Modeling leukocyte trafficking at the human blood-nerve barrier in vitro and in vivo geared towards targeted molecular therapies for peripheral neuroinflammation

Mechanisms of leukocyte infiltration into the endoneurium are pathologically relevant
to immune-mediated neuropathies, such as GBS and CIDP. There are currently no models
that allow direct evaluation of leukocyte trafficking in peripheral nerves in vivo.
Developing a near-physiological human in vitro BNB has been essential to understanding
possible mechanisms of leukocyte-endothelial interactions in vivo 40]. Static and flow-dependent leukocyte trafficking models have provided significant
insight into the normal and pathologic mediators of leukocyte subpopulation entry
to tissues in different disease states 41], 42]. Direct comparison of flow-dependent (in which endothelial cells are exposed to shear
stress) and static blood-brain barrier models showed enhanced endothelial cell differentiation
and restrictive barrier characteristics that are further enhanced by culturing endothelial
cells on modified hollow chambers with microscopic pores that mimic capillaries 43], 44]. Furthermore, recent studies have demonstrated the importance of flow and shear forces
on leukocyte trafficking via induction of selectin-dependent rolling (which is absent
in static models), chemokine-dependent integrin activation and arrest, as well as
generation of mechanical signals that directly influence biochemical responses of
endothelial cells to facilitate leukocyte extravasation such as cytoskeletal alterations
during diapedesis 45]. In addition to the importance of flow and shear forces to endothelial cell-leukocyte
interactions, there is phenotypic and functional heterogeneity between endothelial
cells from different species and tissues, as well as between macrovascular and microvascular
endothelial cells from within the same tissue 46]–52].

Ultrastructural examination of epineurial macrovessels in peripheral nerves demonstrates
lack of electron dense intercellular tight junctions that are present in endoneurial
microvessels 28], 29]. Endoneurial microvessels also specifically express alkaline phosphatase 47]. While GLUT-1 and p-glycoprotein expressions are highly conserved in several species
(e.g., human, cow, guinea pig, and rat), MCT-1 expression has been demonstrated by
human endoneurial endothelial cells (and possibly rabbit), with absent expression
in the mouse and rat 32]. Leukocyte trafficking in peripheral nerves predominantly occurs across endoneurial
microvessels in immune-mediated peripheral neuropathies such as GBS and CIDP and their
rodent animal models, as well as animal models of neuropathic pain, rather than epineurial
macrovessels 23], 53]. The preferential trafficking of activated leukocytes across microvascular endothelium
(capillaries and post-capillary venules) rather than macrovascular endothelium (e.g.,
arterioles and venules) during inflammation is evolutionally conserved down to the
hagfish dermis 50]. It is therefore paramount to directly study homeostatic leukocyte-BNB endothelial
cell interactions relevant for peripheral nerve immune surveillance, as well as pathologic
leukocyte trafficking relevant to human peripheral neuroinflammation in vitro using
a human BNB cell line incorporating near normal physiological flow and shear forces,
and not simply extrapolate from other vascular barriers such as the blood-brain barrier
or use endothelial cells from different species or tissues that may differentially
express proinflammatory cytokines, chemokines, and cell adhesion molecules.

Leukocyte trafficking across the BNB can be studied in real time using a flow-dependent
leukocyte trafficking assay that entails a parallel plate flow chamber coupled to
time-lapse video microscopy 40], 54]–58] (Fig. 2). Primary human endoneurial endothelial cells (pHEndECs) that form the restrictive
human BNB have been isolated from the sciatic nerves of decedent patients for culture
and used in leukocyte trafficking assay experiments. There is evidence that pHEndECs
retain essential molecular and biophysical characteristics of a restrictive microvascular
barrier in vitro when cultured in specialized culture medium supplemented with growth
factors for up to eight passages 6], 32]. These cells can be seeded on rat-tail collagen-coated CellBIND® petri dishes that
facilitate their attachment, proliferation, and contact inhibition. Confluent endothelial
cell layers are ready for use in experiments after 5–7 days in culture, forming an
in vitro BNB barrier 40], 58].

Fig. 2. Flow-dependent in vitro blood-nerve barrier model. This illustration depicts a set-up
for the flow-dependent in vitro blood-nerve barrier model system that allows direct
visualization of human leukocyte trafficking at the BNB in real time using time-lapse
video microscopy. Leukocytes are infused at physiologically relevant flow rates and
leukocyte-endothelial interactions captured by sequential digital photomicrographs
that are merged into videos and analyzed under different experimental conditions.
Black arrows depict the direction of leukocyte flow

The in vitro BNB may be treated with physiological concentrations of proinflammatory
cytokines (e.g., tissue necrosis factor-? and interferon-?) up to 48 h prior to experiments
to promote leukocyte trafficking without passively altering transendothelial resistance
or inducing endothelial cell death 40]. Studying basal leukocyte migration without exogenous cytokine treatment of the BNB
may provide insights relevant to the determinants and signaling pathways relevant
to peripheral nerve immune surveillance in normal, healthy individuals. This process
may be compromised in patients with human immunodeficiency virus infection with peripheral
neuropathies. Cytokine-mediated endoneurial endothelial cell activation using concentrations
of tissue necrosis factor-? and interferon-? within the range demonstrated in patients’
sera was associated with de novo or increased expression of chemokines such as CCL2,
CXCL2-3, CXCL8, CXCL9, CXCL10, and CXCL11, as well as selectins and cell adhesion
molecules such as E-selectin, P-selectin, ICAM-1, VCAM-1, and the alternatively spliced
fibronectin variant, connecting segment-1 (a counterligand for leukocyte ?
₄
integrin) 40]. These chemokines have been shown to mediate monocyte/macrophage (CCL2), neutrophil
(CXCL2-3, CXCL8), and CD4+ T-helper 1 lymphocyte (CXCL9, CXCL10, CXCL11) transmigration
59]. The repertoire of cell adhesion molecules expressed supports leukocyte rolling,
adhesion and transmigration in vitro and in vivo based on studies from human brain
and non-neural microvessels.

Peripheral blood mononuclear leukocytes (PBMLs) from untreated patients with immune-mediated
neuropathies such as GBS or CIDP, or patients with chronic neuropathic pain can be
infused across the BNB model at predefined flow rates or shear forces that mimic estimated
capillary hemodynamics in peripheral nerves. In this model system, firmly adherent
PBMLs become noticeable within seconds and start to form adherent clusters with transmigration
after about 5 min, with an increase in the number of attached and transmigrated PBMLs
over time, reaching a peak between 20 and 25 min 58]. In order to elucidate the molecular determinants and role(s) of different signaling
mechanisms on leukocyte trafficking at the BNB at different stages of the cascade,
function neutralizing monoclonal antibodies or small molecular antagonists can be
added to the leukocytes or applied to the BNB prior to performing the assay. The effect
of leukocyte activation and different stages of a disease on leukocyte trafficking
may also be studied using leukocytes obtained from different patients with the same
inflammatory neuropathy characterized for activation markers by flow cytometry and
leukocytes obtained from the same patient at disease onset, peak severity, and during
the recovery phases, respectively.

Based on our review of hundreds of videos generated in the Neuromuscular Immunopathology
Research Laboratory using PBMLs from healthy donors and untreated patients with immune-mediated
neuropathies, our initial observations of leukocyte trafficking at the human BNB in
vitro are as follows: as hypothesized by the multi-step paradigm of leukocyte extravasation
at vascular beds, leukocyte trafficking at the human BNB consists of rolling, attachment
with or without post-adhesion locomotion to sites presumed to express high concentrations
of chemoattractant molecules (haptotaxis), firm adhesion concentrated at intercellular
membranes, and transmigration via the paracellular route. Endothelial cell cytokine
activation is a stronger stimulus for leukocyte trafficking than systemic leukocyte
activation that occurs in GBS. GBS patient-derived mononuclear leukocytes preferentially
adhere to the BNB in vitro (Additional file 1: Video 1), dependent on ?
_M
integrin-ICAM-1 mediated signaling (Additional file 1: Video 2), with CD14
⁺
monocytes being the most prevalent adherent leukocyte subpopulation 40]. Monocytes/macrophages are the most prevalent leukocyte subpopulation observed within
the endoneurium in peripheral nerve biopsies of human GBS and its representative animal
model, experimental autoimmune neuritis (EAN) 60], 61], further validating the utility of the flow-dependent in vitro BNB model to understand
pathogenic leukocyte trafficking in peripheral nerves in vivo.

By using pHEndECs as well as patient-derived PBMLs with estimates of capillary flow
rates and alterations in the endoneurial microenvironment in vivo, this in vitro human
BNB model currently provides the most physiologic means to study human peripheral
nerve inflammation pathogenesis as a tool to guide the discovery and potential efficacy
of specific molecular-based target drugs to treat immune-mediated neuropathies such
as GBS and CIDP. Our previous work demonstrated the differential expression ?
_M
integrin (CD11b) on GBS patient leukocyte subpopulations and its critical role in
leukocyte transmigration via ICAM-1 at the BNB using this model 40], suggesting potential efficacy in GBS that is currently being tested in animal models.
There are no leukocyte trafficking assays published using other human primary or immortalized
endoneurial endothelial cells.Static leukocyte assays using CCR2-expressing human
acute monocytic leukemia cell line and primary or immortalized rat endoneurial endothelial
cells 62], 63] provide some insights relevant to understanding peripheral neuroinflammation but
neither utilize patient-derived leukocytes nor a human BNB model, limiting translational
potential due to likely interspecies differences and the inadequacies of static transmigration
assays.

Limitations of the described flow-dependent human in vitro BNB model include the use
of a parallel plate system rather than a hollow, capillary-like, microtube chamber
that allows more accurate application of physiological shear forces; the absence of
an abluminal compartment to apply chemoattractant molecules and collect transmigrated
leukocytes for downstream analyses; suspension of patient-derived leukocytes in medium
with less viscosity than circulating blood, potential alterations (e.g., downregulation
in chemokine receptors) with cryopreserved patient-derived leukocytes ex vivo, potential
changes in BNB endothelial biology with in vitro culture; and unknown effects of endothelial
cell culture without potentially supportive Schwann cells and pericytes. Despite these
limitations, this model provides an avenue to determine whether endothelial cytokine
activation or leukocyte activation state influences leukocyte rolling velocities and
probability of firm leukocyte arrest, determine which chemokine ligand-receptor pairs
are crucial for leukocyte subpopulation arrest and integrin activation, as well as
subsequent transmigration at the BNB, determine the differential roles of ICAM-1,
VCAM-1, and fibronectin connecting segment-1 in specific leukocyte subpopulation adhesion
as well as the signaling mechanisms involved in leukocyte docking at intercellular
junctions and paracellular migration at the human BNB relevant to normal immune surveillance
and pathogenic neuroinflammation.