Mechanobiology of Leukocyte Adhesion

MECHANOBIOLOGY OF LEUKOCYTE ADHESION

BRYAN LAUCK BENSON

Submitted in partial fulﬁllment of the requirements

For the degree of Doctor of Philosophy

Thesis Advisor: Alex Yee-Chen Huang, MD, PhD

Department of Pathology

CASE WESTERN RESERVE UNIVERSITY

January, 2019 Case Western Reserve University

School of Graduate Studies

We hereby approve the dissertation1 of

Bryan Lauck Benson

candidate for the degree of

Doctor of Philosophy

George Dubyak, PhD Committee Chair

Justin Lathia, PhD

Clive Hamlin, PhD

Umut Gurkan, PhD

Richard Ransohoﬀ, MD

Alex Huang, MD, PhD

August 16th, 2018

1We certify that written approval has been obtained for any proprietary material contained therein. Acknowledgements

This thesis represents the contributions of many people, amongst whom I am just one minor player.

My parents Lauck and Mary Lynn gave me unconditional love and support my entire life, and gave me the resources and freedom to build things at home. This ob- session with building is evident in the microﬂuidic devices. My grandmother LeGrace and uncle Lloyd supported me with perspective and encouragement in navigating academia.

Rachael took me in as a college freshman with no experience, and skillfully bal- anced promoting and nurturing my love of science with instilling rigor and discipline.

She guided me from complete newcomer to the conclusion of an independent research product and continues to be a valued mentor today.

Joaquin introduced me to a style of science using careful reading of the literature leading to wild ideas, and pairing a Zenlike approach to repetition with unbridled intellectual curiosity. This remains my style today, and is what I try to instill in my own students.

The Shipman Society and Nerd Herd breathed vitality into my college experience and pushed me to achieve more. I am forever thankful to Kim Elliott who advocated for me to join that group. Steve Allen, fellow Shipmate and future Co-PI, introduced me to the idea of the MSTP and gave me the example to follow in pursuing this training program.

The CWRU MSTP took a chance with a Psychology undergraduate wanting to become a translational biomedical researcher, and my MSTP class made this whole

iii process fun, mixing ”the couch that lived” silliness with peer role models of excellence.

Cliﬀ Harding has been a steadfastly excellent MSTP director, a needed voice of reason, and strong representative for us to the broader the university community. Kathy,

Crista, and Jane have kept me shipshape and Bristol fashion throughout this voyage.

Richard courageously accepted me into his lab after just a brief, unscheduled meeting. He models the approach I one day hope to have the necessary wisdom and patience to emulate: accepting the extreme complexity of biology, taking each constituent piece in turn to give it full attention, and rejecting generalizations. At times he also has taken on the diﬃcult but necessary role of scientiﬁc spiritual advisor, for which I am very grateful.

Richard surrounded himself and ﬁlled his lab with excellent people. Among these many excellent scientists, I have special thanks for Roo, Anna, LiPing, and Bunny for teaching me bench science from the ground up and remaining friends. Kate, likewise, was a critical friend in need, informal advisor, and did much to keep my head aﬂoat.

Birgit conspired with me to start the microﬂuidics investigation, taught me cell culture technique (about which I remain admittedly somewhat stuck up, superior, and persnickety), and gave me unending enthusiastic encouragement. She has remained a staunch supporter and loyal friend throughout, protecting me from unhelpful negative inﬂuences while also being singularly unafraid to give needed honest constructive feedback.

Saurabh Vyawahare provided an excellent microﬂuidics bootcamp course that ultimately enabled this crazy project. Without this course and his later technical assistance, none of this could have come together.

iv Judy Drazba patiently introduced me to the wonderful worlds of confocal and multiphoton microscopy, two cornerstones of every phase of this project.

Alex took me into the lab when he already had four other graduate students, and even let me keep my ongoing project. He gave me the resources and independence to grow as a researcher in my own right, even when it would have beneﬁted him to guide me in other directions. Above all, he exempliﬁed grace under pressure through multiple stressful situations, a quality I greatly admire and hope to achieve with much practice.

Those other four students: AT, Fred, Dixon, and Deb, were welcoming and kind almost to a fault, taught me a new work ethic, and slowly (feet kicking) dragged me into the world of T lymphocyte biology. I look forward to collaborations with them in the future, or just saying ”I knew them when.” Lifer Huang Lab members Jay,

Saada, and Dave are trusted conﬁdants, helped keep me sane within the sometimes chaotic lab environment, and provided much needed guidance and help on countless occasions. Peter was a much needed comrade during the peak of my hours in lab and pushed me to stay on task, while also getting me onto the ski hill.

Joseph and Steven guided me through the worlds of synthetic and molecular biology, taught me Gibson cloning and CRISPR, are genuinely good people, and are fun and loyal friends outside of lab despite our mutually impossible schedules.

Anna HB kept me going many times when I considered quitting, and beyond being a trusted friend is also an excellent collaborator and scientiﬁc sounding board. The project also owes almost its whole existence to her and her mentor Diana graciously allowing our extended use of their treasured microscope.

v Graham in Boston gave valuable input and camaraderie as part of the ”Beebs” team. Brooke, Suzanne, and Mari brought in a fun collaboration and also oﬀered some much needed eleventh hour emotional support.

Susi exhibited saintlike patience with two rounds of fellowship grant applications with late-breaking changes. Kathy M exuded both professionalism and cool by sub- mitting that late-breaking application from a beach via smartphone. Christy kept me on track within my degree program, with similar patience in asking my neurotic questions.

Alan and Neil took interest in my work, supported my morale, and are true academics who refreshingly enjoy just talking about science and medicine.

Justin, Umut, and Clive gave valuable project, meta-project, and life advice and will hopefully remain collaborators in the future.

George was an unfailing pillar of support, especially when times got tough, and an ideal committee chair. He also on more than one occasion provided last-minute needed reagents after 5 on Friday.

Collaborators and consultants Tracy Handel, Raﬁck Sekaly, Jim McGrath, Pulak

Nath, Lance Munn, and Derek Abbott kept the direction of the project in bounds and provided valuable advice and reagents. Dr. Rae-Grant introduced me to the clinical world of MS treatment and gave valuable pointers on what “you research guys” should be doing.

The MSTP, CTSTP, Neurodegeneration T32 with Wenquan and Xiongwei, and the NINDS, especially Jim Koenig, believed in me and the project enough to give essential support.

vi Libby, my ﬁrst student at a time when I knew very little, contributed substantially to our understanding of CXCL12 expression by BBB endothelial cells. Candi worked tirelessly to open a new front on PIEZO1 biology in sickle erythrocytes, which we hope to revisit when our understanding of PIEZO1-integrin relationships is more complete.

Ben was a great help in the early days of the microﬂuidic adhesion experiment quan- tiﬁcation as we scaled up into more and more devices around a circular manifold.

Hannah brought fresh life into the project, and did tremendous service by finding the one decent PIEZO1 antibody, out of a field of over a dozen, and by generating the PIEZO1 expression lentiviral vectors. Lucy followed me down the rabbit hole of trying (unsuccessfully) to procedurally generate complex biomimetic microfluidic networks using scaling laws, and more practically developed a streamlined approach to generating devices and performed much of the immunofluorescence staining optimization. Luis took over the monotony of creating new PIEZO1 truncation mutants when I had to take a step back for family, and by rights should take most of the credit for identifying the critical interaction domain. Ali pushed me to work at a simply ridiculous pace during the final crunch of dissertation writing and data collection, and accomplished the incredible task of completing a figure using CRISPR knockouts in primary human T cells within a span of about three weeks - something we had been trying for two years. Each of these students taught me a great deal about how to be a good mentor, and are also to be thanked for putting up with my unorthodox mentorship style.

Susan and Mike took many trips out to Cleveland to vitally support the ﬁnal stretch with a new baby. That new baby Arthur was always ready with a smile and

vii gave this all a renewed sense of purpose. Finally, my partner in crime and unfailingly patient wife Anna quite simply gave me the strength (and sometimes the necessary push) to get up every morning and try again.

viii Table of Contents

Acknowledgements iii

List of Figures xiii

Abstract xiv

1 Leukocyte traﬃcking: a clinical problem 1

1.1 Selectins and selectin ligands ...... 4

1.2 Integrins ...... 8

1.3 Chemokines and chemokine receptors ...... 11

1.4 Clinical implications ...... 14

2 The environment of leukocyte adhesion is microﬂuidic 17

2.1 Chemokines and convection ...... 20

2.2 Intravascular signals ...... 25

2.3 Leukocyte intravascular crawling ...... 30

2.3.1 Neutrophils ...... 31

2.3.2 Monocytes ...... 32

2.3.3 Lymphocytes ...... 32

2.3.4 Conclusions ...... 34

3 Leukocyte adhesion requires erythrocyte-driven forces 35

3.1 Abstract ...... 35

3.2 Introduction ...... 36

ix 3.3 Results and Discussion ...... 39

3.3.1 Observation and simulation of in vivo post-capillary venules . 39

3.3.2 Design of the biomimetic microﬂuidic devices ...... 42

3.3.3 Creation of an automated data analysis workﬂow ...... 45

3.3.4 Automatic cell segmentation ...... 48

3.3.5 Eﬃcient adhesion in expansions requires cell-cell forcing . . . . 49

3.3.6 Expansion-induced leukocyte adhesion is dependent on sidewall

contact and ICAM-1 ...... 52

3.3.7 Eﬀects of erythrocytes on increasing adhesion eﬃciency are not

attributable to viscosity ...... 59

3.3.8 In channels of physiological scale, expansion phenomena domi-

nate over shear stress in determining relative adhesion probability 62

3.4 Conclusions ...... 62

3.5 Methods ...... 67

4 PIEZO1 forms a mechanosensitive complex with high aﬃnity LFA-1

on T lymphocytes 76

4.1 Introduction ...... 76

4.2 Results ...... 79

4.2.1 PIEZO1 is expressed and functional on human T lymphocytes 79

4.2.2 PIEZO1 colocalizes with high-aﬃnity integrins on chemokine-

activated crawling leukocytes ...... 82

4.2.3 PIEZO1 interacts with the alpha subunits of leukocyte integrins 84

x 4.2.4 PIEZO1 interacts with ITGAL via a conserved membrane-proximal

region within putative transmembrane helix 2 ...... 85

4.2.5 PIEZO1 is required to coordinate crawling motility ...... 90

4.3 Discussion ...... 92

4.4 Materials and methods ...... 94

5 Conclusions and future directions 101

5.1 Structural relationship between PIEZO1 and ITGAL ...... 102

5.2 Downstream signaling from PIEZO1 ...... 103

5.3 Next steps in vitro ...... 105

5.4 Piezo1 contribution to immunity in vivo ...... 105

xi List of Figures

1.1 The leukocyte adhesion cascade ...... 2

2.1 Laminar Flow in vivo and in vitro ...... 20

2.2 Chemokine structural evolution ...... 21

2.3 Prototypical chemokine structure ...... 23

2.4 CCL5 oligomers bound to GAGs ...... 24

2.5 Conﬂuent TY10 culture in a microﬂuidic device ...... 27

2.6 NFκB p65 nuclear translocation after 90 minutes selective stimulation 28

2.7 ICAM-1 induction after 18 hours selective stimulation ...... 29

3.1 Mimicry of in vivo extensional stress ...... 40

3.2 Layout of devices used in this study ...... 43

3.3 Data handling workﬂow ...... 47

3.4 Automated cell segmentation ...... 49

3.5 Cell-cell forcing is required for adhesion ...... 51

3.6 Spatial statistics of leukocyte adhesion ...... 55

3.7 LFA-1 / ICAM-1 interaction underlies the expansion eﬀect ...... 58

3.8 Leukocyte adhesion in physiological scale channels ...... 61

4.1 T lymphocytes express functional PIEZO1...... 80

4.2 PIEZO1 colocalizes with surface high-aﬃnity LFA-1 on crawling leuko-

cytes ...... 83

4.3 PIEZO1 physically interacts with the alpha chains of leukocyte integrins 86

4.4 PIEZO1 does not interact with MYH9 ...... 87

xii 4.5 PIEZO1 interaction with ITGAL is dependent on a conserved region

of transmembrane helix 2 and adjacent intracellular residues. . . . . 89

4.6 PIEZO1 transmembrane topology ...... 91

4.7 PIEZO1 is required for crawling motility ...... 93

5.1 Model of Piezo1 interplay with Rac1 and Myosin II ...... 104

xiii Mechanobiology of Leukocyte Adhesion

Abstract

BRYAN LAUCK BENSON

Splitting their lives between the chaotic, dynamic environment within the bloodstream and the calm, stable waters of tissues, leukocytes face unique challenges in integrating and responding to mechanical forces across their lifespan. The bloodstream aﬀords leukocytes rapid transit. In order to be useful, leukocytes must exit from the bloodstream and enter lymphoid organs for antigen surveillance or sites of inﬂammation to respond to threats. This process, extravasation via the leukocyte adhesion cascade, consists of at least 8 steps and each involves dynamic interactions with plasma, erythrocytes, platelets, other leukocytes, endothelial cells, and cells of the tissue parenchyma.

To navigate this process, leukocytes make use of an array of signaling, cytoskeletal, motor, and adhesion molecules, and emergent in the literature is the understanding that each of these has different functions in different contexts: intravascular vs. extravascular, in different tissues, and in different leukocyte subsets. Of these, by far the most important are the integrins, which bind to adhesion molecules on endothelia and are critically involved in every step of the cascade. Blockade of integrins with the monoclonal antibodies natalizumab and efalizumab results in almost complete ablation of leukocyte transmigration into tissue and marked clinical benefit in autoimmune diseases, but also carries risks of rare but potentially fatal progressive

xiv multifocal leukencephalopathy.

The focus of the present work is on understanding the responses of leukocytes to physical forces in the adhesion cascade, with emphasis on shear stress, erythrocyte collisions, and leukocyte cytoskeleton-coupled motor protein activity. We first generated biomimetic in vitro models of the post-capillary venule, the anatomical site of leukocyte adhesion. These models afforded us the experimental control needed to demonstrate three insights. First, chemokines and cytokines cannot spread laterally within the convective context of the bloodstream, and therefore inflammation must be propagated by extravascular means. Second, collisions with erythrocytes are the most significant forces acting on leukocytes in post-capillary venules, and are essential for efficient adhesion. Finally, the mechanosensitive ion channel PIEZO1 is necessary for a feedback loop that transduces myosin activity and allows coordinated release of high affinity integrin bonds during leukocyte intravascular crawling.

xv Chapter 1

Leukocyte traﬃcking: a clinical problem

Leukocyte trafficking plays a vital role in mediating innate and adaptive host responses to infection and malignancy, while also contributing to pathological autoimmune and inflammatory disorders. Correspondingly, tremendous research effort has been directed at uncovering the principles and molecular mechanisms underlying leukocyte trafficking. Particularly close attention has been paid to the leukocyte adhesion cascade, outlined in Figure 1.1. This adhesion cascade can be conceptualized as a series of multiple steps. Progression from one step to the next is driven by molecular signals and mechanical forces, and each step is reversible. The leukocyte adhesion cascade begins prior to the capture in Figure 1.1 with margination to the vessel wall, a process promoted by collision interactions with erythrocytes and transitional flow as leukocytes exit from capillary to post-capillary venule (see Chapter

3). Here, I introduce and review the steps of the leukocyte adhesion cascade, with

1 Figure 1.1: The Leukocyte Adhesion Cascade.

Adapted with permission from Klaus Ley, Carlo Laudanna, Myron I. Cybul- sky, and Sussan Nourshargh: Getting to the Site of Inﬂammation: The Leukocyte Adhesion Cascade Updated. Nature Reviews Immunology 7(9) (Sept. 2007), 678–689. doi: 10.1038/nri2156.

special focus on the blood-brain barrier (BBB) endothelium and its application to

our disease of interest, multiple sclerosis (MS).

Fundamentally, all steps of the leukocyte adhesion cascade can be conceptualized

as the coordinated establishment and breaking of strong intermolecular contacts. Dur-

ing rolling, transient catch-bond interactions between selectins and selectin ligands

tether leukocytes to the wall and control their speed. This interaction allows leuko-

cytes to sense immobilized signals, such as chemokines, presented on the endothe-

lial surface, and downstream signaling from selectins promotes chemokine receptor

plasma membrane localization to enhance this eﬀect.2 After rolling, the subsequent steps of leukocyte adhesion are dominated by integrins on the leukocyte surface, which pair with immunoglobulin-like cell adhesion molecules (IgCAM), junctional adhesion

2 molecules, and extracellular matrix proteins to form stable adhesive interactions. Ini- tial arrest of leukocytes on the endothelial wall is mediated by activation of surface integrins, either via an inside-out pathway driven by chemokine receptor signaling3 or an outside-in pathway driven by suﬃcient IgCAM density and mechanical forces.4

After initial arrest, leukocytes deform and spread across the endothelial wall, while simultaneously strengthening their adhesive contacts in response to forces.5–7 Once leukocytes have ﬁrmly arrested, spread and established themselves on the endothelial wall, they must subsequently crawl to sites of potential transmigration. Migration on the essentially two-dimensional surface of endothelial cells while also resisting blood

flow requires coordination of cell polarity, coupling of motor and cytoskeletal proteins with high affinity integrins, and detachment of high affinity integrins on the trailing edge8,9 . This is in contrast to integrin-independent migration of leukocytes in most tissues that lack strong convection, such as lymph nodes.10

The interactions of selectins and integrins with their ligands, as well as chemokines with their receptors, has resulted in a simplified model in which selectins interact with mucin-like adhesion molecules to mediate rolling, and integrins, upon activation by chemokines, bind to immunoglobulin-like adhesion molecules (Ig-CAM) to mediate firm adhesion or sticking. Under this model, cell type and tissue specificity is achieved by combinatorial pairings of selectins with mucin-like molecules, integrins with Ig-CAM, and chemokine expression with cell type specific chemokine receptor expression.11 It is therefore worthwhile to provide a brief introduction of selectins, integrins, and their putative ligands, as well as chemokines and chemokine receptors, with special attention given to those molecules known to be expressed at the

3 blood-brain barrier (BBB) under pathological and homeostatic conditions.

1.1 Selectins and selectin ligands

Lectins are proteins that bind carbohydrates. A subtype of lectins, C-type lectins,

requires calcium to coordinate the carbohydrate-protein interaction. Selectins are

a subset of C-type lectins, and three examples are known: E(ndothelial)-selectin,

L(eukocyte)-selectin, and P(latelet)-selectin. Each selectin is a monomer encoded

by a single gene, with multiple domains in series: an N-terminal lectin domain, an

epidermal growth factor-like (EGF) domain, consensus repeats with similarity to com-

plement receptor, a single transmembrane span, and a C-terminal cytoplasmic tail.

All domains are well-conserved between species.12 Identiﬁcation of real selectin ligands

is not trivial. All selectins recognize a minimal fucosylated, sialylated tetrasaccharide

called sialylLewisX (SLeX ),13 which is added as a post-translational modiﬁcation.

However, the binding aﬃnity to SLeX alone is very low, which is fortunate because

motifs such as SLeX are common on membrane-bound proteins.14 Further modi-

ﬁcations are therefore necessary for physiologically relevant binding and to ensure

selectivity. For example, sulfation of three tyrosine residues on p-selectin glycopro-

tein ligand 1 (PSGL-1) is necessary for eﬃcient binding to P-selectin.15 Unfortunately,

these aﬃnity diﬀerences between physiological vs. non-physiological ligands may not

be reﬂected in in vitro assays of leukocyte rolling or adhesion on puriﬁed substrates,

leading to many false positives.16

Further complicating the picture is the fact that post-translational modiﬁcations,

such as fucosylation, are regulated events. For example, PSGL-1 is always active on

4 neutrophils and monocytes, but is held in an inactive, un-fucosylated state on T-cells.

Only when activated in certain contexts will T-cells express fucosyltransferase VII,

the rate-limiting enzyme for functionalization of PSGL-1.17,18 This level of regulation

increases the chances of missing a relevant selectin ligand if the correct cell pair is

not observed in the correct context. Not only the lectin domain, but also the EGF

domain and complement receptor consensus repeats aﬀect ligand binding, speciﬁcity,

affinity, and kinetics.19,20 These domains also mediate the ‘catch’ and ‘slip’ phenomena, whereby this interaction is first strengthened and then weakened as shear stress increases.21 These extracellular domains are moderately conserved between lectins, reflecting their overlapping ligand affinities.12 In contrast, the transmembrane span- ning and intracellular domains, which influence intracellular trafficking and signaling functions, differ significantly between lectins.12 E-selectin is so named because of

its expression on endothelial cells. Under physiologic conditions, it is expressed at

low or undetectable levels, and is induced in a transcription-dependent manner by

activation with cytokines. This expression allows trapping of leukocytes expressing

E-selectin ligands. Well-documented ligands for E-selectin in humans include PSGL-1

and CD44.22 The seductively-named E-selectin ligand (ESL-1) is called golgi glyco-

protein 1 in humans and its expression on the surface of human leukocytes has not

been convincingly demonstrated.23 PSGL-1 and CD44 binding to E-selectin on en-

24,25 dothelial cells can induce αLβ2 integrin-dependent slow rolling in neutrophils and

26 TH1 activated T-cells. In terms of overall cell distribution of ligands, PSGL-1 is expressed strongly on myeloid lineage cells in active form, moderately on T-cells in inactive form, and weakly on B cells in an inactive form.27 CD44, in addition to being

5 28 expressed on neutrophils and TH1 T-cells, is expressed by all B cells, and all myeloid

cells;29 therefore E-selectin, in principle, could aﬀect rolling of any of these cell types.

Like cells from other endothelia, human brain microvascular endothelial cells express

E-selectin after activation with TNF in vitro.30–32 In pathologic studies, higher expres-

sion of E-selectin has been found in microvascular endothelial cells within MS lesions

vs. healthy tissue33 and in soluble form in the CSF of MS patients.34,35 However, E-

selectin does not appear to be critical in the progression of experimental autoimmune

encephalitis (EAE),36,37 putting the directionality of this association into question. P-

selectin is held in platelet alpha granules and in endothelial cell Weibel-Palade bodies,

where it can be rapidly presented upon activation. Its only known ligand is PSGL-1.

In mice, microvascular endothelial cells must transcribe P-selectin de novo in response

to activating cytokines.38 This occurs in EAE and aﬀects leukocyte rolling.39–42 How-

ever, the eﬀect of P-selectin blockade on EAE clinical course tends to be zero.36,37,42 In

humans, P-selectin transcription is probably unresponsive to cytokine stimulation.43

One study found that autoreactive CD8+ T-cells in MS patients expressed functional

PSGL-1, rolled and stuck more to endothelia.44 However, this eﬀect could also be

mediated by E-selectin. One study showed that P-selectin is expressed on chorodial

and subarachnoid endothelia, and is important in recruitment to meningeal blood

vessels for immune surveillance, but is not expressed in the parenchymal microvascu-

lature, either under physiological conditions or in the setting of MS.45 In summary,

P-selectin in humans probably contributes to physiological immune surveillance, but

not directly to inﬂammation or parenchymal lesions in MS. L-selectin is expressed on

almost all leukocytes and not on endothelium. Its interaction with addressins is crit-

6 ically important in traﬃcking of lymphocytes to high endothelial venules (HEVs).46

On BBB endothelium, potential ligands include mucosal vascular addressin cell ad-

hesion molecule 1 (MAdCAM-1), glycosylation-dependent cell adhesion molecule 1

(GlyCAM-1), CD34, and PSGL-1. MAdCAM-1 is expressed on choroid plexus ep-

ithelium, but not at the BBB.47 Ectopic expression of MAdCAM-1 did not aﬀect

EAE in mice,48 and blockade of α4β7 integrin, a MAdCAM-1 receptor, did not aﬀect

EAE in rhesus macaques.49 Further, blockade of α4β7 does not appear to aﬀect homeostatic immune surveillance of the CNS.50 GlyCAM-1 is not expressed in humans.51

Cultured human brain endothelial cells do express CD34.52 However, only certain

glycoforms of CD34 bind to L-selectin,53 and it is not yet known whether these are

found on BBB endothelium. PSGL-1 is expressed in inactive form by HUVECs and

may be converted to its active form by cytokine stimulation of these cells,54 but this

cannot necessarily be extrapolated to BBB cells. The most well-supported expression

of active PSGL-1 on the endothelium is by myeloid cells and TH1 T-cells to medi-

ate secondary capture of other leukocytes by L-selectin.55 However, PSGL-1 deﬁcient

mice exhibit a similar clinical course of EAE vs. wild type mice.56,57 In summary, the

BBB endothelium tends to express a narrower range of selectins and selectin ligands

than peripheral tissues. Much of the evidence is conﬂicting and unsatisfactory, but

there appears to be a reproducible induction of E-selectin expression on BBB en-

dothelial cells in response to cytokine activation. This can mediate interactions with

and signaling to neutrophils and activated T-cells via PSGL-1 and CD44. Evidence

for other cell types and ligands is scant. Many attempts have been made to perturb

selectin-ligand interactions in EAE with no consistent eﬀect. These molecules may

7 be important for eﬃcient responses to pathogens, but seem to be dispensable in the context of overwhelming EAE inﬂammation.

1.2 Integrins

Integrins are created by heterodimerization of one alpha and one beta chain. There are 18 known alpha chains and 8 known beta chains, which combine in a pseudo- combinatorial manner to create 24 known integrins.58 These molecules have wide- ranging functions for cell interaction with the extracellular matrix, a subset of which includes leukocyte interactions with vascular factors. Of these, leukocyte interactions with Ig-CAMs constitute a further subset. There are eight integrins specific to leukocytes,59 of which three are considered to be especially relevant to leukocyte trafficking and inflammation at the blood brain barrier: αLβ2 integrin (LFA-1),

60 αM β2 (Mac-1, CR3), and α4β1 (VLA-4). Integrins participate not only in sticking upon activation, but also contribute to rolling, locomotion, and diapedesis.1 Integrins bind to IgCAMs, of which intercellular adhesion molecule 1 (ICAM-1) and vascular CAM 1 (VCAM-1) are upregulated on BBB endothelium in inflammation and are the most studied. ICAM-2 is also expressed at the BBB endothelium, but is not upregulated with inflammation.61 ICAM-3 (CD50) is expressed on infiltrating leukocytes, but not endothelium in MS lesions.62 Integrin ligand CAMs associated with inter-endothelial junctions are important in later steps of transmigration, are fancifully named junctional adhesion molecules (JAMs) A, B, and C, and bind to

LFA-1, MAC-1, and VLA-4, respectively, to facilitate diapedesis.1 Other CAMs such as platelet endothelial cell adhesion molecule 1 (PECAM-1, also known as CD31),

8 which acts via homophilic interactions between leukocytes and endothelial cells, and

activated leukocyte CAM (ALCAM) are probably important in neuroinﬂammation,

but are not integrin ligands.60,63,64

αLβ2 integrin, also known as lymphocyte function-associated antigen 1 (LFA-1),

is expressed by all leukocytes and can recognize ICAMs 1-3 and JAM-A.1,59 ICAM-1

is constitutively expressed by human brain endothelial cells, and its expression is in-

creased by activation with cytokines such as TNF.30,65 LFA-1 may exist in multiple

activation states. Low and medium aﬃnity states can support rolling.24–26 LFA-1

can be rapidly activated to high aﬃnity form by presentation of chemokines such as

CXCL12 on the luminal surface,66–68 causing sticking to the endothelium. Beyond sticking, LFA-1 appears to be necessary to support the function of a population of

CX3CR1 positive monocytes that patrols tissues under physiologic conditions.69 At the BBB, it (along with Mac-1) appears to be particularly important for neutrophils.70

Blockade or knockout of αL integrin has unpredictable eﬀects in EAE, sometimes reducing disease burden, and sometimes leading to fulminant disease: LFA-1 on encephalitogenic T cells appears to promote EAE, whereas LFA-1 on other immune cells tends to reduce it.71 This may be due to diﬀerences in integrin dependence between T cell subsets: TH1 cells appear to require VLA-4, whereas TH17 cells require LFA-1 to enter the central nervous system (CNS).72 Similar transcriptional programs between

73 TH17 and Treg subsets may underlie this eﬀect, as fewer T regulatory cells reach the CNS in LFA-1 deﬁcient mice.74

αM β2 integrin, also known as Mac-1, CR3, and CD11b/CD18 has been studied much less extensively with regard to BBB inﬁltration vs. LFA-1 and VLA-4. It

9 shares the β2 chain with LFA-1 and has been ascribed many of the same functions in terms of leukocyte transmigration.1 However, interpretations of Mac-1 knockout are complicated by its many other roles. For example, it is expressed by microglia,75 where it mediates clearance of myelin debris by directly recognizing myelin basic protein.76 Meanwhile, it is also integral in activation of microglia in the context of

BBB breakdown and exposure to blood proteins such as ﬁbrin.77

α4β1 integrin, also known as very late antigen 4 (VLA-4), is expressed primarily by monocytes and T cells, and not on neutrophils. Its main ligand appears to be

VCAM-1, which is upregulated on brain endothelium by inflammatory cytokines.30 It also binds the connecting segment (CS1) epitope on fibronectin.78 VLA-4 can mediate leukocyte rolling via VCAM-1.79 This rolling can be converted rapidly to tethering by increasing VLA-4 affinity for VCAM-1 and/or clustering.80,81 Capture via VLA-4 may also occur almost instantly, without prior rolling, in encephalitogenic T cells.82 In

1992, Yednock and colleagues tested antibodies against a number of molecules mediating leukocyte-endothelial interactions, including selectins and integrins, ﬁnding that only antibodies against α4 or β1 integrin blocked leukocyte adherence to inﬂamed endothelia in mouse brain tissue in vitro.83 Then, presumably because therapy against

59,84 β1 integrin is not feasible given its use in many contexts, they tested infusion of an antibody against α4 integrin and found it prevented EAE. The effectiveness of α4 integrin blockade was replicated the next year.85 This eventually became natalizumab,86 a strong treatment for MS.87 The therapeutic benefit of this treatment appears to derive primarily from its interruption of leukocyte-endothelial interactions, although it may also have effects on T cell proliferation and migration through tissue.86

10 1.3 Chemokines and chemokine receptors

The name ’chemokine’ is a portmanteau of ”chemotactic” and ”cytokine.” As such, chemokines play diverse roles both as chemoattractants and as cytokines controlling growth and diﬀerentiation. Structurally, chemokines are a superfamily of small proteins of typically less than 10 kilodaltons in size which are stabilized by disulﬁde linkages (the details of chemokine structural biology are further explored in chapter 2). In humans, the chemokine superfamily contains 48 members, which bind in overlapping fashion to 23 chemokine receptors.

Chemokines The 48 chemokine ligands in humans can be classiﬁed in diﬀerent ways.

One classification is based on spacing between cysteines close to the N terminus: C-C chemokines have two adjacent cysteines, C-X-C chemokines have a residue in between, and so on. However, probably more useful is a classification based on evolutionary origin and function. Under this schema, there are three types of chemokines: ”cluster” chemokines, ”mini-cluster” chemokines, and ”non-cluster” chemokines.88 Non-cluster chemokines are the most ancient, the most conserved between species, and the least promiscuous, binding to only one or two receptors. These primarily function as homeostatic chemokines, expressed constitutively by certain tissues. Canonical examples are the oldest chemokine, CXCL12, a non-cluster chemokine expressed homeostatically in bone marrow, and the mini cluster chemokines CCL21 and CCL19 expressed homeostatically in high endothelial venules. In contradistinction, cluster chemokines are relatively new, often arising and multiplying in number through tandem gene duplication events after speciation.89 These cluster chemokines often differ significantly

11 in context and detailed function between species, but their expression is generally in-

duced only by inﬂammatory stimuli. These ”inﬂammatory” chemokines, due to their

history, are much more promiscuous. For example, CCL5 can bind and activate CCRs

1, 3, and 5, and conversely, the receptor CXCR2 has 7 chemokine ligands.90 Rapid

evolution of inﬂammatory chemokines makes identiﬁcation of orthologs between even

somewhat closely-related species like mice and humans diﬃcult. This can restrict

applicability of ﬁndings in model organisms to humans and may obscure potentially

interesting contributions of these inﬂammatory chemokines to pathological processes.

Nevertheless, dependence of inﬂammatory processes on expression of inﬂammatory

chemokines is frequently established. It is therefore notable that studies of EAE

and MS consistently point to critical roles for the homeostatic chemokines CXCL12,

CCL19, and CCL21.91

In adult humans and mice, CXCL12 serves as a homeostatic chemokine not only for

the adaptive immune system, but also the nervous system. CXCL12 within the brain

parenchyma guides chemotaxis of neural progenitor cells92–94 and modulates their cell

cycle.95 This appears to play a critical role in neural stem cell migration to repair

sites of CNS injury.96–98 However, CXCL12 expression in brain is also associated with

inﬂammatory insult: CXCL12 expression is upregulated in MS.99,100 This association

appears to be causal, as CXCL12-mediated leukocyte recruitment determines the

severity of EAE.101 Critical to this inﬂammatory context is redistribution of CXCL12

from the basolateral to the apical surface of microvascular endothelial cells of the post-

capillary venule, a process which may be partially controlled by ACKR3.102,103 On the abluminal surface of the endothelium in perivascular spaces, CXCL12 restricts

12 leukocyte entry into the CNS, whereas on the luminal surface of the endothelium,

CXCL12 can signal for arrest and recruitment of peripheral blood leukocytes.66,104

CCR7 ligands CCL19 and CCR21, traditionally associated with the high endothe-

lial venules of secondary lymphoid organs, also appear to contribute to EAE and MS.

Incidentally, CCR7 is also necessary for inﬁltration of T acute lymphoblastic leukemia

into CNS.105 In EAE, CCL19106 and CCL21107 can be induced in inﬂammatory le-

sions, and contribute to leukocyte arrest.107 In MS, CCR7 expression is associated

with perivascular antigen presenting cells of potentially myeloid origin, as well as

inﬁltrating central memory T lymphocytes.108

Chemokine Receptors Of the 23 chemokine receptors, 18 are sometimes described

as Gαi coupled receptors containing a intracellular DRY motif, and 5 are receptors

without canonical G protein coupled function, sometimes described as ’decoy’ recep-

tors. However, this characterization is clearly an oversimpliﬁcation for both receptor

types. In the case of the former, there are additional linkages such as Gαo, and β-

Arrestin, allowing for biased signaling.109,110 These same linkages can allow for signaling of ’decoy’ receptors: for example, the β-arrestin pathway is downstream of both

CXCR4 and ACKR3 (formerly CXCR7), and so CXCL12 binding to ACKR3 can lead to activation of p38 MAPK and ERK.111–113 This is a signiﬁcant growth factor for endothelial cells114 and endothelial progenitors.115–117 Such signaling may actually precede G protein coupled signaling in the evolution of chemokine receptors: our current evolutionary understanding of ACKR3 is that it arose from adrenomedullin receptor (ADMR) and it precedes CXCR4.118 This suggests further roles beyond simple ligand sequestration. However, both its predecessor ADMR and its successor

13 CXCR4 are G-protein coupled receptors and the function of the ancestral ACKR3 is not known. It may have lost G-protein signaling if duplication events (creation of CXCR4) made its signaling function redundant. In humans, ACKR3 still shares most homology (34% identity) with ADMR, 33% identity with angiotensin receptor type I, 29% identity with delta and kappa opioid receptors, and 28% identity with

CXCR4. This may represent more than an evolutionary footnote, as ACKR3 has been reported to be a functional receptor for intermediate opiate peptides in mice,119 signaling via β-arrestin to help regulate glucocorticoid secretion. Aside from roles in chemotaxis, growth, and differentiation, the most well-known and well-studied role for chemokine receptors is in triggering integrin inside-out activation and thus arrest of leukocytes on high endothelial venules or sites of inflammation.120 The details of chemokine presentation in this context are discussed in Chapter 2. While generally accepted, this model may not always be true for all endothelia. Effector T cells, in

4,121 particular, have been shown to be able to arrest independently of Gαi. Never- theless, these same studies still identiﬁed critical roles for Gαi signaling in overall transmigration across the endothelium, implicating chemokine receptor signaling in driving intraluminal crawling and paracellular chemotaxis.

1.4 Clinical implications

Clinically, the leukocyte adhesion cascade is important in a variety of diseases, but special attention is paid in two ﬁelds: autoimmunity and cancer.

To take the latter ﬁrst, an emerging issue in using engineered T lymphocytes to treat cancers is relatively poor traﬃcking to solid tumors.122 In this case, the clinical

14 goal would be to increase trafficking. Engineered T lymphocytes propagated in vitro express sufficient chemokine receptors and integrins on their cell surfaces, yet often fail to efficiently extravasate within neoplastic vasculature. A greater understanding of factors beyond chemokines and integrins is therefore clearly of use in future generations of engineered T lymphocyte therapies.

In autoimmunity, with MS as the leading example, therapies targeting leukocyte trafficking are demonstrably effective. The first of these therapies to reach patients was natalizumab, which was discussed earlier. In psoriasis, efalizumab, an antibody against αL integrin, had demonstrable clinical benefit but was removed from the mar- ket due to concerns over progressive multifocal leukencephalopathy (PML). That this rare but fatal complication occurs in both cases of anti-integrin therapy suggests a common pathogenesis. Currently, it is believed that lack of JC virus control within the CNS is to blame. This, in turn, is probably due to effects of anti-integrin therapy on memory T lymphocyte reactivation and homeostatic surveillance of CNS.123

The importance of αL and α4β1 in CNS immune surveillance are underscored by the lack of PML cases reported with another anti-integrin therapy, vedolizumab. Unlike natalizumab, which reacts with all α4 integrins, vedolizumab targets α4β7integrins speciﬁcally and is FDA approved for treatment of ulcerative colitis and Crohn’s disease.124 Beside the risk of PML, anti-integrin therapy also suﬀers from potential escape phenomena. For example, in MS, it is reported that natalizumab therapy can induce

125 myeloma cell adhesion molecule (MCAM) expression in TH17 cells.

Another strategy targeting lymphocyte traﬃcking is to prevent egress from lymph nodes. This is the approach that yielded ﬁngolimod, a sphingosine-1-phosphate (S1P)

15 receptor modulator. Fingolimod is FDA-approved to treat multiple sclerosis, and interferes with S1P-dependent egress of central memory T lymphocytes, and in partic-

126 ular the TH17 subset, in secondary lymphoid organs. While this different approach is theoretically more T lymphocyte specific, fingolimod is associated with increased risk of various infections, and also predisposes to PML.127 An important question left unresolved is whether increased risk of JC virus reactivation and PML is an obligate component of any therapy that interferes with leukocyte trafficking, and in particular, T central memory homeostatic surveillance. Given that PML is rare, despite overwhelming JC virus seropositivity in the adult population, it seems likely that additional modulating factors and host susceptibilities are involved. This provides sufficient hope that further exploration of the leukocyte adhesion cascade and corresponding development of new therapies may, in the future, allow separation of the therapeutic benefits in autoimmunity from the risks of PML.

16 Chapter 2

The environment of leukocyte

adhesion is microﬂuidic

The principal anatomical site for the leukocyte adhesion cascade is the post-capillary

venule (PCV) both in the case of homeostatic adhesion and transmigration through

high endothelial venules in secondary lymphoid organs, and in the case of activated

endothelia within inﬂamed tissues. In humans, the deﬁnition of PCV is tautological

with vessel size. Nevertheless, it is reasonable to propose a range of 16-50 micrometers,

with a characteristic size of 25 micrometers.128 The consequences of this scale are numerous, profound, and challenge intuitions about ﬂuid behaviors that are formed through our daily experiences with macro-scale ﬂows. A dimensionless quantity, the

Reynolds number, helps us to bridge the gap between our macro-scale understanding and the micro-scale reality of PCV ﬂuid physics. The ratio of inertial to viscous forces, it is deﬁned for circular pipes, such as vessels, as:

QD Re = νA

17 Where Re is the Reynolds number, Q is the volumetric rate of ﬂow, D is the

hydraulic diameter of the vessel, ν is the kinematic viscosity, and A is the cross

sectional area of the vessel.

Using known values for human PCVs:129,130

QD Re = νA Q = 280 pL/s = 2.8 ∗ 10−13 m3/s

D = 24 µm = 2.4 ∗ 10−5 m

A = π ∗ 1.2 ∗ 10−5 m2 = 4.5 ∗ 10−10 m2

η 1.9∗10−3 P a∗s −6 2 ν = ρ = 1060 kg/m3 = 1.8 ∗ 10 m /s

2.8∗10−13m3/s ∗ 2.4∗10−5m −3 Re = 1.8∗10−6 m2/s ∗ 4.5∗10−10 m2 = 8.3 ∗ 10

Most daily experiences with ﬂuids, typically water, involve much larger Reynolds

numbers: for instance, the Reynolds number for a human swimming in water is about

106. Our understanding of the microvasculature is therefore always threatened by the

misapplication of our large-scale intuitions. Chief among these is an over-emphasis

on turbulence. The phenomenon of turbulence occurs typically above a Re of 4000,

or in limited instances down to a Reynolds number of about 2000. Only in extremely

artiﬁcial and contrived scenarios can it be observed within the microﬂuidic regime.131

To prevent these errors, it is helpful to imagine the microvasculature in an embodied context. Were we to scale the PCV to our scale according to the size of leukocytes, this would mean imagining ourselves in a pipe of about 5.5 meters in diameter, with an average ﬂuid velocity of 143 m/s or about 300 MPH. To equal the same Reynolds

numberas in PCVs, then, would require ν = 9.5∗104 m2/s, or a material 5∗1010 more

viscous than blood at the same density. The closest common material is pitch, of the

18 (in)famous pitch drop experiment.132 This situation is borderline inconceivable, but underscores the important point that leukocytes are not able to exert meaningful independent motion within the bloodstream unless they are tethered to the wall. Within the conﬁnes of this embodied example, it is also necessary to keep in mind the scaling of motor strength: just as ants are proportionally much stronger than humans, so too are leukocytes proportionally much stronger than ants. Like Antaeus, leukocytes are powerless aloft within the bloodstream, but tremendously powerful once tethered: forces from the rush of blood in PCVs are signiﬁcant, with shear stress in the tens of pico-newtons and collisions with erythrocytes in the hundreds of pico-newtons,133 but the force generated by a single myosin molecule is 3-4 pico-newtons134 and the forces generated by whole cells are in the tens of thousands of pico-newtons.135

This example explains why leukocytes must be pushed against the wall by collisions with erythrocytes, and must then remain dynamically tethered via selectin interactions in order to meaningfully respond to danger signals, such as chemokines.

Even the smallest separation would mean complete inability of the leukocyte to act on the information, just as a human swimmer within the above 300 mile-per-hour river of pitch would be powerless to reach the shore.

The afferent limb of this signaling loop is affected similarly to the efferent limb.

Due to the dominance of viscosity in this system, the laminar nature of the flow, and the relative rates of convection versus diffusion, soluble signals are not able to meaningfully traverse the vessel lumen (Figure 2.1). Thus, just as the responding leukocytes must be in contact with the vessel wall, so also must the signals be immobilized there against the force of blood flow.

19 Figure 2.1: Laminar Flow in vivo and in vitro. Left panel: Evans Blue dye injected into mesenteric veins of living rabbits, demonstrating that laminar flow conditions exist due to low Reynolds number in vivo. Adapted under CC BY-NC 3.0 US from E. P. W. Helps and D. A. McDonald: Observations on Laminar Flow in Veins. The Journal of Physiology 124(3) (June 28, 1954), 631–639. Right panel: Trypan Blue dye injected into a three inlet microfluidic device fabricated to match physiological scale of post-capillary venules, demonstrating identical laminar flow principles.

2.1 Chemokine structural biology reﬂects dual

convective contexts

Chemokines, a superfamily of 48 genes in humans, are critical in the development,

homeostasis, and host defense of every tissue in the body.89 In diﬀerent contexts,

the same chemokine can serve multiple roles: for example, the chemokine CXCL12 is

incorporated into the developmental processes of the heart and vasculature,137,138 central nervous system,139,140 gonads,141 hematopoietic system,142 and thymus.143 Com-

parative biological studies suggest that most of these functions represent the cooption

of newly-appearing chemokines by evolutionarily older systems. CXCL12, potentially

the evolutionarily oldest chemokine,118,144 while being critical to circulatory and ner-

vous system development in higher order vertebrates, is not present in lampreys or

tunicates (Figure 2.2),145 which sport well-deﬁned circulatory and nervous systems.

20 Figure 2.2: Chemokine receptor evolution, demonstrating the origins of CXCR4 and CXCR7 (and their ligand CXCL12) coeval with the emergence of adaptive immunity in Lamprey, but after the origins of the central nervous system and circulatory system in Amphioxus and Tunicates and well after the origins of hemocytes in invertebrates.

Adapted with permission from Baubak Bajoghli: Evolution and Function of Chemokine Receptors in the Immune System of Lower Vertebrates. European Journal of Immunology 43(7) (), 1686–1692. doi: 10.1002/eji.201343557.

The system that arises as the closest correlate of chemokine emergence is adaptive immunity. While adaptive immunity in lampreys represents convergent evolution146 and is thus not ancestral to RAG-based immunity in humans, nonetheless, CXCR4 is expressed on lymphocytes and is critical to the development of thymic structures in both adaptive systems.143,145

Eﬀective adaptive immune surveillance is predicated on two mechanisms: ﬁrst, the concentration of antigens from larger territories of tissue into nodal points, and second, the transport and recirculation of adaptive immune cells to scan those disparate points for antigens.147 If we (provisionally) believe that the cardinal indispens-

21 able biological function of chemokines is to allow adaptive immunity, then from the very beginning this implies two convective contexts: the stable tissue interstitium that allows diﬀusive gradients and ameoboid motion of leukocytes, and the dynamic vasculature where soluble signals are immediately whisked away and forgotten and leukocytes must be in contact with the vessel wall. These dual functions are evident in essential structural features of chemokines (Figure 2.3).

First are the structural features of chemokines as extracellular signaling molecules.

Chemokines are small proteins, typically less than 10 kDa, facilitating rapid diﬀusion within tissues. The N terminus, which plays an important role in receptor interaction, is stabilized against degradation by (typically) two disulﬁde bridges.

Second, most chemokines readily bind to glycosaminoglycans (GAGs) on endothelial surfaces (Figure 2.4), which immobilizes them against the wall in spite of rapid convective flow. Removal of GAGs, such as heparan sulfate, results in a failure of chemokine-mediated integrin activation and sticking of leukocytes to activated endothelium under flow conditions.149 This binding to GAGs is associated with another property of chemokines: the ability to form multimers.150 Oligomeriza- tion of chemokines increases binding affinity to GAGs via cooperativity.151 Mutant forms of the chemokines CCL2, CCL4 and CCL5 that cannot bind GAGs fail to elicit intraperitoneal recruitment in mice, and mutant forms that cannot oligomer- ize have a similar effect.152 Campanella and colleagues demonstrated that convective washing removes engineered monomeric, but not oligomeric forms of CXCL10, with consequences for transmigration.153 In contrast, monomeric and non-GAG-binding forms retain chemoattractant ability in in vitro assays without fluid flow in which

22 Figure 2.3: Structure of protoypical chemokine. The N terminus is flexible and mediates ligand binding. It is stabilized by two disulfide bridges (yellow): one to the 30s loop between the first two of three antiparallel beta strands, and the last to the 50s loop or final beta strand. After the beta strand section, the chemokine character- istically folds and finishes with an overlapping alpha helix. This helix stabilizes the beta sheet, participates in receptor binding, and promotes binding to GAGs. Used under CC BY 3.0 from Leonard T. Nguyen and Hans J. Vogel: Structural Perspec- tives on Antimicrobial Chemokines. Frontiers in Immunology 3 (2012), 384. doi: 10.3389/fimmu.2012.00384

23 Figure 2.4: Oligomers of the chemokine CCL5 bound to a stereotypical glycosaminoglycan. Used with permission from Xu Wang, Caroline Watson, Joshua S. Sharp, Tracy M. Handel, and James H. Prestegard: Oligomeric Structure of the Chemokine CCL5/RANTES from NMR, MS, and SAXS Data. Structure 19(8) (Aug. 10, 2011), 1138–1148. doi: 10.1016/j.str.2011.06.001

chemokines and leukocytes are placed on opposite sides of a membrane-supported endothelial cell monolayer.152,154

This supports a schema in which the transition between the convective, immobilized, and calm, diﬀusive environments for chemokines occurs at the level of the endothelial basement membrane. In vivo, inter-endothelial junctions experience neg- ligible ﬂow, and the last step of transmigration is driven by soluble gradients of chemokines such as CCL5.4 In fact, counter-intuitively, careful studies have demonstrated that shear potentiates transmigration of adherent leukocytes by increasing time probing inter-endothelial junctions.156,157 Yet further biological support for this

24 schema comes from the ﬁnding that engineered monomeric and dimeric forms of CCL5

lose the ability to cause leukocyte arrest via CCR1, while retaining the ability to guide

CCR5-mediated chemotaxis of leukocytes.158 Arrest of CD4+ T cells was restored in

moving from a dimeric to a tetrameric form of CCL5, while eﬃcient monocyte arrest

required wild-type CCL5, which can form even larger multimers.155

The details of multimerization diﬀer between the C-C and C-X-C classes of chemokines.

C-X-C chemokines, of which the most well-studied is CXCL8 (IL-8) typically form globular dimers, which do not extend to higher-order structures.148 C-C chemokines preferentially form elongated dimers, which can themselves polymerize. GAGs may serve as scaffolds for oligomerization of C-C chemokines (Figure 2.4), which may in turn increase affinity for GAGs.155 CXCL12 appears to be an exception to this paradigm: despite its classification by primary structure as a C-X-C chemokine, it is most homologous with CCL5 and is capable of forming not only C-X-C dimers, but also C-C type dimers and potentially higher-order polymers.159 This picture is yet further complicated by the finding that chemokines can form heterodimers, with changes in signaling function.160 Chemokines may also be presented directly to leuko-

cytes without immobilization in order to drive chemotaxis after initial adhesion in

certain contexts, such as for eﬀector T lymphocytes.121

2.2 Intravascular signals are likely propagated by

extracellular means

This diﬃculty of signaling within the bloodstream brings up the question of how such

signals are coordinated and transduced from inﬂamed parenchyma to endothelial

25 surface. The question gains even greater relevance in consideration that leukocytes, especially neutrophils, exhibit intravascular crawling behavior that is directed towards inﬂammatory foci.161,162 Multiple alternative models exist, which may not be mutually exclusive.

The ﬁrst, and simplest, is that chemokines diﬀuse within inter-endothelial junctions and become immobilized on or near the endothelial surface by binding to GAGs.

For this model to be feasible, the characteristic pore size of the endothelial junctions and basement membrane must be large enough for chemokines to pass eﬃciently. This varies between tissue beds: on the one end, nervous endothelial barriers at homeostasis have pores of less that one nanometer in diameter, whereas liver sinusoids, where the most robust directed crawling has been demonstrated, exhibit pores of almost a quarter of a micron.163 It also varies across physiological state: it is established that pore sizes increase in the context of inﬂammation,164 particularly in brain.165,166

The second is that chemokines may be actively translocated from the abluminal to the luminal surfaces of endothelial cells. This is almost certainly true,167 but makes explaining directed crawling phenomena more diﬃcult: the surface area of endothelial cells is many times larger than that of leukocytes, so the abluminal gradient would have to be faithfully maintained during the transcytosis process.

The third, intriguing proposition is that endothelial cells themselves may propa- gate inﬂammatory signals to one another via gap junctions or direct contact.168–170

To test this hypothesis, we generated three inlet microfluidic devices and exploited principles of laminar flow to selectively expose a subset of cultured brain microvascular endothelial cells to an inflammatory stimulus. We hypothesized that endothelial

26 Figure 2.5: Confluent TY10 culture in a microfluidic device. Phase contrast image demonstrating a confluent monolayer of TY10 brain microvascular endothelial cells cultured under physiological flow conditions in a microfluidic chip.

cells on the opposite, non-exposed side of the channel would exhibit NFκB signaling as a result of cell-cell spreading of inﬂammatory stimulus.

To do this, we first developed a system of culturing brain microvascular endothelial cells within the microfluidic devices under flow conditions (Figure 2.5). Briefly, conditionally-immortalized human brain microvascular endothelial cells (TY10)171 were grown at 33◦C in collagen I coated flasks to 50% confluence, then trypsinized and isolated. Meanwhile, microfluidic devices with main channel height 25 microns and width 220 microns were coated with 1 mg/mL Collagen IV (Sigma C5533). TY10 were then introduced into the device at a concentration of 25 million cells per mL and allowed to settle over two to three hours. During this time, careful control of the cell distribution and media level was maintained to optimize cell density within

27 Figure 2.6: NFκB p65 Nuclear translocation after 90 minutes selective stimulation with 10 ng/mL TNF. Left panel: CMFDA dye indicating which cells received selective stimulation. Center panel: NFκB p65, demonstrating robust nuclear translocation in selectively stimulated cells (right side of channel) but cytoplasmic p65 in unstimulated cells (left side of channel). Right panel: overlay. the device. Once seeding was satisfactory, devices were fed using gravity-driven ﬂow of endothelial cell media (Lonza EGM-2 in MCDB 131) at the TY10 diﬀerentiation

◦ temperature of 37 C in a cell culture incubator with 5% CO2. Under these conditions,

TY10s migrated and spread out to evenly cover the channels and establish junctions with one another. Once cultures were established over at least 48 hours, we then stimulated the cells selectively with 10 nanograms per mL of human tumor necrosis factor (TNF). To verify the laminar ﬂow streams, we also marked the stimulated

ﬂow lane with CMFDA dye. Consistent with the timecourse in other cell lines, this rapidly (within 30 minutes) and robustly induced nuclear translocation of NFκB p65, as assayed by immunoﬂuorescence (Figure 2.6).

Having established that we could selectively and robustly induce NFκB signaling over the short term, we then carried out the same stimulation over a period of

18 hours, and also assessed induction of ICAM-1 expression downstream of NFκB.

During this time, selective stimulation was maintained, as demonstrated by CMFDA

28 Figure 2.7: ICAM-1 induction after 18 hours selective stimulation. A) CMFDA dye indicating the cells that were exposed to TNF. In this case, the middle lane of the three-inlet device became blocked, so the CMFDA evenly bisects the channel left- right. B) NFκB p65 staining, demonstrating weak nuclear translocation, as expected for desynchronized signaling, after 18 hours in the directly stimulated cells on the left side of the channel, but cytoplasmic p65 on unstimulated cells. C) ICAM-1 immunoﬂuorescence, demonstrating robust upregulation on directly stimulated cells only. D) Overlay.

29 staining (Figure 2.7A). NFκB p65 translocation to nucleus was weak on the stim-

ulated side (Figure 2.7B), which is expected after 18 hours of desynchronization.172

In refutation of our hypothesis, p65 remained cytoplasmic on cells that were not di-

rectly stimulated (Figure 2.7B). Furthermore, ICAM-1 upregulation was restricted to

directly stimulated cells (Figure 2.7C).

While not deﬁnitive, this suggests that intravascular signals presented to leuko-

cytes are either a direct transmission of extravascular information by diﬀusion through

inter-endothelial gaps, and/or a reﬂection of extravascular information via chemokine

transcytosis and induction of inﬂammatory gene expression on endothelial cells by

abluminal signals from the parenchyma. Further careful study and comparison of

endothelial cell responses in microﬂuidic devices with vs. without access to an exper-

imentally manipulable convection-free abluminal compartment would likely be deﬁni-

tive (see Conclusions and future directions).

2.3 Leukocyte intravascular crawling

With the aforementioned extreme physical environment and challenges to establishing

signals, it is remarkable that leukocytes, as opposed to immediately transmigrating

into tissue and following soluble signals through the extracellular matrix to their tar-

gets, instead exhibit robust intravascular crawling behavior. This behavior is observed

amongst a variety of diﬀerent cell types, including neutrophils,8 monocytes,69 NKT cells,173 and T lymphocytes.9 Crawling represents a potentially attractive therapeutic avenue for three reasons. First, detailed mechanisms of crawling diﬀer between the leukocyte subsets studied so far: neutrophils, monocytes, and T lymphocytes. This

30 may aﬀord greater selectivity as compared with anti-integrin therapy, particularly

efalizumab which will aﬀect all leukocytes. Second, while crawling is required for

an eﬃcient immune response, it is not an absolute requirement. This may open a

therapeutic window if, for example, an anti-crawling therapy were to allow homeo-

static scanning of brain for JC virus to be weak but suﬃcient, while dampening MS

inﬂammatory activity below a clinically worrisome level.

2.3.1 Neutrophils

Of the various leukocyte subsets, neutrophils have been the most extensively studied

in relation to intravascular crawling. While crawling had long been observed in vitro,

Phillipson and colleagues ﬁrst reported in vivo intravascular crawling of neutrophils,8

and demonstrated that for these cells, while LFA-1 was required for ﬁrm adhesion, it

was not required for crawling. Instead, Mac-1 was required for crawling. Neutrophils

were also the cell type used to establish the potential for directed crawling in liver161

and in cremaster.162

Numerous studies in neutrophils have identiﬁed Ras family GTPases and associ-

ated factors as critical in coordinating intravascular crawling: Vav1, a Rac1 GEF, is

essential in neutrophil crawling,174 which may require LFA-1 ligation.175 The Rac1

GEF P-Rex1 is necessary for eﬃcient crawling and resistance to shear stress176 and

the downstream target, Pak, is also required.177

Meanwhile, RhoA, which is sometimes antagonistic to Rac1, is also critically involved in LFA-1 high aﬃnity state induction.178 It also plays a role in neutrophil

ﬂattening in reaction to shear stress during intravascular crawling.179

Finally, Cdc42 also has a role in neutrophil intravascular crawling in vivo by

31 coordinating their polarity via their PSGL-1 dependent interaction with platelets.180

2.3.2 Monocytes

One of the earliest reports of leukocye intravascular crawling in vivo was the de-

+ 69 scription by Auﬀray et al. of a circulating population of CX3CR1 monocytes.

These cells were found to be dependent on LFA-1 for adhesion to the endothelium and crawled constitutively: in the absence of a specific inflammatory stimulus, they remained essentially resident within the vascular lumen. They were especially con- centrated in the dermis and mesentery and were the first to enter pathological tissues, before even the recruitment of neutrophils. Emphasizing the cell type specificity of crawling mechanisms, Shulman and colleagues demonstrated that intravascular crawling macrophages formed podosomes enriched with phospho-tyrosines and F-actin, a process dependent on Src.181 This was in contrast to T lymphocytes where none of the above were true.

2.3.3 Lymphocytes

In 2009, Bartholomáusand colleagues9 demonstrated intravascular crawling of autoreactive T lymphocytes in vivo. In this scenario, the authors did not detect a distinct directionality vis àvis any focal inflammatory stimulus, but did note a predilection towards crawling against the flow direction, suggesting further cell type specificity of effects. Reinforcing this point, T lymphocytes orient against fluid flow in vitro, an effect enhanced by T cell activation ex vivo, whereas T lymphoblasts (HSB2) and neutrophils do not.182

Also in 2009, Shulman et al. published the results of an extensive in vitro inves-

32 tigation of T lymphocyte crawling motility under ﬂow. This is coined the ”millipede

model” and established many important principles.181 First, high aﬃnity LFA-1 is

required: intermediate aﬃnity LFA-1 cannot mediate crawling. Second, primary T

lymphocytes formed distributed dots of high aﬃnity LFA-1, as opposed to the diﬀuse

focal zone observed in T lymphoblasts.183 Third, these contacts must be turned over

and they colocalize with talin. Fourth, myosin contractility is required for crawling

but does not induce high aﬃnity LFA-1. Fifth, on endothelial monolayers, T lym-

phocytes probe with both adhesive and invasive ﬁlopodia that are enhanced by shear

stress and require Cdc42 signaling, which is required for crawling under ﬂow but not

static conditions. Finally, chemokines were critical in generating high aﬃnity LFA-1

foci.

Extending these ﬁndings, Park used locked high aﬃnity LFA-1 on T lymphoblasts

in vivo and demonstrated enhanced binding to HEVs, but decreased crawling and unaffected interstitial migration, consistent with a model of required turnover of high affinity LFA-1 for crawling, and no significant involvement of integrins in interstitial migration.184

In terms of integrin selectivity, crawling on T lymphocytes appears to be depen-

dent on LFA-1 interaction with ICAMs 1 and 2, and does not require Mac-1 as in

neutrophils. Meanwhile, VLA-4 interactions with VCAM1, while necessary for ﬁrm

adhesion, and probably necessary to be turned over with crawling, do not play a

direct role in crawling motility.185

As in neutrophils, numerous reports suggest a role for GTPases in T lymphocyte

crawling. Rac1 activity is repeatably involved in promoting both adhesion and crawl-

33 ing, with Rac1 overactivity increasing adhesion, crawling, and polarization,186 while deﬁciency leads to less adhesion, less crawling, and less assembly of F-actin.187 RhoA is also involved, coordinating leading and trailing edges during crawling.188 Finally,

Rap1 may be involved in LFA-1 activity, and therefore inﬂuence crawling, whereas it does not aﬀect VLA-4.189

2.3.4 Conclusions

At the present moment, the phenomenon of leukocyte intravascular crawling is explained in unsatisfying teleological terms. To wit, the crawling mechanism is often described as a search for ’hot spots’ of transmigration. What these hot spots might represent, and how their presence is signaled intravascularly to migrating leukocytes within the convective context of flowing blood, are not known. Alternatively, crawling has been described as a process serving an unknown purpose, promoted by chemokines but with random spatial orientation. In this way, the meta-scientific observation is that it represents a Rorschach test, reflective of the investigator’s belief in fate vs. free will, and determinism vs. random chance.

34 Chapter 3

Leukocyte adhesion requires erythrocyte-driven forces

This chapter has been previously published as:

Bryan L. Benson, Lucy Li, Jay T. Myers, R. Dixon Dorand, Umut A. Gurkan, Alex

Y. Huang, and Richard M. Ransohoﬀ: Biomimetic Post-Capillary Venule Expansions for Leukocyte Adhesion Studies. Scientiﬁc Reports 8(1) (June 19, 2018), 9328. doi:

10.1038/s41598-018-27566-z

3.1 Abstract

Leukocyte adhesion and extravasation are maximal near the transition from capillary to post-capillary venule, and are strongly influenced by a confluence of scale- dependent physical effects. Mimicking the scale of physiological vessels using in vitro microfluidic systems allows the capture of these effects on leukocyte adhesion assays, but imposes practical limits on reproducibility and reliable quantification. Here we

35 present a microﬂuidic platform that provides multiple (54–512) technical replicates

within a 15-minute sample collection time, coupled with an automated computer vi-

sion analysis pipeline that captures leukocyte adhesion probabilities as a function of

shear and extensional stresses. We report that in post-capillary channels of physiolog-

ical scale, eﬃcient leukocyte adhesion requires erythrocytes forcing leukocytes against

the wall, a phenomenon that is promoted by the transitional ﬂow in post-capillary

venule expansions and dependent on the adhesion molecule ICAM-1.

3.2 Introduction

The leukocyte adhesion cascade, by which peripheral blood immune cells exit from

ﬂowing blood and migrate into tissues, is highly constrained by biological regulation,

statistics, and biophysics.

First, molecular regulations on the cellular level constrains the eﬃciency of leuko-

cyte adhesion. Defects in integrin expression,190–192 selectin ligand sialylation,193 or integrin affinity upregulation194,195 all lead to severe functional immunodeficiencies, while excessive leukocyte adhesion and accumulation is a contributor to numerous disease states, classically autoimmunity, but also including atherosclerosis,196 acute lung injury,197 and sickle cell disease.198 This regulation occurs with high statistical efficiency. In healthy physiology, 25 billion leukocytes traverse the 100,000 kilometers of vasculature in the body, one circuit per minute, with significant adhesion occurring only in homeostatic sites such as bone marrow, high endothelial venules of secondary lymphoid sites, thymus, spleen, and liver.199–203

The biophysical context in which these processes are regulated is also challenging.

36 More than half of the vasculature length is made up of capillaries - vessels that are

physically smaller than the 8–15 micrometer diameter of leukocytes - where adhesion

would result in plugging of the capillary blood ﬂow and local ischemia. Yet, given the

speed of blood ﬂow, adhesive interactions between leukocytes and endothelium must

occur within fractions of a second204 for any adhesion to be spatially relevant. Despite

these challenges, leukocyte adhesion is largely restricted to post-capillary venules

(PCVs),205,206 vessels of about 20–50 micrometers in diameter that immediately follow

capillaries.

Such eﬃcient regulation on a single-cell basis in a stochastic biological system

requires multiple overlapping factors to achieve. For example, while shear stress

has a significant influence on leukocyte adhesion, it alone is not sufficient to predict

the locations where such adhesion occurs.207 Likewise, while endothelial cell adhe-

sion molecule expression is maximal in PCVs of 25 micrometer diameter,208 it is not

constrained to zero in capillaries. Consequently, multiple levels of regulation may

act in a coordinated fashion to control leukocyte adhesion. In vivo, inﬂammation

triggers multiple coordinated changes, such as increasing adhesion molecule,209,210

and arrest chemokine expression,211 the endothelial barrier becoming more permeable

to inﬂammatory signals from the parenchyma,212 vascular geometry changing from

more capillary-like to more PCV-like,213,214 and erythrocyte aggregation promoting

margination of leukocytes to the vascular wall.215,216

Therefore, to maximize physiological relevance, for the study of leukocyte adhe-

sion, microfluidic models should closely mimic the scale and fluid dynamics of in vivo vessels. Currently, the majority of microfluidic leukocyte adhesion assays use

37 device channels with a height of 100 micrometers or larger.104,217,218 While microfluidic designs featuring in vivo capillary and PCV scale have been used successfully to investigate scale-dependent effects on cell migration, mechanics, and capillary transit,219–222 designs specifically aimed at investigating the process of leukocyte capture from flowing blood are much fewer in number.218,223 The common feature of these smaller channel devices is low numbers of observed adherent cells, precluding their use to quantify the effect of perturbations on leukocyte adhesion. Such low numbers are counter to the predictions of some physical models that postulate higher leukocyte adhesion probability in smaller channels.224 However, other models show that the smaller vessels with decreasing channel height resulted in significant partial occlusion by leukocytes at vessel sizes below 50 micrometers, increasing peak shear stress on the cell surface,225 and shrinking cell-surface contact area.226 Finally, the small perimeter of these near-physiological channels limits the surface area available for leukocyte interaction with the channel surface.

To address these challenges, we took a two-pronged approach: (1) directly visualize leukocyte arrest behavior in vivo to design microﬂuidic devices that recapitulate the in vivo forces on leukocytes, and (2) leverage simple computer vision techniques and repeating device designs to overcome the statistical challenges of low numbers and variable leukocyte adhesion events in microchannels.

In vivo imaging of leukocytes suggested that arrest was most prominent at areas of sudden vessel volume expansion. We hypothesized that mimicking such expansions would increase numbers of adherent cells in our microﬂuidic assay. We expected that these leukocytes would arrest at the channel side walls due to margination induced by

38 such expansions.227 Early experiments suggested that extensional stress at expansions was critical. Therefore, we used computational ﬂuid dynamics (CFD) simulations to create devices that approximate the rheological stresses experienced by leukocytes in

PVC expansions.

Here, we introduce devices and an analysis pipeline that makes possible statistically valid studies of leukocyte adhesion in microﬂuidic channels of physiological scale.

We use these devices to investigate the eﬀects of rheological stresses at the capillary to PCV transition on leukocyte adhesion probability, with the hypothesis that extensional stresses will increase leukocyte adhesion probability. Our results indicate that interactions with other ﬂowing cells within the microchannel, which are governed by rheological stresses, rather than rheological stresses per se, most strongly determine leukocyte adhesion probability.

3.3 Results and Discussion

3.3.1 Observation and simulation of in vivo post-capillary

venules

To directly observe the patterns of leukocyte adhesion in vivo, we implanted cranial

windows into Ubiquitin-GFP mice,228 injected vessel dye, and used two-photon laser

scanning microscopy (2PLSM) to capture the behavior of leukocytes and endothelial

cells in real time. We ﬁrst sought to make a systematic study, collecting 455 x 455 x

55 micrometer ﬁelds of view every 20 seconds (3D + time). However, while we were

successful in capturing overall leukocyte adhesion under diﬀerent conditions, leuko-

39 Figure 3.1: Mimicry of in vivo extensional stress (a) A snapshot of in vivo two-photon microscopy imaging of inflamed pial vessels in a ubiquitin-GFP mouse (Supplemental Movie 1). Red arrows: Flow direction. Yellow arrows: Stably adherent leukocytes appear to be preferentially located after volume expansions, either at capillary to post- capillary venule (PCV) transition (top left arrow) or at expansions within a vessel (other three arrows). Scale bar: 25 micrometers. (b) Computational fluid dynamics simulation of capillary to PCV transition, demonstrating wall-and-rolling relative tensor basis, and extensional stress incurred as the leukocyte enters the PCV. (c) CFD of biomimetic device that varies extensional stress by varying expansion length, while keeping shear stress equivalent. Scale bar in lower right shows 250 micrometers. Numbers indicate hyperbola lengths, with relative length (RL) = hyperbola length / 150 micrometers. A shorter length corresponds to a higher extensional rate. cyte crawling motility was too fast relative to the sizes of vessels for precise adhesion locations to be identified as a function of rheological stresses. We therefore switched to real-time observation of leukocyte-endothelial interactions using resonance scanning and a 2D+time scanning mode. With this method, it became apparent that leukocytes were preferentially adhering in regions of expansion (Figure 3.1A)

To mimic these expansions in vitro, we turned to CFD simulations of a stereotypical capillary-PCV junction, as we were not able to directly measure three-dimensional

40 flow characteristics with sufficient spatiotemporal resolution in vivo. We chose to model a stereotypical capillary to PCV intersection using a 90-degree intersection between an 8 micrometer capillary and a 25 micrometer PCV. In this model, we used as inputs established volume rate of flow vs. diameter data from human conjunctival vessels,129 assuming conservation of mass and adjusting the diameter of the pre-junction PCV to compensate for the capillary side flow. Resultant shear stress and pressures were close matches for reported physiological values. We rotated the simulation outputs to a leukocyte-centric coordinate system at each location: m1, the direction along which a leukocyte would roll, m2, the direction along the wall, and m3, the axis of contact between the leukocyte and the wall. This was done using the following method:

ANSYS Fluent provides velocity and velocity gradients in Cartesian basis (x,y,z):

E3 = {e1, e2, e3} (1)

To describe ﬂow dynamics as relevant to rolling leukocytes, we must describe these

velocity gradients relative to a new basis that is wall-relative:

M3 = {m1, m2, m3} (2)

We start with the wall distance at each cell. The gradient of wall distance provides

one basis vector:

c = ∇d (3)

m3 =c ˆ (4)

Then get the cross product of v and c, a vector tangential to the wall, orthogonal to

both velocity and wall distance:

b = v × c (5)

41 ˆ m2 = b (6)

The wall-relative rolling direction is the cross product of the wall gradient and wall

tangent vectors:

m1 = m2 × m3 (7)

M3 deﬁnes the wall-relative basis that can be used to compute a transformation

matrix Q:    m1 · e1, m1 · e2, m1 · e3,        Q =  m2 · e1, m2 · e2 m2 · e3  (8)       m3 · e1 m3 · e2 m3 · e3 The second order velocity gradient tensor S is then transformed to the new basis via:

[S0] = [Q][S][QT ] (9)

Extensional strain rates are then quantiﬁed as Sii for extensional rates in mi, where

i = 1 : 3

For extensional strain, which we took as a quantification of the expansion effect, the rate in the m2 direction best captured the expansion phenomenon as observed in vivo, with a defined region of extension occurring at the exit from capillary to PCV.

Figure 3.1 shows these extensional rates experienced by leukocytes, as well as the rotation of basis.

3.3.2 Design of the biomimetic microﬂuidic devices

We created three devices to systematically investigate the eﬀect of channel expansions on leukocyte adhesion, progressing from near-physiological to physiological scale (Fig- ure 3.2).

The ﬁrst device (Figure 3.2A) featured a circular manifold with 54 channels, each

42 Figure 3.2: A) Device used in Figure 3.6: 54 straight ROIs around a circular manifold. Narrow inlets are 50 micrometers and expansions are 250 micrometers. Length 4 expansions are used on all. B) Device used in Figure 3.5, 3.7, 3.8A and 3.8B: 240 ROIs in counterbalanced order around a circular manifold. ROIs are Length 1, 2, and 4 expansions with short straight sections, and a length 4 expansion with long straight section. Narrow and expanded channels have the same dimensions as in A. Branching using the planar adaptation of Murray’s law reduces nonspeciﬁc adhesion at branch points C) Device used in Figure 3.8C and 3.8D: 512 ROIs constituting a 16 micrometer narrow section and either 48 or 64 micrometer expanded section, in 256 rows of 2 in counterbalanced order. As in B, a branching pattern is critical in reducing nonspeciﬁc adhesion.

43 containing one region of interest (ROI), taking oﬀ directly from a circular manifold.

Due to issues with clogging of channel inlets under certain conditions, pre- and post- expansion widths were set at a constant 50 and 250 micrometers, respectively. Chan- nel height was set at 25 micrometers. We used hyperbolic expansions terminated at an angle significantly less than 90 degrees in order to avoid the confounding issues of stagnation points or vortices immediately post-expansion. Following each expansion was a long straight section, over which time the effects of the expansion could relax, so that adhesion could be investigated under established flow that represents the status quo.

We made two major modiﬁcations in the second device (Figure 3.2B). First, we added ROIs in which the relative strength of the expansion is varied by varying the length of the expansion, keeping all other parameters constant (Figure 3.1C).

These ROIs were added in series, three to a channel in counterbalanced order, for a total of 240 ROIs per device. The shortest (relative length 1, at 150 micrometers) expansion provided 20 1/s extensional strain rate in whole blood, whereas the longest

(relative length 4, at 600 micrometers) expansion provided 5 1/s. The intermediate

(relative length 2, at 300 micrometers) expansion provided 10 1/s. This allowed us to investigate the eﬀects of expansion independent of shear stress, which is identical at

2.4 dyne/cm2 under typical operating pressure across all designs. The second major change was a branching inlet pattern using the planar adaptation of Murray’s law.229

This modiﬁcation dramatically reduced the occurrence of channel inlet blockade.

In the third device (Figure 3.2C) we investigated whether the reduction in clogging from the branching inlet pattern would allow us to use channels of physiological

44 scale. These devices were 20 micrometers in height, featuring 16 micrometer narrow channels to expansions of 48 and 60 micrometers. These expansions were arrayed two per channel, in counterbalanced order, for a total of 512 ROIs per device.

With the emphasis placed on throughput, the assembled chips consisted of a single layer of PDMS bonded to cover glass, coated with immobilized adhesion molecule

ICAM1, chemokine CXCL12, and glycosaminoglycan (GAG)-bearing proteoglycan

SDC4. These molecules were chosen due to the near-uniform expression of the cognate integrin LFA-1 and chemokine receptor CXCR4 on peripheral blood leukocytes. SDC4 was chosen as the most representative proteoglycan expressed on endothelial cells.

Gravity was used to drive flow, using P1000 filter pipet tips as reservoirs set to a specific fluid height.

3.3.3 Creation of an automated data analysis workﬂow

After sample and device preparation time, each experimental run required only 15 minutes to complete. Therefore, data acquisition throughput met its practical limit only by the constraints of device construction steps such as cleaning and bonding. To generate statistically meaningful data from this volume of raw imaging data required workﬂow automation. Of particular interest to us was the ability to spatially register each leukocyte so that the patterns of cell-speciﬁc leukocyte adhesion probabilities in response to rheological stresses could be observed. A tile-scanning microscope was used to image the entire chip surface at the beginning and end of collection.

Due to out-of-plane fluorescence from cells in the inlet region and blockade of fluorescence in experiments with red blood cells, there were significant inhomogeneities in background fluorescence between tiles (Figure 3.3A). Nevertheless, channel edges

45 were still clearly discernible. Attempts to use the raw ﬂuorescence data to register devices to the template failed. To allow registration, we consulted the computer vision literature and Kovesi’s method of calculating phase congruency,230 which, unlike other edge detection methods, is spatially faithful and contrast-invariant. After initial

filtering, phase congruency images efficiently detected channel edges as well as flowing and adherent leukocytes (Figure 3.3A). This image could further be subdivided into leukocytes versus channel edges by multiplying the phase congruency estimate by a constant and the sine of the phase angle: leukocytes, as bright objects, have phase angles approaching π/2, whereas the channel edges are dark, with phase angles approaching -π/2.

With divided phase congruency images in hand, we could then eﬃciently align the microscopy data to our CAD device designs. This started with a rigid manual alignment, which was easily performed conjointly with visual inspection of the data to ensure that there were not major abnormalities such as chip delamination or large manufacturing defects. During this time, lanes blocked with PDMS debris were marked for exclusion from further analysis. Overall, 240 out 21,030 ROIs (1.14%) were excluded in this manner. Of these, exclusion rate was highest for the ﬁrst device, not designed with Murray’s law in mind, with 94/2646 ROIs excluded (3.55%).

Manual alignment provided an initial estimate but resulted in minor registration errors. Due to the deformability of PDMS, devices frequently exhibited slight stretch from one side to the other. This stretch is not significant enough at the level of a single ROI to affect fluid dynamics but is significant enough across a millimeters- wide device to affect registration. Therefore, we turned to affine registration using

46 Figure 3.3: Data handling workflow. A) Raw data. Top left: tiled microscopy images show adherent leukocytes (green arrow), but also have large inhomogeneities due to differences in background illumination between tiles (yellow line). Some lanes are blocked and must be excluded (red X). Bottom right: data after phase congruency calculation. This isolates important features, removing effects of background inho- mogeneity. B) Registration of phase contrast images to the template. Top left: initial manual rigid registration between the phase congruency map (magenta) and the device template (green). Affine registration using Mattes mutual information results in alignment between the data and the template (bottom right). C) Raw images of cells (green) superimposed with the device template (magenta), demonstrating quality of the resultant alignment. Wall-adherent and rolling leukocytes abut the wall delineated on the template. D) Example of ROI from media (LCIS) condition. E) Example of ROI from RBC condition, demonstrating sidewall-adherent leukocytes, a center adherent leukocyte, flowing leukocytes, and rolling leukocytes.

47 Mattes mutual information, which generally resulted in excellent alignment between the microscopy data and the device template (Figure 3.3B&C). Once aligned to a common template space, collecting ROI-by-ROI data across devices and participants became trivial, applying a simple program to align ROIs to each other using known geometric transformations from the device layout.

3.3.4 Automatic cell segmentation

The volume of ROIs to analyze precluded manual counting from being practical. We therefore developed a method for automatic segmentation, outlined in Figure 3.4.

First, raw images are high-pass ﬁltered to isolate cells, which are high frequency data, from low frequency data such as channel geometries and background illumination. Then phase congruency and phase angles are calculated using two separate wavelengths. Phase congruency at the shorter wavelength identiﬁes cells (Figure

3.4B) but does not robustly identify the full extent of flowing cells, preventing morphological analysis from distinguishing flowing vs. adherent cells. To circumvent this issue, we combined this information with the phase angle (Figure 3.4C) calculated at a longer wavelength. This robustly identifies the extent of flowing cells but is too sensitive to noise to be used as a standalone segmentation method. The resulting reconstruction (Figure 3.4D) using both data robustly identifies cells at a range of intensities and identifies the full contiguous regions of flowing cells. From this starting point, simple morphological filtering then isolates adherent cells (Figure 3.4E&F).

This algorithm performed better than trained human operators. Overall correspon- dence was 85% for the algorithm vs. human operators in spot checks, but subsequent re-examination demonstrated that this was attributable to deterioration of human

48 Figure 3.4: Automated Cell Segmentation. A) Raw data, demonstrating background, ROIs, adherent, and flowing leukocytes. B) Phase congruency of data. Note discontinuities in phase congruency of flowing leukocytes. C) Phase angle of data. This eliminates discontinuities but is too noisy to use as standalone. D) Initial segmentation, using seed points from phase congruency and region growing using phase angle to reconstruct the image. E) Segmentation after morphological filtering to remove flowing and rolling cells, filter artefacts, and platelets. F) Overlay of segmented cells (outlined in green) over the raw data (magenta). counting performance over several continuous hours.

3.3.5 Eﬃcient adhesion in expansions requires cell-cell

forcing

With the data acquisition and analysis pipeline established, we performed leukocyte adhesion assays using freshly-isolated peripheral blood mononuclear cells (PBMC) from healthy human donors, using immobilized recombinant human CXCL12 (rhCXCL12)

49 and recombinant human Fc-ICAM1 fusion protein (rhFc-ICAM1) as substrates. To

maintain physiological pH, cells were resuspended in live cell imaging solution (LCIS),

a HEPES buﬀered, calcium-containing saline, plus bovine serum albumin (BSA) and

a nondiabetic amount of D-glucose (5 mM).

Initial experiments with typical methods of cell preparation used in the ﬁeld, with

isolated PBMC in otherwise cell-free media showed robust adhesion focused in the

initial expansion regions of the ROIs (Figure 3.5A&D, left), but close inspection of

time lapse imaging data revealed that the majority of adhesion was driven not by sin-

gle leukocytes, but by doublets or larger aggregates of cells, which formed nucleation

points for further secondary capture (Fjgure 3.5E). Under these conditions, adhesion

probability did not have a linear response with time, but rather was exponentially in-

creased by adhesion of a prior leukocyte,231,232 a process termed “secondary capture”.

Although clumping adhesion has long been described for cancer metastasis, and we observed occasional leukocyte doublets in circulation by intravital 2PLSM, in general leukocyte adhesion is not driven by aggregates in vivo. Thus, we suspected that the adhesion as observed in our device was most likely an artifact of the cell preparation method. To test this, we passed the prepared cells through a ﬁlter to enforce a single cell suspension. Under these conditions, clumping adhesion and secondary capture were all but eliminated. Interestingly, the apparent biasing towards the expansion was eliminated, but so was overall adhesion (Figure 3.5B&D, center).

From these data, we hypothesized that in such small channels, cell-cell forcing due to collisions may be necessary to increase contact area and overcome the eﬀects of focally high shear stress, allowing ﬁrm adhesion to be established. In vivo, leukocytes

50 Figure 3.5: Efficient leukocyte adhesion requires cell-cell forcing at physiological scale. A-C) Heatmaps showing leukocyte adhesion probability distribution across length 4 ROIs. A) Unfiltered leukocytes in media, demonstrating a bias towards the front of the ROI. B) Filtered leukocytes in media. C) Filtered leukocytes in media reconstituted with erythrocytes at physiological hematocrit, demonstrating redistribution to the sidewalls. D) Overall adhesion probabilities for leukocytes in the three conditions, demonstrating a profound lack of adhesion for filtered leukocytes in media. Each point represents averaged data from one participant, N=5 donors. Boxes indicate mean and standard error. E) Leukocyte adhesion over time with unfiltered leukocytes. Time progresses from left to right in 5-minute increments. Initial adhesion is seeded by small clumps or doublets, which form nuclei for later secondary capture of leukocytes.

51 do not circulate in free plasma but are constantly surrounded in large numbers by

red blood cells (RBCs). Collisions with RBCs within the vasculature could create

the necessary strong forces on leukocytes to allow adhesion. We therefore performed

leukocyte adhesion assays on ﬁltered PBMCs to which RBCs were reintroduced at

50% of the total volume. These RBCs were isolated by centrifugation, resulting in

a packed volume approximately 10% higher than true clinical hematocrit: thus we

expect the 50% volume ratio to represent a hematocrit of approximately 45%, which

is within physiological range, and corresponds to the value we used for our CFD

simulations. Re-introduction of RBCs restored leukocyte adhesion (Figure 3.5D)

while strongly shifting probability density from the center of the channel to the side

walls (Figure 3.5C). On an ROI-by-ROI basis, this also all but eliminated secondary

capture eﬀects and thus decreased variability in mean adhesion between ROIs. The

inclusion of RBCs also led to the induction of ICAM-1 mediated slow rolling, which

is visible on the sidewalls of the device (Supplemental Movie 2), an eﬀect consistent

with predictions of computational studies.233,234

3.3.6 Expansion-induced leukocyte adhesion is dependent

on sidewall contact and ICAM-1

To further characterize the eﬀect of the expansion on leukocyte adhesion, we investigated leukocyte adhesion patterns in ROIs in which we coupled an expansion to a long, straight section over which the expansion eﬀect would be extinguished. In this platform, we observed leukocyte adhesion in response to a factorial combination of adhesion molecule ICAM1, chemokine CXCL12, and heparan sulfate proteogly-

52 can SDC4, which is widely expressed by endothelial cells and can bind and immobilize CXCL12.154,235 In simple coating experiments using AlexaFluor 647 labeled

CXCL12,236 we observed that SDC4 increased the amount of immobilized CXCL12, and changed the distribution from uniform to somewhat punctate.

Leukocyte adhesion probability in all the ICAM-1 containing conditions are demonstrated in Figure 3.6A. As in the shorter channel, sidewall adhesion was dominant.

However, two broad trends were noticeable. First, adhesion was most robust on the sidewalls immediately following the expansion. Second, adhesion in the center of the channel remained steady, or even slightly increasing over the length of the ROIs.

As Figure 3.6B demonstrates, overall adhesion was dependent on ICAM1 F (1, 38) =

60.9, p < 0.001, SDC4 F (1, 38) = 6.45, p = 0.015, and CXCL12 F (1, 38) = 4.23, p =

0.047, as assessed by a linear model using the presence of ICAM1, SDC4, and CXCL12 as independent variables, and donor ID as a covariate of no interest to account for global donor eﬀects. Each data point represents the mean number of adherent cells per device. One device was used per condition per donor. Under the hypothesis that the SDC4 eﬀect would be dependent on its ability to immobilize CXCL12, we created an additional model to test for an interaction between CXCL12 and SDC4 model terms, but surprisingly this was not the case F (1, 37) = 0.86, p = 0.36.

We then sought to further characterize the eﬀects of the coating molecules on adhesion to the sidewall versus the center of the channel. To achieve this, we dilated the image of the channel edge slightly to make a sub-ROI encompassing wall adherent leukocytes. Leukocytes in the remainder were taken to be mid-channel, i.e. top and bottom wall adherent. This approach (Figure 3.6C) more clearly demonstrated the

53 54 Figure 3.6: Spatial statistics of leukocyte adhesion as a function of coatings. A) Leukocyte adhesion distribution heatmap for all ICAM-1 containing conditions in the straight channel device (Figure 3.2A). Flow direction is left to right. These con- sist of a length 4 expansion and an extended straight section to capture relaxation of the expansion effect. Wall associated adhesion peaks immediately after the initial expansion, and again at the collision-inducing point where the expansions end. Chan- nel center adhesion subtly increases over the length of the channel. B) Mean number of adherent cells per ROI in response to factorial combinations of CXCL12, SDC4, and ICAM1 in the coating, demonstrating additive effects. Each point represents mean numbers for a single participant, N=6 donors. Boxes indicate mean and standard error. C) Sub-ROI quantification using a mask for wall-associated vs. center of channel leukocytes as a function of flow distance along the ROI, demonstrating relaxation of the expansion effect for wall-associated but not channel center leukocytes. Wall associated adhesion probability peaks immediately after the expansion, then declines across the ROI until the junction with the constriction induces further collisions between erythrocytes and leukocytes, generating a second peak of adhesion. In contrast, channel center adhesion gradually increases across the length of the channel. D) Validation of the sub-ROI quantification method by quantifying across the cross section of the ROI: Wall associated adhesion is restricted to the sidewalls, whereas channel center adhesion peaks in the channel center. E) Linear model fits across the ROI cross section for the relative contributions of CXCL12, SDC4, and ICAM1 to adhesion. As expected, the contribution of adhesion molecule ICAM1 is most significant in the channel center, where shear stress is highest. F) Linear model fits along the ROI for coating contributions to leukocyte adhesion on the sidewall. The initial spike of adhesion at the expansion is strongly ICAM-1 dependent. G) Linear model fits along the ROI for leukocyte adhesion in the center of the channel. Adhesion immediately after the expansion is dependent on all three coatings, whereas ICAM-1 becomes more dominant as the expansion effect relaxes.

55 trends seen in the heatmap: adhesion probability for wall associated leukocytes peaks sharply after the expansion and then diminishes as ﬂow becomes established. This comes in contrast to channel center adhesion, which exhibits a small transient peak at the expansion, and then overall steadily increases as ﬂow stabilizes. Wall associated adhesion then peaks again at the beginning of the expansion, as wall associated leukocytes are again forced to collide with overriding erythrocytes. Adhesion within constrictions and expansions themselves was minimal.

Looking at the channel cross section served to validate the sub-ROI approach

(Figure 3.6D): wall-associated cell adhesion probability peaked sharply at the sidewalls, with a tail attributable to the sidewalls of the expansion and constriction, while channel center adhesion was zero at the walls.

With these definitions in hand, we then performed spatial parametric modeling to test whether the effects of adhesion molecules were dependent on transitional vs. established flow. As initial validation, we looked at these effects across the center of the channel for leukocytes that were not wall adherent (Figure 3.6E). As expected, leukocyte adhesion under these conditions was strongly ICAM-1 dependent, with a peak at the center of the channel where shear stress is maximal, and the leukocyte is more dependent on firm adhesive contacts to avoid detachment. The effects for

ICAM-1 dominated over those of CXCL12 and SDC4 in this ﬁne-grained analysis but

CXCL12 did reach a small but signiﬁcant positive peak in the center of the channel.

Having obtained these results from the cross section of the channel, we then analyzed patterns down the entire length of the ROI in response to expansion, ﬂow establishment, and constriction.

56 Given that the expansion is an additional factor that influences adhesion but is not captured by our statistical model, we expected that in modeling, the effects of coating molecules would be lowest immediately after the expansion, i.e. that the expansion effect would be captured in a larger error (unexplained variance) term within this region. To our great surprise, this was not the case. As shown in Figure 3.6F, the effect of ICAM-1 coating was greatest immediately after the expansion. Therefore, while the effect is driven by erythrocytes, it is dependent on ICAM-1. For these wall- associated leukocytes, the effects of CXCL12 and SDC4 failed to reach statistical significance at any point.

To further contextualize this unexpected result, we fit the same model for leukocytes in the center of the channel (Figure 3.6G). In contradistinction to wall adherent leukocytes, all three coatings contributed to channel center adhesion immediately after the expansion, with the ICAM1 effect being relatively minimal. This changed over the length of the ROI as flow stabilized and the dependence on ICAM1 steadily increased.

To further test this phenomenon, and to establish the sensitivity of our microﬂuidic models, we tested the eﬀects of various functional treatments on adhesion (Figure

3.7A). For this, we chose to manipulate the integrin LFA-1, for which ICAM-1 is the ligand, by both the activating antibody mAb24 that stabilizes the active state of the integrin, and AF1730, which blocks its binding to ICAM-1. In addition to this, we investigated the eﬀects of one chemokine receptor signaling pathway by blocking

Gαi activation with pertussis toxin, and the eﬀect of glycosaminoglycans on SDC4 by cleaving them with heparinase III. These eﬀects were interrogated using paired t tests

57 Figure 3.7: Leukocyte adhesion as a function of treatments and varying extensional stresses A) Mean number of adherent cells per ROI in response to LFA-1 activating antibody mAb24, LFA-1 blockade antibody AF1730, pertussis toxin, and Heparinase III. Of these, only AF1730 had a significant effect. Each point is the mean value for one participant. N=5 donors. Boxes indicate mean and standard error. Isotype controls and untreated channels are averaged into ‘baseline’ for display clarity purposes – statistics were calculated for antibodies vs. paired isotype. B) Spatial linear model fit for the contribution of AF1730 to adhesion probability for wall adherent leukocytes in straight channels. The strongest effect of AF1730 blockade of LFA-1 is observed immediately after the expansion. for each treatment versus its isotype control (in the case of anti LFA-1 antibodies) or untreated, baseline adhesion (in the case of pertussis toxin and heparinase III).

Since there were no differences between untreated and isotype control adhesion, these control conditions are summarized as ‘baseline’ for simplicity in Figure 3.7A. AF1730 treatment reduced adhesion significantly, t(4) = 4.24, p = 0.013, while mAb24 had no effect, t(4) = 0.65, p = 0.55, consistent with maximal or near-maximal LFA-

1 activation in this model. Pertussis toxin did not show any eﬀect on adhesion, t(4) = 0.41, p = 0.7, which is consistent with other published studies showing that

4,121 leukocyte arrest is not Gαi dependent, especially in vivo. Finally, heparinase III did not decrease adhesion, t(4) = 0.34, p = 0.7, but it also failed to reverse the

SDC4-dependent increase in CXCL12 immobilization under our treatment protocol.

58 In this set of experiments we also assessed whether LFA-1 blockade by AF1730 would mirror the observed ICAM-1 dependence of wall associated adhesion in response to the expansion by fitting a model to AF1730 vs. isotype data from straight channels in the devices. This model used AF1730 treatment in our covariate matrix, along with donor as a covariate of no interest. To our amazement, the effect of AF1730 was almost entirely restricted to the initial expansion and the area immediately after the expansion (Figure 3.7B), indicating that LFA-1-ICAM1 interactions mediate the effect of post-capillary venule expansion mimicry in our model.

3.3.7 Eﬀects of erythrocytes on increasing adhesion

eﬃciency are not attributable to viscosity

By resuspending leukocytes in the presence of a physiological concentration of erythrocytes, we also increase the viscosity of the cell suspension. To prevent any confounding effect of varying viscosity on shear stress, we carefully chose a gravity-driven, pressure- controlled model rather than a syringe-pump driven, volume-controlled model. How- ever, simple kinetic effects of speed could also potentially allow leukocytes more time to establish firm adhesive interactions. To address this concern, we investigated leukocyte adhesion in media with and without erythrocytes, and added to the clear media the polymer polyvinylpyrrolidone of 40 kDa molecular weight (PVP-40). PVP-40 is an established method of increasing media viscosity, and we were able to recreate typical viscosity of leukocyte-erythrocyte suspensions using a final concentration of 10% weight/weight PVP-40. For this form of PVP-40, that constitutes a 2.5 micromolar concentration that is osmotically insignificant.

59 Increasing media viscosity with PVP-40 did not rescue the leukocyte adhesion deﬁcit in clear media, nor did it have any statistically signiﬁcant impact on adhesion,

F (1, 6) = 0.74, p = 0.42, Figure 3.8A.

Version 2 of the device (Figure 3.2B) was used to assess whether higher extensional rates induced by shorter expansions would further increase adhesion. As evident in

Figure 3.8B, this was not the case – length 1 ROIs with the strongest expansion forces had lower adhesion than length 4 ROIs. We believe this to be due to the longer length and therefore greater territory for adhesion in length 4 ROIs. Simple line fitting to account for the extra length is unfortunately not informative, since the variable length occurs in the higher shear expansion and constriction regions, and is therefore impossible to normalize. Of further interest to us in establishing a reliable model for quantification, we also tested signal-noise ratios in the different types, using the observable effects of coating vs. no coating and RBC vs. LCIS. These did not differ repeatably between types, indicating that all ROIs are equally capable for biological experimentation.

We tested these same eﬀects with the same samples in a third device design, using channels at the larger end of true physiological scale and observed the same eﬀect, although overall adhesion in these devices was lower still (Figure 3.8C).

60 Figure 3.8: Leukocyte adhesion as a function of media viscosity in channels of physiological scale A) Mean number of adherent cells per ROI for leukocytes resuspended in media (LCIS), media with viscosity experimentally increased using polyvinylpyrrolidone (PVP-40), and media reconstituted with erythrocytes (RBC). Signiﬁcant adhesion only occurs in the presence of erythrocytes. Data are from the devices displayed in Figure 3.2B. Each point is the mean value for one participant. N=3 donors. Boxes indicate mean and standard error. B) The same data as in A, demonstrating ROI length-adhesion relationship. Length 1 ROIs with the highest extensional stress did not induce a higher adhesion probability than length 2 or length 4 with the lowest adhesion probability. C) Mean number of adherent cells per ROI in the same conditions as in A, with the same donors as in A, but using a device of 16 micrometer channels, expanding to 48 vs. 64 micrometer expansions (Figure 3.2C). D) Leukocyte adhesion probabilities as a function of wall collision and shear stress in devices at physiological scale. Shear stress is highest in the center of the 48 micrometer expansions and lowest at the walls of the 64 micrometer expansions. However, critically, shear stress is higher on the wall of the 48 micrometer expansion than in the center of the channel in the 64 micrometer expansion, demonstrating that induced wall collisions can moderate the eﬀects of higher shear stress in channels of physiological scale.

61 3.3.8 In channels of physiological scale, expansion

phenomena dominate over shear stress in determining

relative adhesion probability

Thanks to progress in our device designs (Figure 3.2), we were able to create a leukocyte adhesion model that features post-capillary expansion mimicry on the upper end of physiological scale (48 and 64 micrometers). These expansions were specifically designed to test the relative effects of shear stress versus expansion. Within these ROIs, the shear stresses are (in descending order): 48 micrometer sidewall >64 micrometer center >48 micrometer center >64 micrometer sidewall. Therefore, if shear stress is the principal determinant of adhesion probability at this scale, adhesion would be predicted to be higher in the center of the 64 micrometer ROIs than on the sidewalls of the 48 micrometer ROIs. However, this was not the case (Figure 3.8D). Despite higher shear stress on the walls of the 48 micrometer ROIs, adhesion here matches that of the center of the 64 micrometer ROIs, demonstrating that the effects of expansion and erythrocyte collisions can counteract higher baseline and adhesion-induced peak shear stresses in these channels at physiological scale.

3.4 Conclusions

Here we have presented a device and analysis method for systematic investigation of leukocyte adhesion in response to rheological forces at the capillary to post-capillary venule transition.

Initial experiments with leukocytes in media alone without erythrocytes sug-

62 gested that extensional stress itself could modulate leukocyte adhesion, independent

of margination. However, this eﬀect was later found to be dependent on leukocyte-

leukocyte doublets or aggregates, as pre-ﬁltered leukocytes did not show such an

eﬀect, and exhibited very low adhesion probability. Notably, we do not believe this

is due to cell depletion, since manual counting demonstrated no detectable change in

cell number with ﬁltration, and inclusion of erythrocytes together with ﬁltered cells

recovered the deﬁcit. We propose that this phenomenon may be due to a jamming

eﬀect as anisotropic particles, such as leukocyte doublets or aggregates, are reoriented

under conditions of extensional ﬂow.237

Inclusion of erythrocytes at physiological hematocrit induced marked margination

and slow rolling of leukocytes on the channel sidewalls, and greatly increased leuko-

cyte adhesion probability while signiﬁcantly shifting adhesion from the center to the

sides of the micro-channel. Additionally, the presence of erythrocytes signiﬁcantly

attenuated the eﬀects of secondary capture of leukocytes.

Integrating across the three cell conditions - raw leukocytes, ﬁltered leukocytes,

and ﬁltered leukocytes with erythrocytes - by far the strongest predictor of leuko-

cyte adhesion was forcing against the wall by cell-cell interactions, either as a result

of aggregate jamming (as in the case of unﬁltered leukocytes) or by intravascular

collisions between erythrocytes and leukocytes. The forces exerted on marginated

leukocytes by erythrocytes are overwhelming, on the order of several hundred pico-

newtons.133 In comparison with these effects of cell-cell interactions, the effects of shear and extensional rates and stresses per se were less significant. This point was further underscored by the inability of increasing viscosity to rescue the adhesion

63 deﬁcit of leukocytes in erythrocyte-free media.

Though extensional ﬂow induced cell-cell forcing at the sidewalls, we did not ob-

serve a strong response of adhesion to increasing extensional stress. In fact, adhesion

was highest in length 4 ROIs with the lowest extensional stress, versus length 1 ROIs

with the highest. This is potentially attributable to the larger physical size of length

4 ROIs, and therefore larger potential capture area. A linear ﬁt of adhesion vs. ROI

length does show length 1 ROIs to give higher-than-expected adhesion. However,

because the variable length lies in the higher-shear and disturbed ﬂow regions of the

expansion and constriction, such an analysis is not informative. Future, more exhaus-

tive studies of geometry versus adhesion could potentially deﬁne whether the eﬀect

exhibits a threshold phenomenon, a response to increasing extensional stress, or both.

Convergent data from spatial parametric modeling of the eﬀects of coating, adhesion-

stimulating molecules CXCL12, SDC4, and ICAM1 and the LFA-1 blocking antibody

AF1730 led to the surprising observation that the erythrocyte-dependent eﬀects of the

expansions were critically dependent on interactions between LFA-1 and ICAM1. This

could potentially be attributable to increased surface contact area from compression,

inherent mechanosensitivity of integrins238 and/or to integrin-linked mechanosensitive pathways. These intriguing observations deserve further study.

Scaling biological assays down to microscale always introduces significant new statistical challenges, and this problem is compounded in the case of leukocyte adhesion with flowing media, where the rheological effects of adherent leukocytes become significant as channel scale approaches the same order of magnitude as leukocyte diameter. To overcome these problems, we greatly increased the number of observations

64 and employed computer vision to allow us to make analyses that were not possible previously (e.g. 512 ROIs per device in the ﬁnal design).

The inclusion of erythrocytes is necessary for adhesion assays at this scale. They increase adhesion probability to detectable level in the smallest channels and essentially eliminate secondary capture. We also observed fewer clogging events within the microchannels in the presence of erythrocytes. Altogether, the inclusion of erythrocytes in the leukocyte adhesion assay increased reliability, physiological validity, and statistical robustness.

There are several limitations and caveats of this work. First, our model uses the traditional photolithography, glass-and-PDMS Whitesides technique for device construction, generating essentially two-dimensional patterns. While this approach is expedient, it does not allow us to recapitulate the full three-dimensional rheological milieu at the capillary to post-capillary venule transition. However, we predict that in a fully three-dimensional system, where erythrocytes flow over newly entrant leukocytes, the effects of cell-cell contacts will be further increased, since streamlines from the joining capillary are extended across the vessel wall. Further technical advances in microfluidic device construction are necessary before such a junction can be reliably constructed at this microscale.

Second, our analysis workflow currently necessitates isolating leukocytes from whole blood in order to stain them with cell tracker dyes. Erythrocytes must then be re-introduced later in sample processing. While this has the potential to reduce experimental variance due to differences in plasma protein expression, and allow cell type specificity of effects (e.g. using isolated CD4+ T lymphocytes) it comes with risks.

65 First, it introduces cells to signiﬁcant stress, which can cause cell death, especially

in neutrophils. Second, ﬂuorescent staining of cells may increase their stiﬀness.239

The speciﬁc staining protocol used should reduce this eﬀect,240 and margination and

rolling were preserved for unstained cells in our system (Supplemental Movie 2). Nev-

ertheless, an increase in stiﬀness would be predicted to promote margination. Last,

while it may reduce experimental variance between donors, depriving cells of plasma

proteins may signiﬁcantly aﬀect their behavior.

Finally, our assay used immobilized adhesion molecules and chemokines rather

than inﬂamed endothelial cells as an adhesion substrate. This was a deliberate de-

cision, made in order to dissect potential leukocyte-intrinsic responses to rheological

stresses. The inclusion of endothelial cells would have confounded results of eﬀects of

shear and extensional stress on leukocytes, versus as moderating factors in inﬂamma-

tory activation of endothelial cells. With these data established, future experiments

could include inﬂamed endothelial cells as the adhesion substrate.

Despite these caveats, we believe that the present study represents a signiﬁcant

ﬁrst step in addressing the practical challenges inherent in scaling leukocyte adhe-

sion assays down to physiological scale. We also believe that our results call for a

reconceptualization of the eﬀects of rheological stresses on leukocyte adhesion at in vivo scale. Whereas shear stress is the dominant factor determining leukocyte adhesion probability in larger channels, the eﬀects of cell-cell interactions play a more important role in smaller channels.

66 3.5 Methods

Special ethics statement As detailed below, the research involving human participants and animal subjects adhered to all institutional, local, state, federal, and international laws and guidelines concerning responsible and ethical conduct of research. Research involving mice was approved by the Case Western Reserve Uni- versity Animal Resource Center (IACUC Approval Number: 2015-0118), which is accredited by the Association for Assessment and Accreditation of Laboratory Ani- mal Care (AAALAC). Research activities strictly followed the procedures and protocols outlined in that approval. Research involving human samples was approved by the University Hospitals Cleveland Medical Center Institutional Review Board

(UHCMC IRB Approval Number: 08-15-17), which is accredited by the Association for the Accreditation of Human Research Protection Programs (AAHRPP). Our research activities strictly followed the procedures and protocols therein. Most of the blood samples were obtained de-identiﬁed from the Case Western Reserve Univer- sity Hematopoietic Biorepository & Cellular Therapy Shared Resource (HBCTSR), which also acts in accordance with its own UHCMC IRB approved protocols. In- formed consent was obtained by the HBCTSR for those participants. Other samples were obtained under our IRB-approved protocol. Informed consent was obtained by the experimenter, recorded, and stored separately in a secured location.

In vivo imaging C57BL/6-Tg(UBC-GFP)30Scha/J (Stock #004353) mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Animals were housed, bred, and handled in the Case Western Reserve University Animal Resource Center

67 (IACUC) in accordance with approved Institutional Animal Care and Use Committee experimental protocols (IACUC Approval Number: 2015-0118) and maintained on a

Teklad 2018 S alfalfa free diet (Harlan Lab, USA) starting two weeks before imaging to reduce autofluorescence. Animals were given open cranial window preparations as previously described.241,242 Animals were imaged within five days of cranial window implantation to preserve inflamed vessel architecture. Mice were anesthetized with nebulized isoflurane (3% induction, 2% maintenance) in 1:1 O2:air and placed in a stereotactic holder. Mice were then placed in a custom environmental chamber and maintained at a constant animal temperature of 37◦C via both heating pads and other environmental controls. Temperature, anesthetic depth, and breathing rate

(approximately 40–80 breaths per minute) were monitored to ensure animal health and comfort. A Leica SP5 confocal microscope equipped with a 20x water immer- sion lens (Leica HCX-APO-L, N.A. 1.0) and a tunable 16 W Ti/Sapphire IR laser tuned to 860 nm (Chameleon Coherent, Inc.) was used for intravital 2PSLM imaging.

XYT images with an XY dimension of 456x456 micrometers were obtained at 71ms intervals. The data were analyzed using Imaris (BitPlane, Inc.) to generate images

Device design and modeling 2D masks for photolithography and 3D models of devices for computational ﬂuid dynamics (CFD) simulations were created using Open-

SCAD (www.openscad.org). 3D model geometries were imported into ICEM CFD

(ANSYS Inc., Canonsburg, PA), blocked, and meshed. Meshes were exported for

Fluent (ANSYS) and fluid flow was modeled using incompressible flow, with an apparent viscosity130 of 1.85 cP, as a compromise value between 1.83 for vessels of 8 micrometer diameter, and 1.92 for vessels of 25 micrometer diameter. We used a

68 pressure-based solver (PISO scheme, second order pressure, QUICK momentum). A realizable K-Epsilon model was incorporated after solution convergence in order to load the wall distance parameter into memory for our Fluent user-defined function in C. This did not result in any significant changes to model outcome, as consistent with the laminar nature of microfluidic systems.

Photolithography 300 silicon test wafers (University Wafer, Catalog #447, Boston,

MA) were rinsed extensively with acetone and isopropyl alcohol, then placed on a hotplate at 300◦C for a minimum of two hours to drive oﬀ water. Wafers were then coated with SU8 2025 photoresist (MicroChem Corp., Westborough, MA) and spun on a WS-400-6NPP spin coater (Laurell Technologies Corporation, North Wales,

PA) using empirical protocols. After soft bake, photoresist was exposed on a Karl

Suss MJB3 mask aligner (Suss MicroTec, Garching bei München, Deutschland) using transparency masks (printed by CAD/Art Services, Bandon, OR) mounted on a quartz slide (Chemglass Life Sciences, Vineland, NJ). Wafers were then post-exposure baked according to MicroChem protocol and developed using SU8 Developer (Mi- croChem). The thickness of resultant channel masks was confirmed at multiple locations using a NewView 7300 optical profilometer (Zygo Corp., Middlefield, CT), and re-verified after device assembly with fluorescence imaging. To facilitate later release of molded devices, wafers were treated with 1% v/v Trichloro(1H,1H,2H,2H- perfluorooctyl)silane (Sigma-Aldrich Corp., St. Louis, MO) in hexanes for 5-30 minutes, dependent on ambient humidity, at room temperature and washed extensively with isopropyl alcohol before overnight hard bake at 80◦C in a hybridization oven.

69 PDMS layer Polydimethylsiloxane (PDMS) Sylgard 184 (Dow Corning, Auburn,

MI) was mixed at a 9:1 ratio of base to curing agent, centrifuged to remove bubbles, inverted extensively to mix base and curing agent, then centrifuged again before pouring over molds. PDMS on molds was then degassed for 4 hours in a vacuum desiccator before placing in a room temperature hybridization oven. The oven was then ramped to 80◦Cand the PDMS baked overnight. After baking, the oven was turned oﬀ and allowed to return to room temperature before the PDMS and mold were removed. Devices were manually cut out using a craft knife. The center inlet was punched with a 1500 micrometer PDMS Port Creator (Cor Solutions, Ithaca,

NY). The outer edge was cut under a dissecting microscope using a scalpel. Devices were pre-cleaned using Magic greener tape. (3M, Catalog # 812, St. Paul, MN), and sonicated in 70% ethanol in a Bransonic (Danbury, CT) CPX1800H sonicator at

70% power for 15 minutes. PDMS layers were then placed within a biosafety cabinet, removed from the pouch, allowed to dry, and stored with tape on the channel side until bonding.

Glass 40 mm round cover glasses (Warner Instruments, Catalog # CS-40R15, Ham- den, CT) were sonicated at 100% power for 60 minutes in water. Sonicated cover glasses were treated with 1 molar potassium hydroxide for 1 hour, rinsed extensively in tap water, then de-ionized water, and ﬁnally washed in 70% ethanol before being allowed to dry in a biosafety cabinet and transferred to a sealed pouch until bonding.

Bonding Bonding was achieved by treating PDMS and glass in a Plasma Etch (Car- son City, NV) PE-100-RIE at 30 W x 30 seconds @ 20 sccm O2 and 200 mtorr. PDMS and glass halves of the devices were then placed together in light conformal contact.

70 Devices were then placed in a sterilization pouch and autoclaved for 40 minutes at

121◦C. The autoclave was left closed overnight, such that the devices remained at

80◦C for an additional 6+ hours to ensure maximal biocompatibility.243

Device coating Recombinant human CXCL12 and recombinant human Fc-ICAM1 chimeric fusion protein were obtained from R&D Systems (Minneapolis, MN). SDC4

, expressed in mammalian cells so as to have glycosaminoglycans, was obtained from

Adipogen Corporation (San Diego, CA). These molecules adsorbed onto the device walls at concentrations of 20, 80, and 50 micrograms per milliliter, for CXCL12,

ICAM1, and SDC4, respectively, in phosphate buﬀered saline (PBS) for a minimum of four hours in a 37◦C incubator. After this incubation, device channels were blocked with 5% BSA in saline for a minimum of 30 minutes.

Inhibitor Treatment Pertussis toxin (Sigma Aldrich) was used at a ﬁnal concentration of 100 nanograms per milliliter. Blocking antibodies AF1730 (R&D Systems) and mAb24 (Biolegend) were used at 20 micrograms per milliliter, as were mouse

IgG1 isotype (GeneTex) and goat IgG isotype (R&D Systems). Heparinase III (R&D

Systems) was used at 40 micrograms per milliliter. All inhibitors were added just before the experiment start, except for Heparinase III which was added to the initial coating solution.

Blood collection and handling Peripheral blood from healthy human donors was obtained fresh in heparinized tubes. Most blood was obtained de-identiﬁed from the

HBCTSR. Some blood samples were obtained directly, using identical procedures and reagents, after acquiring informed consent, under our University Hospitals Cleveland

Medical Center Institutional Review Board (IRB) approved protocol sample and data

71 handling. Blood was centrifuged at 200 g for 10 minutes at 20◦C to sediment cells.

Buﬀy coats were carefully aspirated from the top of the packed layer, mixed 1:1 with phosphate buﬀered saline, and then layered over FicollPaque PLUS 1.077 (GE). Cells over density media were centrifuged at 400 g for 40 minutes at 20◦C without brakes.

The interface was isolated, diluted with PBS, and centrifuged again at 200 g for 10 minutes at 20◦C. During this time, cells were counted. After this centrifugation, cells were resuspended in 500 microliters of RPMI 1640 with 10% fetal bovine serum (FBS) and stained with CFDA-SE according to optimized protocols.240 After the washing steps in this protocol, cells were ﬁnally resuspended at 4 million per mL in live cell imaging solution (LCIS, Thermo Fisher Scientiﬁc) supplemented with 5% BSA and

50 mM D-glucose. In indicated cases, cells were passed through a 35 micrometer cell strainer. Then, cells were mixed 1:1 with either LCIS, LCIS with 20% w/w PVP-

40 (Sigma Aldrich Corporation) for a ﬁnal concentration of 10% w/w (2.5 mM), or packed red blood cells drawn from the bottom of the centrifuged collection tube.

Such mixing occurred immediately before experiments began to prevent erythrocyte aggregation. Cells were maintained at 37◦C during this ﬁnal preparation.

/textitIn vitro Microscopy Images were acquired on a Leica DMI 6000B with a heated environmental chamber stabilized at 37◦C (Okolab USA, Burlingame, CA).

The objective used was Leica 5X NA 0.15 HCX PL S APO. Tiles were collected using

L5 ﬁlter cube with 20% overlap between tiles, as leukocytes were still ﬂowing.

Image analysis All images were stitched using MIST,244 which is freely available as part of FIJI. Stitched images were then subjected to a custom-made MATLAB

(MathWorks, Natick, MA) pipeline: for fast registration, images were downsampled

72 to 18% of original size, then bandpass ﬁltered using an order 8, 20 micrometer cutoﬀ

lowpass and an order 1, 250 micrometer cutoﬀ highpass Butterworth ﬁlter. Phase con-

gruency was then calculated using the Kovesi algorithm,230 for which the MATLAB function is freely available. Input image contrast ranges were separately optimized for LCIS vs. RBC. Phase congruency images were then multiplied by the sine of the phase angle, or negative sine of the phase angle, depending on context (LCIS vs. RBC) to isolate channel edges. Resultant cleaned phase congruency masks were provided an initial manual alignment using control points, a step incorporated with data cleaning. Data cleaning took the form of marking and automatically excluding blocked lanes, or lanes with significant manufacturing debris. Devices with more than ten percent excluded lanes were excluded entirely. Further refined alignment of the data to the templates was accomplished using affine transform with Mattes mutual information. ROIs, with a 15 micrometer buffer zone, were extracted and rotated to common space, where cells were segmented automatically.

Automatic cell segmentation Raw images were transformed to template space, then highpass filtered using an order 4, 25 micrometer cutoff Butterworth filter to isolate cell information. Phase congruency and phase angle of the filtered image were calculated using Kovesi’s algorithm, using a minimum wavelength of 4 micrometers for phase congruency and 8 micrometers for phase angle. Seed regions were identified by thresholding the phase congruency image at phase congruency greater than zero.

Potential regions for expansion were identiﬁed by median ﬁltering the phase angle image using a 5 micrometer kernel, then automatically thresholding using Otsu’s method. Morphological reconstruction using these seeds and regions was performed

73 using imreconstruct in MATLAB. This method robustly identified both flowing and adherent leukocytes, a small number of contaminating platelets, and rings of high pass filter artifact. This was further filtered to adherent leukocytes by enforcing an

Euler number of 1 to remove rings, a major axis length between 5 and 25 micrometers to exclude ﬂowing and rolling leukocytes, and an area above 30 square micrometers to exclude platelets. Centroids of the resulting segmented areas provided cell position and cell number on a cell-by-cell and ROI-by-ROI basis. Cell positions were visualized by kernel density estimation with an 8 micrometer Gaussian kernel.

Statistics Analysis of cell numbers was performed in R245 in R Studio.246 Initial attempts at quantification used an ROI-by-ROI basis, with negative binomial hurdle models via the pscl package247 and diagnostic rootograms via the countreg package248 to handle the over-dispersed and zero-inflated count data. While this approach was viable, it was ultimately more robust to use mean number of adherent cells per ROI per device. These means were log-transformed to normality to satisfy linear model assumptions. Data were fitted using linear models as functions of experimental conditions and donor identification number as a covariate of no interest to account for donor effects. These models were analyzed with analysis of variance (ANOVA). In

Figure 3.7, paired samples t-tests for antibodies mAb24 and AF1730 were performed separately versus their isotype controls mouse IgG1 and goat IgG, respectively. Re- sults of statistical tests are reported in the text with the type of test (e.g. F or t), followed by degrees of freedom in parentheses, the test statistic, and the p value.

One device was used per condition. For spatial parametric mapping in Figures 3.6 and 3.7, mean number of adherent cells were binned into 100 micrometer wide bins

74 for analysis along straight ROI lengths, and 20 micrometer wide bins for analysis across straight ROI widths. These means were log transformed to normality, with an addition of one to account for zero data. These were then input into identical models as for overall means: linear models using donor identiﬁcation number as a covariate of no interest. Beta terms, representing the least squares estimate of the contribution of the factors of interest to adhesion, were then reported. Additional 2D spatial parametric maps were generated in the same conditions, using 25 micrometer

Gaussian-kernel smoothing of the data. These resulted in the same conclusions, but were not eﬃcient to present visually. Plots were generated using package ggplot2.249

Reports were generated using knitr.250

Data Availability Statement The datasets generated and analyzed in this study are available from the corresponding author upon request.

75 Chapter 4

PIEZO1 forms a mechanosensitive complex with high aﬃnity LFA-1 on T lymphocytes

This chapter was contributed to by: Bryan L Benson, Joseph K Rathkey, Luis Correa,

Alicia V Aguilar, Lucy Li, Jay T Myers, Jeﬀrey A Tomalka, Hannah R Kennelley,

Derek W Abbott, Richard M Ransohoﬀ, and Alex Y Huang

4.1 Introduction

T lymphocytes, long distinguished as keystones of adaptive immunity, are also sin- gular in the number and diversity of mechanical environments they encounter. In secondary lymphoid organs such as lymph nodes, T lymphocytes exist in a relatively static, matrix-deﬁned tissue similar to what is experienced by most of the body’s cells. Having scanned the lymph node for antigen, the T lymphocyte will then exit

76 into the unique mechanical environment of the lymphatic circulation, surrounded by

a relatively acellular and matrixless medium and subjected to pulsatile ﬂuid stress.251

From this point, the cell can enter another lymph node, or be carried eventually to the

venous circulation.252 Once inside the bloodstream, T lymphocytes are surrounded by erythrocytes which subject them to compressive forces,133 and also experience yet

another array of ﬂuid dynamic environments: from turbulent and pulsatile ﬂow in

the aorta, through laminar but high shear stress ﬂow in capillaries, to expansion

into post-capillary venules (PCVs) into a lower shear stress laminar regime. Once

in PCVs, ﬂuid forces, erythrocyte compression, adhesion molecule expression, and

chemokines conspire to drive re-entry into parenchymal sites of immune surveillance.

Across the lifespan of a na¨ıve T lymphocyte, roughly ten years in humans,253 this

biomechanical context switching may occur thousands of times. For central or eﬀec-

tor T lymphocytes, though their lifespan is shorter, these switches can include the

additional contexts of virtually any tissue in the body254 with their idiosyncratic ar-

chitectures and elasticities, and for central nervous system (CNS) scanning memory

T lymphocytes255 the route is further complicated by the pulsatile nature of ﬂow in

PCVs within the CNS.256 Despite this range of mechanical contexts experienced by

T lymphocytes, relatively few published works have addressed the question of how

T lymphocytes navigate these disparate environments. Having recently discovered

that compressive forces induced by transitional ﬂow in PCVs promote adhesion in

an integrin-dependent manner,128 we sought next to determine the molecular basis

for this eﬀect, and potentially to begin elucidating how T lymphocytes sense force

context. Our investigations immediately focused on the candidate mechanosensitive

77 ion channel PIEZO1 (formerly FAM38A) on the weight of the following observations:

First, T lymphocytes were identiﬁed in multiple publicly-available RNA-seq and mi-

croarray datasets as among the highest expressing cells of PIEZO1 transcripts. Sec-

ond, we hypothesized that the T lymphocyte mechanosensor should aﬀect integrins,

and PIEZO1 was identiﬁed as an integrin activator in a high-throughput screening

assay for genes that rescue H-Ras mediated integrin affinity suppression.257 This work also demonstrated that the effects were dependent on the calcium-sensitive protease calpain 2 and talin. Third, it is established that PIEZO1 mediates endothelial cell responses to blood flow shear stress.258 These responses are necessary for normal development of blood vasculature, one of the contexts for T lymphocytes. This work further reinforced that PIEZO1 can modulate integrin affinity by inducing calpain

2-mediated cleavage of talin. Fourth, genetic studies have suggested that PIEZO1 de-

ficiency in humans leads to defects in architecture of the lymphatic vasculature,259,260 another of the critical biomechanical contexts for T lymphocytes. Fifth, efficient integrin-dependent T lymphocyte crawling motility was shown to require extracellular calcium influx and modulation of integrins via calpain 2. This was hypothesized by the authors to be due to a mechanosensitive channel, though the identity of the channel was not identified in that work.261 Sixth, though limited to the study of CHO cells in static environments, Hung et al. demonstrated that PIEZO1 potentiates

Myosin II activity, both by allowing calcium inﬂux, and by releasing Rac1 mediated suppression.222 Given that Myosin II activity is critical in leukocyte crawling motility, we hypothesized that PIEZO1 may play a role in this context.262,263

By thus focusing on PIEZO1 as the candidate mechanosensor on T lymphocytes,

78 we conﬁrmed that PIEZO1 is expressed and highly functional on T lymphocytes, that its function depends on extracellular calcium, and that it coordinates with the alpha chains of T lymphocyte integrin alpha/beta heterodimers. Focused work on the association of PIEZO1 with ITGAL, the alpha subunit of LFA-1, demonstrated that PIEZO1 preferentially associates with activated alpha integrin subunits through a conserved amphipathic 8 amino acid motif that spans the border between the cytoplasm and inner leaﬂet of the plasma membrane. These residues are near the

N terminus, a region not yet accurately resolved in published structural studies of

PIEZO1, but which presumably constitutes the extreme outer end of the PIEZO1 lever arm. Finally, genetic ablation of PIEZO1 expression using CRISPR/Cas9 on primary human T lymphocytes demonstrated that normal crawling motility requires

PIEZO1.

4.2 Results

4.2.1 PIEZO1 is expressed and functional on human T

lymphocytes

To follow up the suggestion of the publicly-available mRNA expression databases, we

ﬁrst sought to establish whether PIEZO1 is truly expressed on T lymphocytes. To this end, we employed a three-pronged approach to verify PIEZO1 expression at the transcriptional, protein, and functional levels.

First, we tested whether freshly isolated T lymphocytes from healthy donors would respond to the PIEZO1 activator Yoda1264 with a calcium response, as measured by

79 Figure 4.1: T lymphocytes express functional PIEZO1. A) T lymphocytes were stained with Fluo-4 and resuspended in HEPES buffered physiological saline containing 1 mM calcium or EGTA, then stimulated with Yoda1 or DMSO vehicle at indicated concentrations. Mean fluorescence intensity versus time is reported. B) Schematic outlining the experimental procedures used to validate antibodies and Yoda1/Indo-1 functional verification approach C) Calcium response to Yoda1 and green fluorescence of CEM cells transduced with PIEZO1-1591-mNeonGreen lentiviral construct and then electroporated with negative control crRNP. D) Yoda-1 induced calcium response and green fluorescence of PIEZO1-1591-mNeonGreen CEM cells electroporated with a crRNP targeted to exon 4 of PIEZO1. E) Western blot demonstrating elimination of the >250 kDa PIEZO1 band in electroporated CEM cells sorted for silent Yoda1 calcium response, but not in control crRNP electroporated cells.

80 Fluo-4 signal via microscopy. We exposed the cells to two diﬀerent concentrations

of Yoda1, in the presence or absence of extracellular calcium, and observed a dose-

dependent calcium signal that is dependent on extracellular calcium (Figure 4.1A). Ki-

netically, this calcium increase was slower than for erythrocytes (Supplemental Movies

1 & 2). Dependence of the calcium increase on extracellular calcium lay in contrast to

previous suggestions of endoplasmic reticulum localization of PIEZO1.257 To establish

the sensitivity and speciﬁcity of this functional assay, we transduced CEM leukemia T

cells with a PIEZO1 mNeonGreen fusion protein construct using lentiviral particles.

This fusion protein was adapted from the functionally validated PIEZO1-1591-eGFP

developed by Cox and colleagues.265 Subsequently, we knocked out both this exogenous gene, as well as the endogenous PIEZO1 loci using electroporated CRISPR/Cas9 with guides against exon 4 and exon 6 of PIEZO1 (Figure 4.1B for schematic). Using the ratiometric calcium-sensitive dye Indo-1 and saturating concentrations of Yoda1 in calcium-containing media, we assessed the resulting cells on ﬂow cytometry for calcium signal and green ﬂuorescence. CEM cells electroporated with a control guide were primarily green and calcium sensitive (Figure 4.1C). After electroporation with

Cas9 and guides against PIEZO1, three populations emerged (Figure 4.1D): calcium- sensitive green cells unaﬀected by the Cas9 (top right quadrant), calcium sensitive non-green cells, in which the exogenous PIEZO1 fusion protein had been removed but not endogenous PIEZO1 (top left quadrant), and calcium-insensitive non-green cells in which both the endogenous and exogenous PIEZO1 had been genetically ablated

(bottom left quadrant). We sorted and harvested these cells, then assayed PIEZO1 expression via western blot and demonstrated loss of PIEZO1 protein (Figure 4.1E).

81 Using this system, we also tested all commercially available anti-human PIEZO1 antibodies (11 in total) for both immunofluorescence and western blot, and established that only Abcam 82336 is sensitive to PIEZO1 via immunofluorescence, and only Ab- cam 82336 and Abbexa 321198 are sensitive to PIEZO1 via western blot. Notably, the widely-used Proteintech anti-PIEZO1 antibody generated only a nonspecific band at about 300 kDa in our hands and was insensitive to PIEZO1 via immunofluorescence.

4.2.2 PIEZO1 colocalizes with high-aﬃnity integrins on

chemokine-activated crawling leukocytes

Having conﬁrmed that T lymphocytes express functional PIEZO1, and that one antibody was accurate via immunoﬂuorescence, we next assayed the subcellular localization of PIEZO1 under various conditions. In resting primary human T lymphocytes,

PIEZO1 exists in vesicles that lie close to the membrane (Figure 4.2A). This pattern did not co-localize with KDEL, further supporting the interpretation that PIEZO1 is not located on the endoplasmic reticulum in T lymphocytes (data not shown).

Plating of the primary T lymphocytes on immobilized ICAM-1 without chemokine stimulation did not result in redistribution of PIEZO1 (Figure 4.2B), however, exposure to immobilized ICAM-1 and CXCL12 together resulted in polarization of the T lymphocytes and redistribution of PIEZO1 from vesicular structures to the plasma membrane, particularly in a region suggestive of the high-aﬃnity LFA-1 focal zone,266 but with additional puncta detectable in the leading edge and uropod (Figure 4.2C and inset).

To follow up this ﬁnding, we then quantiﬁed colocalization of PIEZO1, assayed

82 Figure 4.2: Chemokines induce translocation of PIEZO1 from intracellular vesicles to surface high-affinity LFA-1. A) Confocal immunofluorescence image of unstimulated primary human T lymphocytes. Magenta: anti-PIEZO1. Cyan: phalloidin. Single plane through cell center – no contact patch to image. B) Image of primary human T lymphocytes on an ICAM-1 coated surface. Single plane through cell center. C) Image of primary human T lymphocytes on ICAM-1 and CXCL12 coated surface. Single plane across the cell-glass contact patch. Inset: PIEZO1 staining on topmost cell, shown alone for clarity, indicating additional puncta in the leading edge and uropod. D) Nyquist resolution confocal image of PIEZO1-1591-mNeonGreen in CEM cells on CXCL12 + ICAM-1 coated glass. E) ITGB2 staining using MAB1730 of the same cells as in D. F) Correlation of mNeonGreen and MAB1730 signal, showing weak but significant correlation. G) Summary statistics of comparative colocaliztion for HI- 111, which binds LFA-1 in an inactive conformation and does not colocalize, MAB 1730, which marks all LFA-1 and weakly colocalizes, and MEM 148, which binds only high affinity LFA-1 and colocalizes most strongly.

83 using the PIEZO1-mNeonGreen fusion construct in CEM cells to prevent issues of antibody cross-reactivity, versus three conformation-speciﬁc antibodies to LFA-1: HI-

111, which is specific for the closed, inactive form of LFA-1; MEM 148, which is specific for the open, active form of LFA-1; and MAB 1730, which is not conformation specific. We plated CEM cells on immobilized ICAM-1 and CXCL12 and performed confocal microscopy at Nyquist resolution (Figure 4.2D). MAB1730, which stains all conformation of LFA-1, demonstrated robust plasma membrane staining as expected

(Figure 4.2E) and colocalized weakly but signiﬁcantly with PIEZO1 (Figure 4.2F).

Meanwhile, HI-111 did not colocalize signiﬁcantly with PIEZO1, whereas MEM148 colocalized more strongly (Figure 4.2G). Taken together, these data suggest that

PIEZO1 speciﬁcally colocalizes with high-aﬃnity integrins on chemokine-activated lymphocytes.

4.2.3 PIEZO1 interacts with the alpha subunits of

leukocyte integrins

Given this subcellular colocalization, we next asked whether PIEZO1 physically interacts with LFA-1. Given the scarcity of good PIEZO1 antibodies, we used the

PIEZO1-1591-eGFP fusion protein265 as bait for interactors, precipitated using GFP

Trap nanobody sepharose beads, and compared versus PIEZO1-IRES-eGFP as control (Figure 4.3A for schematic). When heterologously expressed in HEK 293T cells, this fusion construct co-precipitated with integrin alpha L (ITGAL, Figure 4.3B), as well as integrins alpha M (ITGAM) and alpha 4 (ITGA4) (Figure 4.3C) whereas free

GFP did not. Untagged PIEZO1 co-precipitated with ITGAL, and the reciprocal IP

84 demonstrated that ITGAL can pull down PIEZO1 as well (data not shown).

LFA-1 exists as a heterodimer, and HEK293T cells express suﬃcient beta integrins endogenously that the interaction could be mediated by either subunit in this system.

A series of combinatorial experiments using overexpression of ITGAL and ITGB2 was non-informative. To address which might be the primary partner, we performed a different experiment, expressing each protein in different cell types and then pooling lysates prior to immunoprecipitation. In this system, PIEZO1 precipitated ITGAL above baseline levels only when ITGAL was alone in the lysate without ITGB2, and ITGB2 was not precipitated at all (Figure 4.3D). Given the increase in physical distance between ITGAL and ITGB2 subunits with activation, this, taken together with the colocalization data, suggests that PIEZO1 interacts specifically with the alpha subunits of activated leukocyte integrins.

4.2.4 PIEZO1 interacts with ITGAL via a conserved

membrane-proximal region within putative

transmembrane helix 2

Having established the association of PIEZO1 and ITGAL, we next sought to identify the region of PIEZO1 responsible for the interaction. We hypothesized that PIEZO1 would likely interact with ITGAL using a site in the distal end of its lever arm. At the time of these experiments, the published structure of PIEZO1 had suggested that the N terminus lay close to the central pore, and that the transmembrane domains radiated outward in a shoelace-like fashion before connecting back to the C terminal pore-determining domains267 This structure identiﬁed a region between 1351 and

85 Figure 4.3: PIEZO1 physically interacts with the alpha chains of leukocyte integrins A) Schematic depicting the experimental approach. B) PIEZO1-1591-eGFP associates with ITGAL. C) PIEZO1-1591-eGFP associates with ITGAM and ITGA4. D) Comparison of co-expression versus combined lysate approaches for immunoprecipitation of ITGAL and ITGB2 using PIEZO1-1591-eGFP. Left lane: co-expression of all three proteins results in all three co-immunoprecipitating. Rightmost lanes: expressing the proteins in separate cells, then combining lysates combinatorially, reveals that PIEZO1 associates primarily with ITGAL, not ITGB2. Furthermore, association of ITGAL and ITGB2 prevents association of ITGAL with PIEZO1.

86 Figure 4.4: PIEZO1 does not interact with MYH9. MYH9-mCherry was expressed heterologously in HEK293T cells and can be diﬀerentiated from endogenous MYH9 due to the molecular weight shift of the fusion protein. PIEZO1 did not immunopre- cipitate either the endogenous or exogenous MYH9.

1591 as being most distal, so we initially focused on mutant versions of PIEZO1 that queried this and adjacent regions. To our surprise, no truncation mutants of

PIEZO1 failed to bind ITGAL – not even the mutually exclusive mutants 1:800-eGFP and 1351:1591-eGFP. This confusing result could have been explained multiple ways, such as the presence of multiple intermolecular contact points or nonspeciﬁc binding.

However, further experimentation demonstrated that the 1351:1591 fragment, but not the 1:800 fragment, could associate with diﬀerentially tagged PIEZO1 trimers (data not shown). Since PIEZO1 is endogenously expressed at a low level in wild type HEK

293T cells,268 we believe this is the explanation for the data.

Having thus excluded the region around 1351-1591, we continued further N terminal by halves, creating a series of truncations in regions of relatively low conservation across species, and placing the C-terminal eGFP tag as well as possible within re-

87 88 Figure 4.5: PIEZO1 interaction with ITGAL is dependent on a conserved region of transmembrane helix 2 and adjacent intracellular residues. A) All truncation mutants targeting the middle of PIEZO1 bind ITGAL. 1351-1591 independently binds ITGAL, despite not overlapping with other truncation mutants. B) Further N terminal PIEZO1 truncation mutants all bind ITGAL, with the exception of 1:24. C) Further narrowing of the putative PIEZO1-ITGAL interaction site D) PIEZO1 1:48 binds ITGAL weakly, and PIEZO1 1:51 binds ITGAL equivalently to other mutants. Mutants shorter than 1:48 do not bind ITGAL. E) Superimposition of the truncation mutants (vertical lines, binding indicated by green color) over a multiple sequence alignment of PIEZO1 showing conservation, amino acid type, and membrane topology. Top bar: within membrane regions are colored brown, extracellular regions are colored blue, and intracellular regions are colored red. The consensus of Guo et al. 2017 and Zhao et al. 2018 is shown by coloring both and setting opacity to 50 F) Amino acids 42:49 of PIEZO1 are suﬃcient for precipitation of ITGAL, but not when presented in scrambled order.

gions established by Coste and colleagues to be membrane-internal.269 By doing so,

we eventually demonstrated that the extreme N terminal end, 1:24, did not medi-

ate binding to ITGAL, whereas 1:51 did (Figure 4.5B). This narrowed the potential

span considerably but seemed to contradict our hypothesis about the location of the

interaction site. At this time, however, a new report of PIEZO1 structure emerged,

suggesting that the distal lever arm of PIEZO1 is not contributed by later amino

acids, but by the extreme N terminus,270 which was soon conﬁrmed by a convergent,

independently-generated structure.271 We further dissected the region from 1:24 to

1:51, and demonstrated that the region between 41 and 51 seemed to be required

(Figure 4.5C). Narrowing even further, the truncation mutant 1:43 does not bind, the mutant 1:48 binds weakly, and the mutant 1:51 binds equally or better as compared with other truncation mutants of PIEZO1 (Figure 4.5D). Superimposing these data with the conservation and putative transmembrane topology of PIEZO1 as reported

89 by Guo et al.270 and Zhao et al.271 reveals that the most likely interaction residues are the sequence PWFPGPTR (in humans) and PWLPGPSR (in mice). This region spans the interface between the inner membrane leaﬂet and immediately adjacent cytoplasm (Figure 4.5E), and is thus positioned to interact with the GFFKR motif of alpha integrins which is dissociated from beta integrins upon activation.3,272

Further experiments are underway to reﬁne the interaction region on ITGAL. Fi- nally, we tested whether this region is suﬃcient for binding to ITGAL. To determine this, we used biotinylated PWFPGPTR peptide, versus scrambled control peptide

WPFPGPRT, to precipitate ITGAL using streptavidin beads and conﬁrmed that the

PIEZO1-derived peptide, but not the scrambled peptide, precipitates ITGAL (Figure

4.5F). The location of this site, as well as the putative transmembrane regions made by cross-referencing available biochemical and structural data of PIEZO1, is shown in Figure 4.6.

4.2.5 PIEZO1 is required to coordinate crawling motility

Having thus established a plausble biochemical link between PIEZO1 and activated

LFA-1, we next sought to determine whether PIEZO1 deficiency carries a discernible phenotype in the adhesion cascade. We focused our efforts first on leukocyte crawling, given the results of our immunofluorescence investigations, and the previously established role of PIEZO1 in confined migration of CHO cells. We used recombinant

Cas9 protein complexed with CRISPR RNAs and electroporated these into primary human T lymphocytes to knock out PIEZO1. This generated suﬃcient knockout ef-

ﬁciency to test a phenotype (Figure 4.7A). This manipulation signiﬁcantly reduced mean crawling displacement over time (Figure 4.7B), t = −2.13, p < 0.05. Together

90 Figure 4.6: PIEZO1 transmembrane topology. The topology of the 38 PIEZO1 transmembrane domains was assembled by cross-referencing Guo & Mackinnon,270 Zhao et al.,271 and Coste et al.269 The location of the binding motif required for interaction with ITGAL is indicated

91 with this signiﬁcant reduction in motility, we also observed induction of grotesque protrusions from cells (Figure 4.7C). These protrusions appeared to be attempting crawling, with the remainder of the cell staying behind. Ultimately, these cells extruded their plasma membranes too far and appeared to die. To better reinforce whether our crawling phenotype was dependent on PIEZO1 knockout, we investigated the change in crawling speed (versus each donor’s average) as a function of

PIEZO1 knockout eﬃciency, assayed as Yoda1 nonresponsiveness. This produced a signiﬁcant anticorrelation (Figure 4.7D), F (1, 17) = 4.61, p < 0.05, whereas a similar test for cell viability did not.

4.3 Discussion

While roles of PIEZO1 in modulating integrin aﬃnity and coordinating cell migration have been established in the past, the present study advances our knowledge of PIEZO1 and T lymphocyte biology in three important ways: First, most existing conceptual models of PIEZO1 function ascribe it as a ﬂoating entity in the membrane.

Our finding that PIEZO1 specifically associates with high-affinity LFA-1 suggests that

PIEZO1 may instead be transducing tension on other proteins. There are numerous examples of extracellular signaling axes that are myosin and cross-linking dependent.

In fact, during the preparation of this manuscript, another group demonstrated that

PIEZO1 plays a role in myosin-dependent calcium influx during T lymphocyte activation in vitro.273 Given the importance of LFA-1 to T lymphocyte activation in vitro, it is possible that LFA-1 plays a role in this effect. Second, we have identified a new potential molecular target for modulating the leukocyte adhesion cascade. Targeting

92 Figure 4.7: PIEZO1 is required for LFA-1 dependent crawling motility. A) Percentage of Yoda1 nonresponsive cells under different CRISPR conditions, as assayed by Indo-1 flow cytometry. In the NT (control gRNA) and WT (wild type) conditions, approximately 10% of cells fail to respond to Yoda1, presumably because of its high EC50. The percentage of Yoda1 non-responsive cells is increased by electroporation of PIEZO1-targeted gRNAs, indicating knockout of PIEZO1 in these cells. N=3 donors, 4 P1 guides, 2 NT guides, and 1 WT. B) Median crawling speed (Displacement/time) in the three CRISPR conditions, demonstrating that PIEZO1 knockout reduces crawling. C) Representative time-lapse brightfield image of a PIEZO1 knockout T lymphocyte attempting to crawl. While NT and WT cells sent out occasional thin projections before crawling, P1 cells extruded large sections of cytoplasm into strange and grotesque shapes that, in this case, doubled back on itself. D) Response effect of PIEZO1 knockout efficiency (as assayed by Yoda1 nonresponsiveness) on crawling speed. Speeds are adjusted to each donor’s mean to account for high donor to donor variability.

93 crawling, as opposed to adhesion altogether, may allow the eﬃciency of pathological inﬂammatory responses to be dampened while still permitting homeostatic immunity.

The feasibility of this approach will also depend on the expression of PIEZO1 on other cell types. Third, we believe the unique morphological phenotype of PIEZO1 knockout T lymphocyte attempting to crawl may prompt new and fruitful investigation into the regulation of cytoskeletal proteins during T lymphocyte crawling.

There are important caveats with this study. First, the phenotype has been established in vitro but not in vivo. Follow up experiments are ongoing to establish whether Piezo1 deﬁciency in mice aﬀects homing to lymph nodes. Second, the biochemical work was almost all performed using a fusion protein in HEK 293T cells.

While this was necessary given the lack of sensitivity of PIEZO1 antibodies, it does increase the risk of false positives. Further characterization into the structural basis of the interaction of ITGAL with PIEZO1 using NMR and parallel biochemical approaches is in the planning stage. Finally, the link between PIEZO1 and integrins remains to be established. Given the prior suggestion of an eﬀect of modulating Rac1, we propose that PIEZO1 activation will allosterically lead to de-activation of Rac1, allowing LFA-1 to leave high aﬃnity state.

4.4 Materials and methods

Ethics Statement As detailed below, the research involving human participants and animal subjects adhered to all institutional, local, state, federal, and international laws and guidelines concerning responsible and ethical conduct of research.

Research involving mice was approved by the Case Western Reserve University An-

94 imal Resource Center (IACUC Approval Number: 2015-0118), which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care

(AAALAC). Research activities strictly followed the procedures and protocols outlined in that approval. Research involving human samples was approved by the Uni- versity Hospitals Cleveland Medical Center Institutional Review Board (UHCMC IRB

Approval Number: 08-15-17), which is accredited by the Association for the Accred- itation of Human Research Protection Programs (AAHRPP). Our research activities strictly followed the procedures and protocols therein. Most of the blood samples were obtained de-identiﬁed from the Case Western Reserve University Hematopoietic

Biorepository & Cellular Therapy Shared Resource (HBCTSR), which also acts in accordance with its own UHCMC IRB approved protocols. Informed consent was obtained by the HBCTSR for those participants. Other samples were obtained under our IRB-approved protocol. Informed consent was obtained by the experimenter, recorded, and stored separately in a secured location.

Cell Lines CEM and Jurkat human T leukemia lines were obtained from ATCC and checked regularly for Mycoplasma spp. contamination using the Venor GeM My- coplasma Detection Kit (Sigma Aldrich). These cells were maintained in T cell media

(TCM): RPMI-1640 supplemented with 10% Fetal Bovine Serum, 10 mM HEPES, 1 mM Sodium Pyruvate, and non-essential amino acids. HEK 293T cells were maintained in DMEM with 15% SuperCalf Serum (Gemini), 1% penicillin/streptomycin

(Gibco) and supplemented with non-essential amino acids (Gibco), L-glutamine (Gibco) and HEPES. Transient transfections for overexpression were performed using calcium phosphate.

95 Plasmids and cloning All constructs were prepared from plasmids generously provided by the Gottlieb lab, and adapted using Gibson cloning. mNeonGreen274 source plasmid was purchased from Allele. Human ITGAL (Addgene plasmid # 8630),275

ITGAM (Addgene plasmid # 8631),276 and ITGB2 (Addgene plasmid # 8640)277 were gifts from Timothy Springer. Human ITGA4 was a gift from Chinten James

Lim (Addgene plasmid # 80016).278 MYH9 was a gift from Venkaiah Betapudi (Ad- dgene plasmid # 35687).279 Lentiviral and truncation constructs were prepared using

NEBuilder (New England Biolabs). Cloning was veriﬁed using Sanger sequencing.

Final plasmids were sequenced completely by the MGH CCIB DNA Core for veriﬁ- cation.

Lentiviral transduction N=8-10 x 6 cm plates of HEK293T cells at optimal density were transfected using calcium phosphate and supernatants were harvested at 24 hours. Pooled supernatants were centrifuged at 150,000 g for 2 hours and resuspended in 200 microliters of PBS. These supernatants were allowed to bind to the surface of non-tissue culture treated plates that had been coated with Retronectin (Clontech) and blocked with 2% BSA in PBS. Cells were then added and centrifuged at 800 g for 2 hours at 32◦C.

Western blotting and immunoprecipitation Cells were lysed in an ice-cold buﬀer of 140 mM NaCl, 1% NP-40, 0.5% deoxycholate, pH=7.45 (at 20◦C), with protease inhibitor cocktail (Sigma-Aldrich P8340) and PMSF (Acros Organics). Lysate was spun at 12,000 g for 10 minutes at 4◦C to clear debris, then split for whole cell lysate vs. immunoprecipitation analysis. Lysates were combined with 4X Laemmli buﬀer and heated at 42◦C for 30 minutes. Immunoprecipitations were performed using

96 GFP-Trap using the manufacturer’s protocol, ending with addition of 4X Laemmli buﬀer and heating to 42◦C for 30 minutes. SDS-PAGE was performed using Nu-

PAGE 3-8% Tris-Acetate and 4-12% Bis-Tris gels (Thermo Fisher), depending on the molecular weight of the proteins under study, and transferred to 0.45 micron PVDF membranes (EMD Millipore). Blocking buﬀer and resuspension buﬀer for all antibodies was 5% BSA in 0.1% TBS-Tween 20, pH=7.6 (at 20◦C). Blots were imaged using a LAS 4000 (GE) after exposure to chemiluminescence substrate (Advansta).

Antibodies Cell Signaling Technologies was the source for anti-eGFP (D5.1), anti-

ITGA4 (4600), anti-ITGAM (D6X1N) anti-MYH9 (3403), anti-rabbit (7074), and anti-mouse (7076) antibodies. Antibodies to ITGAL (EP1285Y and HI111) and

PIEZO1 (ab82336) were obtained from Abcam. Anti-goat antibody was leftover stock from Santa Cruz Biotechnologies. MAB1730 and AF1730 against ITGB2 were obtained from R&D systems. MEM 148 against ITGB2 was obtained from Thermo

Fisher Scientiﬁc. GFP Trap was obtained from Chromotek GmbH. Recombinant protein G Sepharose beads were from Life Technologies. Streptavidin beads were from

Thermo Scientiﬁc.

Drugs and Dyes Ionomycin, Indo-1, and FITC-phalloidin were obtained from Cay- man Chemical. CFDA-SE and Fluo-4 were obtained from Molecular Probes. Yoda1 was obtained from Tocris Biosciences.

Peptides Peptides were prepared by Genscript.

Immunoﬂuorescence T lymphocytes were allowed to adhere to uncoated, ICAM-

1 coated or CXCL12 and ICAM-1 coated wells of an 8-well chamber glass (Nunc)

97 for 30 minutes, then ﬁxed in 4% para-formaldehyde (PFA) for 10 minutes, washed in PBS, and blocked with a mixture of 0.25% Triton X100, 5% donkey serum, and

5% autologous human serum for 45 minutes prior to addition of primary antibodies.

These were incubated overnight at 4◦C, washed three times with PBS, and replaced with donkey secondary antibodies (Jackson ImmunoResearch) against the appropriate primary species. Images were obtained on a Leica SP5 DMI 6000B confocal using argon 488-nm laser, and helium-neon 594-nm and 633-nm lasers with a Leica 506192

HCX PL APO lambda blue 63x/1.4 oil objective with 0.17-mm glass correction. For detection, a 12-bit Photomultiplier tubes were used with Leica LAS AF acquisition software. Deconvolution was done with Huygens Professional 16.10 using CMLE with

SNR 5 and visualized using maximum intensity projections in Huygens Professional.

No deconvolution was performed for colocalization analyses, which were performed using FIJI.

Microﬂuidic Devices Preparation of microﬂuidic devices has previously been described.128 Devices were coated using proteins as follows: recombinant human CXCL12 and recombinant human Fc-ICAM1 chimeric fusion protein were obtained from R&D

Systems (Minneapolis, MN). SDC4, expressed in mammalian cells so as to have glycosaminoglycans, was obtained from Adipogen Corporation (San Diego, CA). These molecules adsorbed onto the device walls at concentrations of 20, 80, and 50 micrograms per milliliter, for CXCL12, ICAM1, and SDC4, respectively, in phosphate buﬀered saline (PBS) for a minimum of four hours in a 37◦C incubator. After this incubation, device channels were blocked with 5% BSA in saline for a minimum of 30 minutes. PBMC isolation was performed as previously described.128 T lymphocytes

98 were isolated from PBMC using an EasySep Human T Cell Isolation Kit (STEMCELL

Technologies, Cambridge, MA) according to the manufacturer’s protocol.

T Lymphocyte Stimulation In indicated cases, T lymphocytes were stimulated

for 48 hours using functional antibodies anti-CD3 (OKT3, Miltenyi) and anti-CD28

(BD Pharmingen) pre-immobilized on non-tissue culture treated 12 well plates at con-

centrations of 100 micrograms per microliter. Stimulation was veriﬁed by formation

of activation clusters and increase of metabolic demand.

T Lymphocyte CRISPR Genetic knockout of PIEZO1 and ITGAL were per-

formed according to established protocols.280 Brieﬂy, crRNA:tracrRNA duplexes were formed at a 100 micromolar concentration in a 1:1 molar ratio in nuclease free duplex buﬀer (Integrated DNA Technologies; IDT) by heating to 95◦C and ramping 0.2◦C/s

to 4◦C. These duplexes were then complexed with recombinant Cas9, generously pro-

vided by Dr. Chris Jeans (QB3 macroLab, UC Berkeley) at a 3:1 molar ratio, and

incubated for a minimum of 37◦C for 15 minutes. Complexes were generated en masse

and stored at 4 degrees for a maximum of one month. For each reaction, 1 micro-

liter of a 100 micromolar stock of electroporation enhancer oligonucleotide (IDT) was

added. Lymphocytes were resuspended at 20 microliters per reaction in Lonza P3

buﬀer with variable density, for a maximum of 10 million cells per well, transferred to

16-well strips, and electroporated on an Amaxa 4D Nucleofector (Lonza) using pulse

code EH100. Cells were then immediately transferred to pre-warmed T cell media

without antibiotics.

Flow cytometry All sorting experiments were performed in Hank’s basic salt so-

lution (HBSS) with added calcium at 2 mM or, in indicated cases, added EGTA at

99 2 mM. Temperature was maintained at 37◦C when adding Yoda1 at 25 micromolar concentration. To sort CEM cells for PIEZO1 knockout, a BD FACS Aria SORP was used with UV 379 and UV 535 ﬁlters for Indo-1 and B515 for mNeonGreen. To assess CRISPR knockout eﬃciency in primary T lymphocytes, identical conditions were used, but on a BD Fortessa. Ionomycin was used as a positive control at 1 micromolar.

Live Cell Imaging: Crawling Assay Cells were resuspended to 20 thousand cells per well in live cell imaging solution (LCIS, Thermo Fisher Scientiﬁc) supplemented with 5% BSA and 50 mM D-glucose and passed through a 35 micrometer cell strainer before plating on Nunc 8 well chamber glass coated with CXCL12 and ICAM1 as described above. To this solution, 2 micrograms per mL of Acridine Orange (Cayman

Chemical) was added to track nuclei. Cells were allowed to settle before imaging for one hour at 15 second intervals on a Leica SP5 DMI 6000B confocal using argon 488- nm laser and 10x dry objective with a custom-built heated environmental chamber stabilized at 37◦C.

Image analysis Spots representing lymphocyte nuclei were tracked using Imaris

(BitPlane). Spots with excessive speed, IE known to be ﬂoating in convective motion, were removed using the same threshold for all experiments. Tracks of less than 150 seconds were excluded.

Statistics Analysis was performed in R245 in R Studio.246 Results of statistical tests are reported in the text with the type of test (e.g. F or t), followed by degrees of freedom in parentheses, the test statistic, and the p value. Plots were generated using package ggplot2.249 Reports were generated using knitr.250

100 Chapter 5

Conclusions and future directions

The conclusions of the present set of studies are relatively straightforward.

First, scale dependence is a critical consideration in any biological process, but especially so when physical forces are concerned. Signaling modes that seem plausible given intuitive understanding of fluid flows at the macroscale are demonstrably impossible at the microscale of post-capillary venules. Meanwhile, bringing microfluidic leukocyte adhesion models from ’small’ to ’physiologically small’ produced a qual- itative change in adhesion phenomena and exposed a requirement for erythrocyte interactions in inducing efficient adhesion. Surprisingly, this effect was dependent on ICAM-1 interaction with LFA-1, through a presently unknown mechanism. Fur- ther experimentation manipulating erythrocyte and leukocyte deformabilities may shed light on whether the compression mechanism is the most critical component, or whether there is another factor delivered by the erythrocytes. Chief among the latter: it is plausible that low-level expression of adhesion molecules such as ICAM4 on erythrocytes may subtly enhance the compression effect in the transition to PCVs.

101 Second, the mechanosensor PIEZO1 appears to play a critical role in T lymphocyte biology, not only in activation, but also in migration. To do so, it appears to sense forces not, as has been suggested, by ﬂoating in the membrane and sensing stretch, but rather by complexing directly with high aﬃnity integrins. This interaction and its consequences should be the focus of the next investigations, since this represents an exciting candidate for mediating a host of mechanotransductive phenomena in T lymphocytes and is a potential therapeutic target.

5.1 Structural relationship between PIEZO1 and

ITGAL

While a promising biochemical link has been established between PIEZO1 and alpha integrins, with a focus on ITGAL, further exploration of this link should focus on determination of the structural basis. Generation of mimetic peptides did not result in recapitulation of PIEZO1 knockout phenotype at low peptide concentrations. Given the lack of knowledge about the structural basis of the interaction, this could be due to a true lack of effect, or due to a failure of the peptides to adequately disrupt the interaction. The already-made 1:143 truncation mutant of PIEZO1 would serve as an appropriate starting point, given that the transmembrane regions of PIEZO1 exist in functional groups of four:270,271 143 corresponds to the cytoplasmic end of the fourth helix. The prediction is that PIEZO1 will specifically interact with activated ITGAL via a latch mechanism, corresponding to a published report of activation-induced retrofolding of ITGAL GFFKR into the inner membrane leaflet.272

102 5.2 Downstream signaling from PIEZO1

Likely numerous important links are missing between PIEZO1 and control of integrin affinity in crawling T lymphocytes. To follow off the pathway proposed by Hung et al.,222 a recently-published single-excitation cAMP reporter281 could be co-transfected with PIEZO1 CRISPR RNPs and used to assess cAMP activity in crawling leukocytes under static and flow conditions. This plasmid is available on Addgene. To this, further exploration could take the form of using inhibitors and activators of PKA,

RhoA, and Rac1 to try to recapitulate and rescue the PIEZO1 knockout phenotype.

Available Rac1 and RhoA activation assays can be used in conjunction with immobilized plate-bound CXCL12 and ICAM1, or with recombinant protein G dynabeads with Fc-ICAM1 and SDC4 GAGs used to link CXCL12.

An additional quesiton of vital importance is the type of force that is activating

PIEZO1 during crawling. The trimeric structure and implication of binding integrins on each pole tend to suggest that failure of the T lymphocyte to coherently apply traction in one direction, IE after the high aﬃnity integrin complex has passed un- derneath the MTOC, would lead to PIEZO1 activation. To test this hypothesis, one approach would be to measure the vectors of forces exerted by T lymphocytes during crawling.282

Finally, signaling investigations of PIEZO1 are currently hampered by the lack of good antibodies. Development of a high quality antibody to an extracellular loop of

PIEZO1, preferably one conserved with mouse Piezo1, should begin forthwith.

103 Figure 5.1: Model of Piezo1 interplay with Rac1 and Myosin II, adapted under CC BY NC ND from Wei-Chien Hung, Jessica R. Yang, Christopher L. Yankaskas, Bin Sheng Wong, Pei-Hsun Wu, Carlos Pardo-Pastor, Selma A. Serra, Meng-Jung Chi- ang, Zhizhan Gu, Denis Wirtz, Miguel A. Valverde, Joy T. Yang, Jin Zhang, and Konstantinos Konstantopoulos: Conﬁnement Sensing and Signal Optimization via Piezo1/PKA and Myosin II Pathways. Cell Reports 15(7) (May 17, 2016), 1430– 1441. doi: 10.1016/j.celrep.2016.04.035

104 5.3 Extension of in vitro ﬁndings with

microﬂuidic endothelial cell cultures

Given the known role of PIEZO1 in sensing substrate stiﬀness,283 the use of glass as a substrate for adhesion studies is admittedly suspect, if expedient. Integration of the technique for culture of TY10 brain microvascular endothelial cells with the leukocyte crawling assays should allow a more physiological investigation into the role of PIEZO1 in leukocyte intravascular crawling. This may represent the best compromise between ease of manipulability and validity of results going forward.

5.4 Piezo1 contribution to immunity in vivo

Ultimately, a demonstration of the relevance of Piezo1 in mice will be necessary. Given that Piezo1 has a role in T lymphocyte activation in vitro,273 Investigations should start first with basic characterization of T lymphocyte development in thymus. The current mouse line, Cd4 Cre, Piezo1 fl/fl will not be helpful in assaying double negative stages, but should be useful starting in the double positive stages, through single positive, and ultimate differentiation. Given the prior literature on the importance of

LFA-1 to T lymphocyte activation in vitro versus in vivo, and the robust presence of

both Cd4 and Cd8 positive T lymphocytes in these mice, we predict that there will be

only subtle shifts in lineages, rather than a full blown immunosuppressed phenotype.

Next is an assessment of the contribution of Piezo1 to T lymphocyte traﬃcking.

This should start with the simplest experiment possible: isolation of na¨ıve Cd4 and

Cd8 T lymphocytes, staining with CFDA-SE, and quantiﬁcation of traﬃcking behav-

105 ior to lymph nodes using intravital imaging and flow cytometry. Care must be taken with the flow cytometry assay: prior to harvest, blockade of integrins must be performed to dissociate inefficiently crawling cells from the high endothelial vasculature.

The methods used in Park et al., 2010184 should be followed carefully. This likely means, for practical purposes, ignoring some traditional sites such as spleen, bone marrow, and blood, since affecting lymphocyte crawling has no demonstrable effect in these areas, despite a strong effect on lymph nodes.

Finally, a disease model that demonstrates functional consequence of Piezo1 de-

ficiency should be identified. Most likely, this will be EAE. If mouse numbers are sufficient, both primary and adoptive transfer EAE should be attempted. The focus of the first round of both primary and adoptive transfer EAE experiments should be assaying T lymphocyte proliferation and cytokine production, in order to correlate in vivo phenotype with the published in vitro findings. Once this is established, then adoptive transfer should be attempted. If Cd4 Cre+, Piezo1 fl/fl mice fail to generate an appropriate response to MOG 35-55 antigen, then Mog-specific T lymphocyte lines should be expanded and Piezo1 deleted using CRISPR RNPs prior to transfer to control for initial activation. Labeling of these cells and intravital imaging of pial vasculature 48-60 hours after induction should be performed, using Bartholomäuset al. as a guide.9

106 Complete References

1. Klaus Ley, Carlo Laudanna, Myron I. Cybulsky, and Sussan Nourshargh: Getting

to the Site of Inﬂammation: The Leukocyte Adhesion Cascade Updated. Nature

Reviews Immunology 7(9) (Sept. 2007), 678–689. doi: 10.1038/nri2156.

2. David Askew, Charles A. Su, Deborah S. Barkauskas, R. Dixon Dorand, Jay Myers,

Rachel Liou, Joseph Nthale, and Alex Y. Huang: Transient Surface CCR5

Expression by Naive CD8+ T Cells within Inﬂamed Lymph Nodes Is Dependent on

High Endothelial Venule Interaction and Augments Th Cell-Dependent Memory

Response. Journal of Immunology (Baltimore, Md.: 1950) 196(9) (May 2016),

3653–3664. doi: 10.4049/jimmunol.1501176.

3. Olga Vinogradova, Algirdas Velyvis, Asta Velyviene, Bin Hu, Thomas A. Haas,

Edward F. Plow, and Jun Qin: A Structural Mechanism of Integrin αIIbβ3

“Inside-Out” Activation as Regulated by Its Cytoplasmic Face. Cell 110(5)

(Sept. 6, 2002), 587–597. doi: 10.1016/S0092-8674(02)00906-6.

4. Taylor H. Schreiber, Vera Shinder, Derek W. Cain, Ronen Alon, and

Robert Sackstein: Shear Flow–Dependent Integration of Apical and Subendothelial

Chemokines in T-Cell Transmigration: Implications for Locomotion and the

Multistep Paradigm. Blood 109(4) (Feb. 15, 2007), 1381–1386. doi:

10.1182/blood-2006-07-032995.

5. Hwee San Lek, Vicky L. Morrison, Michael Conneely, Paul A. Campbell,

David McGloin, Stefanie Kliche, Colin Watts, Alan Prescott, and

Susanna C. Fagerholm: The Spontaneously Adhesive Leukocyte

Function-Associated Antigen-1 (LFA-1) Integrin in Eﬀector T Cells Mediates Rapid

Actin- and Calmodulin-Dependent Adhesion Strengthening to Ligand under Shear

107 Flow. Journal of Biological Chemistry (Apr. 12, 2013), jbc.M112.430918. doi:

10.1074/jbc.M112.430918.

6. Ronen Alon and Michael L. Dustin: Force as a Facilitator of Integrin

Conformational Changes during Leukocyte Arrest on Blood Vessels and

Antigen-Presenting Cells. Immunity 26(1) (Jan. 1, 2007), 17–27. doi:

10.1016/j.immuni.2007.01.002.

7. Ronen Alon, Sara W. Feigelson, Eugenia Manevich, David M. Rose, Julia Schmitz,

Darryl R. Overby, Eitan Winter, Valentin Grabovsky, Vera Shinder,

Benjamin D. Matthews, Maya Sokolovsky-Eisenberg, Donald E. Ingber,

Martin Benoit, and Mark H. Ginsberg: Α4β1-Dependent Adhesion Strengthening

under Mechanical Strain Is Regulated by Paxillin Association with the

Α4-Cytoplasmic Domain. The Journal of Cell Biology 171(6) (Dec. 19, 2005),

1073–1084. doi: 10.1083/jcb.200503155.

8. Mia Phillipson, Bryan Heit, Pina Colarusso, Lixin Liu, Christie M. Ballantyne, and

Paul Kubes: Intraluminal Crawling of Neutrophils to Emigration Sites: A

Molecularly Distinct Process from Adhesion in the Recruitment Cascade. The

Journal of Experimental Medicine 203(12) (Nov. 27, 2006), 2569–2575. doi:

10.1084/jem.20060925.

9. Ingo Bartholom¨aus,Naoto Kawakami, Francesca Odoardi, Christian Schl¨ager,

Djordje Miljkovic, Joachim W. Ellwart, Wolfgang E. F. Klinkert,

Cassandra Fl¨ugel-Koch, Thomas B. Issekutz, Hartmut Wekerle, and

Alexander Fl¨ugel:Eﬀector T Cell Interactions with Meningeal Vascular Structures

in Nascent Autoimmune CNS Lesions. Nature 462(7269) (Nov. 5, 2009), 94–98. doi:

10.1038/nature08478.

108 10. Tim L¨ammermann,Bernhard L. Bader, Susan J. Monkley, Tim Worbs,

Roland Wedlich-S¨oldner,Karin Hirsch, Markus Keller, Reinhold F¨orster,

David R. Critchley, Reinhard F¨assler, and Michael Sixt: Rapid Leukocyte

Migration by Integrin-Independent Flowing and Squeezing. Nature 453(7191) (May

2008), 51–55. doi: 10.1038/nature06887.

11. T A Springer: Traﬃc Signals on Endothelium for Lymphocyte Recirculation and

Leukocyte Emigration. Annual Review of Physiology 57(1) (1995), 827–872. doi:

10.1146/annurev.ph.57.030195.004143.

12. Klaus Ley: The Role of Selectins in Inﬂammation and Disease. Trends in Molecular

Medicine 9(6) (June 2003), 263–268. doi: 10.1016/S1471-4914(03)00071-6.

13. C. Foxall, S. R. Watson, D. Dowbenko, C. Fennie, L. A. Lasky, M. Kiso,

A. Hasegawa, D. Asa, and B. K. Brandley: The Three Members of the Selectin

Receptor Family Recognize a Common Carbohydrate Epitope, the Sialyl Lewis(x)

Oligosaccharide. The Journal of Cell Biology 117(4) (May 15, 1992), 895–902. doi:

10.1083/jcb.117.4.895.

14. Leszek Poppe, Gregory S. Brown, John S. Philo, Pandurang V. Nikrad, and

Bhavana H. Shah: Conformation of sLex Tetrasaccharide, Free in Solution and

Bound to E-, P-, and L-Selectin†,‡. Journal of the American Chemical Society

119(7) (Feb. 1, 1997), 1727–1736. doi: 10.1021/ja9610702.

15. Fugang Li, Patricia P. Wilkins, Suzanne Crawley, Jasminder Weinstein,

Richard D. Cummings, and Rodger P. McEver: Post-Translational Modiﬁcations of

Recombinant P-Selectin Glycoprotein Ligand-1 Required for Binding to P- and

E-Selectin. Journal of Biological Chemistry 271(6) (Sept. 2, 1996), 3255–3264. doi:

10.1074/jbc.271.6.3255.

109 16. A Varki: Selectin Ligands: Will the Real Ones Please Stand Up? Journal of Clinical

Investigation 99(2) (Jan. 15, 1997), 158–162.

17. Guo-Yun Chen, Hirotaka Osada, Luis F. Santamaria-Babi, and Reiji Kannagi:

Interaction of GATA-3/T-Bet Transcription Factors Regulates Expression of Sialyl

Lewis X Homing Receptors on Th1/Th2 Lymphocytes. Proceedings of the National

Academy of Sciences 103(45) (July 11, 2006), 16894–16899. doi:

10.1073/pnas.0607926103.

18. Amy J. Wagers, Christopher M. Waters, Lloyd M. Stoolman, and

Geoﬀrey S. Kansas: Interleukin 12 and Interleukin 4 Control T Cell Adhesion to

Endothelial Selectins through Opposite Eﬀects on Α1,3-Fucosyltransferase VII Gene

Expression. The Journal of Experimental Medicine 188(12) (Dec. 21, 1998),

2225–2231. doi: 10.1084/jem.188.12.2225.

19. Luthur Siu-Lun Cheung, Phrabha S. Raman, Eric M. Balzer, Denis Wirtz, and

Konstantinos Konstantopoulos: Biophysics of Selectin–Ligand Interactions in

Inﬂammation and Cancer. Physical Biology 8(1) (Feb. 1, 2011), 015013. doi:

10.1088/1478-3975/8/1/015013.

20. William S Somers, Jin Tang, Gray D Shaw, and Raymond T Camphausen: Insights

into the Molecular Basis of Leukocyte Tethering and Rolling Revealed by

Structures of P- and E-Selectin Bound to SLeX and PSGL-1. Cell 103(3) (Oct. 27,

2000), 467–479. doi: 10.1016/S0092-8674(00)00138-0.

21. Tadayuki Yago, Jianhua Wu, C. Diana Wey, Arkadiusz G. Klopocki, Cheng Zhu,

and Rodger P. McEver: Catch Bonds Govern Adhesion through L-Selectin at

Threshold Shear. The Journal of Cell Biology 166(6) (Sept. 13, 2004), 913–923.

doi: 10.1083/jcb.200403144.

110 22. Andr´esHidalgo, Anna J. Peired, Martin K. Wild, Dietmar Vestweber, and

Paul S. Frenette: Complete Identiﬁcation of E-Selectin Ligands on Neutrophils

Reveals Distinct Functions of PSGL-1, ESL-1, and CD44. Immunity 26(4)

(Apr. 27, 2007), 477–489. doi: 10.1016/j.immuni.2007.03.011.

23. Alexander Zarbock, Klaus Ley, Rodger P. McEver, and Andr´esHidalgo: Leukocyte

Ligands for Endothelial Selectins: Specialized Glycoconjugates That Mediate

Rolling and Signaling under Flow. Blood 118(26) (Dec. 22, 2011), 6743–6751. doi:

10.1182/blood-2011-07-343566.

24. Alexander Zarbock, Cliﬀord A. Lowell, and Klaus Ley: Spleen Tyrosine Kinase Syk

Is Necessary for E-Selectin-Induced αLβ2 Integrin-Mediated Rolling on Intercellular

Adhesion Molecule-1. Immunity 26(6) (June 22, 2007), 773–783. doi:

10.1016/j.immuni.2007.04.011.

25. Tadayuki Yago, Bojing Shao, Jonathan J. Miner, Longbiao Yao,

Arkadiusz G. Klopocki, Kenichiro Maeda, K. Mark Coggeshall, and

Rodger P. McEver: E-Selectin Engages PSGL-1 and CD44 through a Common

Signaling Pathway to Induce Integrin αLβ2-Mediated Slow Leukocyte Rolling. Blood

116(3) (July 22, 2010), 485–494. doi: 10.1182/blood-2009-12-259556.

26. Maria Nácher, Ana BelénBlázquez, Bojing Shao, Adela Matesanz,

Colette Prophete, M. Cecilia Berin, Paul S. Frenette, and Andr´esHidalgo:

Physiological Contribution of CD44 as a Ligand for E-Selectin during Inﬂammatory

T-Cell Recruitment. The American Journal of Pathology 178(5) (May 2011),

2437–2446. doi: 10.1016/j.ajpath.2011.01.039.

27. Gloria Vachino, Xiao-Jia Chang, Geertruida M. Veldman, Ravindra Kumar,

Dianne Sako, Lynette A. Fouser, Michael C. Berndt, and Dale A. Cumming:

111 P-Selectin Glycoprotein Ligand-1 Is the Major Counter-Receptor for P-Selectin on

Stimulated T Cells and Is Widely Distributed in Non-Functional Form on Many

Lymphocytic Cells. Journal of Biological Chemistry 270(37) (Sept. 15, 1995),

21966–21974. doi: 10.1074/jbc.270.37.21966.

28. Daniela Fr¨olich, Daniela Blaβfeld, Karin Reiter, Claudia Giesecke,

Capucine Daridon, Henrik E. Mei, Gerd R. Burmester, David M. Goldenberg,

Abdulagabar Salama, and Thomas D¨orner:The Anti-CD74 Humanized Monoclonal

Antibody, Milatuzumab, Which Targets the Invariant Chain of MHC II Complexes,

Alters B-Cell Proliferation, Migration, and Adhesion Molecule Expression. Arthritis

Research & Therapy 14(2) (Mar. 9, 2012), R54. doi: 10.1186/ar3767.

29. Adil I. Khan, Steven M. Kerfoot, Bryan Heit, Lixin Liu, Graciela Andonegui,

Brian Ruﬀell, Pauline Johnson, and Paul Kubes: Role of CD44 and Hyaluronan in

Neutrophil Recruitment. The Journal of Immunology 173(12) (Dec. 15, 2004),

7594–7601. doi: 10.4049/jimmunol.173.12.7594.

30. Monique F Stins, Floyd Gilles, and Kwang Sik Kim: Selective Expression of

Adhesion Molecules on Human Brain Microvascular Endothelial Cells. Journal of

Neuroimmunology 76(1–2) (June 1997), 81–90. doi:

10.1016/S0165-5728(97)00036-2.

31. Carine Fillebeen, B´en´edicteDehouck, Monique Bena¨ıssa,Isabelle Dhennin-Duthille,

Rom´eoCecchelli, and Annick Pierce: Tumor Necrosis Factor-α Increases Lactoferrin

Transcytosis Through the Blood-Brain Barrier. Journal of Neurochemistry 73(6)

(Dec. 1, 1999), 2491–2500. doi: 10.1046/j.1471-4159.1999.0732491.x.

32. Jonathan M. Weiss, Sherry A. Downie, William D. Lyman, and Joan W. Berman:

Astrocyte-Derived Monocyte-Chemoattractant Protein-1 Directs the

112 Transmigration of Leukocytes Across a Model of the Human Blood-Brain Barrier.

The Journal of Immunology 161(12) (Dec. 15, 1998), 6896–6903.

33. Ruth Washington, Jeﬀrey Burton, Robert F. Todd, Walter Newman,

Ljubisa Dragovic, and Paula Dore-Duﬀy: Expression of Immunologically Relevant

Endothelial Cell Activation Antigens on Isolated Central Neurvous System

Microvessels from Patients with Multiple Sclerosis. Annals of Neurology 35(1)

(Jan. 1, 1994), 89–97. doi: 10.1002/ana.410350114.

34. G. V. McDonnell, S. A. McMillan, J. P. Douglas, A. G. Droogan, and

S. A. Hawkins: Serum Soluble Adhesion Molecules in Multiple Sclerosis: Raised

sVCAM-1, sICAM-1 and sE-Selectin in Primary Progressive Disease. Journal of

Neurology 246(2) (Feb. 1999), 87–92.

35. Bettina Kuenz, Andreas Lutterotti, Michael Khalil, Rainer Ehling, Claudia Gneiss,

Florian Deisenhammer, Markus Reindl, and Thomas Berger: Plasma Levels of

Soluble Adhesion Molecules sPECAM-1, sP-Selectin and sE-Selectin Are Associated

with Relapsing-Remitting Disease Course of Multiple Sclerosis. Journal of

Neuroimmunology 167(1–2) (Oct. 2005), 143–149. doi:

10.1016/j.jneuroim.2005.06.019.

36. Axinia D¨oring,Martin Wild, Dietmar Vestweber, Urban Deutsch, and

Britta Engelhardt: E- and P-Selectin Are Not Required for the Development of

Experimental Autoimmune Encephalomyelitis in C57BL/6 and SJL Mice. Journal

of Immunology (Baltimore, Md.: 1950) 179(12) (Dec. 15, 2007), 8470–8479.

37. Britta Engelhardt, Dietmar Vestweber, Rupert Hallmann, and Martina Schulz: E-

and P-Selectin Are Not Involved in the Recruitment of Inﬂammatory Cells Across

113 the Blood-Brain Barrier in Experimental Autoimmune Encephalomyelitis. Blood

90(11) (Dec. 1, 1997), 4459–4472.

38. F. J. Barkalow, M. J. Goodman, M. E. Gerritsen, and T. N. Mayadas: Brain

Endothelium Lack One of Two Pathways of P-Selectin-Mediated Neutrophil

Adhesion. Blood 88(12) (Dec. 15, 1996), 4585–4593.

39. Laura Piccio, Barbara Rossi, Elio Scarpini, Carlo Laudanna, Cinzia Giagulli,

Andrew C. Issekutz, Dietmar Vestweber, Eugene C. Butcher, and

Gabriela Constantin: Molecular Mechanisms Involved in Lymphocyte Recruitment

in Inﬂamed Brain Microvessels: Critical Roles for P-Selectin Glycoprotein Ligand-1

and Heterotrimeric Gi-Linked Receptors. The Journal of Immunology 168(4)

(Feb. 15, 2002), 1940–1949. doi: 10.4049/jimmunol.168.4.1940.

40. Steven M. Kerfoot and Paul Kubes: Overlapping Roles of P-Selectin and Α4

Integrin to Recruit Leukocytes to the Central Nervous System in Experimental

Autoimmune Encephalomyelitis. The Journal of Immunology 169(2) (July 15,

2002), 1000–1006. doi: 10.4049/jimmunol.169.2.1000.

41. Michael D. Carrithers, Irene Visintin, Suk J. Kang, and Charles A. Janeway:

Diﬀerential Adhesion Molecule Requirements for Immune Surveillance and

Inﬂammatory Recruitment. Brain 123(6) (Jan. 6, 2000), 1092–1101. doi:

10.1093/brain/123.6.1092.

42. Steven M. Kerfoot, M. Ursula Norman, Benoit M. Lapointe, Claudine S. Bonder,

Lori Zbytnuik, and Paul Kubes: Reevaluation of P-Selectin and Alpha 4 Integrin as

Targets for the Treatment of Experimental Autoimmune Encephalomyelitis. Journal

of Immunology (Baltimore, Md.: 1950) 176(10) (May 15, 2006), 6225–6234.

114 43. J. Pan, L. Xia, and R. P. McEver: Comparison of Promoters for the Murine and

Human P-Selectin Genes Suggests Species-Speciﬁc and Conserved Mechanisms for

Transcriptional Regulation in Endothelial Cells. The Journal of Biological

Chemistry 273(16) (Apr. 17, 1998), 10058–10067.

44. Luca Battistini, Laura Piccio, Barbara Rossi, Simona Bach, Simona Galgani,

Claudio Gasperini, Linda Ottoboni, Donatella Ciabini, Maria D. Caramia,

Giorgio Bernardi, Carlo Laudanna, Elio Scarpini, Rodger P. McEver,

Eugene C. Butcher, Giovanna Borsellino, and Gabriela Constantin: CD8+ T Cells

from Patients with Acute Multiple Sclerosis Display Selective Increase of

Adhesiveness in Brain Venules: A Critical Role for P-Selectin Glycoprotein

Ligand-1. Blood 101(12) (June 15, 2003), 4775–4782. doi:

10.1182/blood-2002-10-3309.

45. Pia Kivis¨akk,Don J. Mahad, Melissa K. Callahan, Corinna Trebst, Barbara Tucky,

Tao Wei, Lijun Wu, Espen S. Baekkevold, Hans Lassmann, Susan M. Staugaitis,

James J. Campbell, and Richard M. Ransohoﬀ: Human Cerebrospinal Fluid Central

Memory CD4+ T Cells: Evidence for Traﬃcking through Choroid Plexus and

Meninges via P-Selectin. Proceedings of the National Academy of Sciences 100(14)

(Aug. 7, 2003), 8389–8394. doi: 10.1073/pnas.1433000100.

46. Masayuki Miyasaka and Toshiyuki Tanaka: Lymphocyte Traﬃcking across High

Endothelial Venules: Dogmas and Enigmas. Nature Reviews Immunology 4(5) (May

2004), 360–370. doi: 10.1038/nri1354.

47. B. J. Steﬀen, G. Breier, E. C. Butcher, M. Schulz, and B. Engelhardt: ICAM-1,

VCAM-1, and MAdCAM-1 Are Expressed on Choroid Plexus Epithelium but Not

115 Endothelium and Mediate Binding of Lymphocytes in Vitro. The American Journal

of Pathology 148(6) (June 1996), 1819–1838.

48. Axinia Döring,Friederike Pfeiffer, Matthias Meier, Bénédicte Dehouck,

Silke Tauber, Urban Deutsch, and Britta Engelhardt: TET Inducible Expression of

the Α4β7-Integrin Ligand MAdCAM-1 on the Blood-Brain Barrier Does Not

Inﬂuence the Immunopathogenesis of Experimental Autoimmune

Encephalomyelitis. European Journal of Immunology 41(3) (Mar. 2011), 813–821.

doi: 10.1002/eji.201040912.

49. Krista G. Haanstra, Sam O. Hofman, Dave M. Lopes Estˆev˜ao,Erwin L. A. Blezer,

Jan Bauer, Li-Li Yang, Tim Wyant, Vilmos Csizmadia, Bert A. ‘t Hart, and

Eric R. Fedyk: Antagonizing the Α4β1 Integrin, but Not Α4β7, Inhibits Leukocytic

Inﬁltration of the Central Nervous System in Rhesus Monkey Experimental

Autoimmune Encephalomyelitis. The Journal of Immunology 190(5) (Jan. 3, 2013),

1961–1973. doi: 10.4049/jimmunol.1202490.

50. Geert D’Haens, S´everine Vermeire, Harald Vogelsang, Matthieu Allez,

Pierre Desreumaux, Andre Van Gossum, William J. Sandborn,

Daniel C. Baumgart, Richard M. Ransohoﬀ, Gail M. Comer, Alaa Ahmad,

Fabio Cataldi, John Cheng, Robert Clare, Kenneth J. Gorelick,

Annamarie Kaminski, Vivek Pradhan, Sunday Rivers, Matthew O. Sikpi,

Yanhua Zhang, Mina Hassan-Zahraee, Walter Reinisch, and Olaf Stuve: Eﬀect of

PF-00547659 on Central Nervous System Immune Surveillance and Circulating Β7+

T Cells in Crohn’s Disease: Report of the TOSCA Study. Journal of Crohn’s &

Colitis 12(2) (Jan. 24, 2018), 188–196. doi: 10.1093/ecco-jcc/jjx128.

116 51. Lone K. Rasmussen, Laust B. Johnsen, Torben E. Petersen, and Esben S. Sørensen:

Human GlyCAM-1 mRNA Is Expressed in the Mammary Gland as Splicing

Variants and Encodes Various Aberrant Truncated Proteins. Immunology Letters

83(1) (Aug. 1, 2002), 73–75. doi: 10.1016/S0165-2478(02)00084-6.

52. Elizaveta Fasler-Kan, Claudia Suenderhauf, Natasha Barteneva, Birk Poller,

Daniel Gygax, and J¨orgHuwyler: Cytokine Signaling in the Human Brain Capillary

Endothelial Cell Line hCMEC/D3. Brain Research 1354 (Oct. 1, 2010), 15–22. doi:

10.1016/j.brainres.2010.07.077.

53. Gerard Hernandez Mir, Jari Helin, Kari-Pekka Skarp, Richard D. Cummings,

Antti M¨akitie,Risto Renkonen, and Anne Lepp¨anen:Glycoforms of Human

Endothelial CD34 That Bind L-Selectin Carry Sulfated Sialyl Lewis x Capped O-

and N-Glycans. Blood 114(3) (July 16, 2009), 733–741. doi:

10.1182/blood-2009-03-210237.

54. Paula da Costa Martins, Juan-Jes´usGarc´ıa-Vallejo, Johannes V. van Thienen,

Mar Fernandez-Borja, Janine M. van Gils, Cora Beckers, Anton J. Horrevoets,

Peter L. Hordijk, and Jaap-Jan Zwaginga: P-Selectin Glycoprotein Ligand-1 Is

Expressed on Endothelial Cells and Mediates Monocyte Adhesion to Activated

Endothelium. Arteriosclerosis, Thrombosis, and Vascular Biology 27(5) (May

2007), 1023–1029. doi: 10.1161/ATVBAHA.107.140442.

55. E. E. Eriksson, X. Xie, J. Werr, P. Thoren, and L. Lindbom: Importance of

Primary Capture and L-Selectin-Dependent Secondary Capture in Leukocyte

Accumulation in Inﬂammation and Atherosclerosis in Vivo. The Journal of

Experimental Medicine 194(2) (July 16, 2001), 205–218.

117 56. Inga Osmers, Daniel C. Bullard, and Scott R. Barnum: PSGL-1 Is Not Required for

Development of Experimental Autoimmune Encephalomyelitis. Journal of

Neuroimmunology 166(1-2) (Sept. 2005), 193–196. doi:

10.1016/j.jneuroim.2005.06.001.

57. Britta Engelhardt, Birgit Kempe, Stephanie Merfeld-Clauss, Melanie Laschinger,

Bruce Furie, Martin K. Wild, and Dietmar Vestweber: P-Selectin Glycoprotein

Ligand 1 Is Not Required for the Development of Experimental Autoimmune

Encephalomyelitis in SJL and C57BL/6 Mice. Journal of Immunology (Baltimore,

Md.: 1950) 175(2) (July 15, 2005), 1267–1275.

58. Richard O. Hynes: Integrins: Bidirectional, Allosteric Signaling Machines. Cell

110(6) (Sept. 20, 2002), 673–687. doi: 10.1016/S0092-8674(02)00971-6.

59. Malgorzata Barczyk, Sergio Carracedo, and Donald Gullberg: Integrins. Cell and

Tissue Research 339(1) (Jan. 1, 2010), 269–280. doi: 10.1007/s00441-009-0834-6.

60. Barbara Rossi, Stefano Angiari, Elena Zenaro, Simona Luciana Budui, and

Gabriela Constantin: Vascular Inﬂammation in Central Nervous System Diseases:

Adhesion Receptors Controlling Leukocyte–Endothelial Interactions. Journal of

Leukocyte Biology 89(4) (Jan. 4, 2011), 539–556. doi: 10.1189/jlb.0710432.

61. J. Greenwood, S. J. Heasman, J. I. Alvarez, A. Prat, R. Lyck, and B. Engelhardt:

Review: Leucocyte–Endothelial Cell Crosstalk at the Blood–Brain Barrier: A

Prerequisite for Successful Immune Cell Entry to the Brain. Neuropathology and

Applied Neurobiology 37(1) (Feb. 1, 2011), 24–39. doi:

10.1111/j.1365-2990.2010.01140.x.

62. L. Bö,J. W. Peterson, S. Mørk, P. A. Hoffman, W. M. Gallatin, R. M. Ransohoff,

and B. D. Trapp: Distribution of Immunoglobulin Superfamily Members ICAM-1,

118 -2, -3, and the Beta 2 Integrin LFA-1 in Multiple Sclerosis Lesions. Journal of

Neuropathology and Experimental Neurology 55(10) (Oct. 1996), 1060–1072.

63. Romain Cayrol, Karolina Wosik, Jennifer L. Berard, Aurore Dodelet-Devillers,

Igal Ifergan, Hania Kebir, Arsalan S. Haqqani, Katharina Kreymborg,

Sebastian Krug, Robert Moumdjian, Alain Bouthillier, Burkhard Becher,

Nathalie Arbour, Samuel David, Danica Stanimirovic, and Alexandre Prat:

Activated Leukocyte Cell Adhesion Molecule Promotes Leukocyte Traﬃcking into

the Central Nervous System. Nature Immunology 9(2) (Feb. 2008), 137–145. doi:

10.1038/ni1551.

64. W. A. Muller, S. A. Weigl, X. Deng, and D. M. Phillips: PECAM-1 Is Required for

Transendothelial Migration of Leukocytes. The Journal of Experimental Medicine

178(2) (Jan. 8, 1993), 449–460. doi: 10.1084/jem.178.2.449.

65. Donald Wong and Katerina Dorovini-Zis: Upregulation of Intercellular Adhesion

Molecule-1 (ICAM-1) Expression in Primary Cultures of Human Brain Microvessel

Endothelial Cells by Cytokines and Lipopolysaccharide. Journal of

Neuroimmunology 39(1–2) (July 1992), 11–21. doi: 10.1016/0165-5728(92)90170-P.

66. James J. Campbell, Joseph Hedrick, Albert Zlotnik, Michael A. Siani,

Darren A. Thompson, and Eugene C. Butcher: Chemokines and the Arrest of

Lymphocytes Rolling Under Flow Conditions. Science 279(5349) (Jan. 16, 1998),

381–384. doi: 10.1126/science.279.5349.381.

67. Gabriela Constantin, Meytham Majeed, Cinzia Giagulli, Laura Piccio, Ji Yun Kim,

Eugene C Butcher, and Carlo Laudanna: Chemokines Trigger Immediate Β2

Integrin Aﬃnity and Mobility Changes: Diﬀerential Regulation and Roles in

119 Lymphocyte Arrest under Flow. Immunity 13(6) (Dec. 1, 2000), 759–769. doi:

10.1016/S1074-7613(00)00074-1.

68. Revital Shamri, Valentin Grabovsky, Jean-Marc Gauguet, Sara Feigelson,

Eugenia Manevich, Waldemar Kolanus, Martyn K. Robinson, Donald E. Staunton,

Ulrich H. von Andrian, and Ronen Alon: Lymphocyte Arrest Requires

Instantaneous Induction of an Extended LFA-1 Conformation Mediated by

Endothelium-Bound Chemokines. Nature Immunology 6(5) (May 2005), 497–506.

doi: 10.1038/ni1194.

69. Cedric Auﬀray, Darin Fogg, Meriem Garfa, Gaelle Elain, Olivier Join-Lambert,

Samer Kayal, Sabine Sarnacki, Ana Cumano, Gregoire Lauvau, and

Frederic Geissmann: Monitoring of Blood Vessels and Tissues by a Population of

Monocytes with Patrolling Behavior. Science 317(5838) (Mar. 8, 2007), 666–670.

doi: 10.1126/science.1142883.

70. Roser Gorina, Ruth Lyck, Dietmar Vestweber, and Britta Engelhardt: Β2

Integrin–Mediated Crawling on Endothelial ICAM-1 and ICAM-2 Is a Prerequisite

for Transcellular Neutrophil Diapedesis across the Inﬂamed Blood–Brain Barrier.

The Journal of Immunology 192(1) (Jan. 1, 2014), 324–337. doi:

10.4049/jimmunol.1300858.

71. Kari J. Dugger, Kurt R. Zinn, Casey Weaver, Daniel C. Bullard, and

Scott R. Barnum: Eﬀector and Suppressor Roles for LFA-1 during the Development

of Experimental Autoimmune Encephalomyelitis. Journal of Neuroimmunology

206(1-2) (Jan. 3, 2009), 22–27. doi: 10.1016/j.jneuroim.2008.10.006.

72. Veit Rothhammer, Sylvia Heink, Franziska Petermann, Rajneesh Srivastava,

Malte C. Claussen, Bernhard Hemmer, and Thomas Korn: Th17 Lymphocytes

120 Traﬃc to the Central Nervous System Independently of Α4 Integrin Expression

during EAE. The Journal of Experimental Medicine 208(12) (Nov. 21, 2011),

2465–2476. doi: 10.1084/jem.20110434.

73. Maha Ayyoub, Florence Deknuydt, Isabelle Raimbaud, Christelle Dousset,

Lucie Leveque, Gilles Bioley, and Danila Valmori: Human Memory FOXP3+ Tregs

Secrete IL-17 Ex Vivo and Constitutively Express the TH17 Lineage-Speciﬁc

Transcription Factor RORγt. Proceedings of the National Academy of Sciences

106(21) (May 26, 2009), 8635–8640. doi: 10.1073/pnas.0900621106.

74. Sandra G¨ultner,Tanja Kuhlmann, Amke Hesse, Jan P. Weber, Constanze Riemer,

Michael Baier, and Andreas Hutloﬀ: Reduced Treg Frequency in LFA-1-Deﬁcient

Mice Allows Enhanced T Eﬀector Diﬀerentiation and Pathology in EAE. European

Journal of Immunology 40(12) (Dec. 1, 2010), 3403–3412. doi:

10.1002/eji.201040576.

75. H. Akiyama and P. L. McGeer: Brain Microglia Constitutively Express Beta-2

Integrins. Journal of Neuroimmunology 30(1) (Nov. 1990), 81–93.

76. Romualdas Stapulionis, Cristiano Luis Pinto Oliveira, Mikkel Carstensen Gjelstrup,

Jan Skov Pedersen, Marianne Elisabet Hokland, Søren Vrønning Hoﬀmann,

Knud Poulsen, Christian Jacobsen, and Thomas Vorup-Jensen: Structural Insight

into the Function of Myelin Basic Protein as a Ligand for Integrin αMβ2. The

Journal of Immunology 180(6) (Mar. 15, 2008), 3946–3956. doi:

10.4049/jimmunol.180.6.3946.

77. Ryan A. Adams, Jan Bauer, Matthew J. Flick, Shoana L. Sikorski, Tal Nuriel,

Hans Lassmann, Jay L. Degen, and Katerina Akassoglou: The Fibrin-Derived

Gamma377-395 Peptide Inhibits Microglia Activation and Suppresses Relapsing

121 Paralysis in Central Nervous System Autoimmune Disease. The Journal of

Experimental Medicine 204(3) (Mar. 19, 2007), 571–582. doi:

10.1084/jem.20061931.

78. E. A. Wayner, A. Garcia-Pardo, M. J. Humphries, J. A. McDonald, and

W. G. Carter: Identiﬁcation and Characterization of the T Lymphocyte Adhesion

Receptor for an Alternative Cell Attachment Domain (CS-1) in Plasma

Fibronectin. The Journal of Cell Biology 109(3) (Sept. 1989), 1321–1330.

79. R. Alon, P. D. Kassner, M. W. Carr, E. B. Finger, M. E. Hemler, and

T. A. Springer: The Integrin VLA-4 Supports Tethering and Rolling in Flow on

VCAM-1. The Journal of Cell Biology 128(6) (Mar. 15, 1995), 1243–1253. doi:

10.1083/jcb.128.6.1243.

80. Jeﬀrey A. DiVietro, David C. Brown, Larry A. Sklar, Richard S. Larson, and

Michael B. Lawrence: Immobilized Stromal Cell-Derived Factor-1α Triggers Rapid

VLA-4 Aﬃnity Increases to Stabilize Lymphocyte Tethers on VCAM-1 and

Subsequently Initiate Firm Adhesion. The Journal of Immunology 178(6) (Mar. 15,

2007), 3903–3911. doi: 10.4049/jimmunol.178.6.3903.

81. Valentin Grabovsky, Sara Feigelson, Chun Chen, Diederik A. Bleijs, Amnon Peled,

Guy Cinamon, Francoise Baleux, Frenando Arenzana-Seisdedos, Tsvee Lapidot,

Yvette van Kooyk, Roy R. Lobb, and Ronen Alon: Subsecond Induction of Α4

Integrin Clustering by Immobilized Chemokines Stimulates Leukocyte Tethering

and Rolling on Endothelial Vascular Cell Adhesion Molecule 1 under Flow

Conditions. The Journal of Experimental Medicine 192(4) (Aug. 21, 2000),

495–506.

122 82. Peter Vajkoczy, Melanie Laschinger, and Britta Engelhardt: Α4-Integrin-VCAM-1

Binding Mediates G Protein-Independent Capture of Encephalitogenic T Cell

Blasts to CNS White Matter Microvessels. Journal of Clinical Investigation 108(4)

(Aug. 15, 2001), 557–565.

83. Ted A. Yednock, Catherine Cannon, Lawrence C. Fritz, Francisco Sanchez-Madrid,

Lawrence Steinman, and Nathan Karin: Prevention of Experimental Autoimmune

Encephalomyelitis by Antibodies against Α4βl Integrin. Nature 356(6364) (Mar. 5,

1992), 63–66. doi: 10.1038/356063a0.

84. R. F¨asslerand M. Meyer: Consequences of Lack of Beta 1 Integrin Gene Expression

in Mice. Genes & Development 9(15) (Jan. 8, 1995), 1896–1908. doi:

10.1101/gad.9.15.1896.

85. Stefan Brocke, Christopher Piercy, Lawrence Steinman, Irving L. Weissman, and

Timo Veromaa: Antibodies to CD44 and Integrin Α4, but Not L-Selectin, Prevent

Central Nervous System Inﬂammation and Experimental Encephalomyelitis by

Blocking Secondary Leukocyte Recruitment. Proceedings of the National Academy

of Sciences 96(12) (Aug. 6, 1999), 6896–6901. doi: 10.1073/pnas.96.12.6896.

86. George P.A. Rice, Hans-Peter Hartung, and Peter A. Calabresi: Anti-[Alpha]4

Integrin Therapy for Multiple Sclerosis: Mechanisms and Rationale. Neurology April

26, 2005 64(8) (2005), 1336–1342. doi: 10.1212/01.WNL.0000158329.30470.D0.

87. Chris H. Polman, Paul W. O’Connor, Eva Havrdova, Michael Hutchinson,

Ludwig Kappos, David H. Miller, J. Theodore Phillips, Fred D. Lublin,

Gavin Giovannoni, Andrzej Wajgt, Martin Toal, Frances Lynn, Michael A. Panzara,

and Alfred W. Sandrock: A Randomized, Placebo-Controlled Trial of Natalizumab

123 for Relapsing Multiple Sclerosis. New England Journal of Medicine 354(9) (2006),

899–910. doi: 10.1056/NEJMoa044397.

88. Albert Zlotnik, Osamu Yoshie, and Hisayuki Nomiyama: The Chemokine and

Chemokine Receptor Superfamilies and Their Molecular Evolution. Genome Biology

7(12) (Dec. 29, 2006), 243. doi: 10.1186/gb-2006-7-12-243.

89. Albert Zlotnik and Osamu Yoshie: The Chemokine Superfamily Revisited.

Immunity 36(5) (May 25, 2012), 705–716. doi: 10.1016/j.immuni.2012.05.008.

90. Hisayuki Nomiyama, Naoki Osada, and Osamu Yoshie: Systematic Classiﬁcation of

Vertebrate Chemokines Based on Conserved Synteny and Evolutionary History.

Genes to Cells 18(1) (Jan. 2013), 1–16. doi: 10.1111/gtc.12013.

91. David W. Holman, Robyn S. Klein, and Richard M. Ransohoﬀ: The Blood-Brain

Barrier, Chemokines and Multiple Sclerosis. Biochimica et biophysica acta 1812(2)

(Feb. 2011), 220–230. doi: 10.1016/j.bbadis.2010.07.019.

92. Borja L. Holgado, Laura Mart´ınez-Muñoz,Juan Antonio Sánchez-Alcañiz,

Pilar Lucas, Vicente Pérez-Garc´ıa,Gema Pérez,JoséMiguel Rodr´ıguez-Frade,

Marta Nieto, Oscar´ Mar´ın,Yolanda R. Carrasco, Ana C. Carrera,

Manuel Alvarez-Dolado,´ and Mario Mellado: CXCL12-Mediated Murine Neural

Progenitor Cell Movement Requires PI3Kβ Activation. Molecular Neurobiology

48(1) (Aug. 1, 2013), 217–231. doi: 10.1007/s12035-013-8451-5.

93. Bastian Sch¨onemeier,Angela Kolodziej, Stefan Schulz, Stefan Jacobs, Volker Hoellt,

and Ralf Stumm: Regional and Cellular Localization of the CXCl12/SDF-1

Chemokine Receptor CXCR7 in the Developing and Adult Rat Brain. The Journal

of Comparative Neurology 510(2) (Sept. 10, 2008), 207–220. doi: 10.1002/cne.21780.

124 94. Erzsebet Kokovay, Susan Goderie, Yue Wang, Steve Lotz, Gang Lin, Yu Sun,

Badrinath Roysam, Qin Shen, and Sally Temple: Adult SVZ Lineage Cells Home to

and Leave the Vascular Niche via Diﬀerential Responses to SDF1/CXCR4

Signaling. Cell Stem Cell 7(2) (Aug. 6, 2010), 163–173. doi:

10.1016/j.stem.2010.05.019.

95. Meizhang Li, Cathleen J. Chang, Justin D. Lathia, Li Wang, Holly L. Pacenta,

Anne Cotleur, and Richard M. Ransohoﬀ: Chemokine Receptor CXCR4 Signaling

Modulates the Growth Factor-Induced Cell Cycle of Self-Renewing and Multipotent

Neural Progenitor Cells. Glia 59(1) (Jan. 2011), 108–118. doi: 10.1002/glia.21080.

96. Kevin S. Carbajal, Juan L. Miranda, Michelle R. Tsukamoto, and Thomas E. Lane:

CXCR4 Signaling Regulates Remyelination by Endogenous Oligodendrocyte

Progenitor Cells in a Viral Model of Demyelination. Glia 59(12) (Dec. 1, 2011),

1813–1821. doi: 10.1002/glia.21225.

97. Bhaskar Saha, Sophie Peron, Kerren Murray, Mohamed Jaber, and

Afsaneh Gaillard: Cortical Lesion Stimulates Adult Subventricular Zone Neural

Progenitor Cell Proliferation and Migration to the Site of Injury. Stem Cell

Research 11(3) (Nov. 2013), 965–977. doi: 10.1016/j.scr.2013.06.006.

98. Jaime Imitola, Khadir Raddassi, Kook In Park, Franz-Josef Mueller, Marta Nieto,

Yang D. Teng, Dan Frenkel, Jianxue Li, Richard L. Sidman, Christopher A. Walsh,

Evan Y. Snyder, and Samia J. Khoury: Directed Migration of Neural Stem Cells to

Sites of CNS Injury by the Stromal Cell-Derived Factor 1alpha/CXC Chemokine

Receptor 4 Pathway. Proceedings of the National Academy of Sciences of the United

States of America 101(52) (Dec. 28, 2004), 18117–18122. doi:

10.1073/pnas.0408258102.

125 99. Markus Krumbholz, Diethilde Theil, Sabine Cepok, Bernhard Hemmer,

Pia Kivis¨akk,Richard M. Ransohoﬀ, Monika Hofbauer, Cinthia Farina,

Tobias Derfuss, Caroline Hartle, Jia Newcombe, Reinhard Hohlfeld, and

Edgar Meinl: Chemokines in Multiple Sclerosis: CXCL12 and CXCL13

up-Regulation Is Diﬀerentially Linked to CNS Immune Cell Recruitment. Brain

129(1) (Jan. 1, 2006), 200–211. doi: 10.1093/brain/awh680.

100. Erin E. McCandless, Laura Piccio, B. Mark Woerner, Robert E. Schmidt,

Joshua B. Rubin, Anne H. Cross, and Robyn S. Klein: Pathological Expression of

CXCL12 at the Blood-Brain Barrier Correlates with Severity of Multiple Sclerosis.

The American Journal of Pathology 172(3) (Mar. 2008), 799–808. doi:

10.2353/ajpath.2008.070918.

101. Erin E. McCandless, Matthew Budde, Jason R. Lees, Denise Dorsey, Eric Lyng,

and Robyn S. Klein: IL-1R Signaling within the Central Nervous System Regulates

CXCL12 Expression at the Blood-Brain Barrier and Disease Severity during

Experimental Autoimmune Encephalomyelitis. The Journal of Immunology 183(1)

(Jan. 7, 2009), 613–620. doi: 10.4049/jimmunol.0802258.

102. Lillian Cruz-Orengo, Ying-Jr Chen, Joong Kim, Denise Dorsey, Sheng-Kwei Song,

and Robyn S Klein: CXCR7 Antagonism Prevents Axonal Injury during

Experimental Autoimmune Encephalomyelitis as Revealed by in Vivo Axial

Diﬀusivity. Journal of Neuroinﬂammation 8(1) (2011), 170. doi:

10.1186/1742-2094-8-170.

103. Lillian Cruz-Orengo, David W. Holman, Denise Dorsey, Liang Zhou, Penglie Zhang,

Melissa Wright, Erin E. McCandless, Jigisha R. Patel, Gary D. Luker,

Dan R. Littman, John H. Russell, and Robyn S. Klein: CXCR7 Inﬂuences

126 Leukocyte Entry into the CNS Parenchyma by Controlling Abluminal CXCL12

Abundance during Autoimmunity. The Journal of Experimental Medicine 208(2)

(Feb. 14, 2011), 327–339. doi: 10.1084/jem.20102010.

104. Shumei Man, Barbara Tucky, Anne Cotleur, Judith Drazba, Yukio Takeshita, and

Richard M. Ransohoﬀ: CXCL12-Induced Monocyte-Endothelial Interactions

Promote Lymphocyte Transmigration Across an in Vitro Blood-Brain Barrier.

Science Translational Medicine 4(119) (Jan. 2, 2012), 119ra14–119ra14. doi:

10.1126/scitranslmed.3003197.

105. Silvia Buonamici, Thomas Trimarchi, Maria Grazia Ruocco, Linsey Reavie,

Severine Cathelin, Brenton G. Mar, Apostolos Klinakis, Yevgeniy Lukyanov,

Jen-Chieh Tseng, Filiz Sen, Eric Gehrie, Mengling Li, Elizabeth Newcomb,

Jiri Zavadil, Daniel Meruelo, Martin Lipp, Sherif Ibrahim, Argiris Efstratiadis,

David Zagzag, Jonathan S. Bromberg, Michael L. Dustin, and Iannis Aifantis:

CCR7 Signalling as an Essential Regulator of CNS Inﬁltration in T-Cell Leukaemia.

Nature 459(7249) (June 2009), 1000–1004. doi: 10.1038/nature08020.

106. Sandra Columba-Cabezas, Barbara Seraﬁni, Elena Ambrosini, and Francesca Aloisi:

Lymphoid Chemokines CCL19 and CCL21 Are Expressed in the Central Nervous

System during Experimental Autoimmune Encephalomyelitis: Implications for the

Maintenance of Chronic Neuroinﬂammation. Brain Pathology (Zurich, Switzerland)

13(1) (Jan. 2003), 38–51.

107. Carsten Alt, Melanie Laschinger, and Britta Engelhardt: Functional Expression of

the Lymphoid Chemokines CCL19 (ELC) and CCL 21 (SLC) at the Blood-Brain

Barrier Suggests Their Involvement in G-Protein-Dependent Lymphocyte

Recruitment into the Central Nervous System during Experimental Autoimmune

127 Encephalomyelitis. European Journal of Immunology 32(8) (Aug. 1, 2002),

2133–2144. doi:

10.1002/1521-4141(200208)32:8h2133::AID-IMMU2133i3.0.CO;2-W.

108. Pia Kivis¨akk,Don J. Mahad, Melissa K. Callahan, Keith Sikora, Corinna Trebst,

Barbara Tucky, Jerome Wujek, Rivka Ravid, Susan M. Staugaitis, Hans Lassmann,

and Richard M. Ransohoﬀ: Expression of CCR7 in Multiple Sclerosis: Implications

for CNS Immunity. Annals of Neurology 55(5) (May 1, 2004), 627–638. doi:

10.1002/ana.20049.

109. Jenny Corbisier, C´elineGal`es,Alexandre Huszagh, Marc Parmentier, and

Jean-Yves Springael: Biased Signaling at Chemokine Receptors. The Journal of

Biological Chemistry 290(15) (Apr. 10, 2015), 9542–9554. doi:

10.1074/jbc.M114.596098.

110. Anne Steen, Olav Larsen, Stefanie Thiele, and Mette M. Rosenkilde: Biased and G

Protein-Independent Signaling of Chemokine Receptors. Frontiers in Immunology 5

(June 23, 2014). doi: 10.3389/ﬁmmu.2014.00277.

111. Veysel Odemis, Karina Boosmann, Andr´eHeinen, Patrick K¨ury,and

J¨urgenEngele: CXCR7 Is an Active Component of SDF-1 Signalling in Astrocytes

and Schwann Cells. Journal of Cell Science 123 (Pt 7 Apr. 1, 2010), 1081–1088.

doi: 10.1242/jcs.062810.

112. Sudarshan Rajagopal, Jihee Kim, Seungkirl Ahn, Stewart Craig,

Christopher M. Lam, Norma P. Gerard, Craig Gerard, and Robert J. Lefkowitz:

β-Arrestin- but Not G Protein-Mediated Signaling by the “Decoy” Receptor

CXCR7. Proceedings of the National Academy of Sciences 107(2) (Dec. 1, 2010),

628–632. doi: 10.1073/pnas.0912852107.

128 113. Yanling Wang, Guangnan Li, Amelia Stanco, Jason E. Long, Dianna Crawford,

Gregory B. Potter, Samuel J. Pleasure, Timothy Behrens, and

John L. R. Rubenstein: CXCR4 and CXCR7 Have Distinct Functions in Regulating

Interneuron Migration. Neuron 69(1) (Jan. 13, 2011), 61–76. doi:

10.1016/j.neuron.2010.12.005.

114. Yang Liu, Eleanor Carson-Walter, and Kevin A. Walter: Chemokine Receptor

CXCR7 Is a Functional Receptor for CXCL12 in Brain Endothelial Cells. PLoS

ONE 9(8) (Aug. 1, 2014). Ed. by Rajesh Mohanraj, e103938. doi:

10.1371/journal.pone.0103938.

115. Xiaoqing Yan, Shaoxi Cai, Xin Xiong, Wei Sun, Xiaozhen Dai, Sijia Chen,

Qunfang Ye, Zhen Song, Qifeng Jiang, and Zhiling Xu: Chemokine Receptor

CXCR7 Mediates Human Endothelial Progenitor Cells Survival, Angiogenesis, but

Not Proliferation. Journal of Cellular Biochemistry 113(4) (Apr. 2012), 1437–1446.

doi: 10.1002/jcb.24015.

116. Xiaozhen Dai, Yi Tan, Shaoxi Cai, Xin Xiong, Lingqiao Wang, Qunfang Ye,

Xiaoqing Yan, Kaiwang Ma, and Lu Cai: The Role of CXCR7 on the Adhesion,

Proliferation and Angiogenesis of Endothelial Progenitor Cells. Journal Of Cellular

And Molecular Medicine 15(6) (June 2011), 1299–1309. doi:

10.1111/j.1582-4934.2011.01301.x.

117. Xiao-Yu Zhang, Chen Su, Zheng Cao, Shi-Yue Xu, Wen-Hao Xia, Wen-Li Xie,

Long Chen, Bing-Bo Yu, Bin Zhang, Yan Wang, and Jun Tao: CXCR7

Upregulation Is Required for Early Endothelial Progenitor Cell-Mediated

Endothelial Repair in Patients with Hypertension. Hypertension 63(2) (Feb. 2014),

383–389. doi: 10.1161/HYPERTENSIONAHA.113.02273.

129 118. Mark E. DeVries, Alyson A. Kelvin, Luoling Xu, Longsi Ran, John Robinson, and

David J. Kelvin: Deﬁning the Origins and Evolution of the Chemokine/Chemokine

Receptor System. The Journal of Immunology 176(1) (Jan. 1, 2006), 401–415. doi:

10.4049/jimmunol.176.1.401.

119. Yuichi Ikeda, Hidetoshi Kumagai, Amber Skach, Makito Sato, and

Masashi Yanagisawa: Modulation of Circadian Glucocorticoid Oscillation via

Adrenal Opioid-CXCR7 Signaling Alters Emotional Behavior. Cell 155(6) (Dec. 5,

2013), 1323–1336. doi: 10.1016/j.cell.2013.10.052.

120. Klaus Ley: Arrest Chemokines. Frontiers in Immunology 5 (Apr. 4, 2014). doi:

10.3389/ﬁmmu.2014.00150.

121. Ziv Shulman, Shmuel J. Cohen, Ben Roediger, Vyacheslav Kalchenko, Rohit Jain,

Valentin Grabovsky, Eugenia Klein, Vera Shinder, Liat Stoler-Barak,

Sara W. Feigelson, Tsipi Meshel, Susanna M. Nurmi, Itamar Goldstein,

Olivier Hartley, Carl G. Gahmberg, Amos Etzioni, Wolfgang Weninger,

Adit Ben-Baruch, and Ronen Alon: Transendothelial Migration of Lymphocytes

Mediated by Intraendothelial Vesicle Stores Rather than by Extracellular

Chemokine Depots. Nature Immunology 13(1) (Jan. 2012), 67–76. doi:

10.1038/ni.2173.

122. Maria Michela D’Aloia, Ilaria Grazia Zizzari, Benedetto Sacchetti, Luca Pierelli, and

Maurizio Alimandi: CAR-T Cells: The Long and Winding Road to Solid Tumors.

Cell Death & Disease 9(3) (Feb. 15, 2018), 282. doi: 10.1038/s41419-018-0278-6.

123. N. Schwab, J.C. Ulzheimer, R.J. Fox, T. Schneider-Hohendorf, B.C. Kieseier,

C.M. Monoranu, S.M. Staugaitis, W. Welch, S. Jilek, R.A. Du Pasquier, W. Br¨uck,

K.V. Toyka, R.M. Ransohoﬀ, and H. Wiendl: Fatal PML Associated with

130 Efalizumab Therapy. Neurology 78(7) (Feb. 14, 2012), 458–467. doi:

10.1212/WNL.0b013e3182478d4b.

124. Brian G. Feagan, Paul Rutgeerts, Bruce E. Sands, Stephen Hanauer,

Jean-FrédéricColombel, William J. Sandborn, Gert Van Assche, Jeffrey Axler,

Hyo-Jong Kim, Silvio Danese, Irving Fox, Catherine Milch, Serap Sankoh,

Tim Wyant, Jing Xu, and Asit Parikh: Vedolizumab as Induction and Maintenance

Therapy for Ulcerative Colitis. New England Journal of Medicine 369(8) (Aug. 22,

2013), 699–710. doi: 10.1056/NEJMoa1215734.

125. Tilman Schneider-Hohendorf, Jan Rossaint, Hema Mohan, Daniel B¨oning,

Johanna Breuer, Tanja Kuhlmann, Catharina C. Gross, Ken Flanagan,

Lydia Sorokin, Dietmar Vestweber, Alexander Zarbock, Nicholas Schwab, and

Heinz Wiendl: VLA-4 Blockade Promotes Diﬀerential Routes into Human CNS

Involving PSGL-1 Rolling of T Cells and MCAM-Adhesion of TH17 Cells. The

Journal of Experimental Medicine 211(9) (Aug. 25, 2014), 1833–1846. doi:

10.1084/jem.20140540.

126. M. Mehling, R. Lindberg, F. Raulf, J. Kuhle, C. Hess, L. Kappos, and

V. Brinkmann: Th17 Central Memory T Cells Are Reduced by FTY720 in Patients

with Multiple Sclerosis. Neurology 75(5) (Aug. 3, 2010), 403–410. doi:

10.1212/WNL.0b013e3181ebdd64.

127. Joseph R. Berger, Bruce A. Cree, Benjamin Greenberg, Bernhard Hemmer,

Brian J. Ward, Victor M. Dong, and Martin Merschhemke: Progressive Multifocal

Leukoencephalopathy after Fingolimod Treatment. Neurology 90(20) (May 15,

2018), e1815–e1821. doi: 10.1212/WNL.0000000000005529.

131 128. Bryan L. Benson, Lucy Li, Jay T. Myers, R. Dixon Dorand, Umut A. Gurkan,

Alex Y. Huang, and Richard M. Ransohoﬀ: Biomimetic Post-Capillary Venule

Expansions for Leukocyte Adhesion Studies. Scientiﬁc Reports 8(1) (June 19,

2018), 9328. doi: 10.1038/s41598-018-27566-z.

129. Aristotle G. Koutsiaris, Sophia V. Tachmitzi, Nick Batis, Maria G. Kotoula,

Constantinos H. Karabatsas, Evagelia Tsironi, and Dimitrios Z. Chatzoulis: Volume

Flow and Wall Shear Stress Quantiﬁcation in the Human Conjunctival Capillaries

and Post-Capillary Venules in Vivo. Biorheology 44(5) (Jan. 1, 2007), 375–386.

130. A. R. Pries, T. W. Secomb, T. Gessner, M. B. Sperandio, J. F. Gross, and

P. Gaehtgens: Resistance to Blood Flow in Microvessels in Vivo. Circulation

Research 75(5) (Nov. 1, 1994), 904–915. doi: 10.1161/01.RES.75.5.904.

131. G. R. Wang, Fang Yang, and Wei Zhao: There Can Be Turbulence in Microﬂuidics

at Low Reynolds Number. Lab on a Chip 14(8) (Apr. 21, 2014), 1452–1458. doi:

10.1039/c3lc51403j.

132. R Edgeworth, B. J. Dalton, and U. T. Parnell: The Pitch Drop Experiment.

European Journal of Physics 5(4) (Oct. 1984), 198.

133. Dmitry A. Fedosov and Gerhard Gompper: White Blood Cell Margination in

Microcirculation. Soft Matter 10(17) (May 7, 2014), 2961–2970. doi:

10.1039/c3sm52860j.

134. J. T. Finer, R. M. Simmons, and J. A. Spudich: Single Myosin Molecule Mechanics:

Piconewton Forces and Nanometre Steps. Nature 368(6467) (Mar. 10, 1994),

113–119. doi: 10.1038/368113a0.

135. Revathi Ananthakrishnan and Allen Ehrlicher: The Forces Behind Cell Movement.

International Journal of Biological Sciences 3(5) (June 1, 2007), 303–317.

132 136. E. P. W. Helps and D. A. McDonald: Observations on Laminar Flow in Veins. The

Journal of Physiology 124(3) (June 28, 1954), 631–639.

137. Frederic Sierro, Christine Biben, Laura Mart´ınez-Mu˜noz,Mario Mellado,

Richard M. Ransohoﬀ, Meizhang Li, Blanche Woehl, Helen Leung, Joanna Groom,

Marcel Batten, Richard P. Harvey, Carlos Mart´ınez-A,Charles R. Mackay, and

Fabienne Mackay: Disrupted Cardiac Development but Normal Hematopoiesis in

Mice Deﬁcient in the Second CXCL12/SDF-1 Receptor, CXCR7. Proceedings of the

National Academy of Sciences 104(37) (Nov. 9, 2007), 14759–14764. doi:

10.1073/pnas.0702229104.

138. Hao Zheng, Guosheng Fu, Tao Dai, and He Huang: Migration of Endothelial

Progenitor Cells Mediated by Stromal Cell-Derived Factor-1alpha/CXCR4 via

PI3K/Akt/eNOS Signal Transduction Pathway. Journal of Cardiovascular

Pharmacology 50(3) (Sept. 2007), 274–280. doi: 10.1097/FJC.0b013e318093ec8f.

139. Y. R. Zou, A. H. Kottmann, M. Kuroda, I. Taniuchi, and D. R. Littman: Function

of the Chemokine Receptor CXCR4 in Haematopoiesis and in Cerebellar

Development. Nature 393(6685) (June 11, 1998), 595–599. doi: 10.1038/31269.

140. Qing Ma, Dan Jones, Paul R. Borghesani, Rosalind A. Segal, Takashi Nagasawa,

Tadamitsu Kishimoto, Roderick T. Bronson, and Timothy A. Springer: Impaired

B-Lymphopoiesis, Myelopoiesis, and Derailed Cerebellar Neuron Migration in

CXCR4- and SDF-1-Deﬁcient Mice. Proceedings of the National Academy of

Sciences 95(16) (Apr. 8, 1998), 9448–9453.

141. Toshiaki Ara, Yuri Nakamura, Takeshi Egawa, Tatsuki Sugiyama, Kuniya Abe,

Tadamitsu Kishimoto, Yasuhisa Matsui, and Takashi Nagasawa: Impaired

Colonization of the Gonads by Primordial Germ Cells in Mice Lacking a

133 Chemokine, Stromal Cell-Derived Factor-1 (SDF-1). Proceedings of the National

Academy of Sciences of the United States of America 100(9) (Apr. 29, 2003),

5319–5323. doi: 10.1073/pnas.0730719100.

142. Tatsuki Sugiyama, Hiroshi Kohara, Mamiko Noda, and Takashi Nagasawa:

Maintenance of the Hematopoietic Stem Cell Pool by CXCL12-CXCR4 Chemokine

Signaling in Bone Marrow Stromal Cell Niches. Immunity 25(6) (Dec. 2006),

977–988. doi: 10.1016/j.immuni.2006.10.016.

143. Jason Plotkin, Susan E. Prockop, Ana Lepique, and Howard T. Petrie: Critical Role

for CXCR4 Signaling in Progenitor Localization and T Cell Diﬀerentiation in the

Postnatal Thymus. Journal of Immunology (Baltimore, Md.: 1950) 171(9) (Nov. 1,

2003), 4521–4527.

144. Hisayuki Nomiyama, Naoki Osada, and Osamu Yoshie: A Family Tree of Vertebrate

Chemokine Receptors for a Uniﬁed Nomenclature. Developmental and Comparative

Immunology 35(7) (July 2011), 705–715. doi: 10.1016/j.dci.2011.01.019.

145. Baubak Bajoghli: Evolution and Function of Chemokine Receptors in the Immune

System of Lower Vertebrates. European Journal of Immunology 43(7) (),

1686–1692. doi: 10.1002/eji.201343557.

146. Martin F. Flajnik and Masanori Kasahara: Origin and Evolution of the Adaptive

Immune System: Genetic Events and Selective Pressures. Nature reviews. Genetics

11(1) (Jan. 2010), 47–59. doi: 10.1038/nrg2703.

147. Marc Baj´enoﬀ,Jackson G. Egen, Hai Qi, Alex Y. C. Huang, Flora Castellino, and

Ronald N. Germain: Highways, Byways and Breadcrumbs: Directing Lymphocyte

Traﬃc in the Lymph Node. Trends in Immunology 28(8) (Aug. 2007), 346–352. doi:

10.1016/j.it.2007.06.005.

134 148. Leonard T. Nguyen and Hans J. Vogel: Structural Perspectives on Antimicrobial

Chemokines. Frontiers in Immunology 3 (2012), 384. doi:

10.3389/ﬁmmu.2012.00384.

149. Xingfeng Bao, E. Ashley Moseman, Hideo Saito, Bronislawa Petryanik,

Aude Thiriot, Shingo Hatakeyama, Yuki Ito, Hiroto Kawashima, Yu Yamaguchi,

John B. Lowe, Ulrich H. von Andrian, and Minoru Fukuda: Endothelial Heparan

Sulfate Controls Chemokine Presentation in Recruitment of Lymphocytes and

Dendritic Cells to Lymph Nodes. Immunity 33(5) (Nov. 24, 2010), 817–829. doi:

10.1016/j.immuni.2010.10.018.

150. A. J. Hoogewerf, G. S. Kuschert, A. E. Proudfoot, F. Borlat, I. Clark-Lewis,

C. A. Power, and T. N. Wells: Glycosaminoglycans Mediate Cell Surface

Oligomerization of Chemokines. Biochemistry 36(44) (Nov. 4, 1997), 13570–13578.

doi: 10.1021/bi971125s.

151. Romain R. Viv`es,Rabia Sadir, Anne Imberty, Anna Rencurosi, and

Hugues Lortat-Jacob: A Kinetics and Modeling Study of RANTES(9-68) Binding

to Heparin Reveals a Mechanism of Cooperative Oligomerization. Biochemistry

41(50) (Dec. 1, 2002), 14779–14789. doi: 10.1021/bi026459i.

152. Tracy M. Handel, Zo¨eJohnson, David H. Rodrigues, Adriana C. Dos Santos,

Rocco Cirillo, Valeria Muzio, Simona Riva, Matthias Mack, Maud D´eruaz,

Fr´ed´ericBorlat, Pierre-Alain Vitte, Timothy N. C. Wells, Mauro M. Teixeira, and

Amanda E. I. Proudfoot: An Engineered Monomer of CCL2 Has Anti-Inﬂammatory

Properties Emphasizing the Importance of Oligomerization for Chemokine Activity

in Vivo. Journal of Leukocyte Biology 84(4) (Oct. 2008), 1101–1108. doi:

10.1189/jlb.0108061.

135 153. Gabriele S. V. Campanella, Jan Grimm, Lindsay A. Manice, Richard A. Colvin,

Benjamin D. Medoﬀ, Gregory R. Wojtkiewicz, Ralph Weissleder, and

Andrew D. Luster: Oligomerization of CXCL10 Is Necessary for Endothelial Cell

Presentation and In Vivo Activity. The Journal of Immunology 177(10) (Nov. 15,

2006), 6991–6998. doi: 10.4049/jimmunol.177.10.6991.

154. Amanda E. I. Proudfoot, Tracy M. Handel, Zo¨eJohnson, Elaine K. Lau,

Patricia LiWang, Ian Clark-Lewis, Fr´ed´ericBorlat, Timothy N. C. Wells, and

Marie H. Kosco-Vilbois: Glycosaminoglycan Binding and Oligomerization Are

Essential for the in Vivo Activity of Certain Chemokines. Proceedings of the

National Academy of Sciences 100(4) (Feb. 18, 2003), 1885–1890. doi:

10.1073/pnas.0334864100.

155. Xu Wang, Caroline Watson, Joshua S. Sharp, Tracy M. Handel, and

James H. Prestegard: Oligomeric Structure of the Chemokine CCL5/RANTES from

NMR, MS, and SAXS Data. Structure 19(8) (Aug. 10, 2011), 1138–1148. doi:

10.1016/j.str.2011.06.001.

156. Guy Cinamon, Vera Shinder, and Ronen Alon: Shear Forces Promote Lymphocyte

Migration across Vascular Endothelium Bearing Apical Chemokines. Nature

Immunology 2(6) (June 2001), 515–522. doi: 10.1038/88710.

157. Guy Cinamon, Vera Shinder, Revital Shamri, and Ronen Alon: Chemoattractant

Signals and Beta 2 Integrin Occupancy at Apical Endothelial Contacts Combine

with Shear Stress Signals to Promote Transendothelial Neutrophil Migration.

Journal of Immunology (Baltimore, Md.: 1950) 173(12) (Dec. 15, 2004), 7282–7291.

158. Thomas Baltus, Kim S. C. Weber, Zo¨eJohnson, Amanda E. I. Proudfoot, and

Christian Weber: Oligomerization of RANTES Is Required for CCR1-Mediated

136 Arrest but Not CCR5-Mediated Transmigration of Leukocytes on Inﬂamed

Endothelium. Blood 102(6) (Sept. 15, 2003), 1985–1988. doi:

10.1182/blood-2003-04-1175.

159. James W. Murphy, Hua Yuan, Yong Kong, Yong Xiong, and Elias J. Lolis:

Heterologous Quaternary Structure of CXCL12 and Its Relationship to the CC

Chemokine Family. Proteins 78(5) (Apr. 2010), 1331–1337. doi: 10.1002/prot.22666.

160. Philipp von Hundelshausen, Stijn M. Agten, Veit Eckardt, Xavier Blanchet,

Martin M. Schmitt, Hans Ippel, Carlos Neideck, Kiril Bidzhekov,

Julian Leberzammer, Kanin Wichapong, Alexander Faussner, Maik Drechsler,

Jochen Grommes, Johanna P. van Geﬀen, He Li, Almudena Ortega-Gomez,

Remco T. A. Megens, Ronald Naumann, Ingrid Dijkgraaf, Gerry A. F. Nicolaes,

Yvonne D¨oring,Oliver Soehnlein, Esther Lutgens, Johan W. M. Heemskerk,

Rory R. Koenen, Kevin H. Mayo, Tilman M. Hackeng, and Christian Weber:

Chemokine Interactome Mapping Enables Tailored Intervention in Acute and

Chronic Inﬂammation. Science Translational Medicine 9(384) (Apr. 5, 2017),

eaah6650. doi: 10.1126/scitranslmed.aah6650.

161. Braedon McDonald, Keir Pittman, Gustavo B. Menezes, Simon A. Hirota,

Ingrid Slaba, Christopher C. M. Waterhouse, Paul L. Beck, Daniel A. Muruve, and

Paul Kubes: Intravascular Danger Signals Guide Neutrophils to Sites of Sterile

Inﬂammation. Science 330(6002) (Oct. 15, 2010), 362–366. doi:

10.1126/science.1195491.

162. Sara Massena, Gustaf Christoﬀersson, Elina Hjertstr¨om, Eyal Zcharia,

Israel Vlodavsky, Nora Ausmees, Charlotte Rolny, Jin-Ping Li, and Mia Phillipson:

A Chemotactic Gradient Sequestered on Endothelial Heparan Sulfate Induces

137 Directional Intraluminal Crawling of Neutrophils. Blood 116(11) (Sept. 16, 2010),

1924–1931. doi: 10.1182/blood-2010-01-266072.

163. Hemant Sarin: Physiologic Upper Limits of Pore Size of Diﬀerent Blood Capillary

Types and Another Perspective on the Dual Pore Theory of Microvascular

Permeability. Journal of Angiogenesis Research 2 (Aug. 11, 2010), 14. doi:

10.1186/2040-2384-2-14.

164. A. Hirata, P. Baluk, T. Fujiwara, and D. M. McDonald: Location of Focal Silver

Staining at Endothelial Gaps in Inﬂamed Venules Examined by Scanning Electron

Microscopy. American Journal of Physiology - Lung Cellular and Molecular

Physiology 269(3) (Sept. 1, 1995), L403–L418.

165. Shuyan Zheng, Ying-Ying Bai, Yinzhi Changyi, Xihui Gao, Wenqing Zhang,

Yuancheng Wang, Lu Zhou, Shenghong Ju, and Cong Li: Multimodal Nanoprobes

Evaluating Physiological Pore Size of Brain Vasculatures in Ischemic Stroke Models.

Advanced Healthcare Materials 3(11) (Nov. 1, 2014), 1909–1918. doi:

10.1002/adhm.201400159.

166. Elizabeth Nance, Fan Zhang, Manoj K. Mishra, Zhi Zhang, Siva P. Kambhampati,

Rangaramanujam M. Kannan, and Sujatha Kannan: Nanoscale Eﬀects in

Dendrimer-Mediated Targeting of Neuroinﬂammation. Biomaterials 101 (Sept.

2016), 96–107. doi: 10.1016/j.biomaterials.2016.05.044.

167. Jim Middleton, Angela M. Patterson, Lucy Gardner, Caroline Schmutz, and

Brian A. Ashton: Leukocyte Extravasation: Chemokine Transport and Presentation

by the Endothelium. Blood 100(12) (Dec. 1, 2002), 3853–3860. doi:

10.1182/blood.V100.12.3853.

138 168. Christoph Alexander Kasper, Isabel Sorg, Christoph Schmutz, Therese Tschon,

Harry Wischnewski, Man Lyang Kim, and C´ecileArrieumerlou: Cell-Cell

Propagation of NF-κB Transcription Factor and MAP Kinase Activation Ampliﬁes

Innate Immunity against Bacterial Infection. Immunity 33(5) (Nov. 24, 2010),

804–816. doi: 10.1016/j.immuni.2010.10.015.

169. Andrea Ablasser, Jonathan L. Schmid-Burgk, Inga Hemmerling, Gabor L. Horvath,

Tobias Schmidt, Eicke Latz, and Veit Hornung: Cell Intrinsic Immunity Spreads to

Bystander Cells via the Intercellular Transfer of cGAMP. Nature 503(7477)

(Nov. 28, 2013), 530–534. doi: 10.1038/nature12640.

170. Juexuan Long, Michael Junkin, Pak Kin Wong, James Hoying, and Pierre Deymier:

Calcium Wave Propagation in Networks of Endothelial Cells: Model-Based

Theoretical and Experimental Study. PLoS Comput Biol 8(12) (Dec. 27, 2012),

e1002847. doi: 10.1371/journal.pcbi.1002847.

171. Yasuteru Sano, Fumitaka Shimizu, Masaaki Abe, Toshihiko Maeda,

Yoko Kashiwamura, Sumio Ohtsuki, Tetsuya Terasaki, Masuo Obinata,

Koji Kajiwara, Masami Fujii, Michiyasu Suzuki, and Takashi Kanda: Establishment

of a New Conditionally Immortalized Human Brain Microvascular Endothelial Cell

Line Retaining an in Vivo Blood-Brain Barrier Function. Journal of Cellular

Physiology 225(2) (Nov. 2010), 519–528. doi: 10.1002/jcp.22232.

172. Sava¸sTay, Jacob J. Hughey, Timothy K. Lee, Tomasz Lipniacki,

Stephen R. Quake, and Markus W. Covert: Single-Cell NF-kappaB Dynamics

Reveal Digital Activation and Analogue Information Processing. Nature 466(7303)

(July 8, 2010), 267–271. doi: 10.1038/nature09145.

139 173. Frederic Geissmann, Thomas O. Cameron, Stephane Sidobre, Natasha Manlongat,

Mitchell Kronenberg, Michael J. Briskin, Michael L. Dustin, and Dan R. Littman:

Intravascular Immune Surveillance by CXCR6+ NKT Cells Patrolling Liver

Sinusoids. PLOS Biology 3(4) (Apr. 5, 2005), e113. doi:

10.1371/journal.pbio.0030113.

174. Mia Phillipson, Bryan Heit, Sean A. Parsons, Bj¨ornPetri, Sarah C. Mullaly,

Pina Colarusso, R. Michael Gower, Gregory Neely, Scott I. Simon, and Paul Kubes:

Vav1 Is Essential for Mechanotactic Crawling and Migration of Neutrophils out of

the Inﬂamed Microvasculature. The Journal of Immunology 182(11) (June 1, 2009),

6870–6878. doi: 10.4049/jimmunol.0803414.

175. Lorena Sánchez-Mart´ın, Noelia Sánchez-Sánchez, M. Dolores Gutiérrez-López,

Ana I. Rojo, Miguel Vicente-Manzanares, Mar´ıaJos´eP´erez-Alvarez,

Paloma S´anchez-Mateos, Xos´eR. Bustelo, Antonio Cuadrado,

Francisco Sánchez-Madrid, JoséLuis Rodr´ıguez-Fernández, and Carlos Cabañas:

Signaling through the Leukocyte Integrin LFA-1 in T Cells Induces a Transient

Activation of Rac-1 That Is Regulated by Vav and PI3K/Akt-1. The Journal of

Biological Chemistry 279(16) (Apr. 16, 2004), 16194–16205. doi:

10.1074/jbc.M400905200.

176. Jan M. Herter, Jan Rossaint, Helena Block, Heidi Welch, and Alexander Zarbock:

Integrin Activation by P-Rex1 Is Required for Selectin-Mediated Slow Leukocyte

Rolling and Intravascular Crawling. Blood 121(12) (Mar. 21, 2013), 2301–2310. doi:

10.1182/blood-2012-09-457085.

177. Asako Itakura, Joseph E. Aslan, Branden T. Kusanto, Kevin G. Phillips,

Juliana E. Porter, Paul K. Newton, Xiaolin Nan, Robert H. Insall,

140 Jonathan Chernoﬀ, and Owen J. T. McCarty: P21-Activated Kinase (PAK)

Regulates Cytoskeletal Reorganization and Directional Migration in Human

Neutrophils. PloS One 8(9) (2013), e73063. doi: 10.1371/journal.pone.0073063.

178. Cinzia Giagulli, Elio Scarpini, Linda Ottoboni, Shuh Narumiya, Eugene C Butcher,

Gabriela Constantin, and Carlo Laudanna: RhoA and ζ PKC Control Distinct

Modalities of LFA-1 Activation by Chemokines: Critical Role of LFA-1 Aﬃnity

Triggering in Lymphocyte In Vivo Homing. Immunity 20(1) (Jan. 1, 2004), 25–35.

doi: 10.1016/S1074-7613(03)00350-9.

179. Noah Fine, Ioannis D. Dimitriou, Jacob Rullo, Mar´ıaJos´eSand´ı,Bj¨ornPetri,

Jack Haitsma, Hisham Ibrahim, Jose La Rose, Michael Glogauer, Paul Kubes,

Myron Cybulsky, and Robert Rottapel: GEF-H1 Is Necessary for Neutrophil Shear

Stress–Induced Migration during Inﬂammation. J Cell Biol 215(1) (Oct. 10, 2016),

107–119. doi: 10.1083/jcb.201603109.

180. Vinatha Sreeramkumar, Jos´eM. Adrover, Ivan Ballesteros, Maria Isabel Cuartero,

Jan Rossaint, Izaskun Bilbao, Maria N´acher, Christophe Pitaval,

Irena Radovanovic, Yoshinori Fukui, Rodger P. McEver, Marie-Dominique Filippi,

Ignacio Lizasoain, Jes´usRuiz-Cabello, Alexander Zarbock, Mar´ıaA. Moro, and

Andr´esHidalgo: Neutrophils Scan for Activated Platelets to Initiate Inﬂammation.

Science 346(6214) (Dec. 5, 2014), 1234–1238. doi: 10.1126/science.1256478.

181. Ziv Shulman, Vera Shinder, Eugenia Klein, Valentin Grabovsky, Orna Yeger,

Erez Geron, Alessio Montresor, Matteo Bolomini-Vittori, Sara W. Feigelson,

Tomas Kirchhausen, Carlo Laudanna, Guy Shakhar, and Ronen Alon: Lymphocyte

Crawling and Transendothelial Migration Require Chemokine Triggering of

141 High-Aﬃnity LFA-1 Integrin. Immunity 30(3) (Mar. 20, 2009), 384–396. doi:

10.1016/j.immuni.2008.12.020.

182. Marie-Pierre Valignat, Olivier Theodoly, Alexia Gucciardi, Nancy Hogg, and

Annemarie C. Lellouch: T Lymphocytes Orient against the Direction of Fluid Flow

during LFA-1-Mediated Migration. Biophysical Journal 104(2) (Jan. 22, 2013),

322–331. doi: 10.1016/j.bpj.2012.12.007.

183. Andrew Smith, Yolanda R. Carrasco, Paula Stanley, Nelly Kieﬀer,

Facundo D. Batista, and Nancy Hogg: A Talin-Dependent LFA-1 Focal Zone Is

Formed by Rapidly Migrating T Lymphocytes. The Journal of Cell Biology 170(1)

(July 4, 2005), 141–151. doi: 10.1083/jcb.200412032.

184. Eun Jeong Park, Ant´onioPeixoto, Yoichi Imai, Ahmad Goodarzi, Guiying Cheng,

Christopher V. Carman, Ulrich H. von Andrian, and Motomu Shimaoka: Distinct

Roles for LFA-1 Aﬃnity Regulation during T-Cell Adhesion, Diapedesis, and

Interstitial Migration in Lymph Nodes. Blood 115(8) (Feb. 25, 2010), 1572–1581.

doi: 10.1182/blood-2009-08-237917.

185. Oliver Steiner, Caroline Coisne, Rom´eoCecchelli, R´emy Boscacci, Urban Deutsch,

Britta Engelhardt, and Ruth Lyck: Diﬀerential Roles for Endothelial ICAM-1,

ICAM-2, and VCAM-1 in Shear-Resistant T Cell Arrest, Polarization, and Directed

Crawling on Blood–Brain Barrier Endothelium. The Journal of Immunology

(Sept. 20, 2010), 0903732. doi: 10.4049/jimmunol.0903732.

186. Mika Shimonaka, Koko Katagiri, Toshinori Nakayama, Naoya Fujita,

Takashi Tsuruo, Osamu Yoshie, and Tatsuo Kinashi: Rap1 Translates Chemokine

Signals to Integrin Activation, Cell Polarization, and Motility across Vascular

142 Endothelium under Flow. The Journal of Cell Biology 161(2) (Apr. 28, 2003),

417–427. doi: 10.1083/jcb.200301133.

187. Mustapha Faroudi, Miroslav Hons, Agnieszka Zachacz, Celine Dumont, Ruth Lyck,

Jens V. Stein, and Victor L. J. Tybulewicz: Critical Roles for Rac GTPases in T

Cell Migration to and within Lymph Nodes. Blood (Jan. 1, 2010), 5536–5547. doi:

10.1182/blood-2010-08-299438.

188. Sarah J. Heasman, Leo M. Carlin, Susan Cox, Tony Ng, and Anne J. Ridley:

Coordinated RhoA Signaling at the Leading Edge and Uropod Is Required for T

Cell Transendothelial Migration. The Journal of Cell Biology 190(4) (Aug. 23,

2010), 553–563. doi: 10.1083/jcb.201002067.

189. Haifa Ghandour, Xavier Cullere, Angeles Alvarez, Francis W. Luscinskas, and

Tanya N. Mayadas: Essential Role for Rap1 GTPase and Its Guanine Exchange

Factor CalDAG-GEFI in LFA-1 but Not VLA-4 Integrin Mediated Human T-Cell

Adhesion. Blood 110(10) (Nov. 15, 2007), 3682–3690. doi:

10.1182/blood-2007-03-077628.

190. Takashi Kei Kishimoto, Nurit Hollander, Thomas M. Roberts, Donald C. Anderson,

and Timothy A. Springer: Heterogeneous Mutations in the β Subunit Common to

the LFA-1, Mac-1, and P150,95 Glycoproteins Cause Leukocyte Adhesion

Deﬁciency. Cell 50(2) (July 17, 1987), 193–202. doi: 10.1016/0092-8674(87)90215-7.

191. Donald C. Anderson, Frank C. Schmalsteig, Milton J. Finegold, Bonnie J. Hughes,

Robert Rothlein, Linda J. Miller, Steve Kohl, Michael F. Tosi, Robert L. Jacobs,

Thomas C. Waldrop, Armond S. Goldman, William T. Shearer, and

Timothy A. Springer: The Severe and Moderate Phenotypes of Heritable Mac-1,

LFA-1 Deﬁciency: Their Quantitative Deﬁnition and Relation to Leukocyte

143 Dysfunction and Clinical Features. The Journal of Infectious Diseases 152(4)

(Oct. 1, 1985), 668–689. doi: 10.1093/infdis/152.4.668.

192. M. Amin Arnaout, Jane Pitt, Harvey J. Cohen, Julian Melamed, Fred S. Rosen,

and Harvey R. Colten: Deﬁciency of a Granulocyte-Membrane Glycoprotein

(Gp150) in a Boy with Recurrent Bacterial Infections. New England Journal of

Medicine 306(12) (Mar. 25, 1982), 693–699. doi: 10.1056/NEJM198203253061201.

193. Amos Etzioni, Moshe Frydman, Shimon Pollack, Israeli Avidor, M. Laurie Phillips,

James C. Paulson, and Ruth Gershoni-Baruch: Recurrent Severe Infections Caused

by a Novel Leukocyte Adhesion Deﬁciency. New England Journal of Medicine

327(25) (Dec. 17, 1992), 1789–1792. doi: 10.1056/NEJM199212173272505.

194. Lena Svensson, Kimberley Howarth, Alison McDowall, Irene Patzak, Rachel Evans,

Siegfried Ussar, Markus Moser, Ayse Metin, Mike Fried, Ian Tomlinson, and

Nancy Hogg: Leukocyte Adhesion Deﬁciency-III Is Caused by Mutations in

KINDLIN3 Aﬀecting Integrin Activation. Nature medicine 15(3) (Mar. 2009),

306–312. doi: 10.1038/nm.1931.

195. Nikolay L Malinin, Li Zhang, Jeongsuk Choi, Alieta Ciocea, Olga Razorenova,

Yan-Qing Ma, Eugene A Podrez, Michael Tosi, Donald P Lennon, Arnold I Caplan,

Susan B Shurin, Edward F Plow, and Tatiana V Byzova: A Point Mutation in

KINDLIN3 Ablates Activation of Three Integrin Subfamilies in Humans. Nature

medicine 15(3) (Mar. 2009), 313–318. doi: 10.1038/nm.1917.

196. I. Joris, T. Zand, J. J. Nunnari, F. J. Krolikowski, and G. Majno: Studies on the

Pathogenesis of Atherosclerosis. I. Adhesion and Emigration of Mononuclear Cells

in the Aorta of Hypercholesterolemic Rats. The American Journal of Pathology

113(3) (Dec. 1983), 341–358.

144 197. M S Mulligan, M J Polley, R J Bayer, M F Nunn, J C Paulson, and P A Ward:

Neutrophil-Dependent Acute Lung Injury. Requirement for P-Selectin (GMP-140).

Journal of Clinical Investigation 90(4) (Oct. 1992), 1600–1607.

198. D.K. Kaul and R.P. Hebbel: Hypoxia/Reoxygenation Causes Inﬂammatory

Response in Transgenic Sickle Mice but Not in Normal Mice. Journal of Clinical

Investigation 106(3) (Aug. 1, 2000), 411–420.

199. Eugene C. Butcher and Louis J. Picker: Lymphocyte Homing and Homeostasis.

Science 272(5258) (1996), 60–66.

200. Jason G. Cyster: Chemokines and Cell Migration in Secondary Lymphoid Organs.

Science 286(5447) (Dec. 10, 1999), 2098–2102. doi: 10.1126/science.286.5447.2098.

201. Mar´ıaCasanova-Acebes, Christophe Pitaval, Linnea A. Weiss,

CésarNombela-Arrieta, RaphaëlChèvre,Noelia A-González, Yuya Kunisaki,

Dachuan Zhang, Nico van Rooijen, Leslie E. Silberstein, Christian Weber,

Takashi Nagasawa, Paul S. Frenette, Antonio Castrillo, and Andr´esHidalgo:

Rhythmic Modulation of the Hematopoietic Niche through Neutrophil Clearance.

Cell 153(5) (May 23, 2013), 1025–1035. doi: 10.1016/j.cell.2013.04.040.

202. Sibylle von Vietinghoﬀ and Klaus Ley: Homeostatic Regulation of Blood Neutrophil

Counts. Journal of immunology (Baltimore, Md. : 1950) 181(8) (Oct. 15, 2008),

5183–5188.

203. Francois E. Mercier, Christine Ragu, and David T. Scadden: The Bone Marrow at

the Crossroads of Blood and Immunity. Nature reviews. Immunology 12(1)

(Dec. 23, 2011), 49–60. doi: 10.1038/nri3132.

145 204. Carlo Laudanna, Ji Yun Kim, Gabriela Constantin, and Eugene C. Butcher: Rapid

Leukocyte Integrin Activation by Chemokines. Immunological Reviews 186(1)

(Aug. 1, 2002), 37–46. doi: 10.1034/j.1600-065X.2002.18604.x.

205. S. D. House and H. H. Lipowsky: Leukocyte-Endothelium Adhesion:

Microhemodynamics in Mesentery of the Cat. Microvascular Research 34(3) (Nov.

1987), 363–379.

206. S. D. House and H. H. Lipowsky: In Vivo Determination of the Force of

Leukocyte-Endothelium Adhesion in the Mesenteric Microvasculature of the Cat.

Circulation Research 63(3) (Sept. 1, 1988), 658–668. doi: 10.1161/01.RES.63.3.658.

207. K. Ley and P. Gaehtgens: Endothelial, Not Hemodynamic, Diﬀerences Are

Responsible for Preferential Leukocyte Rolling in Rat Mesenteric Venules.

Circulation Research 69(4) (Oct. 1, 1991), 1034–1041. doi:

10.1161/01.RES.69.4.1034.

208. Y. Iigo, M. Suematsu, T. Higashida, J. Oheda, K. Matsumoto, Y. Wakabayashi,

Y. Ishimura, M. Miyasaka, and T. Takashi: Constitutive Expression of ICAM-1 in

Rat Microvascular Systems Analyzed by Laser Confocal Microscopy. The American

Journal of Physiology 273 (1 Pt 2 July 1997), H138–147.

209. J. S. Pober, M. A. Gimbrone, L. A. Lapierre, D. L. Mendrick, W. Fiers,

R. Rothlein, and T. A. Springer: Overlapping Patterns of Activation of Human

Endothelial Cells by Interleukin 1, Tumor Necrosis Factor, and Immune Interferon.

The Journal of Immunology 137(6) (Sept. 15, 1986), 1893–1896.

210. Ursula Gotsch, Ute J¨ager,Mara Dominis, and Dietmar Vestweber: Expression of

P-Selectin on Endothelial Cells Is Upregulated by LPS and TNF-α in Vivo. Cell

146 Adhesion and Communication 2(1) (Jan. 1, 1994), 7–14. doi:

10.3109/15419069409014198.

211. M. B. Furie and G. J. Randolph: Chemokines and Tissue Injury. The American

Journal of Pathology 146(6) (June 1995), 1287–1301.

212. J. A. Royall, R. L. Berkow, J. S. Beckman, M. K. Cunningham, S. Matalon, and

B. A. Freeman: Tumor Necrosis Factor and Interleukin 1 Alpha Increase Vascular

Endothelial Permeability. American Journal of Physiology - Lung Cellular and

Molecular Physiology 257(6) (Dec. 1, 1989), L399–L410.

213. Catherine T. K. Le, Grace Laidlaw, Christopher A. Morehouse, Brian Naiman,

Philip Brohawn, Tomas Mustelin, Jane R. Connor, and Donald M. McDonald:

Synergistic Actions of Blocking Angiopoietin-2 and Tumor Necrosis Factor-α in

Suppressing Remodeling of Blood Vessels and Lymphatics in Airway Inﬂammation.

The American Journal of Pathology 185(11) (Nov. 1, 2015), 2949–2968. doi:

10.1016/j.ajpath.2015.07.010.

214. Li-Chin Yao, Peter Baluk, Jennifer Feng, and Donald M. McDonald:

Steroid-Resistant Lymphatic Remodeling in Chronically Inﬂamed Mouse Airways.

The American Journal of Pathology 176(3) (Mar. 1, 2010), 1525–1541. doi:

10.2353/ajpath.2010.090909.

215. A. S. Popel, P. C. Johnson, M. V. Kameneva, and M. A. Wild: Capacity for Red

Blood Cell Aggregation Is Higher in Athletic Mammalian Species than in Sedentary

Species. Journal of Applied Physiology 77(4) (Oct. 1, 1994), 1790–1794.

216. M. J. Pearson and H. H. Lipowsky: Inﬂuence of Erythrocyte Aggregation on

Leukocyte Margination in Postcapillary Venules of Rat Mesentery. American

147 Journal of Physiology. Heart and Circulatory Physiology 279(4) (Oct. 2000),

H1460–1471.

217. M. B. Lawrence, L. V. McIntire, and S. G. Eskin: Eﬀect of Flow on

Polymorphonuclear Leukocyte/Endothelial Cell Adhesion. Blood 70(5) (Nov. 1,

1987), 1284–1290.

218. Michael L. Smith, Markus Sperandio, Elena V. Galkina, and Klaus Ley:

Autoperfused Mouse Flow Chamber Reveals Synergistic Neutrophil Accumulation

through P-Selectin and E-Selectin. Journal of Leukocyte Biology 76(5) (Jan. 11,

2004), 985–993. doi: 10.1189/jlb.1003483.

219. Hawa-Racine Thiam, Pablo Vargas, Nicolas Carpi, Carolina Lage Crespo,

Matthew Raab, Emmanuel Terriac, Megan C. King, Jordan Jacobelli,

Arthur S. Alberts, Theresia Stradal, Ana-Maria Lennon-Dumenil, and

Matthieu Piel: Perinuclear Arp2/3-Driven Actin Polymerization Enables Nuclear

Deformation to Facilitate Cell Migration through Complex Environments. Nature

Communications 7 (Mar. 15, 2016), 10997. doi: 10.1038/ncomms10997.

220. Sylvain Gabriele, Anne-Marie Benoliel, Pierre Bongrand, and Olivier Th´eodoly:

Microﬂuidic Investigation Reveals Distinct Roles for Actin Cytoskeleton and

Myosin II Activity in Capillary Leukocyte Traﬃcking. Biophysical Journal 96(10)

(May 20, 2009), 4308–4318. doi: 10.1016/j.bpj.2009.02.037.

221. Janina R. Lange, Claus Metzner, Sebastian Richter, Werner Schneider,

Monika Spermann, Thorsten Kolb, Graeme Whyte, and Ben Fabry: Unbiased

High-Precision Cell Mechanical Measurements with Microconstrictions. Biophysical

Journal 112(7) (Apr. 11, 2017), 1472–1480. doi: 10.1016/j.bpj.2017.02.018.

148 222. Wei-Chien Hung, Jessica R. Yang, Christopher L. Yankaskas, Bin Sheng Wong,

Pei-Hsun Wu, Carlos Pardo-Pastor, Selma A. Serra, Meng-Jung Chiang,

Zhizhan Gu, Denis Wirtz, Miguel A. Valverde, Joy T. Yang, Jin Zhang, and

Konstantinos Konstantopoulos: Conﬁnement Sensing and Signal Optimization via

Piezo1/PKA and Myosin II Pathways. Cell Reports 15(7) (May 17, 2016),

1430–1441. doi: 10.1016/j.celrep.2016.04.035.

223. Edgar Gutierrez and Alex Groisman: Quantitative Measurements of the Strength of

Adhesion of Human Neutrophils to a Substratum in a Microﬂuidic Device.

Analytical Chemistry 79(6) (Mar. 1, 2007), 2249–2258. doi: 10.1021/ac061703n.

224. Damir B. Khismatullin and George A. Truskey: A 3D Numerical Study of the

Eﬀect of Channel Height on Leukocyte Deformation and Adhesion in Parallel-Plate

Flow Chambers. Microvascular Research 68(3) (Nov. 2004), 188–202. doi:

10.1016/j.mvr.2004.07.003.

225. X. Lei, M. B. Lawrence, and C. Dong: Inﬂuence of Cell Deformation on Leukocyte

Rolling Adhesion in Shear Flow. Journal of Biomechanical Engineering 121(6)

(Dec. 1, 1999), 636–643. doi: 10.1115/1.2800866.

226. Phillip A. Coghill, Erin K. Kesselhuth, Eddie A. Shimp, Damir B. Khismatullin,

and David W. Schmidtke: Eﬀects of Microﬂuidic Channel Geometry on Leukocyte

Rolling Assays. Biomedical Microdevices 15(1) (Feb. 1, 2013), 183–193. doi:

10.1007/s10544-012-9715-y.

227. Abhishek Jain and Lance L. Munn: Determinants of Leukocyte Margination in

Rectangular Microchannels. PLoS ONE 4(9) (Sept. 21, 2009), e7104. doi:

10.1371/journal.pone.0007104.

149 228. B. C. Schaefer, M. L. Schaefer, J. W. Kappler, P. Marrack, and R. M. Kedl:

Observation of Antigen-Dependent CD8+ T-Cell/ Dendritic Cell Interactions in

Vivo. Cellular Immunology 214(2) (Dec. 15, 2001), 110–122. doi:

10.1006/cimm.2001.1895.

229. David Stephenson, Alexander Patronis, David M. Holland, and

Duncan A. Lockerby: Generalizing Murray’s Law: An Optimization Principle for

Fluidic Networks of Arbitrary Shape and Scale. Journal of Applied Physics 118(17)

(Nov. 6, 2015), 174302. doi: 10.1063/1.4935288.

230. Peter Kovesi: Image Features from Phase Congruency. Videre: Journal of computer

vision research 1(3) (1999), 1–26.

231. B Walcheck, K L Moore, R P McEver, and T K Kishimoto: Neutrophil-Neutrophil

Interactions under Hydrodynamic Shear Stress Involve L-Selectin and PSGL-1. A

Mechanism That Ampliﬁes Initial Leukocyte Accumulation of P-Selectin in Vitro.

Journal of Clinical Investigation 98(5) (Sept. 1, 1996), 1081–1087.

232. R. F. Bargatze, S. Kurk, E. C. Butcher, and M. A. Jutila: Neutrophils Roll on

Adherent Neutrophils Bound to Cytokine-Induced Endothelial Cells via L-Selectin

on the Rolling Cells. The Journal of Experimental Medicine 180(5) (Nov. 1, 1994),

1785–1792.

233. Cristiano Migliorini, YueHong Qian, Hudong Chen, Edward B. Brown,

Rakesh K. Jain, and Lance L. Munn: Red Blood Cells Augment Leukocyte Rolling

in a Virtual Blood Vessel. Biophysical Journal 83(4) (Oct. 2002), 1834–1841. doi:

10.1016/S0006-3495(02)73948-9.

150 234. Prithu Sundd, Maria K. Pospieszalska, Luthur Siu-Lun Cheung,

Konstantinos Konstantopoulos, and Klaus Ley: Biomechanics of Leukocyte Rolling.

Biorheology 48(1) (Jan. 1, 2011), 1–35. doi: 10.3233/BIR-2011-0579.

235. T. M. Handel, Z. Johnson, S. E. Crown, E. K. Lau, M. Sweeney, and

A. E. Proudfoot: Regulation of Protein Function by Glycosaminoglycans—as

Exempliﬁed by Chemokines. Annual Review of Biochemistry 74(1) (2005), 385–410.

doi: 10.1146/annurev.biochem.72.121801.161747.

236. Tetsuya Kawamura, Bryan Stephens, Ling Qin, Xin Yin, Michael R. Dores,

Thomas H. Smith, Neil Grimsey, Ruben Abagyan, JoAnn Trejo, Irina Kufareva,

Mark M. Fuster, Catherina L. Salanga, and Tracy M. Handel: A General Method

for Site Speciﬁc Fluorescent Labeling of Recombinant Chemokines. PLOS ONE

9(1) (Jan. 28, 2014), e81454. doi: 10.1371/journal.pone.0081454.

237. Martin Trebbin, Dagmar Steinhauser, Jan Perlich, Adeline Buﬀet,

Stephan V. Roth, Walter Zimmermann, Julian Thiele, and Stephan F¨orster:

Anisotropic Particles Align Perpendicular to the Flow Direction in Narrow

Microchannels. Proceedings of the National Academy of Sciences 110(17) (Apr. 23,

2013), 6706–6711. doi: 10.1073/pnas.1219340110.

238. Akira Katsumi, A. Wayne Orr, Eleni Tzima, and Martin Alexander Schwartz:

Integrins in Mechanotransduction. Journal of Biological Chemistry 279(13)

(Mar. 26, 2004), 12001–12004. doi: 10.1074/jbc.R300038200.

239. Valentin Lulevich, Yi-Ping Shih, Su Hao Lo, and Gang-yu Liu: Cell Tracing Dyes

Signiﬁcantly Change Single Cell Mechanics. The journal of physical chemistry. B

113(18) (May 7, 2009), 6511–6519. doi: 10.1021/jp8103358.

151 240. Benjamin J. C. Quah and Christopher R. Parish: New and Improved Methods for

Measuring Lymphocyte Proliferation in Vitro and in Vivo Using CFSE-like

Fluorescent Dyes. Journal of Immunological Methods 379(1) (May 31, 2012), 1–14.

doi: 10.1016/j.jim.2012.02.012.

241. Deborah S. Barkauskas, Teresa A. Evans, Jay Myers, Agne Petrosiute, Jerry Silver,

and Alex Y. Huang: Extravascular CX3CR1+ Cells Extend Intravascular Dendritic

Processes into Intact Central Nervous System Vessel Lumen. Microscopy and

Microanalysis: The Oﬃcial Journal of Microscopy Society of America, Microbeam

Analysis Society, Microscopical Society of Canada 19(4) (Aug. 2013), 778–790. doi:

10.1017/S1431927613000482.

242. R. Dixon Dorand, Deborah S. Barkauskas, Teresa A. Evans, Agne Petrosiute, and

Alex Y. Huang: Comparison of Intravital Thinned Skull and Cranial Window

Approaches to Study CNS Immunobiology in the Mouse Cortex. Intravital 3(2)

(May 2014). doi: 10.4161/intv.29728.

243. Larry J. Millet, Matthew E. Stewart, Jonathan V. Sweedler, Ralph G. Nuzzo, and

Martha U. Gillette: Microﬂuidic Devices for Culturing Primary Mammalian

Neurons at Low Densities. Lab on a Chip 7(8) (July 24, 2007), 987–994. doi:

10.1039/B705266A.

244. Joe Chalfoun, Michael Majurski, Tim Blattner, Kiran Bhadriraju, Walid Keyrouz,

Peter Bajcsy, and Mary Brady: MIST: Accurate and Scalable Microscopy Image

Stitching Tool with Stage Modeling and Error Minimization. Scientiﬁc Reports 7

(July 10, 2017). doi: 10.1038/s41598-017-04567-y.

245. R Core Team: R: A Language and Environment for Statistical Computing. Vienna,

Austria: R Foundation for Statistical Computing, 2013.

152 246. RStudio Team: RStudio: Integrated Development Environment for R. Boston, MA:

RStudio, Inc., 2015.

247. Achim Zeileis, Christian Kleiber, and Simon Jackman: Regression Models for Count

Data in R. Journal of Statistical Software 27(8) (2008).

248. Christian Kleiber and Achim Zeileis: Visualizing Count Data Regressions Using

Rootograms. The American Statistician 70(3) (July 2, 2016), 296–303. doi:

10.1080/00031305.2016.1173590.

249. Hadley Wickham: Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag

New York, 2009.

250. Yihui Xie: Dynamic Documents with R and Knitr. 2nd. Boca Raton, Florida:

Chapman and Hall/CRC, 2015.

251. J. Brandon Dixon, Steven T. Greiner, Anatoliy A. Gashev, Gerard L. Cote,

James E. Moore, and David C. Zawieja: Lymph Flow, Shear Stress, and

Lymphocyte Velocity in Rat Mesenteric Prenodal Lymphatics. Microcirculation

(New York, N.Y.: 1994) 13(7) (2006), 597–610. doi: 10.1080/10739680600893909.

252. Jean-Philippe Girard, Christine Moussion, and Reinhold F¨orster:HEVs,

Lymphatics and Homeostatic Immune Cell Traﬃcking in Lymph Nodes. Nature

Reviews. Immunology 12(11) (Nov. 2012), 762–773. doi: 10.1038/nri3298.

253. Rob J. De Boer and Alan S. Perelson: Quantifying T Lymphocyte Turnover.

Journal of theoretical biology 327 (June 21, 2013), 45–87. doi:

10.1016/j.jtbi.2012.12.025.

254. Joseph J. C. Thome, Naomi Yudanin, Yoshiaki Ohmura, Masaru Kubota,

Boris Grinshpun, Taheri Sathaliyawala, Tomoaki Kato, Harvey Lerner,

153 Yufeng Shen, and Donna L. Farber: Spatial Map of Human T Cell

Compartmentalization and Maintenance over Decades of Life. Cell 159(4) (Nov. 6,

2014), 814–828. doi: 10.1016/j.cell.2014.10.026.

255. Britta Engelhardt and Richard M. Ransohoﬀ: The Ins and Outs of T-Lymphocyte

Traﬃcking to the CNS: Anatomical Sites and Molecular Mechanisms. Trends in

Immunology 26(9) (Sept. 2005), 485–495. doi: 10.1016/j.it.2005.07.004.

256. Mark E. Wagshul, Per K. Eide, and Joseph R. Madsen: The Pulsating Brain: A

Review of Experimental and Clinical Studies of Intracranial Pulsatility. Fluids and

Barriers of the CNS 8 (Jan. 18, 2011), 5. doi: 10.1186/2045-8118-8-5.

257. Brian J. McHugh, Robert Buttery, Yatish Lad, Stephen Banks,

Christopher Haslett, and Tariq Sethi: Integrin Activation by Fam38A Uses a Novel

Mechanism of R-Ras Targeting to the Endoplasmic Reticulum. Journal of Cell

Science 123(1) (Jan. 1, 2010), 51–61. doi: 10.1242/jcs.056424.

258. Jing Li, Bing Hou, Sarka Tumova, Katsuhiko Muraki, Alexander Bruns,

Melanie J. Ludlow, Alicia Sedo, Adam J. Hyman, Lynn McKeown,

Richard S. Young, Nadira Y. Yuldasheva, Yasser Majeed, Lesley A. Wilson,

Baptiste Rode, Marc A. Bailey, Hyejeong R. Kim, Zhaojun Fu,

Deborah A. L. Carter, Jan Bilton, Helen Imrie, Paul Ajuh, T. Neil Dear,

Richard M. Cubbon, Mark T. Kearney, K. Raj Prasad, Paul C. Evans,

Justin F. X. Ainscough, and David J. Beech: Piezo1 Integration of Vascular

Architecture with Physiological Force. Nature 515(7526) (Nov. 13, 2014), 279–282.

doi: 10.1038/nature13701.

259. Viktor Lukacs, Jayanti Mathur, Rong Mao, Pinar Bayrak-Toydemir,

Melinda Procter, Stuart M. Cahalan, Helen J. Kim, Michael Bandell, Nicola Longo,

154 Ronald W. Day, David A. Stevenson, Ardem Patapoutian, and Bryan L. Krock:

Impaired PIEZO1 Function in Patients with a Novel Autosomal Recessive

Congenital Lymphatic Dysplasia. Nature Communications 6 (Sept. 21, 2015), 8329.

doi: 10.1038/ncomms9329.

260. Elisavet Fotiou, Silvia Martin-Almedina, Michael A. Simpson, Shin Lin,

Kristiana Gordon, Glen Brice, Giles Atton, Iona Jeﬀery, David C. Rees,

Cyril Mignot, Julie Vogt, Tessa Homfray, Michael P. Snyder, Stanley G. Rockson,

Steve Jeﬀery, Peter S. Mortimer, Sahar Mansour, and Pia Ostergaard: Novel

Mutations in PIEZO1 Cause an Autosomal Recessive Generalized Lymphatic

Dysplasia with Non-Immune Hydrops Fetalis. Nature Communications 6 (Sept. 3,

2015), ncomms9085. doi: 10.1038/ncomms9085.

261. Lena Svensson, Alison McDowall, Katherine M. Giles, Paula Stanley, Stefan Feske,

and Nancy Hogg: Calpain 2 Controls Turnover of LFA-1 Adhesions on Migrating T

Lymphocytes. PLOS ONE 5(11) (Nov. 30, 2010), e15090. doi:

10.1371/journal.pone.0015090.

262. Nicole A. Morin, Patrick W. Oakes, Young-Min Hyun, Dooyoung Lee,

Y. Eugene Chin, Michael R. King, Timothy A. Springer, Motomu Shimaoka,

Jay X. Tang, Jonathan S. Reichner, and Minsoo Kim: Nonmuscle Myosin Heavy

Chain IIA Mediates Integrin LFA-1 de-Adhesion during T Lymphocyte Migration.

Journal of Experimental Medicine 205(1) (Jan. 21, 2008), 195–205. doi:

10.1084/jem.20071543.

263. R. J. Eddy, L. M. Pierini, F. Matsumura, and F. R. Maxﬁeld: Ca2+-Dependent

Myosin II Activation Is Required for Uropod Retraction during Neutrophil

Migration. Journal of Cell Science 113 ( Pt 7) (Apr. 2000), 1287–1298.

155 264. Ruhma Syeda, Jie Xu, Adrienne E. Dubin, Bertrand Coste, Jayanti Mathur,

Truc Huynh, Jason Matzen, Jianmin Lao, David C. Tully, Ingo H. Engels,

H. Michael Petrassi, Andrew M. Schumacher, Mauricio Montal, Michael Bandell,

and Ardem Patapoutian: Chemical Activation of the Mechanotransduction Channel

Piezo1. eLife 4 (May 22, 2015), e07369. doi: 10.7554/eLife.07369.

265. Charles D. Cox, Chilman Bae, Lynn Ziegler, Silas Hartley,

Vesna Nikolova-Krstevski, Paul R. Rohde, Chai-Ann Ng, Frederick Sachs,

Philip A. Gottlieb, and Boris Martinac: Removal of the Mechanoprotective

Inﬂuence of the Cytoskeleton Reveals PIEZO1 Is Gated by Bilayer Tension. Nature

Communications 7 (Jan. 20, 2016), 10366. doi: 10.1038/ncomms10366.

266. Rachel Evans, Irene Patzak, Lena Svensson, Katia De Filippo, Kristian Jones,

Alison McDowall, and Nancy Hogg: Integrins in Immunity. Journal of Cell Science

122(2) (Jan. 15, 2009), 215–225. doi: 10.1242/jcs.019117.

267. Jingpeng Ge, Wanqiu Li, Qiancheng Zhao, Ningning Li, Maofei Chen, Peng Zhi,

Ruochong Li, Ning Gao, Bailong Xiao, and Maojun Yang: Architecture of the

Mammalian Mechanosensitive Piezo1 Channel. Nature advance online

publication (Sept. 21, 2015). doi: 10.1038/nature15247.

268. Adrienne E. Dubin, Swetha Murthy, Amanda H. Lewis, Lucie Brosse,

Stuart M. Cahalan, J¨orgGrandl, Bertrand Coste, and Ardem Patapoutian:

Endogenous Piezo1 Can Confound Mechanically Activated Channel Identiﬁcation

and Characterization. Neuron 94(2) (Apr. 19, 2017), 266–270.e3. doi:

10.1016/j.neuron.2017.03.039.

269. Bertrand Coste, Swetha E. Murthy, Jayanti Mathur, Manuela Schmidt,

Yasmine Mechioukhi, Patrick Delmas, and Ardem Patapoutian: Piezo1 Ion Channel

156 Pore Properties Are Dictated by C-Terminal Region. Nature Communications 6

(May 26, 2015), 7223. doi: 10.1038/ncomms8223.

270. Yusong R. Guo and Roderick MacKinnon: Structure-Based Membrane Dome

Mechanism for Piezo Mechanosensitivity. eLife 6 (Dec. 12, 2017), e33660. doi:

10.7554/eLife.33660.

271. Qiancheng Zhao, Heng Zhou, Shaopeng Chi, Yanfeng Wang, Jianhua Wang,

Jie Geng, Kun Wu, Wenhao Liu, Tingxin Zhang, Meng-Qiu Dong, Jiawei Wang,

Xueming Li, and Bailong Xiao: Structure and Mechanogating Mechanism of the

Piezo1 Channel. Nature 554(7693) (Feb. 2018), 487–492. doi: 10.1038/nature25743.

272. Wahyu Surya, Yan Li, Oscar Millet, Tammo Diercks, and Jaume Torres:

Transmembrane and Juxtamembrane Structure of αL Integrin in Bicelles. PLOS

ONE 8(9) (Sept. 12, 2013), e74281. doi: 10.1371/journal.pone.0074281.

273. Chinky Shiu Chen Liu, Deblina Raychaudhuri, Barnali Paul,

Yogaditya Chakrabarty, Amrit Raj Ghosh, Oindrila Rahaman, Arindam Talukdar,

and Dipyaman Ganguly: Cutting Edge: Piezo1 Mechanosensors Optimize Human T

Cell Activation. Journal of Immunology (Baltimore, Md.: 1950) 200(4) (Feb. 15,

2018), 1255–1260. doi: 10.4049/jimmunol.1701118.

274. Nathan C. Shaner, Gerard G. Lambert, Andrew Chammas, Yuhui Ni,

Paula J. Cranﬁll, Michelle A. Baird, Brittney R. Sell, John R. Allen,

Richard N. Day, Maria Israelsson, Michael W. Davidson, and Jiwu Wang: A Bright

Monomeric Green Fluorescent Protein Derived from Branchiostoma Lanceolatum.

Nature methods 10(5) (May 2013). doi: 10.1038/nmeth.2413.

275. R. S. Larson, A. L. Corbi, L. Berman, and T. Springer: Primary Structure of the

Leukocyte Function-Associated Molecule-1 Alpha Subunit: An Integrin with an

157 Embedded Domain Deﬁning a Protein Superfamily. The Journal of Cell Biology

108(2) (Feb. 1989), 703–712.

276. A. L. Corbi, T. K. Kishimoto, L. J. Miller, and T. A. Springer: The Human

Leukocyte Adhesion Glycoprotein Mac-1 (Complement Receptor Type 3, CD11b)

Alpha Subunit. Cloning, Primary Structure, and Relation to the Integrins, von

Willebrand Factor and Factor B. The Journal of Biological Chemistry 263(25)

(Sept. 5, 1988), 12403–12411.

277. T. K. Kishimoto, K. O’Connor, A. Lee, T. M. Roberts, and T. A. Springer: Cloning

of the Beta Subunit of the Leukocyte Adhesion Proteins: Homology to an

Extracellular Matrix Receptor Deﬁnes a Novel Supergene Family. Cell 48(4)

(Feb. 27, 1987), 681–690.

278. Chi-Chao Liu, Pascal Leclair, Shyong Quin Yap, and Chinten James Lim: The

Membrane-Proximal KXGFFKR Motif of α-Integrin Mediates Chemoresistance.

Molecular and Cellular Biology 33(21) (Nov. 2013), 4334–4345. doi:

10.1128/MCB.00580-13.

279. Natalya G. Dulyaninova, Reniqua P. House, Venkaiah Betapudi, and

Anne R. Bresnick: Myosin-IIA Heavy-Chain Phosphorylation Regulates the

Motility of MDA-MB-231 Carcinoma Cells. Molecular Biology of the Cell 18(8)

(Aug. 2007), 3144–3155. doi: 10.1091/mbc.e06-11-1056.

280. Akiko Seki and Sascha Rutz: Optimized RNP Transfection for Highly Eﬃcient

CRISPR/Cas9-Mediated Gene Knockout in Primary T Cells. Journal of

Experimental Medicine (Feb. 7, 2018), jem.20171626. doi: 10.1084/jem.20171626.

158 281. Christopher R. Hackley, Esteban O. Mazzoni, and Justin Blau: cAMPr: A

Single-Wavelength Fluorescent Sensor for Cyclic AMP. Sci. Signal. 11(520)

(Mar. 6, 2018), eaah3738. doi: 10.1126/scisignal.aah3738.

282. William J. Polacheck and Christopher S. Chen: Measuring Cell-Generated Forces:

A Guide to the Available Tools. Nature methods 13(5) (Apr. 28, 2016), 415–423.

doi: 10.1038/nmeth.3834.

283. Medha M. Pathak, Jamison L. Nourse, Truc Tran, Jennifer Hwe, Janahan Arulmoli,

Dai Trang T. Le, Elena Bernardis, Lisa A. Flanagan, and Francesco Tombola:

Stretch-Activated Ion Channel Piezo1 Directs Lineage Choice in Human Neural

Stem Cells. Proceedings of the National Academy of Sciences of the United States of

America 111(45) (Nov. 11, 2014), 16148–16153. doi: 10.1073/pnas.1409802111.

159