the wells of a 24 well plate. Compounds such as LPA or Forskolin, or Calcein blue-labeled mouse PMNs, pretreated with DMSO or PMA, were added to pericyte monolayers for 1 hour. Cells were fixed and imaged. The area of pericyte gaps was measured. (E) Representative images showing the change in size of gaps between pericytes upon stimulation of DMSO- or PMA-pretreated PMNs. Prominent pericyte gaps are indicated by white X’s. Note: The same amount of DMSO- or PMA-treated PMNs were added onto pericyte monolayers. However, due to the change in pericyte morphology and PMA-treated PMN transmigration, the density of both PMNs and pericytes looks lower in the PMA-treated group. Bar = 50 µm. n = 4–5. T test as indicated. *P<0.05, ** or ^##P<0.01, ***P<0.001.

https://doi.org/10.1371/journal.pone.0045499.g008

Pericyte Relaxation Facilitates PMN Transmigration both in vitro and in vivo

Our data are consistent with a model in which PMN interaction with pericytes induces pericyte relaxation and a consequent expansion of the gaps between pericytes, thereby facilitating passage of PMNs across the pericyte sheath. We wished to determine whether agents that affect pericyte contractility in vivo may affect PMN passage across the pericyte layer. Two drugs that have been used clinically to induce vasoconstriction or vasodilation are Norepinephrine (NE) and Tolazoline, respectively. NE is an agonist [41] of α-adrenergic receptors in pericytes and smooth muscle cells, whereas Tolazoline is an antagonist [42]. We first examined the response of pericytes to these agents in pericyte cultures. Our data demonstrated that pericytes maintained their actin stress fibers and focal adhesions in response to NE but lost these structures in the presence of Tolazoline (Figure 9A). NE activated RhoA and slightly increased the impedance of pericyte monolayers, whereas Tolazoline decreased RhoA activity and decreased the impedance of cultures (Figure 9B, 9C). Image analysis of pericyte monolayers revealed that gaps between pericytes were enlarged by the presence of Tolazoline whereas NE did not induce or expand the gaps between pericytes (Figure 9D, 9E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 9. Effect of vasoactive drugs, NE and Tolazoline, on pericyte morphology and permeability of pericyte monolayers.

(A) Mouse primary pericytes were treated with buffer, NE or Tolazoline. Cells were fixed and stained for F-actin, paxillin and nuclei. Bar = 10 µm. (B) NE increased RhoA activity in pericytes while Tolazoline inhibited it. NE or Tolazoline were added to pericytes and RhoA activity was measured in a pull-down assay. n = 5. T test. *P<0.05 and **P<0.01, compared to vehicle. (C) Effects of NE, Tolazoline or NE + Tolazoline on the impedance of a pericyte monolayer were compared. On the left, the relative impedance was shown for pericytes responding to different treatments. Representative real-time read-outs were shown on the right. The relative impedance at 1 hour after adding a chemical (indicated by the red vertical line) was normalized to the impedance when reagents or cells were added (indicated by the black vertical line). n = 4–5. T test as indicated. P<0.05 and **P<0.01. (D) Change in area of pericyte gaps in culture. Pericytes were grown to confluence on FN-coated coverslips in the wells of a 24 well plate. NE, tolazoline, or NE plus Tolazoline were added to pericyte monolayers for 1 hour. Cells were fixed and imaged. The area of pericyte gaps was measured. n = 5. T test as indicated. **P<0.01. (E) Representative images showing the change in size of pericyte gaps upon stimulation of vehicle buffer, NE or NE plus Tolazoline. Pericyte gaps are indicated by red *’s. Bar = 50 µm.

https://doi.org/10.1371/journal.pone.0045499.g009

We next examined the ability of PMNs to pass across pericyte monolayers grown on FN-coated filters of a Boyden transwell system. Due to strong adhesion to the pericytes and the filters, migration of PMA-activated PMNs toward IL-8 in the lower chamber was less than control PMNs (See the first two bars in Figure 10A). However, as discussed above, engagement of activated PMNs expands pericyte gaps. Pre-exposure of pericyte monolayers to activated PMNs induced an increased transmigration (six fold) of a second batch of Calcein-AM-labeled resting PMNs. This response was diminished when the pericyte layers expressed CA RhoA (Figure 10A). In addition, more PMNs passed through pericyte layers pretreated with relaxants such as Forskolin and Tolazoline than through layers treated with NE or vehicles (e.g. DMSO or PBS) (Figure 10B). These findings indicate that migration of PMNs across pericyte monolayers is promoted by inhibiting pericyte contractility, which results in the expansion of gaps between pericytes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 10. Relaxation of pericytes facilitates PMN transmigration in vitro.

(A) Pericyte engagement with PMA-activated PMNs enhanced PMN passage across a pericyte monolayer. n = 5. Note: Due to strong adhesion to the pericytes and the filters, migration of PMA-activated PMNs toward IL-8 in the lower chamber was less than that of control PMNs (See the first two bars). (B) PMN transmigration through a pericyte monolayer in the presence of agents that affect pericyte contractility. n = 4. *P<0.05 and **P<0.01 as indicated.

https://doi.org/10.1371/journal.pone.0045499.g010

To examine whether modulating the contractility of pericytes affects PMN transmigration in vivo, we topically applied NE or Tolazoline to exteriorized cremaster muscles which had been pre-stimulated with IL-1β. Treatment with Tolazoline significantly enlarged pericyte gaps (Figure 11A). In addition, Tolazoline induced an expansion of the LERs and caused these to become thinner (i.e. there was decreased collagen fluorescence intensity) (Figure 11B, 11C). Examining the distribution of PMNs, we observed that Tolazoline released PMNs from the space between ECs and pericytes, reducing the number of PMNs within the vessel wall (Figure 11D, 11E) but increasing the number of extravasated leukocytes (Figure 11D, 11F). In contrast, NE decreased the area of pericyte gaps and LERs, compared to a saline control (Figure 11A, 11B). NE tended to increase the number of PMNs in the space between ECs and pericytes but decrease the number of PMNs that successfully completed extravasation (Figure 11D–11F). These in vivo data further support that relaxation of pericytes expands the gaps between them, as well as the adjacent LERs, to facilitate PMN passage across inflamed venular walls.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 11. Relaxation of pericytes facilitates PMN transmigration in vivo.

(A) Change in area of pericyte gaps in IL-1β-stimulated venular walls exposed to Tolazoline or NE. (B) Change in area of collagen IV LERs in IL-1β-stimulated venules treated with Tolazoline or NE. (C) Change in MFI of collagen IV LERs in IL-1β-stimulated venular walls treated with Tolazoline or NE. (D) Representative images of IL-1β-stimulated cremaster muscle venules illustrating the effect of Tolazoline and NE on PMN transmigration in vivo. The top panels show 2D images from the middle sections of these venules, stained as indicated to reveal transmigrating and extravasated PMNs. Higher magnifications of the boxed areas (marked with Arabic numerals) are displayed on the right. The bottom 3D images are reconstructions of the same venules stained as indicated to reveal extravasated PMNs. (E) Number of PMNs in the process of transmigrating in IL-1β-stimulated cremasters treated with saline, Tolazoline or NE. (F) Number of extravasated PMNs in the perivascular tissues 100 µm around a 200 µm vessel segment in IL-1β-injected mouse cremaster muscles. 7–8 mice and 15 vessel segments from each mouse were analyzed in each group of the experiments displayed. Bar = 10 µm in the zoom-in views, and = 50 µm in other images. ANOVA plus multiple comparisons were performed. Comparisons with saline control were displayed by *P<0.05, **P<0.01, and ***P<0.001. Other comparisons were indicated as ^##P<0.01 and ^###P<0.001.

https://doi.org/10.1371/journal.pone.0045499.g011