Experimental Approaches to Evaluate Leukocyte-Endothelial Cell Interactions in Sepsis and Inflammation

Abstract

Sepsis is a life-threatening syndrome of organ dysfunction caused by a dysregulated host response to infection characterized by excessive neutrophil infiltration into vital organs. In sepsis, patients often die of organ failure and therapies directed against endothelial cell dysfunction and tissue damage are important targets for treatment of this disease. Novel approaches are required to understand the underlying pathophysiology of neutrophil dysregulation and neutrophil-endothelial cell interactions that play a critical role in the early course of organ damage and disruption of endothelial protective barrier. Here we review methodologies that our laboratories have employed to study neutrophil-endothelial interaction and endothelial barrier function in in vivo and in vitro models of sepsis. We will focus on in vivo rodent models of sepsis and in vitro tools that use human cell culture models under static conditions and the more physiologically relevant biomimetic microfluidic assays. This Methods paper is based on our presentation in the Master Class Symposium at the 41^st Annual Conference on Shock 2018.

Introduction

Sepsis is a clinical syndrome defined as life-threatening organ dysfunction caused by dysregulated host response to infection (Sepsis-3,(1)). Sepsis is a major health issue with over 1.7 million cases/year and >250,000 deaths/year in the US (2–4). Patients with sepsis often die of organ failure and therapies directed against endothelial cell dysfunction and tissue damage are important targets for treatment of this disease (5–8). Neutrophil dysregulation and neutrophil-endothelial cell interactions have a critical role in the early course of organ damage, but the underlying pathophysiology involved in neutrophil-endothelial cell interactions and disruption of endothelial protective barriers has yet to be fully delineated (6, 9). Here we present methodologies that we have employed to study neutrophil-endothelial interaction and endothelial barrier function in our in vivo and in vitro models of sepsis. We will focus on rodent models of sepsis and in vitro tools that use human cell culture models under static conditions and the more physiologically relevant biomimetic microfluidic assays.

The Leukocyte Adhesion Cascade

Key to sepsis-induced tissue damage is the excessive migration of activated neutrophils across the vascular endothelium (10, 11). Neutrophil recruitment during sepsis is a multi-step process that includes five discrete steps: capture/attachment, rolling, firm arrest, spreading and extravasation/migration. Each step of this process requires crosstalk between neutrophils and endothelial cells to orchestrate this dynamic phenomenon. Specifically, circulating neutrophils tether and roll along the vessel wall by establishing transient selectin-mediated interactions with endothelial cells. This initial contact facilitates the binding of neutrophil integrins (β2 and/or β1 integrins) to counter-receptors on the endothelium, which allows firm adhesion of neutrophils to endothelium and resistance to disruptive hemodynamic shear forces. The expression and avidity of β2-integrins on neutrophils is increased by transient rolling on endothelium through contact with inflammatory mediators (e.g. Interleukin-8 (IL-8) and platelet-activating factor (PAF)) expressed on the surface of endothelial cells (10). On endothelial cells, selectins (e.g. P & E-selectin) are responsible for neutrophil capture and rolling, while adhesion molecules ICAM-1, VCAM-1, PECAM-1, and JAM-C (junctional adhesion molecule-C) are critical regulators of neutrophil firm attachment and migration (12–14). Ultimately, arrested neutrophils extravasate to inflamed tissues across endothelial cells via a multi-step process controlled by concurrent chemoattractant-dependent signals, adhesive events and hemodynamic shear forces (7, 8, 15–17).

Neutrophil-endothelium interaction is influenced by endothelial cell phenotypes as well as disease states and the shift in neutrophil phenotypes. Endothelial cell heterogeneity in different organs is well-described and organ-specific variations in endothelial structure and function lead to differential mechanisms of neutrophil recruitment. Studies show that in vivo heterogeneity is in part dictated by intrinsic signals that are differentially expressed across organs and species, as a result of which neutrophil recruitment to discrete organs may be regulated differently (18–20). For example, recruitment of neutrophils to the lung and kidney require β2-integrin-ICAM-1 interactions (19). In contrast, neutrophil migration in the liver sinusoids is selectin-independent and neutrophils are recruited by chemokines and CD44 binding (16, 19). In addition to endothelial cell heterogeneity, several different subtypes of neutrophils have been described under various pathologic conditions. Different neutrophil subpopulations have been identified based on functional activities, organ location, and morphological characteristics such as cell density, ability to generate neutrophil extracellular traps (NETS), variations in receptor expression, bactericidal activity, and survival time (21–23). With the advent of new technologies, such as RNA-Seq, unbiased transcriptome-wide profiling can now be used to identify immune cell diversity and specific disease-related gene signatures. Further single cell RNA-seq enables the identification of distinct cell types through single cell transcriptome profiling that can be used to identify distinct cell types in sepsis (24). Thus, neutrophil-endothelial cell interactions are dependent on organ specificity and types of neutrophil subpopulations involved.

Methodology

In vivo Methods to Study Neutrophil Migration

Neutrophil-endothelial interaction and endothelial barrier function can be studied in rodent models of sepsis. One of the best-characterized models is polymicrobial sepsis induced by cecal ligation and puncture (CLP). We have used this model in both rats and mice (25–31). In this model, rodents develop significant lung injury characterized by neutrophil accumulation, increased pulmonary vascular endothelial permeability, and tissue edema. Using this model, neutrophil-endothelial interaction can be examined by monitoring neutrophil influx into the lung and pulmonary adhesion molecule expression. Intravital microscopy offers a more direct visualization and quantification of neutrophil interaction with the vascular endothelium.

Neutrophil influx into tissues is analyzed by measuring myeloperoxidase activity in tissues or by immunofluorescence techniques. Myeloperoxidase is an enzyme stored in azurophilic granules of neutrophils. To illustrate the types of information that can be obtained using these techniques, we present studies that examined the effect of Protein Kinase C-delta (PKCδ) inhibition on neutrophil migration into the lung following CLP-induced sepsis. As shown in Figure 1A, immunohistochemical detection of MPO in lung tissue sections obtained 24 hr post-sham surgery showed low numbers of MPO positive cells. In contrast, 24hrs post CLP surgery there was a significant increase in MPO-positive cells in the lung that was reduced in septic rats treated with a PKCδ inhibitor (28). Neutrophil influx into organs can also be determined by measuring MPO enzymatic activity in lung tissue homogenates using the O-dianisidine MPO assay (32). Similar to what we observed employing the immunohistochemical technique, we found that 24hrs post sham surgery, there was minimal MPO activity in lung tissue homogenates (Figure 1B). In contrast, CLP surgery induced a substantial inflammatory response leading to increased lung injury and greater MPO activity, which was reduced by administration of the PKCδ inhibitor (26). Cell migration into the lung can also be monitored by analysis of bronchoalveolar lavage fluid (BALF) for specific immune cell types. It should be noted that more neutrophils are found in BALF in response to direct pulmonary injury such as pneumonia rather than in response to indirect pulmonary injury, i.e. CLP-induced sepsis.

Figure 1: Sepsis-induced neutrophil migration into the lung.

A. Immunohistochemical detection of MPO in representative lung tissue sections 24 hr post-surgery (n = 6–8 animals/group). Sham Surgery: only a few MPO-positive cells are seen in each field. Sepsis: following CLP surgery and intratracheal (IT) administration of PBS vehicle, sepsis induces the infiltration of numerous MPO-positive cells throughout the lung parenchyma. Sepsis + PKCδ-TAT Inhibitor: following CLP surgery and IT administration of the PKCδ-inhibitor, there was significant reduction of sepsis-induced, MPO-positive cell numbers in the lung. Original magnification, 400×. Reprinted with permission from Journal of Leukocyte Biology 2016; 100:1027–35 (28).

B: Lung MPO activity in rats following sham surgery with intratracheal (IT) instillation of 400 μl of PBS vehicle, Sepsis: following CLP surgery and IT administration of PBS vehicle, and Sepsis + PKCδ Inhibitor: following CLP surgery and IT administration of the PKCδ-inhibitor. Tissue was harvested at 24 hours post-surgery. Data are expressed as means ± S.E.M. (n = 4 rats/group). *P < 0.001 versus sham, **P<0.01 versus CLP + vehicle. Adapted from Mondrinos et al 2015 (26).

An important step in neutrophil migration is neutrophil adhesion to vascular endothelium. ICAM-1 and VCAM-1 are key adhesion molecules for neutrophil firm attachment and migration through vascular endothelium. During inflammation or infection, these adhesion molecules are upregulated on activation of vascular endothelium by proinflammatory cytokines (25, 33). Pulmonary adhesion molecule expression can be determined in lung tissue slices. We have employed immunohistochemistry with fluorescence visualization to access overall levels and distribution of ICAM-1 in response to CLP-induced sepsis. As shown in Figure 2, using the immunohistochemistry technique, we observed little ICAM-1 expression in the lungs from sham-surgery rats. In contrast, sepsis produced widespread ICAM-1 localization throughout the parenchyma and alveolar walls, while PKCδ inhibition decreased ICAM-1 expression but the expression was still appreciably higher than sham surgery animals (25). We have also employed an AEC (3-amino-9-ethyl-carbazole) chromogen–based detection to visualize ICAM-1 in the context of the overall histological appearance of the tissue and to identify cellular localization in the lung. For chromogen-based visualization of ICAM-1 and VCAM-1 staining, HRP-conjugated secondary antibodies are applied and the stains developed using the AEC chromogen (25).

Figure 2: ICAM-1 Expression in Sepsis-Induced Lung Injury.

Immunohistochemical detection of ICAM-1 expression in representative lung tissue sections. After primary antibody incubation, ICAM-1 is visualized using an Alexa Fluor 488-conjugated secondary antibody (green), with DAPI counterstaining (blue). Representative lung tissue sections from 24 hours after surgery are shown (n =4 rats per group). In the sham surgery group, levels of ICAM-1 are barely detectable. In the CLP +PBS group, widespread and intense sepsis-induced ICAM-1 staining throughout the lung parenchyma is observed. In the CLP+PKCδ-TAT group, marked reduction in sepsis-induced ICAM-1 expression is observed, with some small patches of staining seen in alveoli. Scale bars: 100 mm (right column). Adapted from Mondrinos et al 2014 (25).

Intravital Microscopy

Intravital microscopy provides a powerful technique for study of the microcirculation and permits continuous, direct, real-time measurement of multiple parameters including flow, leukocyte-endothelial interaction and permeability. We and others have used intravital microscopy to study mechanisms and clinical manifestations of a number of pathologies (34, 35). Intravital microscopy can be used to study the inflammatory response in various organs including the closed cranial window model to study the blood-brain barrier (36), cremaster muscle model (35), mesentery model (37), and lung model (38). In several models, leukocyte rolling and adhesion is measured by i.v. injection of Rhodamine-6G (0.4 mg/kg body weight) to fluorescently label leukocytes (39). The development of LysM-GFP mice in which endogenous circulating neutrophils, and to a lesser extent monocytes and macrophages, are fluorescent positive has enhanced the ability to track these immune cells in vivo in response to infection or injury (40). Several other cytoplasmic and nuclear fluorescent protein-based reporters and gene-transfer-free cell-labelling techniques for tagging neutrophils and other blood cells have also been used but toxicity and/or dilution upon cell division of these dyes should be carefully considered (41). Advances in microscopy with the use of 2-photon imaging and spinning disk confocal microscopy enables the study of immune responses in organs such as lung, liver and kidneys (40, 41).

In vivo methods to study lung permeability

During sepsis, damage to the pulmonary endothelial barrier results in increased permeability, precipitating an influx of protein-rich fluid that results in tissue edema and impaired gas exchange (42). In animal models, lung permeability and tissue edema can be determined by multiple methods including variations in lung weight, tissue accumulation of the dye Evans blue, and increased protein levels in BALF. The wet-to-dry lung weight ratio can be used as an index of lung water accumulation during sepsis-induced lung injury (43). Total lung water is determined by weighing the lung immediately after organ harvest (wet weight) and then by weighing the lung tissue after heating it for 72 hrs in a gravity convection oven at 60°C (dry weight). The wet-to-dry ratio is calculated by dividing the wet weight by the dry weight.

The uptake of Evans blue dye is often used to measure vascular permeability or plasma extravasation. This technique makes use of the albumin binding properties of the dye. Under anesthesia, Evans blue is injected into the jugular vein 30 min prior to termination of the experiment. The lungs are perfused to remove excess dye and the concentration of the dye in lung tissue is quantitated in tissue homogenates after correction for contaminating heme pigments. In our rat model of sepsis-induced lung injury, 24 hrs post-surgery, there was a 3 fold increase in Evans blue extravasation in septic rats as compared to sham surgery rats (n=3-4, P< 0.05, unpublished data). Measurement of BALF total protein concentration is also used as an index of lung permeability and capillary leak. In our studies, we found that 24 hrs post CLP surgery, total BALF protein content increased over 2.5-fold as compared to Sham surgery rats (27).

Vascular permeability can also be determined using intravital microscopy by estimating leakage of FITC-albumin (50mg/kg) across the vasculature by measuring the ratio of fluorescence intensity in the venules to the perivenular interstitium at regular time intervals using the following equation (44): $P=\frac{1}{I_{v0}}\frac{V}{S}\frac{{\mathit{dI}}_t}{\mathit{dt}}$ where I_t is the average intensity in the perivenular interstitium, I_v0 is the maximum fluorescence intensity of the venule, V/S is the ratio of venule volume to its surface area. Particle size impacts vascular permeability and to introduce controls for molecule size, permeability of two different color fluorescent tracers (e.g. 40 kDa dextran tagged with FITC and albumin tagged with Texas Red) can be measured simultaneously in one animal during intravital microscopy (44).

In Vitro Methods to Study Human Neutrophil Migration and Endothelial Cell Permeability

Due to the significance of the leukocyte-endothelium interactions, and given the complexity of existing in vivo models of the inflammatory process, several in vitro models have been developed to study various aspects of the leukocyte adhesion cascade in a more controlled environment. Traditional fluidic devices can be classified into two broad categories (45): 1) devices for studying leukocyte rolling and adhesion (e.g. parallel plate flow chamber) and 2) devices for studying leukocyte migration (e.g. transwell chamber). Systems designed to study leukocyte rolling and adhesion (46) have been used to study functional significance of variables such as adhesion molecules and shear forces in the adhesion cascade. A significant drawback of these types of devices is that flow chambers used to study leukocyte rolling and adhesion are simplistic (e.g. straight channels), lack the scale and geometry of the microenvironment, and cannot model transmigration.

Devices to study transmigration are classical 2D models, such as the transwells or Boyden chambers that measure unidirectional migration under static conditions. We have used the Transwell culture system to delineate the role of PKCδ in regulating neutrophil transmigration under static conditions (25). In this system, genetic (PKCδ siRNA) and pharmacological (PKCδ inhibitor) techniques are employed to inhibit PKCδ activity in human pulmonary microvascular endothelial cells (PMVEC). PMVEC are grown in collagen-coated transwell inserts until confluency and then activated with IL-1β. Isolated human neutrophils are added to the upper wells and neutrophil migration into the bottom well is determined in the absence or presence of chemoattractants such as fMetLeuPhe (fMLP) or IL-8. In siRNA experiments, endothelial cells are treated with PKCδ targeted siRNA to selectively deplete PMVEC of PKCδ, but not PKCα, PKCβI/II or PKCζ. PMVEC transfected with percent-GC control siRNA are used as controls (25). Neutrophil migration is significantly reduced in PMVECs depleted of PKCδ (Figure 3). PMVEC treated with the PKCδ inhibitor also showed decreased neutrophil migration indicating endothelial cell PKCδ is an important component of IL-1β-induced transmigration (25). In further studies, we found PKCδ regulation of neutrophil migration is stimulus-dependent (Figure 3). PKCδ activity is required for IL-1β and fMLP-mediated migration (integrin-dependent), but not for IL-8-mediated migration (integrin-independent) (25).

Figure 3: Neutrophil Transmigration In Vitro.

A: Selective depletion of PKCδ by stealth PKCδ siRNA in PMVECs. Controls were PMVECs treated with siRNA with equivalent percentage of GC nucleotide content (GC control) as the stealth PKCδ siRNA. Levels of specific PKC isotypes were determined in cell lysates by immunoblotting with isotype-specific antibodies to PKCδ, PKCα, PKCβI/II, and PKCζ. Blots are representative of three separate Western blotting experiments. B: Neutrophil transmigration through IL-1β–activated PMVECs. PMVECs were grown on Transwell inserts and incubated with IL-1β or buffer after transfection with PKCδ stealth siRNA or siRNA containing equivalent percentage of GC nucleotide content (Control). In a second series of experiments, PMVECs were grown on Transwell inserts and were pretreated with buffer, 10 U/mL IL-1β+vehicle, IL-1β+TAT-TAT control peptide, or IL-1β+PKCδ inhibitor. Neutrophils (1 × 10⁶ per well) were added to the upper well of the Transwell culture system and allowed to migrate for 1.5 hours through PMVECs. C: Role of PKCδ in neutrophil transmigration mediated by IL-8 and fMLP. Chemoattractants fMLP (1 nmol/L) or IL-8 (2 nmol/L) were added to the bottom well of a Transwell culture system. Neutrophils (2 × 10⁶ per well) were added to the upper well and were allowed to migrate for 1.5 hours through PMVECs pretreated with buffer or with 2 μmol/L PKCδ inhibitor. Data are expressed as means±SEM. n = 4 (B, siRNA experiments); n = 3–9 (B, PKCδ inhibitor experiments); n = 6–9 (C). **P < 0.01, ***P < 0.001 versus respective control or controls (B). **P < 0.01 fMLP versus fMLP+PKCδ inhibitor–treated PMVECs; ***P < 0.001 fMLP or IL-8 versus buffer (C). Reprinted with permission from American Journal of Pathology 2014; 184:200-213 (25)

Unfortunately, transwells do not realistically mimic in vivo geometrical features (e.g. successive bifurcations, vascular morphology) and flow conditions (e.g. converging or diverging flows at bifurcations). These devices only measure unidirectional migration, they do not account for in vivo fluid shear and size/topology, and do not represent the microenvironment of living tissues. Importantly, none of these devices realistically reproduces the in vivo microvascular network to model the inflammatory response (e.g. rolling, adhesion, migration) in the vasculature or have been validated against in vivo data. As there are no models that resolve adhesion and migration in a single assay, the understanding of the adhesion cascade has been hindered. The emergence of microfluidic technologies for development of microphysiological systems and ogranoids on a chip has allowed us to better mimic the microenvironment using 3D cell culture systems and extracellular matrix components such as synthetic nanofiber scaffolds, collagens, or Matrigel. Several microfluidic devices have been developed commercially or in academia for basic science studies and drug screening amongst other applications (www.ncats.nih.gov/tissuechip). However, very few of these devices, if any, realistically reproduce the in vivo microvascular network, are specifically designed to model the inflammatory response (e.g. rolling, adhesion, migration) in the vasculature or have been validated against in vivo data.

3D biomimetic microfluidic assay (bMFA)

To address this need, we developed and extensively validated a novel 3D biomimetic microfluidic assay (bMFA) (Figure 4) that realistically reproduces a microvascular network with accurate geometry from microvascular networks observed in vivo (39). This microfluidic system reproduces the entire leukocyte adhesion cascade in a physiologically realistic three-dimensional environment, encompassing circulation, rolling, adhesion and migration of leukocytes into the extra-vascular tissue space under physiologically-relevant flow conditions in a single system (36, 39, 45, 47, 48). A Geographic Information System approach (49) was used to digitize microvascular networks for subsequent generation of synthetic microvascular networks using soft-lithography processes based on microvascular network morphologies obtained from in vivo animal data (45, 47, 48, 50, 51). Similar to in vivo vascular networks, these devices have channels in the range of 15-100 μm. This novel microfluidics device consists of vascular channels in communication with a tissue compartment filled with chemoattractants (e.g. fMLP) via a porous barrier. Endothelial cells cultured in the vascular channels form continuous endothelial lining with a central lumen along the length of the vascular channels (Figure 4). Fluorescent-tagged neutrophils can circulate in the vascular channels and interact with the endothelial cells under physiologic shear conditions (Figure 5). We developed a Computation Fluid Dynamics (CFD) based model to calculate flow parameters (e.g. shear stress) in different vessels of the network (47). The number of adherent cells at each location is quantified by defining cells that do not move for 30 seconds as adherent (Figure 6). The rolling and adhesion patterns of neutrophils in bMFA are similar to those observed in vivo by intravital microscopy (39, 50, 51). For example, both in vivo and in vitro, neutrophils preferentially adhere to activated endothelial cells near bifurcations with rolling and spatial adhesion patterns in close agreement (39). Neutrophil adhesion is minimal in high shear regions (shear rate >120 1/sec) and maximal in low shear regions, indicating that fluidic shear strongly influences cell adhesion in these microvascular networks (39). Thus, we have an integrated microfluidic assay to the leukocyte adhesion cascade in a realistic microvasculature geometry with physiological shear conditions that allows direct observation and quantification of leukocytes rolling, adhesion and migration over time. It should be noted that these studies have primarily used isolated neutrophils and have not explored the role that other blood cells, such as platelets or monocytes, can play in orchestrating neutrophil-endothelial interactions and transmigration (40, 52).

Figure 4: The bMFA mimics a physiologically relevant microvascular environment.

Microvascular network maps obtained in vivo are reproduced on PDMS to assemble the biomimetic microfluidic assay (scale bar 1 cm) (A). Microvascular endothelial cells, WT (scale bar 50 μm) (B) and KI155 (scale bar 50 μm) (C), formed a complete lumen in the vascular channel of bMFA (green indicates F-actin; red indicates cell nuclei). Reprinted with permission from Shock, 51:538-47, 2019 (31).

Figure 5: Overview of the bMFA:

(A) Schematic of the bMFA. The vascular channels (100 μm width; 100 μm height) are separated from the tissue compartment by a 3 μm pore size and 100 μm wide barrier. (B) Fabricated bMFA with a 3 μm pore size barrier. (C) FITC dye perfused bMFA. (D) Confluent culture of endothelial cells in bMFA shows an elongated shape in the direction of the flow. (E) Endothelial cells in bMFA are viable as indicated by Hoechst staining. (F-G) Magnified images of endothelial cells in phase contrast and fluorescence (Scale bar is 500 μm). Dark shadows in panels B and D, as well as the bright fluorescent spot in panel C, are from the tubes connecting the tissue compartment to the syringe pump. Reprinted with permission from Anal. Chem. 2014, 86, 8344–8351; https://pubs.acs.org/doi/10.1021/ac5018716, further permissions related to the material excerpted should be directed to the ACS (39).

Figure 6: Rolling, adhesion and migration of neutrophils in bMFA.

Migration of neutrophils (labeled with fluorescent dye) into the tissue compartment of bMFA after 120 min of continuous flow. Solid arrows in the top right panels show a rolling neutrophil (panels 1 and 2) which becomes adherent (panel 3); dotted arrows in the top right panels show firmly adherent neutrophils. Bottom right panels show a neutrophil migrating from a vascular channel through the barrier into the tissue compartment over time. Reprinted with permission from Anal. Chem. 2014, 86, 8344–8351; https://pubs.acs.org/doi/10.1021/ac5018716, further permissions related to the material excerpted should be directed to the ACS (39).

To illustrate the types of information that can be obtained using this system, we examined the effect of a PKCδ inhibitor on neutrophil adherence and migration utilizing human neutrophils and endothelial cells. Our in vivo studies in a rodent sepsis model indicated a role for PKCδ in regulating neutrophil migration but did not address specific mechanisms or identify specific steps by which PKCδ affects the interaction of neutrophils and endothelial cells during inflammation (25–27, 30). Using the bMFA, the role of PKCδ in regulating temporal and spatial distribution of adhering/migrating neutrophils through endothelium was quantified (28). Employing human HUVEC and neutrophils, we addressed the question as to whether PKCδ regulates β₂-integrin-dependent and/or β₂-integrin-independent pathways under flow conditions. Our studies (Figure 7) demonstrate that there is a time dependent increase in the number of neutrophils that migrate across TNFα-activated endothelial cells in response to fMLP or IL-8 in the tissue compartment. PKCδ inhibition significantly decreased migration in response to either β₂-integrin-dependent (e.g. fMLP) and β₂-integrin-independent (e.g. IL-8) chemoattractants (28). These results are in contrast to our previous results obtained using a static Transwell system whereby PKCδ inhibition decreased fMLP-mediated migration but not IL-8-mediated neutrophil migration (See Figure 3) (25). These conflicting results underscore the importance of in vitro modeling using a realistic 3D system with relevant microvasculature geometry and physiological shear conditions

Figure 7: Human Neutrophil Migration Across Human Endothelial Cells in BMFA.

PKCδ inhibition reduces migration of neutrophils from the vascular channels, across the activated endothelium, into the tissue compartment. There is a general increase in the number of migrated neutrophils across endothelium during the course of experiments: (A) PKCδ inhibition significantly reduces neutrophil transmigration across activated endothelium in response to fMLP. PKCδ inhibition significantly decreases neutrophil transmigration across non-activated endothelial cells in response to fMLP up to 30 min; (B) PKCδ inhibition significantly reduces neutrophil transmigration across activated endothelium in response to IL-8, but by a lesser degree compared to fMLP. Treatment with PKCδ-TAT peptide inhibitor does not result in a significant change in transmigration of neutrophils across non-activated endothelial cells in response to IL-8. (Mean ± SEM; N = 3; * p<0.05, ** p<0.01, *** p<0.001, Two way ANOVA). Reprinted with permission from Journal of Leukocyte Biology 2016; 100:1027-35 (28).

Using this microfluidics system, neutrophil adhesion under different shear rates can be determined. While the shear rate at the inlet of the microfluidics device is fixed, the shear rate in the bMFA varies (15-500 sec-1) at different locations. Using a Computational Fluid Dynamics based model, flow parameters (e.g. shear stress) in the different vessels of the network can be calculated and a plot of shear rate vs. number of adhered cells at different locations can be generated (47). For example, as shown in Figure 8, mouse bone marrow-derived neutrophils preferentially adhere to TNFα-activated mouse lung endothelial cells in channels with low shear rate and near bifurcations and this adhesion is decreased by treatment with the PKCδ inhibitor (28). These types of studies are not possible using traditional in vitro fluidic models lacking realistic vessel geometry and varying flow rates.

Figure 8: Adhesion of Mouse Bone Marrow Neutrophils (BMN) to Mouse Lung Microvascular Endothelial Cells (MLMVEC).

There is a significant increase in adhesion of WT BMN to TNF-α activated WT MLMVEC in the presence of fMLP, especially at low shear rates and near bifurcations. Pharmacological inhibition with the PKCδ-TAT inhibitor (PKCδ-i) significantly reduced the adhesion level of WT BMN to WT MLMVEC to No Treatment levels. (n=3; mean±SEM; *P<0.05, **P<0.01, ***P<0.001, compared to No Treatment; #P<0.05, ##P<0.01, compared to TNF-α, one-way ANOVA). Reprinted with permission from Shock, 51:538-47, 2019 (31).

Measuring Expression of Endothelial Cell Adhesion Molecules under Shear Flow:

Adhesion molecule expression in response to infection and inflammation can be measured in vivo using animal models (Figure 2) and in vitro using monolayer endothelial cell cultures (25, 26). Adhesion molecule expression can also be quantified in the microfluidic device by using antibody-coated microspheres (34). In this technique, fluorescent polystyrene microspheres of desired size (2-10 μm diameter) are coated with the desired mAb (e.g. anti-ICAM-1). Appropriate IgG antibodies are used as controls. In the bMFA, treatment of HUVEC with TNFα results in a significant upregulation of ICAM-1. Similar to our in vivo observations (Figure 2), PKCδ inhibition significantly decreases the expression of ICAM-1 (28).

In Vitro Measurements of Endothelial Cell integrity

The effect of inflammation or sepsis on endothelial permeability can be determined in bMFA under shear flow conditions by measuring the flux of a 40-kDa Texas Red fluorescent dextran (25 μM) from the vascular channel to the tissue compartment and by monitoring changes in transendothelial electrical resistance (TEER). The permeability (P) of dextran across endothelial cells is calculated using : $P=\frac{1}{I_{v_0}}\frac{V}{S}\frac{{\mathit{dI}}_t}{\mathit{dt}}$ where I_t is the average intensity in the tissue compartment, I_v0 is the maximum fluorescence intensity of the vascular channel, and V/S is the ratio of vascular channel volume to its surface area (36). As shown in Figure 9A employing our novel blood-brain barrier on-a-chip (B³C) microfluidic assay containing human brain microvascular endothelial cells (HBMVEC), there is a 3-fold increase in dextran permeability from vascular channels to the tissue compartment following incubation with TNFα. PKCδ inhibition significantly reduces TNF-mediated increases in endothelial permeability (29). These results are in agreement with in vivo observations in our rat model where 24hr post CLP surgery, there is increased Evans blue dye extravasation in the rat brain that is significantly attenuated by PKCδ inhibition (29).

Figure 9: Permeability of Activated Human Brain Microvascular Endothelial Cells (HBMVEC) in our Blood-Brain Barrier On-A-Chip (B³C).

PKCδ inhibition (PKCδ-i) attenuates TNFα-induced permeability increase (a) and TEER decrease (b) in vitro in B³C after 4 h of TNFα activation. Data are presented as mean ± SEM (n = 3). **p < 0.01, *p < 0.05, compared to control and TNFα + PCKδ-i treatment group by ANOVA with Tukey-Kramer post hoc. Reprinted with permission from Journal of Neuroinflammation 2018, 15:309 (29).

In our microfluidic system, TEER is measured using silver chloride electrodes that are inserted on either side of the endothelial cells in the vascular and tissue compartments (29, 36). Baseline TEER is determined and then again following the addition of TNF-α. Employing our novel B³C microfluidic assay containing human brain microvascular endothelial cells (HBMVEC), TEER is measured under shear flow conditions (29). Similar to studies with HUVEC and PMVEC, tight junctional endothelial integrity is significantly enhanced and TEER values increased by more than 2 fold as compared to TEER under static conditions. Treatment with TNFα produces a decrease in TEER indicating decreased EC barrier (Figure 9B). PKCδ inhibition protected HBMVEC from TNFα–mediated increases in EC permeability (29).

While transport of macromolecules larger than 5 nm (e.g. 40 kDa dextran) occurs through the transcellular pathway, regulated in part by cytoskeleton proteins such as actin, TEER is used as an index of current flow via the paracellular route regulated by junctional proteins and via the transcellular route (29). These different regulatory mechanisms may become more prominent depending on the inflammatory stimulus. Compared to TEER, cell-substrate impedance sensing (ECIS) can differentiate between junctional impedance (tightness of cell-cell contacts) and impedance caused by cell-substrate interactions (distance of basal cell membrane to underlying matrix) as well as the contribution of the cell membrane capacitance and can be used to assess cell proliferation and motility (53, 54).

Summary and Future Directions

Leukocyte recruitment and vascular permeability have a critical role in the inflammatory response and development of tissue injury. We have employed established in vivo animal models and novel in vitro biomimetic microfluidic assays employing human and murine cell culture models to study leukocyte dysregulation and leukocyte-endothelial cell interactions in sepsis. Employing a series of varied techniques, we can address questions pertaining to the underlying pathophysiology involved in neutrophil-endothelial cell interactions and disruption of endothelial protective barriers. A strength of these models is the ability to employ both human and murine cells in our in vitro studies to ascertain whether there are species differences in cell adhesion and migration, as well as expression of adhesion molecules and endothelial cell permeability. While these microfluidic and organoid systems provide powerful tools for better understanding the inflammatory response and screening drugs, they do require a moderate level of training for effective use and do not readily account for the effect of variables such as matrix stiffness on the migration process. Furthermore, maintaining these 3D cell culture systems for periods longer than a few weeks may be challenging.

There is an urgent need for the development of models that better represent the human disease and new approaches to screen potential therapeutics for the treatment of sepsis to efficiently predict their response in humans. Established animal models of sepsis combined with emerging biomimetic microfluidic assays could be employed for drug development. However, a clear understanding of how and the degree to which these assays reproduce the biological signals of interest, as well as drug-cell interactions, is critical to their successful deployment in the field of drug discovery. It is therefore critical to decipher omic changes during sepsis to map known response pathways/networks so that in silico models can be used to determine which components of the inflammatory signaling in human cells is preserved in mouse cells to guide further optimization of in vitro assays. We propose that a combination of in vivo, in vitro, omic and in silico modeling approaches will allow us to determine how and to what degree the therapeutics developed and screened in animal models of sepsis and screened in biomimetic microfluidic systems reproduce the inflammatory signal and relevant therapeutic response in humans.