A Comparison Between Cervical and Thoracic Spinal Cord Injury: Critical Level-Dependent Differences in Pathobiology

Home , James Hong

A Comparison Between Cervical and Thoracic Spinal Cord Injury: Critical Level-dependent Differences in Pathobiology

James Yuh-Luen Hong

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Institute of Medical Science University of Toronto

A Comparison Between Cervical and Thoracic Spinal Cord Injury: Critical Level-dependent Differences in Pathobiology

James Yuh-Luen Hong

Doctor of Philosophy

Institute of Medical Science University of Toronto

2020

Abstract Introduction: Despite advances in acute medical and surgical care, there are no effective regenerative or reparative strategies for traumatic spinal cord injury (SCI). Patients with SCI experience a lifetime of neurological impairment, respiratory and cardiovascular distress. While incomplete injuries to the cervical cord are amongst the most common, preclinical and clinical efforts—driven by ease-of-access-have mostly focused on the modeling and treatment of the thoracic injuries. While translational efforts remain difficult, no comparative study of level- dependent SCI pathogenesis exist. In this thesis, I have examined the local, systemic and peripheral responses to a clinically-relevant model of cervical and thoracic SCI with the overarching hypothesis that baseline differences between the levels of injury will result in distinct vascular pathologies following SCI and trigger contrasting profiles of downstream secondary injury locally and systemically. Methods: Female Wistars rats were subjected to either moderate-severe C6-7 or T6-7 SCI, RNA-sequencing, Western blot, ELISAs, Luminex arrays, immunohistochemistry and high-resolution ultrasound were used to characterize the local and peripheral changes in the spinal cord, spleen and plasma. Results: We found reduced baseline levels of collagen, fibronectin

and astrocytic laminin in the cervical relative to the thoracic cord. This corresponded to rapid tissue clearance, increased neuroinflammation and gliosis after cSCI. Secondly, we found that there were profound reductions in the concentration of circulating cytokine and chemokines after cSCI.

Finally, we found a dramatic surge in splenic norepinephrine, cortisol and cleaved caspase-3 and a corresponding reduction in leukocyte trafficking molecules to the spleen after thoracic, but not cervical, SCI. Discussion and Impact: Taken together, this work demonstrates that: 1) vascular disruption after cervical SCI should be a critical acute therapeutic target as it results in irreversible tissue and thereby functional loss following traumatic injury; 2) circulating reductions in cytokines and chemokines are likely due to a pooling at local injury sites; and 3) the level-dependent phenomenon of SCI-immunodeficiency syndrome (SCI-IDS) which has not been consistently shown in patients is reversed in our model of incomplete SCI, suggesting that incomplete injuries result only in local deafferentation and reconciles the reason why SCI-IDS has not been consistently shown in patients.

iii

Acknowledgments

Dr. Michael Fehlings

Thank you for the near decade of mentorship and for providing me with an environment to explore all my scientific curiosities. The breadth of research I was exposed to enabled me to reach new heights in academic productivity and has laid the foundations for many aspects of my scientific interest. It is my hope that someday I can follow in your footsteps and truly make a mark in the field.

Drs. James Eubanks & Isabelle Aubert

Thank you both for the conversations and critical evaluations of my work throughout our numerous committee meetings. At times when I was lost in the big data, these conversations helped me regain focus. I owe you both greatly for your time and hope that one day I can be considered a colleague in the field.

Drs. Vince Tropepe and Phillip Marsden

Thank you both for being a part of my journey, without you I would have never made the transition into my PhD. To Dr. Marsden especially, thank you for taking the time to read and assess both my transfer and my final defense—I am eternally grateful.

Drs. Reaz Vawda and Mahmood Chamankhah

Thank you both for being my primary mentors when I entered the laboratory. Without your initial guidance, I would have not been able to develop my skills at the numerous molecular techniques.

My Parents

Thank you for bringing me into this world, throughout the years I’ve often been ambiguous about what I wanted to do. It was through your support and love and I was able to find the things that I cherished and loved in life, and I hope to someday grow to become the man that you’ve brought me up to be.

To My Friends

You know who you are! Thank you for the fun times in the lab and on conferences, I enjoyed the numerous uncomfortable hikes in dress shoes and flip-flops, the sharing of dreams, the scientific arguments, the relationship advice, and the academic journey. I will cherish the connections I’ve made with all of you and hope to keep in touch.

To Susan

Thank you for being the love of my life. We met at the start of my academic journey, and without your love and support, I would be a very different person today. I look forward to spending the rest of our life together and hope that I can bring you a lifetime of happiness.

Statement of Contributions

For the research described in this thesis, I maintain primary accountability for all results and interpretation of such results. I was principally responsible for the experimental design and execution of the project. However, I acknowledge that I received substantial assistance from lab members who are experts in their respective scientific areas, and I was responsible for training several research students who significantly contributed to the collection of the data displayed in this thesis. Therefore, I wish to formally recognize the scientific contributions of these individuals for their involvement in this research.

Dr. Mahmood Chamankhah collected the thoracic samples for RNA-sequencing, conducted preliminary Western blots and qRT-PCR, trained me in the aforementioned molecular techniques. Involved in the interpretation and discussion of qPCR, Western blot and RNA- sequencing results.

Drs. Jian Wang and Yang Liu executed the animal surgeries in C6-7 and T6-7.

Drs. Dario Righelli and Claudia Angelini involved in the interpretation and discussion of

RNA-sequencing results.

Dr. Anna Badner collected the high-resolution ultrasound data and sectioned numerous samples for immunohistochemistry.

Priscilla Chan sectioned numerous samples for immunohistochemistry.

Dr. Chris Ahuja cultured primary astrocytes and pericytes for in vitro segment of the work.

Alex Chang assisted in RNA-sequencing analysis using command line tools, ran ELISAs, and

Western blots. vi

Mohammad Zavvarian captures confocal images of cords, and stained cords for immunohistochemistry.

Dr. Stefania Forner involved in post-operative behavioral assessments including BBB and grip strength.

Behzad Azad provided substantial post-operative animal care for animals in all experiments throughout my entire thesis.

vii

Abbreviations

ASIA American Spinal Injury Association

ATP Adenosine Triphosphate

BBB Blood Brain Barrier

BBB scale Basso, Beattie, Bresnahan Locomotor Rating Scale BMS Basso Mouse Scale BSCB Blood-Spinal Cord Barrier

+ Ca2 Calcium

Caspase Cysteine Protease, Which Cleaves at an Aspartate Residue

CNS Central Nervous System CORT Corticosterone CSCI Cervical spinal cord injury

CSF Cerebrospinal Fluid

CSPG Chondroitin Sulfate Proteoglycan Da Dalton

EB Evans Blue

ECF Extracellular Fluid

ECM Extracellular Matrix

EGF Epidermal Growth Factor GCSF Granulocyte - colony stimulating factor GMCSF Granulocyte-macrophage colony-stimulating factor HRP Horseradish Peroxidase

IHC Immunohistochemistry

IL1 Interleukin 1 alpha IL1 Interleukin 1 beta

viii

IL10 Interleukin-10 IL12 Interleukin-12 IL17A Interleukin-17A IL18 Interleukin-18 IL2 Interleukin-2 IL4 Interleukin-4 IL5 Interleukin-5 IP10 Interferon gamma-induced protein 10

+ K Potassium

KDa Kilodalton LIX Lipopolysaccharide-induced CXC chemokine MCP1 Monocyte chemoattractant protein-1 MIP2 macrophage inflammatory protein 2

+ Na Sodium NASCIS National Acute Spinal Cord Injury Study NE Norepinephrine NFκB Nuclear Factor Kappa Light Polypeptide Gene Enhancer in B-Cells

NO Nitrous Oxide

NPC Neural Precursor Cell

NVU Neurovascular Unit

OPC Oligodendrocyte Precursor Cell

PAGE Poly Acrylamide Gel Electrophoresis

PBS Phosphate Buffered Saline

PDGF Platelet Derived Growth Factor qRT-PCR Quantitative Real-Time Polymerase Chain Reaction RANTES regulated on activation, normal T cell expressed and secreted Ras Small GTPase Involved in Cell Signaling

ROS Reactive Oxygen Species

SCI Spinal Cord Injury SCI-IDS Spinal cord injury-induced immunodeficiency syndrome SPN Spinal preganglionic neurons TBST Tris-buffered saline with 0.1% Tween-20 TJ Tight Junction

TNFα Tumor Necrosis Factor Alpha

ZO Zona Occludens

Table of Contents

Contents

Acknowledgments...... iv

Statement of Contributions ...... vi

Abbreviations ...... viii

Table of Contents ...... xi

List of Tables ...... xv

List of Figures ...... xvi

Chapter 1 ...... 1

Introduction to Spinal Cord Injury ...... 1

1.1 Incidence and Demography ...... 1

1.2 Prevalence ...... 3

1.3 Etiology ...... 3

1.4 Pathophysiology ...... 4

1.4.1 Primary Injury ...... 6

1.4.2 Secondary Injury ...... 7

Vascular Anatomy of the Spinal Cord ...... 16

2.1.1 Extraspinal Arteries ...... 16

2.1.2 Intraspinal Arteries...... 21

2.1.3 Arterial Hemodynamics ...... 24

2.1.4 Venous Anatomy ...... 24

2.1.5 The Blood Spinal Cord Barrier (BSCB) ...... 27

2.1.6 Lymphatic Drainage in the Spine ...... 30

2.1.7 Level-dependent Differences in Vascular Architecture ...... 31

2.2 Animal models of SCI ...... 36 xi

2.3 Clinical Management of SCI ...... 38

2.3.1 Spine Immobilization and Stabilization ...... 38

2.3.2 Hemodynamic Management ...... 39

2.3.3 Neurological Assessment ...... 40

2.3.4 Operative vs. Nonoperative management and Timing of Decompression ...... 44

2.3.5 Venous Thromboembolism Prophylaxis and Treatment ...... 47

2.3.6 Pharmacological Management of SCI ...... 47

2.3.6.2 Glibenclamide (Glyburide, DiaBeta) ...... 48

Chapter 2 ...... 53

Aims and Hypothesis ...... 53

3.1 Overarching Hypothesis...... 53

3.2 Specific Aims ...... 53

3.3 Publications ...... 53

Chapter 3 ...... 56

Level-dependent differences in the spinal cord following traumatic SCI ...... 56

4.1 Abstract ...... 57

4.2 Introduction ...... 57

4.3 Methods...... 58

4.4 Results & Discussion ...... 64

4.5 Conclusion ...... 74

Chapter 4 ...... 75

Level-dependent differences in the circulating immune factors following SCI ...... 75

5.1 Abstract ...... 76

5.2 Introduction ...... 76

5.3 Methods...... 78

5.3.1 Clip-Compression SCI and Spleen Weight...... 78 xii

5.3.2 Neurobehavioral Assessments ...... 78

5.3.3 Blood Collection and High-Throughput Luminex Assay ...... 80

5.3.4 Clustering and Statistical Analysis ...... 80

5.4 Results ...... 81

5.4.1 Level-specific differences in plasma protein levels after cervical and thoracic laminectomy ...... 81

5.4.2 Level-specific differences in plasma protein levels after cervical and thoracic SCI ...... 83

5.4.3 Temporal Expression Patterns of Plasma Proteins after SCI ...... 85

5.4.4 Functional Classification of Serum Protein after SCI...... 87

5.4.5 Spleen Weight ...... 89

5.5 Discussion ...... 91

5.6 Conclusion ...... 94

Chapter 5 ...... 95

Level-dependent differences in the splenic response following SCI ...... 95

6.1 Abstract ...... 96

6.2 Introduction ...... 96

6.3 Methods...... 98

6.4 Results ...... 101

6.4.1 Level-dependent differences in splenic NE and CORT ...... 101

6.4.2 NE and CORT surge results in increased caspase cleavage in tSCI ...... 104

6.4.3 Reduced leukocyte chemotaxins to the spleen after tSCI ...... 106

6.5 Discussion ...... 108

General Discussion and Future Directions ...... 110

7.1 Summary of Results ...... 110

7.2 Novelty and Impact ...... 111

7.3 General Discussion ...... 111 xiii

7.3.1 Changes in basement membrane constituents after injury ...... 112

7.3.2 Injury level and its impact on the timing of various therapeutic interventions ...112

7.3.3 Impact of tissue preservation in rodents and humans ...... 114

7.3.4 Differences in the immune system of rodents and humans ...... 115

7.3.5 Injury severity between levels ...... 115

7.3.6 Sex considerations ...... 116

7.4 Future Directions ...... 116

7.4.1 Overexpression of basement membrane ECM molecules ...... 116

7.4.2 Inhibition of thrombin targets to reduce intraspinal hemorrhage ...... 117

7.4.3 Acute infusion of mesenchymal stromal cells to attenuate vascular disruption ..117

7.4.4 PKC-β inhibition as a target for attenuating vascular disruption ...... 118

7.4.5 Acute IgG infusion for immunomodulation and rescue of BSCB disruption ...... 120

7.4.6 Remote ischemic preconditioning prior to surgical decompression ...... 121

References ...... 123

xiv

List of Tables

Table 1 Animal Models of SCI ...... 37

Table 2 ASIA Impairment Scale...... 42

Table 3 List of antibodies ...... 63

List of Figures

Figure 1. Annual incidence of SCI across reported countries, states or provinces and regions. .... 2

Figure 2 Pathophysiology of Traumatic Spinal Cord Injury...... 5

Figure 3 Post-spinal cord injury syrinx...... 8

Figure 4 Extraspinal arteries of the spinal cord...... 20

Figure 5 Intraspinal Arteries of the Spinal Cord...... 23

Figure 6 Venous Drainage of the Spinal Cord...... 26

Figure 7 The Cellular Components of the Neurovascular Unit (NVU)...... 29

Figure 8 Cervical and High Thoracic SCIs Disrupt the Outflow of the Sympathetic Nervous System...... 35

Figure 9 Surgical decompression and realignment of the injured spinal cord...... 46

Figure 10 Traumatic SCI induces level-dependent rapid tissue loss and large vessel disruption. 65

Figure 11 Baseline molecular differences between the cervical and thoracic cord center on the composition of the basement membrane...... 68

Figure 12 Traumatic SCI induces a level-specific pathophysiological profile...... 71

Figure 13 In vitro primary cultures of rat astrocytes and pericytes exhibit level-distinct responses when cultured with tissue extract...... 73

Figure 14 BBB and Grip Strength Assessments ...... 79

Figure 15 Heat map and hierarchical cluster analyses reveal six clusters of temporal expression amongst cSham and tSham groups...... 82

Figure 16 Temporal expression of level-specific proteins...... 84

xvi

Figure 17 Heatmap and k-means clustering reveals three major clusters of temporal expression in cSCI and tSCI...... 86

Figure 18 Heat map of functionally-segregated proteins after cSCI and tSCI...... 88

Figure 19 Spleen:mass ratios expressed as fold-change of time-matched laminectomized shams...... 90

Figure 20 Experimental design of Chapter 5...... 102

Figure 21 SCI results in changes to splenic NE and CORT in a level-dependent and time- dependent manner...... 103

Figure 22 SCI results in changes to splenic pro-caspase-3, caspase-3 and cleaved caspase-3 in a level-dependent and time-dependent manner...... 105

Figure 23 SCI induces changes in splenic expression of cytokine/chemokines in a level- dependent and time-dependent manner...... 107

xvii

Chapter 1

Introduction to Spinal Cord Injury

1.1 Incidence and Demography

Global incidence of traumatic SCI in 2007 was estimated to be between 133 thousand and 226 thousand, or 23 cases per million [1,2]. The incidence by country was found to range from 8.0 per million in Spain to 49.1 per million in New Zealand [3] (Figure 1). The Global Burden of Disease Study reported a global estimate of 0.93 million new cases of SCI in 2016 [4]. There is a consistent global trend of higher rates of SCI in males than females, although the ratio varies widely across countries [3]. From 1993 to 2012, the overall incidence of acute spinal cord injury in the United States has slightly increased from 53 cases per 1 million (95%CI: 52-54) to 54 cases per 1 million (95%CI: 53-55). The incidence in the young male population (16-24 years), has decreased from 144 cases per million to 87 cases per million. A similar trend is seen for males age 24-44, decreasing from 96 cases per million to 71 cases per million. Conversely, there has been an increase in incidence for men ages 65-74 from 84 cases per million in 1993 to 131 cases per million in 2012[5]. The National Spinal Cord Injury Statistical Center reports a similar overall incidence rate of 54 cases per million in 2018, increasing from an estimated 40 per million in 2011. The average age at injury increasing from 29 during 1973-1979 to 37.6 in 2000 and 43 in 2018 [6,7]. In a recent analysis of traumatic pediatric cervical spinal fracture patients in the USA, the incidence of SCI was found to have slightly increased between 2003 and 2012, from 2.39% to 3.12% [8]. In Canada, there is a cost of approximately $2.7 billion associated with the approximately 1389 new traumatic SCI patients every year [9]. The initial incidence of traumatic SCI was estimated to be approximately 53 per million in 2010, which includes cases which did not survive to hospital, while the discharge incidence was estimated to be 41 per million [10].

Figure 1. Annual incidence of SCI across reported countries, states or provinces and regions.

The annual incidence of spinal cord injury varies internationally. Republished with permission from the authors and publisher from Figure 2 [11].

1.2 Prevalence

The global prevalence of SCI was estimated to be 27.04 million (24.98-30.15 million) in 2016 [4] . The prevalence in the USA is estimated to be approximately 853 per million (range: 721-4187) [1] or 906 per million [3]. The Canadian prevalence of traumatic SCI has been estimated at 1184 per million [1]. Overall the prevalence of SCI in Canada is approximately 85556 persons, and this figure is split almost evenly across traumatic SCI (51%) and non-traumatic SCI (49%) [10].

1.3 Etiology

The etiology of traumatic SCI varies globally according to local cultural factors, in addition to demographic and economic factors. Generally, younger patients sustain spinal injury through high- energy mechanisms (i.e. MVC, fall from height) while elderly patients endure lower-energy mechanisms (i.e. low falls). Overall, traffic accidents are the primary cause of SCI [3]. In Canada, land transport-related SCI has been reported as the primary etiology at 47%, ranging from 34% to 56% across regions. Similarly, the most common etiology in the United States is land transport at 48% [12]. In developed countries, the incidence of traumatic SCI due to motor vehicle accidents has mostly decreased or stabilized, however this figure is increasing in developing countries. This is likely due an ongoing transition to motorized vehicles in the latter. In comparison, higher- developed regions generally have safer vehicles, roads and more advanced alternative transport infrastructure, although there are significant variations within countries [1]. A pattern displaying elevated rates of violence-related traumatic SCI can be seen through North and South America, Southern Africa and the Middle East. The largest proportion of gunshot-related SCI is present in South Africa, but this etiology is also elevated in Brazil and the United States [1]. Violence was responsible for 15% of SCI in the USA, with 14% gunshot-related and 1% stabbing-related [12]. Greenland has an extremely elevated proportion of suicide-related TSCI relative to other countries, at 23% [1,3]. In the USA, the proportion of SCI related to falls in those older than 65 years

increased from 28% from 1997 -2000 to 66% from 2010-2012[5]. Overall, approximately 23% of traumatic SCI are due to falls in the United States [1], while in Canada, 17% of SCI are due to falls [12]. The widely-reported trend towards low-fall-related SCI in elderly populations is especially important in developed regions with an increasingly aging population [5,13]. Notably, Japan has the highest proportion of tetraplegia in the world, likely due to an ageing population and possible genetic factors [1,14,15].

1.4 Pathophysiology

The timeline of pathological effects after spinal cord injury can be organized into acute, sub-acute, intermediate and chronic phases (Figure 2A-C). The acute phase spans approximately the first 48 hours after injury. The subacute phase lasts from 48h post injury to approximately 14 days post injury, while the intermediate phase lasts 14 days-3 months, and chronic (>3 months). The effects of spinal cord injury can also be divided into primary and secondary mechanisms [16].

Figure 2 Pathophysiology of Traumatic Spinal Cord Injury. a) The initial mechanical trauma to the spinal cord initiates a secondary injury cascade that is characterized in the acute phase (that is, 0–48 hours after injury) by edema, hemorrhage, ischemia, inflammatory cell infiltration, the release of cytotoxic products and cell death. This secondary injury leads to necrosis and/or apoptosis of neurons and glial cells, such as oligodendrocytes, which can lead to demyelination and the loss of neural circuits. b) In the subacute phase (2–4 days after injury), further ischemia occurs owing to ongoing edema, vessel thrombosis and vasospasm. Persistent inflammatory cell infiltration causes further cell death, and cystic microcavities form, as cells and the extracellular architecture of the cord are damaged. In addition, astrocytes proliferate and deposit extracellular matrix molecules into the perilesional area. c) In the intermediate and chronic phases (2 weeks to 6 months), axons continue to degenerate and the

astroglial scar matures to become a potent inhibitor of regeneration. Cystic cavities coalesce to further restrict axonal regrowth and cell migration. Republished with permission from the authors and publisher Figure 3 [11].

1.4.1 Primary Injury

The pathophysiology SCI is highly diverse and dependent on the biological and physical conditions of the primary injury. Primary injury refers to the initial mechanical impact, compression, distraction (stretching), laceration or contusion that results in structural damage [17]. Commonly, however, it is due to contusion with or without accompanying persistent compression from bone fragments due to vertebral burst fractures [18]. The transient or persistent compression of the dural sac causes contusion-type injuries. Particularly in traumatic SCI, the precise mechanism of compression can be difficult to ascertain, as patients may present with a combination of lateral, posterior and anterior compression. Disc herniation, vertebral fracture dislocation and anteriorly located epidural masses can result in anterior compression, while facet dislocation and posteriorly located epidural masses can result in posterior compression [19]. The complete transection of the SC is rarely clinically relevant [20] and for this reason, animal models that focus on contusion have been developed and successful mimic human pathological changes after SCI. Energy transfer through the subdural cerebrospinal fluid (CSF) and the resultant impact on the cord varies according to morphology and thus partially depends on the level of injury as described above [18,20,21].

Across the range of underlying etiologies, mechanical insults cause instantaneous physical changes by destroying neural cells and vascular structures at the site of injury [22].

Both the duration and severity of compression determine the functional outcome and at a critical threshold, significant, irreversible and immediate neurological damage is produced, which

compromises the physiological transmission through the spinal cord. The extent of the primary injury is predictive of functional recovery [16].

1.4.2 Secondary Injury

The secondary injury in SCI is a multifactorial range of physical and biochemical effects progressively damage tissue. These begin immediately after injury and last for several weeks to several months [23,24]. These mechanisms of injury have traditionally been the focus of SCI therapeutics research, as the narrow timeline of primary injury does not allow for a sufficient therapeutic window. The cells that survive the initial injury but are damaged may undergo necrosis or apoptosis [22]. At the injury epicenter, cystic cavities known as syringomyelia (Figure 3) form and spread over time. Astrocytes and pericyte-derived cells surround the cavity to attenuate the spread of the lesion while a fibrous scar (collagen and ECM molecules) and pericytes and is also formed [25]. The secondary injuries are a significant contributor to functional impairment and physical changes in the spinal cord and expand the area of primary injury [26]. Here, a brief overview of secondary injury mechanisms is provided, with a focus on location of injury and vascular disruption.

Figure 3 Post-spinal cord injury syrinx.

T2-weighted MRI of the cervical (parts a–c) and thoracic (parts d–f) spine in sagittal (part a and part d) and axial (parts b, c, e and f) planes shows a post-traumatic syrinx within the spinal cord parenchyma (white arrows) and kyphosis (that is, forward bending) of the thoracic spine at the initial site of the spinal cord injury (SCI; black arrow in part d). The syrinx extends well beyond the mid-thoracic site of the SCI into the high cervical spinal cord, which is associated with neuropathic upper limb pain. Republished with permission from the authors and publisher Figure 7 [11].

1.4.2.1 Hemorrhage

Hemorrhage in the spinal cord can occur due to non-traumatic or traumatic etiology and can be categorized based on compartmental location. Overall, the presence of hemorrhage is predictive of reduced motor function after SCI (Flanders 1996), with toxic effects from the blood itself leading to necrosis, apoptosis and impaired regeneration (Zhong 2008, Whalley 2006, Asano 1980, Losey 2014). Losey (2014) notes that arterial injuries are rare in the arteries supplying the spinal cord and those adjacent to the spinal column. However, the vertebral arteries are frequently injured in cervical injuries, but these do not result in neurological deficits. Rather, intramedullary hemorrhage is believed to be largely responsible for the secondary damage after SCI. Microvascular intramedullary hemorrhage (hematomyelia) can be detected almost immediately after contusion SCI. Notably, as the grey matter (GM), which contains cell bodies, is more heavily vascularized, it displays more hemorrhagic necrosis after SCI [18,27]. Figley [28] notes that hemorrhage is narrowly limited to the immediate site of physical deformation. However, secondary microcirculatory hemorrhage has been reported to occur in nearby tissue, particularly in the cranial and caudal directions, and is linked with secondary lesion expansion [29–32].

1.4.2.2 Primary Ischemia

Ischemia can be defined as insufficient oxygen availability due to reduced/limited blood flow to tissue. Neurons and glial cells in the CNS display high metabolic activity and therefore especially sensitive to disruptions in oxygen and glucose supply. With primary spinal cord injury, the instantaneous mechanical damage to vasculature is often limited to smaller intramedullary vessels and capillaries rather than the larger arteries [16,17,33]. The immediate destruction of microcirculation results in local anoxia and thus directly leads to CNS cell death through lack of oxygen and nutrients [26].

1.4.2.3 Secondary Ischemia and Edema

In the hours after primary injury, local ischemia increases (Fehlings et al. 1989). Beyond the primary injury rupturing blood vessels in the spinal cord, perfusion is further restricted due to vasospasm, which can reduce blood flow by up to 80% [16,34]. Hypoperfusion in the injured spinal cord can also occur due to systemic hypotension as a result of damage to sympathetic circuits [35]. Additionally, thrombosis may play a role in secondary ischemia [19,30]. Ischemia increases local vascular permeability, compromising the integrity of the BSCB [36]. This, in addition to the high colloid pressure of the blood itself, contributes to local edema, which is especially prevalent in contusion injuries relative to transection or laceration injuries. Edema damages the cord by mechanically enlarging internal cavitations, leading to syringomyelia, which are a major barrier to axon regeneration and synaptic formation. Additionally, it has been reported that the increased intrasyringal pressure compresses surrounding tissue, resulting in regional ischemia and a local blood flow reduction of over 80% [20]. This suggests the existence of a detrimental feedback mechanism between ischemia and edema after initial vascular damage.

There is a delayed rise in cytoplasmic levels of arachidonic acid associated with tissue edema and Na+/K+ ATPase inhibition [37]. The activation of intracellular calcium-dependent phospholipase A2 and its subsequent action on membrane phospholipids provides a source of arachidonic acid. Cyclooxygenase 1 and 2 then convert arachidonic acid to Prostaglandin H2, which is further processed by platelets to produce thromboxane A2, which has prothrombic properties, further contributing to hypoxic ischemia [38]. Thromboxane B2 levels have been shown to increase disproportionally to levels of 6-keto-PGF10 in response to SCI, which may contribute to local hypoxia [39]. Furthermore, Sharma et al. [40] showed that inhibition of prostaglandins results in a decreased microvascular permeability after SCI. This suggests that prostaglandins may contribute to edema and associated ischemia. Ischemia has been linked to mitochondrial disfunction through irreversibly increased membrane permeability. The intensity of ischemia corresponds with the

mechanism of cell death, with subacute ischemia allowing for progression of the apoptotic pathway by maintaining a low level of ATP production. In contrast, fulminant ischemia rapidly disrupts mitochondrial function and leads to necrosis [41]. With the progression of the ischemic process, a late-hypoxic state is reached in which the mitochondria permeability increases drastically, leading to swelling and eventual plasma membrane rupture, which irreversibly damages the cell and causes necrotic cell death.

1.4.2.4 Ionic Imbalance

The intracellular accumulation of Na+ and depletion of K+ resulting from failure of the Na+ /K+ ATPase from prolonged ischemia and anoxia results in membrane depolarization and Na+ /Ca2+ exchange, leading to increased intracellular calcium ion concentration. This activates a range of apoptotic enzymes, including phospholipases, calpain, and protein kinase C (PKC) [42]. The glutamate concentration in spinal cord increases excessively after injury, including at the injury site, where it damages the CNS directly through excitotoxic cell death [43,44]. In the spinal cord, uptake of glutamate has been shown to be largely mediated by astroglial rather than neuronal transporters [45]. Excitotoxicity is a result of excessive activation of glutamate receptors, to which neurons and oligodendrocytes are particularly vulnerable, given their expression of all glutamate receptor types [46]. The axonal white matter was believed to be spared from excitotoxicity, due to its lack of synapses within the CNS [47]. However, Xu et al showed that after SCI, glutamate could reach toxic concentrations in white matter [44]. This supports the finding by Agrawal and Fehlings that axonal injury can occur through activation of non-NMDA ionotropic glutamate receptors [48].

Ionotropic glutamate receptor activation and intracellular Na+ accumulation leads to intracellular acidosis and cytotoxic edema [49–52]. Additionally, the increase in cytosolic Na+ increases intracellular Ca2+ concentrations to damaging levels by modulating activity of the Ca2+/Na+ exchanger [53,54]. Intracellular calcium is transported into the mitochondria via the potential driven Ca2+ uniporter. The levels of intramitochondrial Ca2+ remain elevated and continue to

accumulate with successive NMDA receptor stimulation [55,56]. Mitochondrial uptake of cytosolic calcium is linked to mitochondrial dysfunction and resultant cell death, as changes in mitochondrial polarization inhibits ATP production. This metabolic deficiency further exacerbates the hypoxia induced by primary and secondary vascular disfunction. Epstein also suggested a role for calcium dysregulation in reperfusion injury, where Ca2+ may contribute to the formation of membrane-damaging free radicals [57].

1.4.2.5 Oxidative Stress

Reactive oxygen species (ROS) cause apoptosis and necrosis of neural cells through damage of macromolecular cellular structures, including DNA, phospholipids and proteins [58,59]. ROS production peaks 12 hours after SCI and remains elevated before return to baseline levels within 4-5 weeks after SCI [60]. Structural and metabolic dysfunction in mitochondria due to ischemia and elevated intracellular Ca2+ cause ROS production [61]. Additionally, phagocytic leukocytes and microglia contribute to ROS due to increased oxygen consumption and production of superoxide through NADPH oxidase (NOX2) [58,62]. The elevated ROS results in extensive and damaging lipid peroxidation of fatty acids in cell membranes and myelin [63,64]. Notably, oxidative stress may contribute to BSCB dysfunction by promote downregulation of tight junction proteins and activating matrix metalloproteinases [65,66].

1.4.2.6 Neuroinflammation

Inflammation arising from cellular and molecular immunity occurs immediately after SCI and continues for several weeks [67]. The blood-spinal cord barrier (BSCB) is disrupted mechanically and chemically during primary injury. The immediate damage to the blood-spinal-cord-barrier (BSCB) leads to extravasation of large molecules and cells within minutes of injury [68,69] Damage to the BSCB increases the inflammatory response by allowing inflammatory cells into the site of injury [70,71]. Locally, microglia are activated through necrotic by-products from the

primary injury and these release pro-inflammatory cytokines, recruiting neutrophils and monocytes cell from the systemic circulation. These in turn, may further increase the permeability of the BSCB upon extravasation [72,73]. After compression injury in rats, the extent of disturbance to the BSCB was found to be correlated with deficits in functional motor abilities [74].

Initially, the immunological response to lesions in the brain and spinal cord is dominated by microglia, which secrete pro-inflammatory cytokines, recruiting leukocytes, which also upregulate pro-inflammatory mediators. Approximately 24 hours after SCI, neutrophils infiltrate the lesion site. Blood monocytes differentiate into macrophages and remain at the site of injury [75–77]. These chronically release proinflammatory cytokines, chemokines, nitric oxide, ions and proteases, resulting in fibrosis and cell death through apoptosis and necrosis [70,78].

1.4.2.6.1 Leukocytes

Neutrophils are not found natively in the spinal cord. After SCI, they enter the site of injury both passively through microvascular damage and active recruitment to the site of injury, where they phagocytose damaged tissue and produce inflammatory cytokines and proteases, which also serve as chemo-attractants for macrophages [76]. Losey et al. isolated the effects of spinal cord hemorrhage from primary tissue damage through an intramedullary collagenase model, demonstrating that intramedullary hemorrhage in the absence of primary tissue damage is associated with local leukocyte recruitment and that axonal injury continues for several days after the initial period of hemorrhage [79]. Activated neutrophils also induce breakdown of the BSCB through matrix-metalloproteinases, further contributing to extravasation [73,80].

Importantly, edema is related to leukocyte recruitment and infiltration [81] as knockout of leukocyte-expressed CD18 resulted in significantly decreased edema and death in collagenase- induced mouse intracerebral hemorrhage (ICH). However, anatomical differences between the brain and spinal cord vasculature may affect the extrapolation of brain-based ICH studies to SCI.

Importantly, leukocytes are more often recruited through the venous system than the arterial system and that in rats there are significantly more sulcal (central) veins than sulcal arteries [76]. Meanwhile, there are similar proportions of veins and arteries in the rat brain. This implies greater effects of neutrophil extravasation in SCI relative to the cerebral parenchyma. Losey et al. also noted the potential influence of elevated neutrophil recruitment in the spinal cord relative to the brain [79].

Additionally, Wallerian degeneration, the retrograde degradation of neurons away from the primary site of injury, is associated with the infiltration of monocytes. When in direct contact with damaged axons and myelin debris, macrophages undergo a process of polarization from anti- inflammatory to a pro-inflammatory state. In the CNS, macrophages may remain near the area of axonal degeneration for several months [82]. Thus, the increased infiltration of these cells due to hemorrhage, extravasation and a disrupted BSCB contributes to neural degradation and the associated functional deficits [83,84].

1.4.2.6.2 Microglia

Microglia have classically been described as existing in either a resting or activated state, although recently, it has been demonstrated that this relationship is more a spectrum than a dichotomy and that many intermediate types exist [85]. After SCI, microglia migrate to the site of injury, proliferate and phagocytose injured and dying cells. Additionally, they secrete cytokines and proteases that contribute to axonal damage and edema [86]. In the acute phase, microglia are distributed across a spectrum between the M1 and M2 phenotypes. However, with prolonged inflammation, the M1 phenotype becomes more prevalent, which produces more pro- inflammatory cytokines, resulting in a feedback cycle with an increase in M1-oriented microglial phenotypes. Meanwhile, the M2 phenotype is broadly anti-inflammatory and shows neuroprotective effects through increased phagocytic activity, matrix deposition and wound healing [85].

The neuroprotective effects of immune cells in SCI has been widely recognized. Notably, the astrocyte-derived glial scar prevents the spread of the lesion and the macrophages and microglia clear neuronal debris [80,83,85,87,88].

Vascular Anatomy of the Spinal Cord 2.1.1 Extraspinal Arteries

The superficial longitudinal vessels of the spinal cord originate intracranially from the vertebral artery (Figure 4). The subclavian arteries give rise to the vertebral arteries bilaterally, which pass through the transverse foramen of the cervical vertebra at C6 before heading towards the cranial cavity, where descending spinal branches anastomose at the level of the foramen magnum to form the anterior spinal artery (ASA). The vertebral arteries continue and eventually join to form the basilar artery while the ASA descends ventromedially in the anterior median fissure to the conus medullaris [89]. Typically, the vertebral arteries giving rise to the ASA are of different size, with one side being more dominant [90]. The thickness of the ASA varies along the length of the cord from 0.2-0.8 mm on average. It is at its thinnest in the cervical region, widest at the lumbar-sacral region (0.5-0.8mm) [91] and may duplicate itself, particularly in the cervical region [92]. These duplications of the ASA supply their own half of the cord [26,92,93]. Small-caliber lateral branches of the ASA, known as the pial arteries, run peripherally around the anterior and lateral surface of the cord at a concentration of approximately 1.6 per centimeter of cord [26]. These form the pial arterial network, to which the radiculopial arteries contribute [94].

Posteriorly, each branch of the posterior spinal artery (PSA) originates from the ipsilateral vertebral artery or from its posterior inferior cerebellar branch. The two posterior spinal arteries descend in their respective posterolateral sulci on the posterior pial surface of the cord. In some cases, these may decussate to supply the contralateral cord [89]. Additionally, there are bilateral lateral spinal arteries (LSA), which pass anterior to the posterior roots of the spinal nerves, while the PSA is found posterior to these nerve roots. The LSA supply the posterior and lateral aspects and originate intracranially from the vertebral artery or the posterior cerebellar artery [90,95]. All spinal nerve roots have associated radicular or segmental medullary arteries; however, most solely have radicular arteries. Both types of arteries run along nerve roots, but radicular arteries end

before reaching the anterior or posterior spinal arteries; in contrast, larger segmental medullary arteries continue on to supply a segment of the anterior or posterior spinal arteries [94]. Segmental medullary arteries contribute significantly to the ASA and PSA, maintaining adequate flow down the length of these longitudinal vessels.

The segmental arteries from which the segmental medullary arteries arise are derived from varying extraspinal sources, depending on the level. The segmental arteries have three main branches, the first being a spinal branch, which enters the spinal canal through the intervertebral foramina, along with the spinal nerve roots. The relative location of the spinal branch to the nerve roots varies significantly according to vertebral level [96]. The spinal branch further divides into dorsal and vertebral branches (anterior and posterior spinal canal arteries) and a small radicular artery, which bifurcates into the anterior and posterior radicular arteries, supplying the dura and nerve root at every vertebral level [97]. The radicular arteries which continue to the cord surface and supply the pial network on the dorsal surface are termed radiculopial [94]. The anterior and posterior radicular arteries which pass through the dura and supply either the PSA or ASA directly on the cord surface are classified as radiculomedullary arteries (RMA). In adulthood, these reach the cord in a limited number of levels (between 2-14) [98]. Anteriorly, there are a total of between 2 and 17 RMA contributing to the ASA, usually six to eight of significant size [99].

Posteriorly, there are approximately 11-16 radiculomedullary arteries, with the largest joining below the level of the AKA [92]. The segmental arteries and their radiculomedullary branches in the cervical region derive from the ascending cervical artery, whose spinal branches join with those from the vertebral and deep cervical artery. The segmental branches of the vertebral arteries also produce small RMA contributions to the ASA, with the largest usually around the level of C3. The deep cervical artery is helped by branches of the ascending cervical artery and may contribute small radicular feeder branches to the segmental arteries at C7 and C8. Branches of the ascending cervical artery also join the vertebral segmental arteries at C1-C6. The most important however is

a branch of the deep cervical artery, found between C4 and C8, typically following the C6 nerve root. This is known as the artery of cervical enlargement. In the cervical region, the ASA is usually by at least two sizeable anterior radiculomedullary feeder arteries. Spinal branches of the ascending pharyngeal and occipital arteries may also contribute segmental radicular feeder branches to the cervical region [92,99].

In the thoracic region, RMAs arise from the posterior segmental intercostal arteries, which originate from the thoracic aorta [97]. The upper (T3-T7) and mid-thoracic regions of the ASA were historically believed to be solely supplied by smaller anterior radiculomedullary arteries however, Gailloud [100], recognized a consistent significant anterior upper thoracic RMA contributor, naming it the artery of von Haller. Notably, the lower thoracolumbar region is believed to be mostly supplied by a single radiculomedullary artery - the artery radicomedullaris magna, also known as the artery of lumbar enlargement and commonly referred to as the artery of Adamkiewicz (AKA). The AKA has been reported to be between 0.6 and 1.9 mm in diameter [96,101,102]. It originates variably from the left posterior intercostal and lumbar segmental arteries or the intercostal arteries [101–104] and produces both a thin ascending branch and a larger descending branch and forms a characteristic “hairpin” loop caudally before its anastomosis with the ASA [105,106]. It is commonly located on the left side between T8 and L1 [107] and this observation is supported by a recent meta-analysis of anatomical studies of the AKA by Taterra et al.[108]. Below the contribution of the AKA to the ASA, it has been traditionally accepted that there exists little alternative supply to the anterior spinal cord [92]. Alternatively, Lazorthes et al [109] have suggested the importance of the richly anastomosed network found on the vertebrae and extra-dural space of the spinal canal. More recently Griepp et al [110], expand on this model and suggest that the arterial supply of the thoracic and lumbar cord is supported by a collateral network including all of the intercostal and lumbar segmental arteries, several of which contribute RMA branches to the ASA. These are interconnected with the epidural arterial network and small vessels in the paraspinal musculature, which are also proposed to contribute to the ASA. All of

these vessels have significant anastomoses with the cauda equina and sacral arteries [110]. The idea of alternative arterial contributions to the lower two-thirds of the cord is supported by Murikami et al. [111], who showed that interruption of the AKA during total en-bloc spondylectomy in 15 patients did not lead to any neurological defects. Etz et al [112] further support the concept of a collateral network and emphasize the importance of the densely anastomosed vasculature in paraspinous muscle, which was previously noted to be absent at irregular levels [97]. The periconal anastomotic circle typically consists of bilateral anastomoses of the ASA, turning backwards to continue dorsally as the lumbosacral PSA’s. These anastomoses have been recently described to as occasionally occurring within the spinal cord parenchyma [113]. These authors also reported that radicular branches joined mainly with the posterior anastomotic basket and did not observe a high incidence of anterior radiculomedullary feeders below the AKA.

Figure 4 Extraspinal arteries of the spinal cord.

(1) Basilar artery; (2) vertebral artery; (3) anterior spinal artery; (4) posterior spinal arteries; (5) anterior radiculomedullary artery; (6) ascending cervical artery; (7) deep cervical artery; (8) subclavian artery; (9) posterior radiculomedullary artery; (10) segmental arteries (posterior intercostal arteries); (11) great anterior radiculomedullary artery or artery of Adamkiewicz; (12) segmental arteries (lumbar arteries); (13) rami cruciantes. Reused with permission from Figure 1 [92].

2.1.2 Intraspinal Arteries

The intraspinal parenchyma is supplied by both a centrifugal system and a radial arterial system, largely derived from the ASA and PSA, respectively (Figure 5). The capillary beds of the centripetal and centrifugal networks are concentrated especially in the gray matter, due to the higher metabolic demands of the cell bodies, each of which is in usually contact with at least one capillary loop [93,97]. The gray matter displays dense, highly curved and directionally irregular capillary mesh, while white matter capillary beds are typically looser, more rectangular and run parallel to their fibers [99].

The ASA-derived central arterial system supplies the anterior two thirds of the spinal cord. They arteries provide for the anterior horns and the base of the posterior horns. Additionally, they supply the anterolateral white matter, including the pyramidal, spinothalamic, and the dorsal column- medial lemniscus pathway [26,92,99]. Bosmia [99] notes that the different cellular organization of the anterior and posterior grey matter is reflected in their respective vascular supplies. The more uniform posterior gray column features a constant depth of vasculature (largely from the radial system), while the anterior gray matter’s irregular arrangement of cell bodies is supplied by a network of erratically branching vessels from the centrifugal system. Sulcal (central) arteries arise from the ASA, enter the cord anteriorly and run posteriorly along the anterior medial fissure. Once they reach the anterior white commissure, these usually turn to either the left or right side of the cord, typically alternating successively to provide approximately equivalent lateral perfusion in each segment. This occurs most consistently in the cervical and lumbar region compared to the thoracic [99].

Occasionally, a sulcal artery will bifurcate, in the sagittal plane, into both sides of the cord. [92,93,98]. Tator et al. [26] note that this occurs in approximately 19 percent of sulcal arteries. In contrast, previous authors [114] report that each sulcal artery only supplies a single side of the cord. Interestingly, these authors also note that in the lumbosacral region, these may bifurcate into

two branches both entering unilaterally. Turnbull [93] reported that in the cervical region, 5-8 sulcal arteries arise per centimeter from the ASA. This concentration decreases to 2-5 branches per centimeter in the thoracic region before increasing again in the cervical and lumbar regions to 5-12 branches per centimeter. The number and size of sulcal arteries is greater in the cervical and lumbar regions than in the thoracic cord [93]. These observations are also supported by Dommisse [97]. This deficiency is partially resolved with sulcal branches, which extend craniocaudally on the surface of anterior medial fissure and anastomose with neighboring sulcal branches [94] before their lateral entry into the cord. These branches are more significant in the thoracic region, where the individual vertical distance covered by central arteries is larger, at 3.0 centimeters average, compared to 1.7 and 1.2 centimeters in the cervical and lumbosacral regions, respectively [93] The sulcal intramedullary capillaries also overlap craniocaudally with the neighboring sulcal capillaries [26].

Superficially, the ASA and PSA are interconnected via horizontal transverse branches known as coronary perimedullary arches [115] forming a rich network of effective longitudinal and transverse anastomoses known as the pial plexus or vasocorona. The radial arterial system consists of perforant branches of this pial plexus and the of the PSA itself, which supply the parenchymal lateral and dorsal periphery of the cord [26,92,94,98]. This network supplies most of the dorsolateral white matter columns and reaches the periphery of the gray matter laterally [93]. Additionally, a number of the terminal arterioles of these perforant branches directly anastomose with the sulcal arterioles [99], providing a variably effective collateral network for perfusion of the anterior third of the cord.

Figure 5 Intraspinal Arteries of the Spinal Cord.

(1) Posterior spinal arteries; (2) anterior spinal artery; (3) spinal branch; (4) anterior radiculomedullary artery; (5) posterior radiculomedullary artery; (6) central (sulcal) arteries; (7) vasocorona. Reused with permission from Figure 3 [92].

2.1.3 Arterial Hemodynamics

Notably, angiographic studies have demonstrated that the flow in the ASA and PSA can occur in both ascending and descending directions. This may in part due to the bifurcation of the RMA’s providing both a cranial branch and a caudally-directed branch before anastomosis with the PSA and ASA. In fact, it has been suggested that the spinal arteries can be functionally considered a series of anastomotic channels rather than continuous vessels. The area between the cranial and caudal anastomoses of the contributing RMA displays little flow in either direction, creating a so- called “watershed” zone [115]. However, in the cervical ASA, the predominant overall direction of flow is downwards, while below the upper thoracic segments the flow is in the opposite direction [99]. This description of overall caudocranial flow in the caudal ASA are inconsistent with the AKA’s significant provision of mostly downward flow in this region [106]. A reduction in caliber of the ASA above the contributions of the radiculomedullary spinal arteries suggests a mechanical barrier to the reverse direction of downward flow in the cervical region. This feature is not observed in the PSA, although the predominant direction of flow in the posterior cervical region is also downwards, originating from the posterior cerebellar artery. Below the cervical region however, the direction of flow in the PSA is caudocranial, as it is provided from the ASA’s caudal anastomosis and a contributing intersegmental artery [99,116,117].

2.1.4 Venous Anatomy

In contrast to the central arterial system, sulcal veins collect from the medial anterior horns, white matter of the anterior funiculus and anterior gray commissure bilaterally (Figure 6). The radial veins are fed by capillaries in the peripheral gray matter of the lateral horns and the white matter and form a ring on the cord surface. The longitudinal veins are highly anastomosed and contain a network collateral branches. Anteriorly, the anterior medial spinal vein (ASV) follows a similar

course as the ASA, is largest in the lumbosacral region and receives blood from the sulcal veins and veins of the central fissure.

Posteriorly, there are multiple longitudinal spinal veins, known as the posterior spinal veins (PSV), with the medial PSV being the most important and up to two posterolateral spinal veins following a similar course as the PSA. Analogous to the arterial system, the dorsal radial veins supply the PSVs. On their respective aspects of the cord, the medial PSV and ASV drain into the numerous radiculomedullary veins. There are approximately 8-10 radiculomedullary veins anteriorly and 5- 10 posteriorly. The most important of these is the great anterior radiculomedullary vein (GARV), which is spatially and directionally similar to the AKA [118]. These radiculomedullary veins drain to the paravertebral and intervertebral plexuses and eventually the pelvic venous plexuses while the longitudinal veins communicate with the internal vertebral (epidural) venous plexus via the anterior and posterior radiculomedullary veins [92].

Figure 6 Venous Drainage of the Spinal Cord.

(1) Anterior median vein; (2) right deep cervical vein; (3) left deep cervical vein; (4) right vertebral vein; (5) left vertebral vein; (6) subclavian vein; (7) internal jugular vein; (8) left brachiocephalic vein; (9) superior vena cava; (10) accessory hemiazygos vein; (11) intercostal veins; (12) posterior radiculomedullary vein; (13) anterior radiculomedullary vein; (14) azygos vein; (15) hemiazygos vein; (16) lumbar veins; (17) vein of the filum terminale. Reused with permission from Figure 4 [92].

2.1.5 The Blood Spinal Cord Barrier (BSCB)

Endothelial cells in the CNS are characterized by the presence of tight junction proteins between cells and a lack of pinocytic vacuoles. These also have a high number of cytosolic mitochondria, which is suggestive of elevated metabolic requirements for selective active transport. Importantly, these also lack membrane fenestrations [119].

There are differences between the blood-brain barrier (BBB) and BSCB. Notably, the BSCB (Figure 7) displays increased permeability to tracers, and this may be due to reduced levels of tight junction proteins ZO1 and occludin [120].

2.1.5.1 Pericytes

Approximately ¼ of the rat CNS capillary surface is covered with pericytes [121,122], which are separated from the endothelial cells by a basal lamina. These have various important roles in the BSCB and act through regulation of gene expression in endothelial cells and regulation of astrocyte end-foot polarization. They are therefore important in regulating permeability of the BSCB [123,124] and mediate inflammation [125], angiogenesis [126] and capillary diameter [127,128]. They also differentiate into scar tissue after SCI

2.1.5.2 Astrocytes

The CNS capillaries are surrounded by astrocytic end feet, which contribute to the BSCB’s physical barrier and influence the gene expression of epithelial cells through paracrine molecular control. Given their location between capillaries and neurons, these are also heavily involved in ionic, water, amino acid and neurotransmitter homeostasis [129].

2.1.5.3 Basement Membrane

The basement membrane (BM), also known as the basal lamina, is an organized layer of extracellular proteins approximately between 20 and 200 nm thick [130]. Structurally, the endothelial cell basement membrane underlies the endothelium. Meanwhile, the meningeal epithelium contributes to the astroglial basement membrane and this combination is known as the parenchymal basement membrane. In post-capillary venules, the parenchymal basement membrane and the endothelial cell basement membrane are distinguishable due to differing concentrations of laminin [131]. However, in capillaries, these distinct basement membranes fuse to form a single basement membrane [132–134]. The major BM proteins in the BSCB include laminin and collagen IV (stabilized by nidogen), fibronectin, osteonectin, agrin and HSPG2 (perlecan).

2.1.5.4 Collagen IV

The three main isoforms of trimeric Collagen IV are derived from 6 types of α chains. The common isotype in the CNS is formed from two α1 chains and a single α2 chain [130].

2.1.5.5 Laminin

Laminin is a heterotrimeric protein with various isoforms, which are produced by distinct cell types [135]. These isoforms are a result of 11 distinct genes coding for its five different α, three β and three γ subunit forms [136]. A widely implemented nomenclature system identifies isoforms by chain composition, where for example, the configuration α1β1γ1 is abbreviated to laminin 111 [137]. Brain vascular endothelial cells have been shown to produce mostly laminin 411 and 511. Meanwhile, astrocytes produce laminin 111 and 211, and pericytes produce α4 and α5- containing laminins. Laminin isoforms 411 and 511 compose the endothelial BM, while the parenchymal BM is composed of laminin isoforms 111 and 211. The fusion of these two distinct BM layers in capillaries results in a single BM layer composed of 411, 511 and 211 isoforms

[130]. Laminin α4 is believed to promote leukocytic extravasation, while laminin α5 inhibits inflammatory cell infiltration [138]. Laminins have been shown to play a role in maintaining astrocytic end foot structure, regulating tight junction and aquaporin expression and preventing pericytes from differentiating into the harmful contractile stage [139]

Figure 7 The Cellular Components of the Neurovascular Unit (NVU).

The capillary wall contains a single EC layer (green) connected by tight junctions that form the blood-brain barrier along with pericytes (red), that are embedded within the capillary basement membrane (light blue), and nearby astrocyte end feet (yellow). Excitatory neurons (blue) synapse with both vasoactive interneurons (purple), and astrocytes (yellow), who in turn signal to the

capillary to alter blood flow according to the metabolic demands of that brain region. Microglia (indigo) are located in the brain parenchyma and respond to any aversive stimuli to protect the brain. Reused with permission from Figure 2 [140].

2.1.6 Lymphatic Drainage in the Spine

The brain and spinal cord have classically been described as lacking parenchymal lymphatic vessels. The cerebrospinal fluid, produced in the cerebral choroid plexuses, drains into the venous sinuses through arachnoid villi in the subarachnoid space. Additionally, lymphatic drainage through the cribriform plate and dural lymphatic vessels reaches the cervical lymph nodes [141].

Interstitial fluid (ISF) first enters between astrocytic end feet into the basement membrane of capillaries before entering the tunica media, perivascular drainage pathway surrounding arterioles and arteries [141]. Meanwhile, paravascular, Virchow-Robin spaces allow interstitial fluid exchange with the cerebrospinal fluid. This has been termed the glymphatic system [142] and its specific fluid dynamics remain unclear[143].

Recently, a meningeal lymphatic system was described [144], which may share structural features with peripheral lymphatics. There are conflicting reports regarding the presence of lymphatic valves, as Louveau (2015) described a lack of valves and smooth muscle, while Antila (2017) reported their presence in a developmental study [145]. Louveau also described the drainage of CSF from meningeal lymphatics to deep cervical lymph nodes, suggesting a role in immune surveillance [144]. Therefore, the meningeal lymphatic system may modulate the immune response to CNS injury.

Similarly, vertebral lymphatic vessels (vLV) connect to the peripheral lymph nodes, exiting along spinal nerve rami. These are connected to cranial meningeal lymphatic vessels in the upper cervical region [145]. Jacob and colleagues (2019) report that in mice in the thoracolumbar region, these drain towards thoracic mediastinal lymph nodes [146]. These authors also note that, along the

length of the spinal column, vLVs connect with peripheral LV’s at the level of each vertebra, rather than through large longitudinal dorsal or ventral vessels [146]. Additionally, epidural vertebral lymphatics may play a role in immune-cell infiltration of the spinal cord, as inducing lymphangiogenesis was reported to increase infiltration and demyelination in a lysolecithin model of demyelination [146]. Furthermore, surgical removal of lymph nodes in an experimental autoimmune encephalomyelitis mouse model reduced damage to the spinal cord [147].

The roles of the meningeal and vertebral lymphatic vessels have not yet been explored in the context of acute spinal cord injury and may present a novel target for modulating the immune response to SCI.

2.1.7 Level-dependent Differences in Vascular Architecture

Muradov et al show that the extent of axon and oligodendrocyte loss is associated with focal microvascular damage [148]. The grey matter has been found to be more vulnerable to blood flow disturbances after clip-compression SCI as compared to white matter and typically contains relatively smaller vessels [28]. The white matter adjacent to hemorrhagic grey matter is also especially prone to ischemic damage [16]. Overall, the varying arterial and gray matter density throughout the cord has implications for primary and secondary injury after SCI. Regions with reduced vessel density, associated with regions of lesser gray matter, may display reduced focal hemorrhage and its associated secondary effects, including focal ischemic compromise. However, these regions may also be more susceptible to widespread ischemic compromise arising directly from the primary injury. In contrast, areas with increased arterial and microvascular density may display increased focal hemorrhage and secondary injury, while being less susceptible to widespread ischemic compromise from the primary injury.

As noted previously, the ASA is largest in the cervical and lumbar regions, which corresponds with an increased requirement in order to supply the enlarged ganglions at these levels.

Additionally, the central sulcal arteries are both larger and increased in number in the cervical and lumbar regions relative to the thoracic region. The significant narrowing of the spinal cord in the mid-thoracic region corresponds with the narrowing of its major longitudinal arteries and the reduced perfusion requirements from T4 to approximately the level of T9. The relatively reduced arterial supply renders this area more susceptible to ischemic damage upon vascular insult [93,97] Importantly however the collateral circulation in any given area may be as relevant to the local blood supply as the number of arteries [114].

Compared to the thoracic region, the cervical region also has a larger amount gray and white matter and a greater blood supply. The increased blood supply may correspond with greater vascular disruption in the cervical region following SCI. This is supported by greater cavitation and lesion in cervical injuries when compared to thoracic injuries. Additionally, injury at C5 caused neuronal death and ventral horn atrophy over three segments (C6-C8), while injury at C6 impacted two segments (C8-T1) and C7 caused injury over one segmental level (T2) [149]. These results are consistent with the higher vascular demands from elevated grey matter content in the limb enlargements, which would be disproportionately affected by secondary injury stemming from vascular disruption. Notably, due to a decreased amount of pericytes, the BSCB is inherently more permeable in the grey matter of the cervical region as compared to the thoracic region [150].

Whether the spinal cord is compressed dorsally or ventrally, either by mass effect or through a traumatic mechanism, it flattens anteroposteriorly and widens laterally. As a consequence, the perforant branches of the pial plexus running directly in the anteroposterior direction, which supply the anterior and posterior columns, are compressed slightly lengthwise and thus shortened but remain functionally unaltered. In contrast, the lateral perforant branches supplying the lateral columns are stretched lengthwise and flattened, decreasing their diameter. The same effect is observed in the central arterial system, in which the lateral segments are flattened and the anteroposterior segment in the anterior medial fissure is less dramatically affected [93].

Alternatively, in their model of spinal perfusion, Alshareef et al. demonstrate that the reduction in perfusion due to compression is associated with vessel distance from the dura mater and suggest that posterior compression may have greater likelihood of producing ischemic injury when compared to anterior compression [19].

The increased mobility, smaller vertebrae and reduced stabilizing musculature of the cervical spine may make it more susceptible to injury (Figure 8) [18,151]. It is also believed that the cervical region is more susceptible to contusion injury due to it relatively smaller subdural space [20] Of the thoracolumbar junction, Liu also notes that the limited distance between components of the lower thoracic and lumbar spine may cause especially extended tissue damage in this region [152]. However, Vaccaro et al report that that a spinal canal with a larger transverse diameter was predictive of neurologic deficit after burst fractures and thus that larger canals are not necessarily protective of the spinal cord [153].

As discussed previously, the varying hemodynamics among regions of the cord result in several watershed regions with relatively lower perfusion. Notably, the upper thoracic cord between the cervical and lumbar limb enlargements, the anterolateral surface between the coronary perimedullary arches, and the intramedullary gray-white junction between the centripetal and centrifugal arterial systems [99].

The level of injury may affect the systemic inflammatory response and by extension, the local inflammatory environment. Notably, the expression of both pro-inflammatory and anti- inflammatory compounds may be influenced by level of injury. Hong et al (2018) showed reductions in systemic circulatory levels of pro-inflammatory molecules with lesions in the cervical region when compared to other regions. SCI above the mid-thoracic level disrupts connections to sympathetic preganglionic neurons below that level, which control secondary lymphoid tissues and systemic immune function. While this dysfunction may lead to spinal cord injury-induced immune depression syndrome (SCI-IDS), it may also result in reduced chronic

autoimmunity [151,154]. Ibbara et. al demonstrated that the level of injury influences the functions of T cells, which are implicated in the severe infections accompanying SCI-IDS. The type of SCI (complete vs. incomplete) may also be an important factor in the immunologic depression observed after SCI [155] . As recently demonstrated by Hong et al., contusion injuries may increase splenic function [156]. This contrast with the previous studies which used transection injuries and resulted in decreased peripheral immune function.

Figure 8 Cervical and High Thoracic SCIs Disrupt the Outflow of the Sympathetic Nervous System.

Injuries in the cervical and high thoracic cord can disrupt the sympathetic outflow (blue line) to the heart and the peripheral vascular system, while preserving baroreceptor inputs (red line) and parasympathetic output (green line). As a result, parasympathetic innervation to the heart dominates in patients with cervical and upper thoracic injuries, which causes bradycardia and decreased cardiac output. This is further compounded by the loss of peripheral muscular and vascular tone, which promotes a redistribution of blood to the periphery with reduced venous

return. Consequently, patients often experience hypotensive symptoms, particularly with exertion or upright positioning. The parasympathetic–sympathetic imbalance can also allow unchecked reflex spinal sympathetic stimulation as a consequence of noxious triggers (such as bladder distension or pressure sores), which leads to sudden peripheral vasoconstriction and acute hypertension. As a response, parasympathetic outflow above the injury level increases, leading to vasodilation, headaches, sweating and sinus congestion. This dangerous cardiovascular complication of SCI is known as autonomic dysreflexia. S2–S4, sacral levels 2–4. Republished with permission from the authors and publisher [11].

2.2 Animal models of SCI

To-date, several animal models of traumatic SCI have been characterized and have played a quintessential role in deepening our understanding of the SCI pathogenesis. As with any experimental model, the ideal species should possess anatomical and pathological similarities to human SCI and be consistently reproducible. The most commonly used model is that of rat and murine SCI. In these two models, the injury response to SCI is like that of humans, but differences in animal size, molecular signaling, capacity for recovery make direct translation from these rodent models difficult. Large animal models such as porcine and non-human primates, overcome some of these limitations, but cost and special animal housing make it difficult to adopt these models. Nevertheless, both rodent and large animal models should be used in concert prior to taking any preclinical therapeutic paradigm to trial.

Table 1 Animal Models of SCI

The type and location of SCI are key factors in the development of a clinically-relevant and translational animal model. The anatomical and pathophysiological differences between the cervical and thoracic spinal regions are substantial and should be considered. Furthermore, the choice of species and type of model might be useful in answering different scientific questions. Republished with permission from the authors and publisher from Box 2 [11].

Model Description

Contusion models inflict transient, acute injuries through weight- drop, or electromagnetic or pneumatic impactors

Compression models inflict prolonged, acute injuries through calibrated clip-compression and forceps, among others

Transection models Inflict unilateral (partial) or bilateral (complete) lesions

2.3 Clinical Management of SCI

2.3.1 Spine Immobilization and Stabilization

All patients with a suspected spinal injury should have their spine immobilized for transport during emergency transportation and in early in-hospital management until ruled out according to neurological assessment and/or diagnostic imaging, when appropriate. Prehospital immobilization (more accurately termed spinal motion restriction), has the aim of preventing further mechanical damage to cord the and remains standard procedure in most centers. The 2012 AANS/CNS guidelines provide a level II recommendation for spinal immobilization of all trauma patients with a cervical spine or spinal cord injury or a mechanism with potential for SCI. However, trauma patients with a normal neurological evaluation who are awake alert and not intoxicated with no neck pain should not be immobilized. Notably, spinal immobilization is not recommended in patients with penetrating trauma due to increased mortality from delayed resuscitation. Prolonged immobilization also increases morbidity, including pressure sores and airway complications. In awake and symptomatic patients, the 2012 AANS/CNS Guidelines for the management of cervical spinal injuries provide a level I recommendation for radiographic evaluation using high-quality CT. Alternatively, in the absence of high-quality CT imaging, 3-way cervical spine radiographs are recommended. These also provide a level I recommendation against radiographic evaluation of the cervical spine in awake and asymptomatic patients with a normal neurological examination, without neck pain or tenderness and who can complete a range of motion evaluation. Alternatively, the Canadian C-spine Rule determines the need for radiography based first on the following High- Risk Factors: age >65 years, a dangerous mechanism (fall from elevation, axial load to head, high speed MVC, motorized recreational vehicle, bicycle struck or collision) or paresthesia in extremities. In the absence of a high risk factor, the inability to safely assess range of motion or inability to actively rotate the neck 45 degrees both left and right indicates a requirement for radiography [157,158]. The 2018 AANS/CNS guidelines on thoracolumbar spine injuries provide

a grade B recommendation for the use of MRI to assess posterior ligamentous complex integrity when determining the need for surgery, but note that there is insufficient evidence that radiographic findings can predict clinical outcomes in thoracolumbar fractures [159].

2.3.2 Hemodynamic Management

Neurogenic shock may worsen secondary spinal cord injury through hypoperfusion and ischemia. Spinal cord perfusion pressure (SCPP) is defined as the difference between intrathecal pressure (ITP) and mean arterial pressure (MAP). After SCI, a decrease in MAP may occur either through loss of sympathetic vasomotor tone, leading to pooling of blood in the extremities, or interruption of sympathetic input to cardiac fibers, which results in arrhythmias due to exclusive parasympathetic contribution. Furthermore, hypovolemia from concurrent injuries can result in hypotension, although this is usually associated with tachycardia, rather than the bradycardic hypotension commonly developing from neurogenic shock. These complications are most effectively recognized and managed in an intensive care setting [160]. Initially, restoration of normal hemodynamic parameters should be attempted using crystalloid-based volume resuscitation, although this may not be effective if cardiac output is compromised. Vasopressors should be used to obtain normotension, and those with both alpha and beta-adrenergic effects are preferable as the lack of beta-adrenergic stimulation can result in reflex bradycardia. In 2008, The Consortium for Spinal Cord Medicine included a recommendation for either dopamine or norepinephrine in cervical and high thoracic SCI due to dual alpha and beta-adrenergic effects. In lower regions phenylephrine was suggested [161]. With aims to increase SCPP, several other clinical practice guidelines have recommended maintenance of a MAP of 85-90mmHG for up to 5-7 days after SCI [160,162]. Regarding thoracolumbar injuries specifically, the AANS/CNS guidelines found insufficient evidence to recommend for or against maintenance of arterial blood pressure. They instead note that results from pooled (cervical + thoracolumbar) populations indicate that clinicians may choose to maintain MAP > 85mmHg. However, in the 2012

AANS/CNS guidelines on management of cervical injuries, a level III recommendation is provided to maintain MAP between 85 and 90 mmHg in the first week after an acute spinal cord injury to improve cord perfusion [163]. A recent meta-analysis reported that in two prospective studies, neurological outcomes were stable to-improved with management of MAP above 85mmHg and 90mmHg. These authors note that norepinephrine is preferred for cervical and upper thoracic injuries and phenylephrine or norepinephrine for mid-to lower thoracic injuries [164]. In 2016, a crossover evaluation was undertaken in a series of 11 patients with cervical or thoracic SCI, using norepinephrine and dopamine. While maintaining a similar MAP, a significantly higher spinal cord perfusion pressure (SCPP) was obtained using norepinephrine when compared to dopamine. This was achieved through a significantly lower intrathecal pressure (ITC) with the norepinephrine administration. A recent systematic review suggests that dopamine is associated with higher rates of complications than phenylephrine and supports the view that norepinephrine slightly increases perfusion pressure relative to dopamine and phenylephrine [165]. Although no significant differences in neurological outcomes were observed, other reports suggest that cardiogenic complications were more strongly associated with the use of dopamine (OR 8.97, p<0.001) than phenylephrine (OR 5.92, p = 0.004). Both types of vasopressors were associated with increased complications in SCI [166]. This is supported by Readdy et al, who reported that cardiac complications were associated with both dopamine and phenylephrine but that the differences were not statistically significant [167].

2.3.3 Neurological Assessment

The 2018 AANS/CNS guidelines on the evaluation and treatment of patients with thoracolumbar spine trauma contain a Level C recommendation for the use of the Functional Independence measure, Sunnybrook Cord Injury Scale, Frankel Scale for SCI while noting that the American Spinal Injury Association International Standards for Spinal Cord Injury Classification scale (ASIA) has not specifically been validated in thoracolumbar injuries. However, a Level B

recommendation is given on the use of Entry ASIA scale grade, sacral sensation, ankle spasticity, urethral and rectal sphincter function, and AbH motor function in predicting neurological function and outcome in patients with thoracic and lumbar fractures [168]. The 2012 AANS/CNS guidelines on the management of acute cervical spine and spinal cord injury contain a Level II recommendation for the use of ASIA as the preferred tool for assessment of neurological status and a Level I recommendation for the use of the Spinal Cord Independence Measure (SCIM III) for assessment of functional status [169]. Despite insufficient evidence supporting the use of AIS scale in thoracolumbar injury specifically, it remains the most widely used assessment tool for determining the level of injury, extent of injury and functional status. The ASIA system involves sensory and motor examinations to determine the neurological level of injury (NLI) and determine a score on the ASIA Impairment Scale (AIS). The NLI is defined as the most caudal cord segment with intact sensory and antigravity muscle function. The level of sensation on each side is defined as the most caudal intact dermatome for both pinprick (sharp/dull discrimination) and light touch sensation. The motor level is determined by the lowest key muscle function rated at least 3 (full ROM against gravity without resistance). AIS grade A is used to describe complete injuries, in which no sensory or motor function is preserved in the sacral segments S4-S5A. This requires the absence of voluntary rectal sphincter function, no deep rectal pressure sensation and a score of 0 in all the S4-5 sensory tests. AIS grade B describes sensory incomplete injuries, in which sensory but not motor function is preserved below the NLI, including the sacral segments S4-S5 and no motor function is preserved more than 3 levels below the motor level (lowest level with full ROM against gravity) on either side of the body. AIS grade C indicates a motor incomplete injury, in which motor function is preserved below the NLI and more than half of key muscle functions below the single NLI have a muscle grade < 3. AIS grade D indicates a motor incomplete injury in which at least half of key muscle functions below the NLI have a muscle grade > 3. A grade of E refers to normal sensation and motor function in all segments [170].

Table 2 ASIA Impairment Scale.

The American Spinal Injury Association (ASIA) Impairment Scale grade is a global measure of injury severity and is largely based on the concept of sacral sparing (that is, some degree of maintained perineal sensation, voluntary anal contraction and/or great toe flexion indicating an incomplete lesion). The scale is used to determine the grade of spinal cord injury (SCI), which ranges from ASIA Impairment Scale grade A (the most severe injury with complete sensorimotor loss) to ASIA Impairment Scale grade E (the least severe injury with no neurological deficit). Republished with permission from the authors and publisher from Box 2 [11].

ASIA Impairment Scale

Grade Description

A Sensory or motor function below the neurological level (that is, the lowest segment where sensorimotor function is normal on both sides) of injury, including absent sacral function (that is, no voluntary anal contraction, no great toe flexion, and no perineal, genital, anal pinprick or light touch sensation).

B Sensory but not motor function is preserved below the neurological level of injury, including the distal sacral segments (S4– S5). No motor function is present more than

three levels below the neurological level, on either side of the body.

C Motor function below the neurological level of injury (including the distal sacral segments) is preserved, with more than half of the key muscles (that is, elbow flexors and extensors, wrist extensors, finger flexors and abductors, hip flexors, knee extensors, ankle dorsiflexors, long toe extensors and ankle plantar flexors) having a grade of <3 on the ASIA motor score (against gravity without additional resistance).

D Motor function below the neurological level of injury (including the distal sacral segments) is preserved with more than half of the key muscles having a grade of ≥3 (antigravity) on the ASIA motor score.

E Neurologically intact patients (that is, sensorimotor function is normal in all segments) who previously had deficits secondary to a suspected SCI.

2.3.4 Operative vs. Nonoperative management and Timing of Decompression

Surgical fusion and/or decompression are required when injuries to the spine result in instability. Spinal instability may be quantified using several different scales, however in the cervical region the Subaxial Spine Injury Classification (SLIC) scale delineates a need for surgical intervention at a score of 5 or more, while a score of 4 is equivocal and a score of 3 or less should be treated non- surgically. This system demonstrates an inter-rater intraclass reliability coefficient of 0.71 and its use is supported by a level I recommendation from the AANS/CNS 2012 guidelines. These guidelines also recommend use of the Cervical Spine Injury Severity Score (CSISS), which has higher inter-rater reliability (0.883), but is more complicated and may be more challenging to implement clinically [171]. Analogously, in the thoracolumbar region, the Thoracolumbar Spine Injury Classification (TLIC) or AO Spine Classification systems can be used to characterize the injury [172] and predict the need for surgical management. Similarly to the SLIC, the TLIC also defines the need for surgical intervention at a score of 5 or above, with 4 being equivocal and 3 or less treated non-operatively [173]. Thus far, the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) provides the highest-quality evidence supporting early decompression (<24h). In a prospective, multicenter cohort study, patients who had surgery within 24 hours showed a higher rate of 2-or-more point improvement in AIS grades at 6 months follow-up. However, these results should be evaluated in the context of the study’s limitations stemming from the cohort study design and the reported lack of statistical difference between treatment groups in a 1-point grade improvements on the AIS scale [174]. A 2017 systematic review noted that the results and quality of evidence for early surgical decompression were variable depending on level, timing of follow- up and outcome but suggested that sufficient evidence existed to support the improved neurological recovery for cervical SCI patients with early surgical intervention [175]. The 2017 AOSpine Guidelines suggest that surgical decompression within 24 hours be considered as a

treatment option in adults with traumatic central cord syndrome (Figure 9). They also suggest that early surgery be offered to adults with acute SCI regardless of level. The quality of evidence was low, and the strength of these recommendations were weak. Recently, a study applying the AOSpine Subaxial cervical spine trauma classification system concluded that early decompression (<72h) is not required for Type A and F1-3 fractures. In contrast, early surgical treatment was especially beneficial in Type B and Type C/F4 fractures [176].

The 2018 AANS/CNS Guidelines [164] state that in patients with thoracolumbar burst fractures who are neurologically intact, the conflicting evidence is not sufficient to make a recommendation and the discretion of the treating provider be used. These guidelines also state that there is insufficient evidence to recommend for or against surgical intervention in non-burst thoracic or lumbar injuries, and similarly recommend that the decision be at the discretion of the treating physician. The 2012 AANS/CNS Guidelines give a level III recommendation to reduce subaxial cervical fractures or dislocations with the goal of spinal cord decompression/restoration of the spinal canal. Stable immobilization by internal fixation or external immobilization is recommended. If surgical treatment is considered, either anterior or posterior fixation and fusion is acceptable if a particular approach is not required. If these treatment options are not available, prolonged bed rest in traction is the recommended treatment. In patients with ankylosing spondylitis, posterior long segment instrumentation and fusion or a combined dorsal/anterior procedure is recommended as anterior standalone instrumentation and fusion is associated with a failure rate of up to 50%.

Figure 9 Surgical decompression and realignment of the injured spinal cord.

Arrows mark the cervical level 5 (C5)–C6 where the injury is centered. a) Preoperative CT imaging demonstrates a severe C5–C6 fracture dislocation (arrow), with compromise of the central spinal canal. b) Preoperative MRI shows ongoing compression of the spinal cord (arrow) and a bright T2-weighted signal in the surrounding ligaments that is suggestive of disruption. c) Following surgery, including cervical traction, surgical decompression and instrumented fusion anterior and posterior metal hardware can be seen on the CT, and the restoration of appropriate spinal alignment. d) Successful decompression of the spinal cord can be seen on the postoperative MRI. Republished with permission from the authors and publisher from Figure 6 [11].

2.3.5 Venous Thromboembolism Prophylaxis and Treatment

Pulmonary embolism (PE) and deep vein thrombosis (DVT) are commonly associated with spinal cord injuries [35]. The majority of these occur within the first three months after injury. Patients with severe motor deficits following SCI should be treated prophylactically for venous thromboembolism (VTE) using a combination of anticoagulation and pneumatic compression devices. The use of oral coagulation and low dose heparin alone is not recommended. The prophylactic treatment should be initiated within 72 hours and continued for 3 months (level II) [35].

2.3.6 Pharmacological Management of SCI

2.3.6.1 Vasopressors and blood pressure management

Neurogenic shock due to interruption of sympathetic innervation may worsen secondary spinal cord injury due to hypoperfusion and ischemia. Several clinical practice guidelines have recommended maintenance of a MAP of 85-90mmHG for up to 5-7 days after SCI [160,162]. A recent meta-analysis reported that in two prospective studies, neurological outcomes were stable to-improved with management of MAP above 85mmHg and 90mmHg. These authors note that norepinephrine is preferred for cervical and upper thoracic injuries and phenylephrine or norepinephrine for mid-to lower thoracic injuries [164]. The Consortium for Spinal Cord Medicine has recommended dopamine or norepinephrine for cervical and high thoracic SCI due to neurogenic shock requiring alpha and beta adrenergic effects. In lower regions, hypotension may be caused by vasodilation and thus phenylephrine is suggested [177]. A recent systematic review by Yue et. al (2019) suggests that dopamine is associated with higher rates of complications than phenylephrine and that norepinephrine slightly increases perfusion pressure relative to dopamine and phenylephrine [165]. Although no significant differences in neurological outcomes were observed, other reports suggest that cardiogenic complications were more strongly associated with

the use of dopamine (OR 8.97, p<0.001) than phenylephrine (OR 5.92, p = 0.004). Both types of vasopressors were associated with increased complications in SCI [166]. This is supported by Readdy et al, who reported that cardiac complications were associated with both dopamine and phenylephrine but that the differences were not statistically significant [167]. In animal models of SCI, phenylephrine has been associated with more hemorrhage in comparison to controls while decreasing SCBF and PO2. In contrast, norepinephrine increased SCBF and PO2 [178].

2.3.6.2 Glibenclamide (Glyburide, DiaBeta)

Glibenclamide is FDA approved for the treatment of diabetes and blocks the sulfonylurea- receptor-1-(SUR1)-regulated-Ca2+-activated-[ATP]-sensitive non-specific cation channel (NC Ca-ATP). Simard et al. demonstrated its important for capillary fragmentation following SCI and the use of glibenclamide in animal models showed decreased lesion volumes and improved behavioral outcomes [179]. A phase I/II clinical trial administering glibenclamide within 8 hours in patients with cervical SCI (C2-C8) is currently underway and is expected to be complete in 2020 (ID: NCT02524379).

2.3.6.3 Riluzole

Riluzole is a benzothiazole Na+ channel blocker approved for the treatment of amyotrophic lateral sclerosis. Its neuroprotective effects have been widely reported and are achieved by inhibiting Na+ influx, thus reducing the subsequent Ca2+ accumulation. This reduces the activation of pro- inflammatory enzymes, namely calpains and phospholipases. It also reduces the excessive glutamate release responsible for neurotoxicity through inhibition of glutamate release and activation of glutamate transporters [151,180,181].

Notably, although the precise molecular mechanisms are unclear, the ventral horn area is less affected by riluzole than other regions, perhaps because of a lower susceptibility to excitotoxicity arising from a relatively lower number of neurons, lower concentration of glutamate receptors than

the dorsal horn [182]. A 2010 phase I/IIA trial enrolled 36 patients with SCI who received riluzole (28 doses of 50 mg) had significant improvement compared with the control group (n = 36). Incomplete cervical injury showed the highest improvements in motor score, while there was no difference within the thoracic injury group [183]. The ensuing 2014 phase IIB/III trial (NCT01597518) will enroll approximately 351 patients and results are expected in 2021. Riluzole is administered twice daily (100 mg) within the first 24 hours after injury and twice daily (50 mg) for the next 13 days.

2.3.6.4 VX-210

VX-210 is an engineered recombinant variant of C3 transferase that has the ability to cross the spinal cord dura and cell membranes. C3 transferase inactivates RhoA, promoting axonal outgrowth. It has also been shown to reduce the extent of the lesion and improve locomotor function [184,185]. In a phase I/IIa clinical trial (NCT00500812), improvements in American Spinal Injury Association (ASIA) motor scores were especially notable in cervical SCI patients when compared to thoracic SCI [186]. A phase IIb/III trial for cervical SCI was undertaken (NCT02669849), however it was terminated early due to futility in October 2018.

It may be difficult to isolate vascular disruption and neuroprotection as separate therapeutic targets after SCI as several therapeutics in development have several proposed mechanisms of action which may affect both vascular disruption and neuroprotection through immunomodulation. The therapies discussed here which fall into that category include minocycline, hepatocyte growth factor (HGF) and methylprednisolone.

2.3.6.5 Minocycline

Minocycline is a tetracycline antibiotic that has shown anti-inflammatory effects through inhibition of cytokine expression, suppression of microglial activation. Additionally, it may help promote BSCB integrity through inhibition of matrix metalloproteinases [187].

A phase I/II clinical trial completed in 2010 showed improved motor scores in cervical SCI and no improvement in thoracic SCI, with no serious adverse events [188]. A phase III trial (NCT01828203) has been underway since 2013 with an estimated complete date in 2018 and no results currently posted.

2.3.6.6 HGF

Hepatocyte growth factor (HGF) is a c-Met receptor ligand which has been reported to increase neuron survival and decrease oligodendrocyte apoptosis. Additionally, it has been shown to reduce cavitation and mediate angiogenesis in animal models of SCI [189]. Yamane et al. [190,191]reported neuroprotective anti-inflammatory effects through a decrease in cytokine levels, reduced glial scar formation and reduced leukocyte infiltration. Interestingly, HGF may affect the stability of the BSCB by protecting against endothelial cell injury [192,193]. A phase I/II trial (NCT02193334) examining the safety and efficacy of intrathecal HGF in 45 patients was completed in July 2018 however no results have been posted. In this trial, HGF (0.6 mg) was administered every week for 5 weeks, beginning 72-hours after SCI.

2.3.6.7 MPSS

Methylprednisolone is an immunosuppressant glucocorticoid steroid which has been extensively studied in SCI. Its main mechanism inhibits lipid peroxidation resulting from oxidative stress after injury [194]. In preclinical models, it has been shown to upregulate anti-inflammatory cytokines [23]. Leypold et al. report that MPSS significantly decreased intraparenchymal hemorrhage in humans after SCI [195]. Additionally, MPSS reduces neutrophil recruitment to the site of injury, which could reduce their detrimental effects on the integrity of the BSCB [79,196].

However, administration of methylprednisolone is believed to increase the risk of serious side effects. As an immunosuppressant, it has been associated with increases in wound infections and pneumonia [197]. Guidelines have varied on the recommendation for MPSS. The CNS/AANS

2002 guidelines suggest either 24-hour or 48-hour infusion while noting that the evidence of serious infection-related side effects is more consistent than the evidence for neurological improvement. The 2013 CNS/AANS guidelines recommend against the use of MPSS based on lack of approval for SCI by the FDA, the lack of Class I and II evidence supporting clinical benefits of MPSS, and the Class I, I and II evidence of harmful side effects (Hulbert 2013).

In the National Acute Spinal Cord Injury series of trials [198–200] the increased risk of infection- related complications using a high dose 48 hour protocol outweighed potential neurological benefits. The 24h protocol (30 mg/kg bolus + 5.4 mg/kg/h x 23-hour) had a lower rate of complications. In a highly criticized [201,202] subgroup analysis of patients given MPSS within 8 hours of injury, MPSS was found to improve neurological outcomes. This conclusion was subsequently supported by a Cochrane review which found a significantly increased ASIA motor score in patients who received MPSS within 8 hours of injury [203]. Thus, the AOSpine 2016 guidelines includes a weak recommendation that a 24-hour infusion of high-dose MPSS should be offered to adults within 8 hours of injury [204].

The significant side effects of methylprednisolone, including poor wound healing and infection, may vary in severity based on the level of injury. As previously discussed, injuries above the mid- thoracic region reduce peripheral inflammatory function, particularly in contusion and incomplete injuries. Therefore, for low thoracic and lumbar injuries, the effects of methylprednisolone which warrant its limited use may be less prevalent. However, this has not been reported, even in a more recent cohort study examining level-specific outcomes [205].

More trials are needed to definitively establish differences in efficacy between levels of injury for the therapeutics in development; both those relating to neuroprotection generally and vascular dysfunction specifically. The importance and challenges in addressing intra-level and inter-level differences in clinical SCI trials has been noted by several authors [149,152,206].

I examined this topic during my thesis, and co-authored the following manuscript: 1. Vidal, P. M., et al. (2018). "Methylprednisolone treatment enhances early recovery following surgical decompression for degenerative cervical myelopathy without compromise to the systemic immune system." J Neuroinflammation 15(1): 222.

Chapter 2

Aims and Hypothesis

3.1 Overarching Hypothesis

Neurovascular differences between the cervical and thoracic cord will impact the pathophysiology of traumatic spinal cord injury

3.2 Specific Aims

1. Characterize the level-dependent pathophysiological response after SCI and the vascular underpinnings of the phenomenon (Chapter 3)

2. Characterize the level-dependent circulating profile after SCI and its impact on spleen weight (Chapter 4)

3. Characterize the level-dependent splenic response after SCI and how injury etiology impacts peripheral immunity (Chapter 5)

3.3 Publications

Publications directly related to thesis aims: 1. Hong, J., et al. (2019). "Incomplete Spinal Cord Injury Reverses the Level-Dependence of Spinal Cord Injury Immune Deficiency Syndrome." Int J Mol Sci 20(15).

2. Hong, J., et al. (2018). "Level-Specific Differences in Systemic Expression of Pro- and Anti-Inflammatory Cytokines and Chemokines after Spinal Cord Injury." Int J Mol Sci 19(8). 3. Hong, J., et al. (2019). “The neurovascular niche of the cervical spinal cord renders it differentially sensitive to vascular injury and inflammation in comparison to the thoracic spinal cord.” Nature Medicine. Submitted.

Publications related to other components of the thesis: 4. Badner, A., et al. (2018). "Splenic involvement in umbilical cord matrix-derived mesenchymal stromal cell-mediated effects following traumatic spinal cord injury." J Neuroinflammation 15(1): 219.

5. Badner, A., et al. (2019). "Endogenous Interleukin-10 Deficiency Exacerbates Vascular Pathology in Traumatic Cervical Spinal Cord Injury." J Neurotrauma 36(15): 2298-2307.

6. Beldick, S. R., et al. (2018). "Severe-combined immunodeficient rats can be used to generate a model of perinatal hypoxic-ischemic brain injury to facilitate studies of engrafted human neural stem cells." PLoS One 13(11): e0208105.

7. Chio, J. C. T., et al. (2019). "The effects of human immunoglobulin G on enhancing tissue protection and neurobehavioral recovery after traumatic cervical spinal cord injury are mediated through the neurovascular unit." J Neuroinflammation 16(1): 141.

8. Vawda, R., et al. (2019). "Early Intravenous Infusion of Mesenchymal Stromal Cells Exerts a Tissue Source Age-Dependent Beneficial Effect on Neurovascular Integrity and

Neurobehavioral Recovery After Traumatic Cervical Spinal Cord Injury." Stem Cells Transl Med 8(7): 639-649.

9. Vidal, P. M., et al. (2018). "Methylprednisolone treatment enhances early recovery following surgical decompression for degenerative cervical myelopathy without compromise to the systemic immune system." J Neuroinflammation 15(1): 222.

10. Vidal, P. M., et al. (2019). "The changes in systemic monocytes in humans undergoing surgical decompression for degenerative cervical myelopathy may influence clinical neurological recovery." J Neuroimmunol 336: 577024.

Chapter 3 Level-dependent differences in the spinal cord following traumatic SCI

This chapter is adapted from work submitted to Nature Medicine.

Hong, J., et al. (2019). “The neurovascular niche of the cervical spinal cord renders it differentially sensitive to vascular injury and inflammation in comparison to the thoracic spinal cord.” Nature Medicine. Submitted.

4.1 Abstract

The cervical and thoracic spinal cord are anatomically and physiologically distinct, and despite dramatic level-dependent differences in response to therapy, no study has comparatively examined the impact of injury level on the pathogenesis of traumatic SCI. Herein, we report that baseline differences in the molecular constitution of the vascular basement membrane between the injury levels contribute to the distinct neurovascular response following SCI. Reduced collagen, fibronectin and astrocytic laminin levels in the cervical cord may be critical for meeting its increased bioenergetic demands, but in turn render the region susceptible to severe BSCB disruption, gliosis, neuroinflammation and tissue clearance relative to the thoracic cord. This finding highlights the importance of acute attenuation of BSCB disruption as a critical upstream target necessary for tissue preservation after cSCI.

4.2 Introduction

Spinal cord injury (SCI) is a devastating neurological condition that often results in lifelong, severe sensorimotor deficits [3,11]. Traditionally, the pathology of traumatic spinal cord injury has always been largely thought to be identical across the different levels of the spine with the degree of neurological deficit being the key distinguishing phenotype [170]. The majority of preclinical studies of SCI involve thoracic spinal cord models due to ease of animal care in comparison with cervical cord models. However, clinical translational studies of SCI are increasingly involving patients with cervical cord injury due to the relatively higher frequency of injuries at this level, the higher level of impairment with cervical lesions, and the greater sensitivity of clinical outcome measures to detect neurological improvements. To-date, our lab and other researchers have shown that distinct peripheral disruptions between the injury levels [207–210] exists in various SCI etiologies, and that cell therapies are not tailored to suit the distinct rostrocaudal identity of the milieu have limited therapeutic efficacy [211]. Here, we show for the first time that there are

distinct cellular and extracellular matrix (ECM) compositions between the cervical and thoracic spine, particularly in components critical to the neurovascular unit, and that these differences are critical to consider when translating therapeutic approaches for SCI from animal models to man.

4.3 Methods

Sample preparation

For molecular approaches, a 0.5-mm segment of the spinal cord was removed after perfusion with 1xPBS. The tissue was then immediately flash-frozen in liquid nitrogen. For histological approaches, a 0.5-mm segment of the spinal cord was removed after perfusion with 4% PFA. The tissue was post-fixed in 4% PFA in 10% sucrose-PBS for 4-hours, and transferred to 20% for 2- days prior to being mounted in M1-matrix for cryosectioning.

RNA-sequencing

Samples were homogenized in 600 µl of chilled 1xPBS, and a 200 µl aliquot was processed immediately for RNA extraction using the miRVana PARIS kit (Invitrogen) and quantified using Nanodrop (Millipore). The RNA integrity was first assess using the A260/A280 ratio, samples below a value of 2.0 were reprocessed to ensure purity. To further ensure RNA quality, Bioanalyzer was used and samples with RIN < 8 were reprocessed as well. Finally, 2000 ng of RNA was sent for single-end sequencing using an Illumina Genome AnalyzerIIX sequencer using the TruSeq library construction protocol. Samples were multiplexed and randomly distributed across 7 lanes of a single flow cell (SCI n = 5, sham = 3, for each time point, total n = 48). Base- calling and demultiplexing was done using Illumina CASAVA 1.8.1 with default parameters, quality check was done using FastQC [212], and trimming of reads was done using the Cutadapt program [213] and contaminating sequences (phiX, polyA, polyC, rRNA) were filtered out. Alignment was done using Tophat [214] and samples were then quantified using the featureCounts procedure [215] with default parameters using Rattus novegicus v5, in the ENSEMBL GTF

version [216] 5.0.79 keeping only the protein coding gene biotype. Genes with low counts were filtered out using the filtered.data method of the NOISeq R package[217] using the proportion test filter method setting the norm parameter to FALSE. Normalization of reads was done using the RUVs protocol (RUVSeq R package) [218] with k=7. Differential gene expression was done timepoint-by-timepoint using DESeq2 [219] with the design matrix being ~tissue + treatment + tissue:treatment where the group “tissue” contains levels “cervical” and “thoracic” and treatment contains levels “sham” and “injured”. Functional enrichment was done using gProfiler [220] (significant only, no electronic GO annotations, size of category = 3-500, size of query/term = 3, GO, KEGG, Reactome). Genes with constitutive upregulation/downregulation were defined as genes with a log2(fold-change of cSham/tSham) > 0 or < 0 at all 3 time-points. In the comparative baseline analysis of shams, we filtered out non-constitutive expressors due to potential differences in the sham surgery of cervical and thoracic animals as we logically assume level-distinct genes to not undergo temporal change in the uninjured condition. As such, we only compared the DEGs that had opposing constitutive expression to reduce noise. Contrastingly, in the comparative analysis of injured animals, the interaction term controlled for sham variability at each time-point, respectively. As such, we did not opt to only include genes that sustained a constitutive expression across time.

Deconvolution of bulk-sequencing dataset

Deconvolution was done using raw counts from our bulk-sequencing dataset in conjunction with GSE98816 [221] in CIBERSORTx [222,223]. First, a single-cell RNA-seq signature matrix file was created from phenotype-classified matrices provided in GSE98816. This matrix entitled “vascular” was then used to impute the cell fractions of our bulk-sequencing dataset labeled “cervical” or “thoracic”. The following settings were used in both the cervical and thoracic datasets, respectively: enable batch correction, disable quantile normalization, enable run in

absolute mode, 500 permutations. The absolute abundance results were exported into a datasheet and was plotted using GraphPad Prism (GraphPad, La Jolla, USA).

Protein extraction and quantification

Samples were processed as above, and 20 µl of 10X RIPA buffer (Pierce) was added 190 µl of homogenate. The samples were then sonicated for 3 cycles of 10-seconds each, with a 1-minute rest period between the cycles. The samples were then centrifuged at 12,000 g for 15-minutes at 4°C, and the supernatant collected. Protein extract was then quantified using the Pierce microBCA kit, whereby a BSA standard curve is used to generate the linear OD-to-concentration formula.

Western blotting

15 ug of protein was used for Western blotting. Except for the mitochondrial antibodies which require non-denaturing conditions, the proteins were first mixed in with 2x Laemelli loading buffer (Bio-Rad) with 2-mercaptoethanol and boiled at 95°C for 5-minutes. The samples were then loaded into 4-20% gradient gels (Bio-Rad) and run at 250V constant voltage until the dye front reached the edge of the gel cast. The proteins were then transferred at 100V for 1-hour unto 0.2 µm nitrocellulose membrane (Bio-Rad). A reversible Ponceau-S stain was performed and captured on the Bio-Rad Chemidoc scanner to be used for total protein normalization. The Ponceau-S stain was then removed by gentle washing with 1x TBST and the membrane was blocked in 4% skim milk in 1xTBST for 1-hour at room temperature. Finally, the antibodies listed in Supplementary Table 1 were used for overnight 4°C incubation. The next day, the membranes were incubated in anti-rabbit HRP secondary (1:1000, Sigma) and auto-exposed on the Bio-Rad ChemiDoc scanner.

Slot blot

15 ug of protein was used for slot blotting. Proteins were diluted into final volume of 100 µl, and using the slot blotting apparatus (Bio-Rad), the samples were passed through using gravity

filtration. Ponceau-S stain was used for total protein normalization. A reversible Ponceau-S stain was performed and captured on the Bio-Rad ChemiDoc scanner to be used for total protein normalization. The Ponceau-S stain was then removed by gentle washing with 1x TBST and the membrane was blocked in 4% skim milk in 1xTBST for 1-hour at room temperature. Finally, the antibodies listed in Table 1 were used for overnight 4°C incubation. The next day, the membranes were incubated in anti-rabbit or anti-mouse HRP secondary (1:1000, Sigma) and exposed on the Bio-Rad ChemiDoc scanner.

Multiplex dot blot

100 ug of protein was used for multiplexed dot blotting. The ARY005 rat inflammatory proteome profiler kit (R&D) was used according to the manufacturer’s protocols. The final incubation in IR- streptavidin (1:1000, Mandel) was visualized on the IR channel of the Odyssey scanner (LICOR).

3-dimensional ultrasound and Power Doppler

For functional blood vessel quantification, animals were subjected to in vivo very high-resolution ultrasound (VHRUS) and Power Doppler imaging, as previously described [224]. At 24 h after injury, the animals were anesthetized using isoflurane and placed on a Vevo 770 imaging platform (VisualSonics) with a custom-made stabilization frame. The injury was exposed with a midline incision and retraction of the paraspinal muscle layers. Ultrasound gel (Medi-Inn) was placed on the dura mater and scanned with the VHRUS probe (44 MHz) in three-dimensional (3D) B-mode. The 3D B-mode scans were analyzed using Fiji as previously described [225]. Briefly, the bright pixels were delineated by one independent blinded observer within 19 central sagittal image slices, and these slices were used to generate a reproducible lesion volume with the TrakEM2 plugin [226].

Immunohistochemistry

The samples were prepared as described above. The frozen tissue sections were stained with various antibodies diluted with blocking solution made of 1× PBS containing 10% w/v goat serum (Bioshop) and 0.03% Triton X-100 (Thermo Scientific). Antibody, concentration, and purpose are listed in Table 3. The tissue sections were blocked for 1 h in room temperature with a blocking solution. Secondary antibody alone (no primary antibody) served as the negative control. After 3 cycles of 10-minute washes with 1x PBS, the slides were mounted onto the coverslips using Mowiol containing 1:1000 DAPI for nuclear counterstain (Sigma Aldrich).

Cell culture

Primary rat cortical astrocytes (Gibco) and vascular pericytes (ScienCell) were plated in astrocyte medium (1831, ScienCell) and pericyte medium (1201, ScienCell), respectively. The cells were then cultured in 6-well NUNC-coated plates (9.6 cm2, Thermo) for 2-weeks to 80% confluence with weekly media changes. Finally, the cells were subjected to 0.06ug/ul of 3-day post-SCI- conditioned media from either the cervical or the thoracic cord. The SCI-conditioned media was generated by incubating a crushed 0.5-mm segment of the cord in phenol-free, FBS-free DMEM at 37°C for 1-hour with intermittent 10-second vortexes spaced 20-minutes apart. After centrifugation for 15-minutes at 4°C, the protein supernatant was quantified as described above. 3-days after exposure, the cells were fixed in 4% PFA in 1xPBS and incubated overnight at 4°C with the antibodies listed in Table 3. Epifluorescent images were then captured using a confocal microscope (Ti2, Nikon) and quantified using custom thresholding algorithms in Fiji [225].

Statistical Analysis

Statistical analysis was performed on GraphPad Prism 8.0.2 (GraphPad, La Jolla, USA). All groups that passed the Shapiro-Wilk test of normality were subject to a parametric test such as one-way ANOVA or t-test was used with the alpha set to 0.05. All groups that did not, were subject to a non-parametric test such as the Mann-Whitney or Kruskal-Wallis.

Table 3 List of antibodies

Antibody Catalogue # Dilution

Col1a1 Ab34710, Abcam 1:200

F3648 (Sigma), Ab2413 Fn1 1:200 (Abcam)

Lama1 L9393, Sigma 1:500

Ocln 71-1500 (Invitrogen) 1:200

ZO-1 61-7300 (Invitrogen) 1:200

Cldn5 35-2500 (Invitrogen) 1:200

Ab5694 (Abcam), A2547 Acta2 1:1000 (Sigma)

Stat3 9139 (CST) 1:500

pStat3 9145 (CST) 1:500

Lcn2 AF1857 (R&D) 1:200

Gfap G3893 (GFAP) 1:1000

Membrane integrity cocktail Ab110414 (Abcam) 1:1000

Total OXPHOS cocktail Ab110413 (Abcam) 1:1000

4.4 Results & Discussion

As no comparative pathology study exist between cervical (cSCI) and thoracic SCI (tSCI), our first undertaking was to survey the rate of tissue clearance and vascular disruption after SCI. To do this, we used 3-dimensional very high ultrasound imaging with Power doppler (Figure 10). At 3-days post-SCI, the formation of the cystic cavity had already begun in the cervical, but not in the thoracic region. Thereafter, the traumatic cyst formation occurred in an accelerated fashion in cSCI relative to tSCI, plateauing at 2-weeks near its 8-week volume (Figure 10A). Power Doppler imaging surveys functional (i.e. flowing) vessels down to a resolution of 30 um, as such capillaries are excluded from this analysis [224]. Anatomically, the cervical region is far more vascularized per tissue area by both the central sulcal arteries and the posterior spinal arteries [97,227] (Figure 10B). Strikingly, there is prominent large vessel disruption after cSCI but not tSCI, with primary vessel loss localized to the lesion epicenter. Interesting, angiogenesis occurs after 14-days post- cSCI in the surrounding lesion penumbra (Figure 10B). While we cannot ascertain the degree of capillary disruption, given the absence of significant large vessel disruption after tSCI, the majority of previously reported vascular disruption after tSCI must be in the capillary. As tissue clearance and the formation of the syrinx dependent on infiltrating macrophages, neutrophils and local activated microglia [228], the increased large vessel disruption no doubt further contributes to the exponential increase in syrinx volume 7-days post-cSCI.

Figure 10 Traumatic SCI induces level-dependent rapid tissue loss and large vessel disruption.

A) High resolution ultrasound reveals rapid tissue loss in the cervical relative to the thoracic cord following traumatic SCI (SCI, n = 10; Sham, n = 8 per level, ****p < 0.001); B) 3D reconstruction of sagittal Power Doppler imaging shows large vessel disruption in the cervical relative to the thoracic cord (SCI, n = 10; Sham, n = 8 per level, *** p < 0.001, ****p < 0.0001, One-way ANOVA, post-hoc Sidak).

We next sought to determine any molecular underpinnings that could reconcile the distinct pathophysiological trajectories between the injury levels. This was undertaken by examining the transcriptomic signature of the cervical and thoracic cord before and after SCI through RNA- sequencing (Fig. 11). To do this, we evaluated the differences between all cervical and thoracic shams (Fig 11A) and isolated only the transcripts that were uniformly upregulated or downregulated throughout all 4-timepoints (Fig 11B). Despite the increased vascularity in the cervical region, we found a lower expression of basement membrane constituents in the cervical cord, primarily in members of the collagen family and fibronectin (Fig 11C, D). Further, we observed relative higher expression of neurofilament, potassium channel and soluble ligand carriers in the cervical region relative to the thoracic which is likely a reflection of the increased grey-to-white matter ratio in the cervical region (Fig 11E). Of note, these transcripts mapped to biological processes such as collagen fibril formation and vessel development. To confirm these differences at the protein and tissue level, we performed slot blotting and immunohistochemistry for collagen I (encoded by Col1a1; Fig 11F) and fibronectin I (encoded by Fn1; Fig 11G). Similar to the transcriptomic differences, we found a reduced level of collagen in cervical relative to thoracic shams at the protein and tissue level. Further, as a previous study has shown reduced Pdgfrb-expressing mural cells in the cervical and lumbar region relative to the thoracic region, we probed for baseline astrocytic laminin 1 (encoded by Lama1) at the tissue level, which has been shown to be necessary for mural cell localization. To this end, we found a significantly reduced expression of laminin 1 in the cervical grey matter relative to the thoracic tissue (Fig. 11H).

Figure 11 Baseline molecular differences between the cervical and thoracic cord center on the composition of the basement membrane.

A) PCA analysis showing the distinct clustering of cervical and thoracic shams; B) Pie chart showing a small subset of genes that demonstrated constitutive and uniform expression across time; C) List of top 10 upregulated and downregulated genes and their average log2 fold-change; D) Functional enrichment analysis of the differentially expressed genes (up-regulated in red and down-regulated in blue) using GO, KEGG and REACTOME, respectively; E) Functional classification of differentially-expressed genes; F) Protein and tissue level expression of collagen- 11 (n = 5, **p<0.01, Mann-Whitney); E) Protein and tissue level expression of fibronectin 1 (n = 5, **p<0.01, Mann-Whitney); F) Protein and tissue level expression of laminin-1 (n = 5, *p<0.05, **p<0.01, Mann-Whitney).

We next sought to determine how these baseline differences impact SCI progression at the transcript, protein and tissue level. RNA-sequencing demonstrated that the majority of differential gene expression between the levels occurs at 3- and 14-days post-SCI (Figure 12A). To narrow down the targets of interest, we examined the genes that responded contrarily after SCI (Figure 12B). Of these, the majority of them mapped to a surge in extracellular matrix processes in the cervical cord, with many of these terms overlapping with the same baseline downregulations in collagen shown before, there was also strong reduction in the relative expression of mitochondrial transcripts in the cervical relative to the thoracic cord signaling a staggering contrast in the bioenergetic disruption between the levels (Figure 12C). As the literature suggest that astrocytes and pericytes are the main producers of extracellular matrices, we utilized a FACS-sorted RNA- sequencing database of STAT3-mediated astrogliosis markers after SCI and cross-referenced it to our bulk sequencing database (Figure 12D). Indeed, we observed a profound upregulation of these STAT3-driven astrogliosis markers in cervical relative to thoracic SCI.

To validate these transcriptomic differences, we surveyed critical junctional, pericytic, astrocystic, inflammatory and mitochondrial proteins using a Western blot at 3-days post-injury (Figure 12F- H). There was a striking decrease in the junctional markers zona occludens-1 (encoded by Zo1), occludin (encoded by Ocln) and claudin-5 (encoded by Cldn5). These proteins collectively regulate the integrity of the blood-spinal-cord-barrier (BSCB) which ultimately act as the barrier from infiltrating immune cells. Further, the reduction of these markers signals a compromise in the architecture of the neurovascular unit which is composed of endothelial cells, pericytes, and astrocytes. To examine pericytic changes, we probed for alpha smooth muscle actin (aSMA; encoded by Acta2) which is a marker of a contractile phenotype change in pericytes [229]. Studies have shown that when the pericyte becomes contractile, it instigates astrocystic end feet detachment, thereby disrupting the BSCB and also promotes a fibroblast secretion [121,139,230–

232]. We found elevated Acta2 expression after cSCI compared to tSCI. Astrocytic changes were equally contrasting, with the expression of astrogliosis markers lipocalin-2 (encoded by Lcn2), serpin-a3n (encoded by Serpina3n) and glial fibrillary acidic protein (encoded by Gfap) all increase in cervical relative to thoracic SCI. Finally, we utilized a multiplex mitochondrial cocktail to survey the core proteins in complex 1 through 5 of the electron transport chain and found a significant reduction in the expression of ATPase (complex 5), which is critical in the production of energy for the cell.

Figure 12 Traumatic SCI induces a level-specific pathophysiological profile.

A) Number of differentially-expressed genes between cervical and thoracic SCI; B) Number of differentially-expressed genes that are distinct either due to their magnitude of expression in the same direction (red), or opposing direction (blue); C) Functional enrichment using GO of transcripts upregulated and downregulated in cervical relative to thoracic SCI 3-days post-SCI; D)

The number of DEGs that map to STAT3-KO astrocyte-specific astrogliosis (ASA) genes; E) Protein expression of tight junction markers occludin, zona occludens 1, claudin-5 and contractile marker alpha smooth muscle actin 3-days post-SCI; F) Protein expression of transcription factors STAT3 and phospho-STAT3 and astrogliosis markers GFAP and LCN2 3-days post-SCI; G) Expression of inflammatory cytokines and chemokines 3-days post-SCI; H) Protein expression of mitochondrial complexes 3-days after SCI.

To further dissect the cellular basis of these differences, we deconvoluted our bulk-sequencing dataset using an existing murine CNS single-cell database (GSE: 98816)[221,233] in CIBERSORTx [222,223]. Using this method, we are able observe changes in the absolute abundance of cell-specific transcripts (Fig 13A) and see that there were stark level-dependent differences in the abundance of cell-specific transcripts—particularly in pericytes and subtypes of endothelial cells. Finally, we wanted to see if these level-specific differences result in distinct extracellular microenvironments that can impact cell fate. To do this, we used spinal cord extracts isolated 3-days after injury induction. In the absence of cell disruption, we incubated extracts from the cervical and thoracic cord with primary astrocyte and pericyte cultures for 3-days. We then surveyed the expression lipocalin 2 (encoded by Lcn2) and glial fibrillary acid protein (encoded by Gfap) as markers of reactivity in astrocytes (Fig 13B); and collagen-1a1 (encoded by Col1a1) and aSMA (encoded by Acta2) as markers of fibrosis and contractility in Pdgfrb+Ng2+-pericytes (Fig 13C)[234]. In line with our other data, cervical extracts resulted in a significant increase in astrogliosis as evidenced by an increase in the intensity of GFAP and LCN2, and an overall increase in the abundance of collagen-1a1 and aSMA expression in pericytes.

Figure 13 In vitro primary cultures of rat astrocytes and pericytes exhibit level-distinct responses when cultured with tissue extract.

A) Deconvolution of bulk-sequencing using CIBERSORTx and GSE98816 (PC=pericytes, EC=endothelial cell subtypes, aSMC = arterial smooth muscle cells, vSMC = venous smooth muscle cells, capilEC = capillary endothelial cells, aaSMC = arteriolar smooth muscle cells, AC = astrocytes, aEC = arterial endothelial cells, vEC = venous endothelial cells, MG = microglia, FB

= fibroblasts subtype, OL = oligodendrocytes) shown in abstract values that represent the absolute abundance of cell-specific transcripts (* p < 0.05, n = 5, Kruskal-Wallis, post-hoc Sidak); B) Percent area of astrogliosis markers GFAP and LCN2 in primary rat astrocytes 3-days after administration of 3-day post-SCI tissue extract (**p<0.005, n = 9, t-test); C) Percent area of PDGFRB+NG2+aSMA+ and PDGFRB+NG2+Col1a1+ triple-positive primary rat pericytes 3-days after administration of 3-day post-SCI tissue extract (*p<0.05, n = 9, t-test).

4.5 Conclusion

Taken together, we show, for the first time, conclusive evidence that there exist fundamental baseline differences between the cervical and thoracic cord in their neurovascular composition - specifically in the composition of the basement membrane and synaptic features. These changes, render the cervical cord more susceptible to vascular and excitatory injury, and inflammation following traumatic SCI, and can independently contribute to alterations in both endogenous and exogenous cell fate. Due to the accelerated nature of cSCI, neurological deficits as a result of neural tissue loss is imminent and all efforts must be focused on reduces the initial tissue loss within the first 2-weeks. After this point, only functional regeneration or rehabilitation through neuroplasticity will likely yield any visible improvements in sensorimotor function. Thus, to limit irreparable damage to the spinal cord—especially the cervical spine--an early interventional paradigm targeting vascular disruption is essential.

Chapter 4

Level-dependent differences in the circulating immune factors following SCI

This chapter is adapted from my published work

Hong, J., Chang, A., Zavvarian, M. M., Wang, J., Liu, Y., & Fehlings, M. G. (2018). Level- Specific Differences in Systemic Expression of Pro- and Anti-Inflammatory Cytokines and Chemokines after Spinal Cord Injury. Int J Mol Sci, 19(8). https://doi.org/10.3390/ijms19082167

5.1 Abstract

While over half of all spinal cord injuries (SCIs) occur in the cervical region, the majority of preclinical studies have focused on models of thoracic injury. However, these two levels are anatomically distinct—with the cervical region possessing a greater vascular supply, grey-white matter ratio and sympathetic outflow relative to the thoracic region. As such, there exists a significant knowledge gap in the secondary pathology at these levels following SCI. In this study, we characterized the systemic plasma markers of inflammation over time (1, 3, 7, 14, 56 days post- SCI) after moderate-severe, clip-compression cervical and thoracic SCI in the rat. Using high- throughput Luminex panels, we observed a clear level-specific difference in plasma levels of VEGF, leptin, IP10, IL18, GCSF, and fractalkine. Overall, cervical SCI had reduced expressions of both pro- and anti-inflammatory proteins relative to thoracic SCI, likely due to sympathetic dysregulation associated with higher level SCIs. However, contrary to the literature, we did not observe level-dependent splenic atrophy with our incomplete SCI model. This is the first study to compare the systemic plasma-level changes following cervical and thoracic SCI using level- matched and time-matched controls. The results of this study provide the first evidence in support of level-targeted intervention and also challenge the phenomenon of high SCI-induced splenic atrophy in incomplete SCI models.

5.2 Introduction

Traumatic spinal cord injury (SCI)—despite breakthroughs in pre-operative, surgical and post- operative care—continues to be a life-threatening injury, both acutely and chronically [11]. After primary mechanical injury, a dual-edged cascade of inflammatory and vascular events— collectively referred to as the secondary injury phase—ensues [16,26]. While it is difficult to determine the causative mechanism of secondary injury, several mechanisms including vascular disruption [28], glutamate excitoxicity [45,48], lipid peroxidation [235,236], blood-spinal-cord-

barrier disruption [150,237,238] and ionic imbalance [51]have been the focus of therapeutic targeting. The ultimate consequence of these events is apoptosis, neuronal and axonal death, and de/dys-myelination manifesting as grey and white matter loss at the injury epicenter [11].

Preclinical SCI studies thus far, driven by post-operative care requirements and ease-of-use, have most commonly employed tSCI models despite the increased prevalence and incidence of cSCI [149]. A central rationale for specifically investigating cSCI models is the appreciation that critical anatomical differences exist between the cervical and thoracic spinal cord resulting in different pathophysiological responses to injury and treatment [151]. For instance, the cervical spine is composed of smaller vertebrae with increased mobility, has increased central and peripheral vascular supply and flow, a higher gray-white matter ratio, and contains the neural circuitry crucial for respiration, forelimb motion, and sympathetic outflow to the heart. Pathophysiologically, the cervical gray matter vasculature has less pericyte coverage than the thoracic cord, resulting in a blood spinal cord barrier (BSCB) predisposed to increased permeability [150]. Further, in high- thoracic transection models of SCI, removal of spinal sympathetic preganglionic neurons from supraspinal control results in autonomic dysreflexia [239]. This in turn has been shown to instigate immunosuppressive effects—known as SCI-induced immune depression syndrome (SCI-IDS)— that stem directly from early splenocyte death and splenic atrophy due to acute and repeated chronic exposure to glucocorticoids and intrasplenic norepinephrine [154,240].

As cSCI has a direct neurological impact on cardiovascular function and peripheral immunity [160,241,242], we aimed to characterize the temporal profile (3-56 days) of vascular and inflammatory markers after cSCI and tSCI and elucidate any level-specific changes in their expression. Further, as robust spleen-mass changes were observed in the aforementioned transection studies on SCI-IDS, we also evaluated time and sham-normalized spleen-body weight ratios in our model.

5.3 Methods

All animal experiments were approved by the animal care committee at the University Health Network (Toronto, Ontario, Canada) in compliance with the Canadian Council on Animal Care.

5.3.1 Clip-Compression SCI and Spleen Weight

Female adult Wistar rats (12-weeks old, 250-300 g, n = 5/group for injured, n = 3/group for laminectomy and naïve) were used (Charles River Laboratories, Wilmington, MA, http://www.criver.com). Prior to surgery, 0.05 mg/kg of buprenorphine and 5 ml of saline were administered subcutaneously. 1-2% of isoflurane in a 1:1 mixture of O2 and N2O was used for anesthesia, and a laminectomy was performed at C6-7 and T6-7, respectively. Following this, a moderate-severe injury was induced for 1-minute at the cervical or thoracic level as described previously [149]. Until the endpoint (1, 3, 7, 14, 56 days post-SCI), the animals were given subcutaneous buprenorphine (0.05 mg/kg, bid), oral amoxicillin trihydrate/clavulanate potassium (Apotex Pharmaceuticals, Toronto, CA) and subcutaneous saline injections (0.9%, 5 ml sid). Animals were housed individually in cages at 27°C, and their bladders were manually expressed thrice daily until recovery. Prior to sacrifice and perfusion, animal mass and spleens were collected from anesthetized rats and their weight recorded and normalized to their body mass.

5.3.2 Neurobehavioral Assessments

Starting at 7 days post-SCI, weekly forelimb and hindlimb function were assessed with the grip strength meter (SDI Grip Strength System DFM-10, San Diego Instruments, San Diego, http://www.sandiegoinstruments.com) and the BBB Locomotor Rating Scale [243,244] for cSCI and tSCI, respectively (Figure 14).

Figure 14 BBB and Grip Strength Assessments

Hindlimb (BBB) and forelimb (Grip strength) assessments following tSCI and cSCI, respectively. No statistical outliers were detected using Grubb’s test (α > 0.05) indicating good homogeneity of data. Error bars represent SEM.

5.3.3 Blood Collection and High-Throughput Luminex Assay

Blood was collected via a cardiac puncture prior to perfusion using a BD-Vacutainer® Safety- Lok™ blood collection set containing EDTA. The blood samples were kept on ice and immediately centrifuged at 3000 rpm (Eppendorf 5810R) for 10 minutes at 4°C. The plasma (supernatant) was then carefully aspirated and transferred to a Protein Lo-Bind tube (Eppendorf). 100 µl of sample were then sent to Eve Technologies (https://www.evetechnologies.com) for high- throughput Luminex profiling using their rat Discovery Assays™ for cytokine/chemokines (RD27) and vascular injury markers (P1, P2). All proteins that contained interpolated/extrapolated/out-of-range values were removed from the study. The concentration of these proteins was calculated using a standard curve and expressed in pg/ml.

5.3.4 Clustering and Statistical Analysis

Data are presented as mean±SEM and comparisons are presented in order from cervical to thoracic. Heatmap, k-means and hierarchical row clustering (1−Cosine Similarity) was performed using the Morpheus software package from the Broad Institute (https://software.broadinstitute.org/morpheus/). Assessment of normality was performed for each group using the Shapiro-Wilk test of the Rfit package. All protein level comparisons between cSCI and tSCI were performed in GraphPad using either the one-way ANOVA function with post-hoc Sidak’s for multiple corrections (parametric, p-adjusted threshold = 0.05) or Kruskal-Wallis test with post-hoc Dunn’s for multiple corrections (non-parametric, p-adjusted threshold = 0.05). Spleen-body mass ratio comparisons were done in GraphPad using a one-way ANOVA with post- hoc Sidak for multiple corrections (p-adjusted threshold = 0.05).

5.4 Results

Of the 35 proteins surveyed in this study 19 passed our initial filtering criteria, while 16 proteins that contained interpolated, extrapolated or out-of-range values were removed. All comparisons below are presented in order from thoracic to cervical.

5.4.1 Level-specific differences in plasma protein levels after cervical and thoracic laminectomy

To investigate whether there were baseline differences in the expression of any of these proteins after cervical and thoracic laminectomy, heat-mapping and statistical analyses of protein concentrations were carried out with naïve plasma shown as a reference (excluded from cluster analyses). Heat-mapping (Figure 15A) demonstrated several clusters of expression. While several proteins had trending differences (RANTES, p = 0.06 at 14-days; LIX, p = 0.052 at 14-days; and IL10, p = 0.068), two proteins within cluster 5 showed significant differences in the expression of IP10 (56-days, 131.5±11.2 vs. 450.3±15.8 pg/ml, p = 0.003) and IL18 (3-days, 745.1±84.8 vs. 400.2±47.5 pg/ml, p = 0.036). The time-series expression of these three proteins are shown in Figure 15B.

Figure 15 Heat map and hierarchical cluster analyses reveal six clusters of temporal expression amongst cSham and tSham groups.

A) Of the proteins analyzed, two proteins (marked with arrows) within cluster 4 reached statistical significance (IL18 and IP10). Naive data are shown as a baseline reference. Data are shown with relative color coding, with blue associated with the row minimum and red with the row maximum; all data are based on raw concentration in pg/mL; (B) Temporal expression of the two significant level-distinct proteins. Error bars represent SEM. One-way ANOVA, post-hoc Sidak.

5.4.2 Level-specific differences in plasma protein levels after cervical and thoracic SCI

Of the 19 proteins analyzed (Figure 16), 6 showed significant differences at one or more timepoints between time-matched, level-matched, laminectomy-normalized cSCI and tSCI groups (expressed as fold-change to laminectomy). The 6 proteins that showed level-specific differences were VEGF (day 7, -0.1±0.2 vs. 1.325±0.1, p = 0.03), leptin (days 1, 1.2±0.4 vs. -0.05±0.3, p = 0.04; day 56, -0.1±0.4 vs. -2.0±0.4, p = 0.0009), IP10 (day 1, 0.3±0.3 vs. -0.7±0.04, p = 0.02; day 7, -0.891±0.3 vs. 0.4±0.2, p = 0.0005; day 56, 0.9±0.2 vs. -1.0±0.3, p < 0.0001), IL18 (day 56, - 1.5±0.3 vs. -0.05±0.4, p = 0.02), GCSF (day 7, 1.010±0.187 vs. 3.943±1.663, p = 0.006 ), and fractalkine (day 1, 0.3±0.1 vs. -0.6±0.2, p = 0.004). Both of the proteins (IP10 and IL18) that showed level-specific baseline differences were significant after SCI. However, while differences in the 56-day baseline of IP10 contributed to a significant result, the 3-day baseline difference in the IL18 did not.

Figure 16 Temporal expression of level-specific proteins.

Temporal expression profile of the six significant differentially-expressed proteins: VEGF, leptin, IP10, IL18, GCSF and fractalkine. Error bars represent SEM. Two-way ANOVA, post- hoc Sidak.

5.4.3 Temporal Expression Patterns of Plasma Proteins after SCI

To dissect the various temporal expression patterns after cSCI and tSCI, heatmap and k-means cluster analysis was performed (Figure 17). In tSCI, three main clusters were found with cluster 1 showing acute/chronic upregulation with subacute downregulation; cluster 2 showing an acute/subacute upregulation with chronic downregulation; and cluster 3 consisting of a single member showing constitutive upregulation. Similarly, in cSCI three major clusters were defined with cluster 1 showing constitutive downregulation with some acute upregulation; cluster 2 showing proteins with acute/subacute downregulation with chronic upregulation; and cluster 3 showing proteins that had acute/subacute upregulation followed by chronic downregulation.

Figure 17 Heatmap and k-means clustering reveals three major clusters of temporal expression in cSCI and tSCI.

Expression is displayed as log2 (fold-change of laminectomy) with blue indicating downregulation and red indicating upregulation.

5.4.4 Functional Classification of Serum Protein after SCI

Using several reviews and meta-analyses articles [245–252] , pro-inflammatory and anti- inflammatory functions were assigned to each of the 19 proteins to survey the overall inflammatory status after cSCI and tSCI (Figure 18). It is evident that overall, tSCI has increased expression of both pro- and anti-inflammatory proteins over time compared to cSCI. While most of these proteins are strikingly upregulated in the acute phase of thoracic relative to cervical SCI, a few of these differences equilibrated chronically (e.g. IL1b, fractalkine, IL10).

Figure 18 Heat map of functionally-segregated proteins after cSCI and tSCI.

Expression is displayed as log2 (fold-change of laminectomy) with blue indicating downregulation and red indicating upregulation.

5.4.5 Spleen Weight

To identify whether splenic atrophy was observed in our model of incomplete SCI, mass- normalized spleen weights were measured and expressed as a fold-change of time-matched laminectomized shams (Figure 19). While a decrease in weight between injured and time-matched laminectomized shams was seen in both cSCI and tSCI, this did not reach statistical significance. However, a significant increase in spleen weight was observed between 3 and 14 days in cSCI (0.862±0.078 vs. 1.188±0.113, p = 0.03), but only trended for tSCI (p = 0.06).

Figure 19 Spleen:mass ratios expressed as fold-change of time-matched laminectomized shams.

Error bars represent SEM, and means are indicated by +. A significant change (One-way ANOVA, post-hoc Sidak, p = 0.03) in spleen weight was observed in cSCI between day 3 and day 14.

5.5 Discussion

In summary, this is the first study to characterize the temporal plasma expression profile of multiple cytokines, chemokines and growth factors after SCI. It is also the first study to use clinically-relevant models of cSCI and tSCI to determine level-specific differences in the expression of these inflammation-related molecules. This study establishes three main points: 1) time and level-matched laminectomy controls are essential for accurate data interpretation after SCI, 2) there exist both acute and chronic differences in plasma protein expression between cSCI and tSCI, and 3) splenic atrophy is not a robust phenomenon after incomplete cSCI, and as there continues to be evidence of peripheral immune depression after cSCI, it is also not a conclusive diagnostic tool for assessing the state of SCI-IDS.

Inflammation after SCI is considered one of the major drivers of secondary injury and tissue loss, and is often considered a dual-edged sword [253–255]. Our current results examined a small percentage of the cytokine/chemokines/growth factors involved, however due to the spectrum of cells that secrete these factors, we cannot accurately pinpoint the cellular mediators of the temporal and level-specific changes that we have observed. With regards to the cause of level-differences between laminectomized shams, it is likely they are due to the degree of invasiveness and its associated fibrosis above the site of laminectomy and its differential impact on the cord over-time between the two levels.

Studies on five of the six level-distinct proteins have already been conducted in the rodent tSCI model, with IL18 being the only exception with no SCI-associated studies. VEGF is well-known as a potent angiogenic factor that promotes the growth and development of endothelial cells. Our lab was one of the first to study the role of VEGF as a therapeutic agent after acute tSCI [256,257]. In these studies, transcriptionally-enhancing VEGF expression resulted in increased axon preservation, reduced necrosis, and increase in blood vessels that ultimately translated to increased

functional recovery as measured by Catwalk gait analyses. Another study using a contusion tSCI model [258], found that an acute intraspinal infusion of VEGF into the lesion epicenter induced autophagy and reduced inflammation in the spinal cord, ultimately resulting in functional recovery as measured by the BBB motor scale. In this study, they showed that VEGF administration reduced the expression of IL1b, IL10 and TNFα in in vitro cultures of LPS-treated neuro-glia co-cultures. In our study, VEGF was upregulated at day 7 relative to tSCI and this change did indeed coincide with striking systemic reductions in IL1b and IL10. Further, an upregulation of these proteins was observed at 14- and 56-days post-cSCI when VEGF expression returned to baseline. Perhaps one of the major underlying factors of this level-dependent change after cSCI is the impact of cervical injury on the systemic vasculature as autonomic dysreflexia contributes directly to frequent vascular stress. Overall, there is significant evidence to suggest that VEGF therapy would be effective in both acute cSCI and tSCI, with the potential to also reduce chronic inflammation in both models.

As a hormone produced mainly by adipose cells, leptin is crucial for energetic balance in the central nervous system. Previous studies into leptin have shown that it is often upregulated both locally and systemically after tSCI [259–261]. While these acute studies were severely limited by the lack of time-matched controls, we observed a striking upregulation of plasma leptin in tSCI, but not cSCI at 1- and 56-days post-SCI. As leptin is regulated by the sympathetic nervous system, a study has shown that patients with high level SCIs have dysfunctional leptin expression [262]— thus supporting our data. A study [263] that acutely administered purified leptin in a rodent model of tSCI showed increased expression of neuroprotective genes, reduced inflammation and improved BBB, Catwalk and von Frey metrics suggesting that acute and chronic leptin deficiency may be a potent therapeutic target in SCI.

An upregulation in systemic and local IP10 has been demonstrated in both human and rodent SCI [264,265], and while no time-matched controls were used, this upregulation persisted as long as

14 days post-SCI in the murine tSCI model. IP10, is a chemokine secreted by a wide array of immune cells, endothelial cells and fibroblasts in response to IFNγ. In our study, IP10 expression was inversely expressed between the two levels, with cSCI experiencing a peak of expression during the subacute phase (days 3-14), and tSCI in the acute and chronic phases (day 1 and 56). Studies that neutralized the expression of IP10 showed markedly reduced inflammation, apoptosis, tissue loss and showed modestly improved BBB and BMS outcomes [266,267].

While GCSF has had no reported systemic or local expression in the SCI literature, here we find that the expression of GCSF is opposite in cSCI and tSCI. That is, while GCSF is upregulated in tSCI, it is downregulated in cSCI with time-to-time changes also in contrary motion with the exception of day 56. As purified GCSF administration alone and in combination with adipose- and bone-marrow-derived stem and neural stem cells and has been shown to be highly beneficial in rodent and human tSCI—including increased tissue preservation, reduced apoptosis and scarring, and improved BBB, BMS and A [268–275]—such a paradigm may prove to be even more effective in the all stages of cSCI where a deficiency in GCSF is seen.

Receptor knockouts of the fractalkine receptor CX3CR1 have resulted in reduced iNOS+/Ly6Clow/MHCII+/CD11c- macrophages and activated microglia that have reduced expression of IL6 and iNOS. In these studies, the authors observed modest improvements in the BMS score [276–278]. In our study, the systemic fractalkine ligand is significantly upregulated in tSCI at 1-day post-SCI relative to cSCI, with both tSCI and cSCI experience late peaks 56-days post-SCI. Fractalkine is present in both a cell-bound and soluble form, and while both forms are potent chemo-attractants for migrating monocytes, the soluble form is also known to attract T cells. As such, while fractalkine receptor antagonism may be an ideal therapeutic target for acute tSCI, it may also be a valuable target for chronic tSCI and cSCI.

5.6 Conclusion

Two potential mechanism by which cSCI induces an overall decrease in circulating protein are 1) SCI-IDS [251,279,280], and 2) increased cellular localization (and as such cytokine/chemokines) to the site of injury [249,251,281]. SCI-IDS is a phenomenon characterized by rapid splenic atrophy due to repeated bouts of autonomic dysreflexia in higher level injuries. Interestingly, level- dependent splenic atrophy was not observed between our two incomplete models of cSCI and tSCI (Figure 19). The latter cytokine/chemokine “sink” concept is well-supported by the literature, as recent characterizations of cytokine/chemokine profiles in the spinal cord of SCI rodents and individuals show a striking acute and chronically-persistent expression of many pro- and anti- inflammatory cytokines. This, in conjunction with the increased BSCB permeability after cSCI, may well result in the formation of an inflammatory milieu that can be a potent trigger for secondary injury—especially chronic inflammation. All in all, we have shown striking evidence of level-specific differences in the systemic plasma expression of various cytokines and chemokines. Considering these results, preclinical researchers should adapt time-matched laminectomized controls and consider the impact of anatomical level on the therapeutic target of interest.

Chapter 5

Level-dependent differences in the splenic response following SCI

This chapter is adapted from my published work:

Hong, J., Chang, A., Liu, Y., Wang, J., & Fehlings, M. G. (2019). Incomplete Spinal Cord Injury Reverses the Level-Dependence of Spinal Cord Injury Immune Deficiency Syndrome. International Journal of Molecular Sciences, 20(15), 3762. https://doi.org/10.3390/ijms20153762

6.1 Abstract

SCI is associated with an increased susceptibility to infections, such as pneumonia, which is the leading cause of death in these patients. This phenomenon is referred to as SCI-Immune Deficiency Syndrome (SCI-IDS) and has been shown to be more prevalent after high-level transection in preclinical SCI models. Despite the high prevalence of contusion SCIs, the effects of this etiology have not been studied in the context of SCI-IDS. Compared to transection SCIs, which involve a complete loss of supraspinal input and lead to the disinhibition of spinally-generated activity, contusion SCIs may cause significant local deafferentation, but only a partial disruption of sympathetic tone below the level of injury. In this work, we investigate the effects of thoracic (T6- 7) and cervical (C6-7) moderate-severe contusion SCIs on the spleen by characterizing splenic norepinephrine (NE) and cortisol (CORT), caspase-3, and multiple inflammation markers at 3- and 7-days post-SCI. In contrary to the literature, we observe an increase in splenic NE and CORT that correspond to an increase in caspase-3 after thoracic SCI relative to cervical SCI. Further, we found differences in expression of leptin, eotaxin, IP-10 and IL-18 that implicate alterations in splenocyte recruitment and function. These results suggest that incomplete SCI drastically alters the level-dependence of SCI-IDS.

6.2 Introduction

Traumatic SCI is a devastating injury that has been shown to disrupt distal organs through maladaptive coordination of the hypothalamus-pituitary axis and the sympathetic nervous system [209]. Recently, the phenomenon known as SCI-induced immunodeficiency syndrome (SCI-IDS) has been broadly implicated [154,209,239,279,282,283]. SCI-IDS is thought to be the cause of the increased susceptibility of SCI patients to pneumonia. The mechanisms by which the syndrome occurs is thought to be level-dependent, where high-level transection lesions (>T3) cause a complete loss of supraspinal sympathetic flow to the spinal preganglionic neurons (SPNs) that

innervate the adrenal gland and spleen. This loss of sympathetic flow ultimately results in several pathologies including splenic cortisol (CORT) and norepinephrine (NE) surge, severe leukopenia, splenic atrophy, and splenocyte apoptosis. These pathologies further coincide with the depression of circulating NE (adrenal insufficiency), elevation of circulating CORT, and an elevation of splenic NE and CORT 3-7 days post-injury [208,284,285]. A recent study characterized the pathways regulating SCI-IDS and found that while the deafferentation of adrenal glands accounts for the changes in NE and CORT (as it is the major producer), direct innervations did not account entirely for organ atrophy as lymph nodes above the level of transection continue to experience cell loss due to circulating factors [279]. Using adrenalectomy and adrenotransplantation, the study demonstrated that organ atrophy after SCI is mediated through the spleen, and no level-dependence was shown for this phenomenon. At the spleen, the use of beta-2-adrenergic and glucocorticoid receptor antagonists have also been shown to rescue spleen atrophy [285], thus demonstrating that NE and CORT released via the maladapted sympathetic-neuroendocrine adrenal reflex plays a direct role in reducing splenocyte homing and apoptosis.

However, to-date, SCI-IDS has been characterized exclusively in transection models of SCI. Full transections and hemisections are exceedingly rare in the clinic, as most patients suffer from incomplete lesions [3,149,286,287]. Outside of an increased susceptibility to pneumonia and adrenal insufficiency (e.g. requiring catecholamine administration), there exist no other robust parallel phenotypes (e.g. organ atrophy) between animal high-level transection models and cervical/high-thoracic patient populations [279]. Previously, we have reported no signs of splenic atrophy despite an overall reduction in circulating cytokine/chemokines after an incomplete cervical SCI [287]. On the contrary, at 14-days post injury we saw a significant increase in the spleen-body mass ratio relative to thoracic SCI. In the present study, we aimed to characterize splenic cytokine/chemokine, NE and CORT levels and their impact on splenocyte apoptosis after incomplete cervical and thoracic SCI. In contrast to past literature, we demonstrate that incomplete lesions to the lower spine (T6-7; tSCI) result in profound splenocyte apoptosis, and elevations of

both splenic NE and CORT relative to a cervical (C6-7; cSCI) lesion. These data suggest that maintenance of partial supraspinal inputs in incomplete spinal cord lesions may drastically alter the level-dependency of SCI-IDS.

6.3 Methods

All animal experiments were approved by the animal care committee at the University Health Network (Toronto, Ontario, Canada) in compliance with the Canadian Council on Animal Care (Project ID Code: #2212, Date of Approval: 17 May 2017).

4.1. Clip-Compression SCI and Spleen Extraction

The same cohort of female adult Wistar rats (12-weeks old, 250–300 g, N=5/group for injured, N= 3/group for laminectomy) were used as previously described [287]. Prior to surgery, 0.05 mg/kg of buprenorphine and 5 mL of saline were administered subcutaneously. 1–2% of isoflurane in a 1:1 mixture of O2 and N2O was used for anesthesia, and a laminectomy was performed at C6-7 or T6-7. Following the laminectomy, a moderate-severe injury was induced for 1-min at the cervical or thoracic level, as described previously [287]. Until the endpoint (3- and 7-days post- SCI), the animals were given subcutaneous buprenorphine (0.05 mg/kg, bid), oral amoxicillin trihydrate/clavulanate potassium (Apotex Pharmaceuticals, Toronto, ON, Canada) and subcutaneous saline injections (0.9%, 5 mL sid). Animals were housed individually in cages at 27°C, and their bladders were manually expressed thrice daily. Prior to sacrifice and perfusion, spleens were collected from anesthetized rats and their weight was recorded and normalized to their body mass. Spleens were then flash frozen in liquid nitrogen and stored in -80°C.

4.2. Protein Extraction from the Spleen

Frozen spleens were crushed with a mortar and pestle over dry ice with constant pouring of liquid nitrogen. The pellet was collected by swirling the liquid nitrogen and spooning the consolidated

pellet into a 1.5 ml protein Lo-Bind tube (Eppendorf, Germany). The pellets were then mixed with 1xRIPA buffer (Thermo Fisher Scientific, Massachusetts, United States) at a ratio of 4 ml per g of tissue followed by 2 rounds of 10-second sonication to homogenize the tissue. The samples were then centrifuged at 15,000 rpm for 15-minutes and the protein supernatant was taken. The protein was then quantified using the microBCA kit (Thermo Fisher Scientific, United States), where the 562 nm reading was used to map the sample absorbance against known concentrations of bovine serum albumin.

4.3. Quantitative ELISA for Norepinephrine and Corticosterone

Splenic NE and CORT levels were measured using quantitative ELISA kits. 50ug of protein was used to quantify samples using the NE ELISA kit (KA1891, Abnova) and 10ug of protein was used to quantify samples using the CORT ELISA kit (ab108665, Abcam). For both kits, 450 nm was used to quantify the absorbance values of the samples (620 nm was subtracted as background for the NE kit, while 650 nm was subtracted as a background of the CORT kit). The standard curve was fitted to a four-parameter logistic regression on GraphPad Prism as suggested by the manual, and the absorbance values of the samples were mapped to the concentrations based on known concentrations of NE and CORT.

4.3. Western Blot

Thirty ug of protein was used for Western blotting. The proteins were first mixed in with 2x Lamelli loading buffer (Bio-Rad, United States) with 2-mercaptoethanol and boiled at 95°C for 5- minutes. The samples were then loaded into 4-20% gradient gels (Bio-Rad, United States) and run at 250V constant voltage until the dye front reached the edge of the gel cast. The proteins were then transferred at 100V for 1-hour unto 0.2 µm nitrocellulose membrane (Bio-Rad, United States). A reversible Ponceau-S stain was performed and captured on the Bio-Rad ChemiDoc scanner to be used for total protein normalization. The Ponceau-S stain was then removed by gentle washing

100

with 1x TBST and the membrane was blocked in 4% skim milk in 1xTBST for 1-hour at room temperature. Finally, the caspase-3 antibody (9662, 1:1000, Cell Signaling Technology) and cleaved caspase-3 Asp175 antibody (9661, 1:1000, Cell Signaling Technology) were used for overnight 4°C incubation. The next day, the membranes were incubated in anti-rabbit HRP secondary (1:1000, Sigma) and auto-exposed on the Bio-Rad ChemiDoc scanner.

4.4. High-throughput Luminex Array for Cytokines and Chemokines

250 ug of protein were sent to Eve Technologies (https://www.evetechnologies.com) for high- throughput Luminex profiling using their rat Discovery Assays™ for cytokine/chemokines (RD27). All proteins that contained interpolated/extrapolated/out-of-range values were removed from the study. The concentration of these proteins was calculated using a standard curve and expressed in pg/ml.

4.5. Statistical Analysis

Heatmapping was performed using the Morpheus software package from the Broad Institute (Cambridge, USA, https://software.broadinstitute.org/morpheus/) and graphing was done using GraphPad Prism (La Jolla, USA). Assessment of normality was performed for each group using the Shapiro-Wilk test of SPSS (IBM, USA). All comparisons between cSCI and tSCI were performed in SPSS using either the one- or two-way ANOVA function with post-hoc Sidak’s for multiple corrections (parametric, p-adjusted threshold = 0.05) or Kruskal-Wallis test with post-hoc Dunn’s for multiple corrections (non-parametric, p-adjusted threshold = 0.05).

101

6.4 Results

6.4.1 Level-dependent differences in splenic NE and CORT

We determined the kinetics of NE and CORT 3 and 7-days after a contusion-compression spinal cord injury. SCI was induced in rats at the C6-7 or T6-7 level with a modified aneurysm clip according to the experimental layout shown in Figure 20; time-matched laminectomized animals served as surgical controls.

102

Figure 20 Experimental design of Chapter 5.

Two moderate-severe SCIs were induced at C6-7 and T6-7 levels and splenic CORT, NE, and inflammatory profile were assessed 3- and 7-days post-SCI.

Based on the current literature, we expected splenic NE and CORT to be elevated after cSCI relative to tSCI, due to a release of NE from the adrenal medulla. On the contrary, we found reduced splenic NE at both levels 3-days post injury (0.59- and 0.61- fold change relative to control for cSCI and tSCI, respectively), and significantly increased NE levels 7-days after tSCI relative to cSCI (0.44- and 1.24- fold change relative to control for cSCI and tSCI, respectively, respectively; Figure 21A). Similarly, while CORT was elevated after cSCI 3-days post-injury relative to control, it did not reach significance (1.05- and 0.59 fold change relative to control for cSCI and tSCI, respectively; p = 0.46). However, 7-days post-injury, splenic CORT was found to be significantly elevated in tSCI relative to cSCI (1.03- and 2.12- fold change relative to control for cSCI and tSCI, respectively; Figure 21B).

103

Figure 21 SCI results in changes to splenic NE and CORT in a level-dependent and time- dependent manner.

(a) SCI at/below (T6-7) but not above (C6-7) the adrenal gland innervation results in marked elevation of splenic NE at 7-days post-injury (two-way ANOVA, **P = 0.0025, t = 3.970, DF = 15, N=5/group, 95% CI of diff. -1.28 to -0.29, Sidak’s multiple comparison test). (b) tSCI resulted in an elevation of splenic CORT 7-days post-injury relative to cSCI (two-way ANOVA, *P = 0.0301, t = 2.720, DF = 16, N=5/group, 95% CI of diff. -2.08 to -0.10, Sidak’s multiple comparison test). Data shown as mean fold-change relative to time-matched controls (N=3/control group).

104

6.4.2 NE and CORT surge results in increased caspase cleavage in tSCI

Next, we evaluated whether the surge in splenic NE and CORT 7-days post-tSCI resulted in splenocyte apoptosis. As we previously reported, we observed no significant change in the spleen- body-mass ratio after cSCI and tSCI [14]. However, based on the profile of NE and CORT in tSCI, we anticipated an elevation in cleaved caspase-3 7 days post-tSCI. In line with this, we found a profound elevation of pro-caspase (0.91- and 4.6-fold change relative to control for cSCI and tSCI, respectively), caspase (0.31- and 1.18-fold change relative to control for cSCI and tSCI, respectively), and 17-kDa cleaved caspase 3-days post-tSCI relative to cSCI (0.54- and 1.4- fold change relative to control for cSCI and tSCI, respectively; Figure 22A,B,D). Further, we also found an elevation of caspase (0.08- and 1.18- fold change relative to control for cSCI and tSCI, respectively), as well as the 19- (0.87- and 1.79- fold change relative to control for cSCI and tSCI, respectively), and 17-kDa cleaved caspase (0.39- and 1.74- fold change relative to control for cSCI and tSCI, respectively), 7-days post-tSCI relative to cSCI (Figure 22B-D).

105

Figure 22 SCI results in changes to splenic pro-caspase-3, caspase-3 and cleaved caspase-3 in a level-dependent and time-dependent manner.

(a) tSCI displayed a marked upregulation of splenic pro-caspase-3, 3-days post-injury relative to cSCI (two-way ANOVA, *P = 0.0156, t = 3.038, DF = 16, N=5/group, 95% CI of diff. -6.69 to - 0.69. Sidak’s multiple comparison test). (b) tSCI resulted in a marked elevation in splenic caspase-3, 3-days post-injury relative to cSCI (two-way ANOVA, *P = 0.0137, t = 3.099, DF = 16, N=5/group, 95% CI of diff. -1.56 to -0.17, Sidak’s multiple comparison test). Similarly, this difference persisted in a similar manner at 7-day post-injury (two-way ANOVA, *P = 0.0404, t = 2.574, DF = 16, N=5/group, 95% CI of diff. -1.14 to -0.03, Sidak’s multiple comparison test). (c) tSCI, but not cSCI, resulted in elevated splenic cleaved caspase-3 (19-kDa form; two-way ANOVA, *P = 0.0378, t = 2.607, DF = 16, N=5/group, 95% CI of diff. -1.80 to -0.04, Sidak’s multiple comparison test). (d) tSCI resulted in elevation of splenic cleaved caspase-3 (17-kDa form), 3-days post-injury relative to cSCI (two-way ANOVA, *P = 0.0231, t = 2.848, DF = 16, N=5/group, 95% CI of diff. -1.62 to -0.11, Sidak’s multiple comparison test). Similarly, this difference persisted in a similar manner at 7-day post-injury (two-way ANOVA, ***P = 0.0009,

106

t = 4.417, DF = 16, N=5/group, 95% CI of diff. -2.1 to -0.59, Sidak’s multiple comparison test). Data shown as mean fold-change relative to time-matched controls (N=3/control group).

6.4.3 Reduced leukocyte chemotaxins to the spleen after tSCI

Finally, we examined the cytokine and chemokine profile of the spleen after cSCI and tSCI (Figure 23A). In our previous work, we showed a profound reduction in circulating IP10 3- and 7- days following tSCI relative to cSCI, a key molecule in regulating T-cell generation and trafficking. On the contrary, current literature suggests that high-level SCI perturbs lymphocyte—particularly B- cell—homing to the spleen. As we observed stark reversals in both NE/CORT and caspase, we anticipated that tSCI would result in a depletion of leukocyte/lymphocyte chemotaxins in the spleen. Indeed, at 3-days post-tSCI, both IP10 and eotaxin (also known as CCL2, a potent leukocyte chemoattractant) were significantly depressed relative to cSCI, which displayed contrary expression (Figure 23B). Intriguingly, tSCI also displayed significantly increased leptin (0.48- and -0.05 log2 fold change of control for cSCI and tSCI, respectively)—a regulator of energy intake and expenditure known for suppressing splenic lymphocyte function—as well as IL18—an inducer of IFN-γ in T and NK cells—7-days post-injury relative to cSCI (Figure 23B).

107

Figure 23 SCI induces changes in splenic expression of cytokine/chemokines in a level- dependent and time-dependent manner.

(a) Heatmap demonstrating both the individual and the mean log2 fold-change of control after cSCI and tSCI. At 3-days post-injury tSCI had relatively lower levels of leptin compared to cSCI (Kruskal-Wallis, *P = 0.028, t = 4.811, DF = 1, N=5/group, all pairwise comparisons). At 7-days post-injury tSCI had relatively lower levels of eotaxin (Kruskal-Wallis, *P = 0.028, t = 4.840, DF = 1, N=5/group, all pairwise comparisons) and IP10 (Kruskal-Wallis, *P = 0.047, t = 3.938, DF = 1, N=5/group, all pairwise comparisons) compared to cSCI. Contrastingly, tSCI had relatively elevated levels of leptin (Kruskal-Wallis, *P = 0.009, t = 6.818, DF = 1, N=5/group, all pairwise comparisons) and IL18 (Kruskal-Wallis, *P = 0.016, t = 5.771, DF = 1, N=5/group, all pairwise comparisons). (b) Bar graphs showing only the cytokines/chemokines that were significantly different between the levels at 3- and 7-day post-injury. Data shown as mean fold- change of time-matched controls (N=3/control group).

108

6.5 Discussion

Our study is the first study to demonstrate that cervical and thoracic compression/contusion spinal cord lesions may differentially influence the development of SCI-IDS. While other studies have not examined both levels concurrently, there have been studies of low thoracic incomplete lesions that have reported similar results. For instance, Zha et al. found a profound elevation of NE following a chronic T9 contusion, and functional deficiencies in the T-cell population without any signs of splenic atrophy [288]. Similarly, Norden et al. found impaired T cell antiviral immunity follow an acute T9 contusion alongside elevated CORT levels [289]. These studies demonstrate the complexities of sympathetic disruption, and how partial deafferentation of the adrenal and splenic innervations can result in a level-dependency that is unlike that of transection SCI. We posit that in an incomplete injury, the sympathetic tone below the level of the injury may be reduced or altered, but not eliminated. It is this factor, in conjunction with the well-documented plasticity and rewiring, that may result in preserved splenic and adrenal innervation after cSCI that continues to respond to the injury stimulus [283]. However, at the lesion site, there is severe tissue loss that results in cystic cavitation as early as 3-days post-injury. This cavity undoubtedly disrupts the local circuitry and may cause—as seen in our data and the T9 contusion studies above— symptoms that mimic local deafferentation of the preganglionic SPNs that lead to the spleen and adrenal gland.

There are several studies that document the impact of both NE and CORT on IP10, leptin, IL18 and eotaxin. For instance, CORT has been implicated in the inhibition of multiple leukocyte chemoattractants including eotaxin, eotaxin-2, and monocyte-chemotactic protein-4 [290]. While eotaxin serves as a potent chemoattractant and homing signal for leukocytes, IP10 is known to be critically involved in effector T-cell generation and trafficking [291]. IP10-deficient mice have been shown to have decreased proliferation and weaker IFN-γ secretion in response to immune challenge. On the other hand, leptin is a potent regulator of energy intake and expenditure and has

109

also been shown to suppress splenic T-cell function through the activation of corticotropin releasing hormone [260,292,293]. Further, CORT has been shown to promote leptin secretion. Finally, IL18 has been shown to be induced by agonism of B2AR and by elevated serum CORT [294,295]. IL18 exerts its affect by stimulating IFNγ secretion from splenocytes, thus it is likely that its elevation signals an attempt for splenic IFNγ production to induce various T-cell trafficking and generation molecules downstream (including IP10).

Despite adequate controls, a caveat that must be considered is that of anesthesia and clavamox and its potential for differential synergistic / antagonistic effect after cSCI and tSCI. Well no documented effect on sympathetic tone has been shown with clavamox, it is well-studied that induction with isoflurane is accompanied by hypotension, indicating an attenuation of sympathetic tone [296]. The accompanying hypotension may synergize with the systemic hypotension following cSCI (and not tSCI), which may impact circulation and overall physiology to some extent. Despite this, there is no evidence to suggest that there are significant alterations to the splenic perfusion following isoflurane induction.

Taken together, our data shows that an incomplete tSCI, but not cSCI, displays an aberrant sympathetic-neuroendocrine coordination that closely mimics key hallmarks of SCI-IDS. This may suggest two possible scenarios that may complement the current mechanisms proposed by the literature: (1) a partial loss of sympathetic tone below the site of injury is insufficient to instigate deafferentation of the SPNs; (2) SPNs branching from the lesion site are lost or severely disrupted to the degree of deafferentation. Taken together, these mechanisms could explain how tSCI instigates SCI-IDS due to a loss of local circuitry while cSCI remains unaffected due to the preservation of sympathetic tone below the site of injury.

110

General Discussion and Future Directions

7.1 Summary of Results

Taken together, this work has demonstrated that cSCI and tSCI result in drastically different spatiotemporal profiles of injury pathogenesis. In our model of incomplete SCI, we found that cSCI resulted in rapid tissue clearance, increased gliosis and neuroinflammation relative to tSCI. Using RNA-sequencing and single-cell deconvolution, we found that the majority of differential gene expression mapped to collagen, fibronectin and other critical genes for vascular development. This was confirmed using various techniques at the protein, cell and tissue level. To further demonstrate that the level specific injury milieu could alone—in the absence of physiological influences—dictate cell fate of primary astrocytes and pericytes we used isolated protein from an organotypic suspension of injured tissue and demonstrated that the protein extract from 3-day injured cervical cord was enough to induce astrogliosis and contractile pericyte differentiation.

We then studied the level-dependent impact of SCI on circulating cytokines and chemokines and found a profound reduction in the concentrations of most circulating markers after cSCI. As the phenomenon of SCI-IDS was well-characterized in high-level transection models, we hypothesized that this reduction in the circulating cytokine and chemokines were as a result of peripheral immune suppression. However, we demonstrated that splenic atrophy—the classic defining phenotype of SCI-IDS—was absent and in fact, cSCI demonstrated splenomegaly at 14- days post-injury relative to baseline. As such, we sought out to determine whether SCI-IDS was a hallmark of our incomplete SCI model.

To this end, we replicated the readouts that were initially done to survey SCI-IDS in the transection models. Specifically, we examined the splenic NE and CORT levels, the expression of caspase and cleaved caspase in the spleen, and the inflammatory profile of the spleen. Based on the existing literature, we hypothesized that cSCI would induce peripheral immune suppression through a

111

mechanism of decreased sympathetic tone to the celiac ganglion that ultimately innervates the spleen. However, the results demonstrated that thoracic, but not cervical, resulted in profound elevations of splenic NE and CORT levels, enough to induce an elevation of caspase-mediated apoptosis and a reduction in leukocyte chemoattractants.

7.2 Novelty and Impact

No studies to date have examined the local, circulating and peripheral impacts of injuries to the cervical and thoracic region using a clinically-relevant model of incomplete SCI. This thesis work contributes two unique transcriptomic datasets to the SCI field. To our knowledge, this is the only time-controlled, time-series dataset for thoracic SCI, and the only cervical transcriptomic dataset available. This sets the foundations for further research at the single-cell level and provides an invaluable resource for the entire field. In addition to the novelty of the dataset, we provide a comprehensive summary of the major level-dependent transcriptomic changes in the spinal cord and reconcile the impact of these changes on SCI pathobiology at the tissue and protein level. Ultimately, this study refocuses the spotlight on acute management of vascular disruption as the most critical determinant of tissue preservation after traumatic SCI, especially in the setting of cSCI. Furthermore, this work questions the fundamental clinical-relevance of the paradigm of SCI- IDS, which was one of the primary arguments against systemic immune suppression in high-level injuries (e.g. MPSS), and demonstrates that different SCI etiologies must be carefully considered prior to generalizing paradigms across all types of SCI.

7.3 General Discussion

This thesis followed a “paper” format, which had specific discussions of each dataset in Chapters 3-5. As such, this section will offer more in-depth discussion on several aspect of the thesis work

112

that were outside of the content constraints of the original publication, its limitations and new frontiers.

7.3.1 Changes in basement membrane constituents after injury

In Chapter 3, we demonstrated baseline differences in the constitution of the vascular basement membrane of the cervical and thoracic cord. In the same study, we found a profound upregulation of these same markers after cSCI. This spike in collagen and other ECM molecules are well- documented in the literature and are collectively referred to as the formation of the fibrotic scar [231,232,297]. As the rate of tissue clearance due to extensive BSCB disruption was increased in after cSCI, the activation of contractile aSMA+ pericytes has been shown to result in collagenous deposits that contribute to the fibrotic scar. Further, a recent study [232] has demonstrated that selective ablation of pericyte-associate scar can effectively result in functional recovery following SCI. Strikingly, nearly 85% of the markers, including most of the collagen and fibronectin class, were affected by the loss of contractile (referred to as “type A”) pericytes were found to be distinctly upregulated in cSCI relative to tSCI, suggesting that targeting the phenotype switch to contractile pericytes may be a critical target—especially in the setting of cSCI.

7.3.2 Injury level and its impact on the timing of various therapeutic interventions

Early Surgical Decompression

There considerable preclinical and clinical evidence that early surgical decompression after SCI is beneficial to functional recovery and is currently the recommended procedure in the guidelines by the Americian Assocation of Neurological Surgeons, the AOSpine, and the congress of Neurological Surgeons. The procedure involves the realignment of the spinal column, and alleviation of any bony or ligamentous compression on the cord. Decompression of the cord through durotomy and durotomy in addition to duroplasty has also been shown in preclinical

113

models of SCI. While there may be level-dependent differences in the efficacy of this procedure, current evidence is limited by the heterogeneity of SCI (i.e. polytraumatic patients), lack of follow- up, and insufficient power. One of the primary pathogenic mechanisms to be considered here is that of ischemia-reperfusion injury. It is well-known that after a period of prolonged ischemia, reperfusion from decompression may result in a cascade of secondary consequences. Under that circumstance, it may be critical to consider prophylactic approaches to controlling ischemia- reperfusion injury prior to surgical decompression in the setting of cSCI relative to tSCI.

Methylprednisolone

MPSS a synthesized corticosteroid that has been shown to deliver potent anti-inflammatory effects after traumatic SCI. However, there have and continue to be concerns about the increased risk of infections associated with MPSS treatment. Currently, it remains as the only treatment option for acute SCI, and debates centering on dose, administration time, adverse events and efficacy has been at the forefront of clinical debates. While the phenomenon of SCI-IDS questions the use of any immunosuppression in the setting of high-level SCI, the phenomenon itself has not been robustly shown in this thesis nor in clinical settings of incomplete SCI. Moreover, a recent publication found that MPSS does not increase the risk of infections and confers significant short- term benefits when given within the first 8-hour window of SCI. Further, patients with cSCI and reduced baseline injury severity was shown to benefit the most from MPSS treatment. Given these results, AOSpine guidelines currently recommend a 24-hour treatment with intravenous MPSS within the first 8-hours of SCI. The results of my thesis are in line with the level-dependence shown in the clinical data, in that cSCI with its elevated level of acute inflammation relative to tSCI would benefit the most from MPSS. Further, my thesis work suggests that an extension of MPSS treatment in cSCI may be warranted to improve tissue preservation but must be carefully weighed against adverse events associated with prolonged MPSS administration.

Cell Transplantation

114

Most studies to date have used neural precursor cells for neural repair and regeneration after SCI. The general timing of these therapies averaged 1-2 weeks from the time of injury [298–301]. Most studies that exceed this average have shown no concrete evidence of functional improvement. Stratification of these studies into cervical and thoracic paradigms demonstrates that cell survival as well as the degree of functional recovery achieved is significantly worse in cervical relative to thoracic injuries [299,302]. To date, it is not entirely understood what results in this discrepancy, but several mechanisms include differential rostrocaudal identity of the neural precursors (most are derived from the subventricular zone of the brain), the increased excitotoxic milieu of the cervical cord, or the increased systemic severity of cSCI [149,303]. While these are all well- grounded hypotheses, this work demonstrates that the degree of tissue loss at 2-weeks is significantly greater than that of the thoracic cord, and the milieu alone is capable of driving distinct glial and pericystic fates. As such, cell transplantation paradigms that target a subacute injection at 14-days would likely not result in any clinically-meaningful functional recovery. That said, the feasibility of acute intraspinal infusion is challenging as many cervical patients are polytraumatic when admitted to intensive care [304,305]. As such, there needs to be considerable focus on pharmacological therapeutics to attenuate the tissue loss associated with SCI, as the success of the graft has been shown to rely greatly on the presence of tissue and the lack of any significant scar formation [298,306].

7.3.3 Impact of tissue preservation in rodents and humans

While rats have similar vascular anatomy to humans, the location of the corticospinal tracts are entirely distinct. While they are in the dorsal funiculus of rodents, they are instead located in the posterolateral funiculus of humans and non-human primates [307–309]. As such, injury to the corticospinal tract—typical of our incomplete contusion SCIs—only result in temporary locomotor dysfunction, while humans experience lifelong paralysis. As such, it is important to consider the

115

drastic functional implications of tissue preservation when using rodent models, as it may translate to exponential improvements in the quality-of-life of humans.

7.3.4 Differences in the immune system of rodents and humans

As our work involves the study of neuroinflammation and the splenic response to SCI, it is important to note the significant differences in the immune system between humans and rodents. While much of it is still unexplored, the human blood is rich in neutrophils, while rodents are rich in lymphocytes [310,311]. As we explored leukocyte chemoattractants in Chapters 4 and 5, it is important to consider the distinct impact of these well-conserved proteins on the immune systems of these two species. In addition to this, the markers of monocyte polarization—specifically the expression of iNOS on human macrophages is still a matter of contention [312,313]. Due to these many caveats, many studies have begun the use of rodent models with humanized immune systems—which come with their own set of limitations [314,315].

7.3.5 Injury severity between levels

One of the central limitations of this thesis is on the topic of injury severity. While severity generally translates to compression strength (measured in grams force) in the clip model and impact strength (measured in kilodynes) in the contusion model, the cervical and thoracic injuries differ greatly in their response to a similar strength clip [149,316–319]. For instance, graded studies within the thoracic cord have shown that a clip strength of 23 g yields a mild-moderate thoracic injury, while a 35 g clip would yield a moderate-severe injury. Contrastingly, in graded studies within the cervical cord, a 35 g clip at the C6-7 level yields a 100% mortality rate (unpublished data), while a 23 g clip is considered a moderate-severe injury without intraoperative mortality. Logically, injuries to the cervical cord—due to its many contributions to autonomic function—are more drastically influenced by increases in injury force. And thus, for this work, we used injuries of similar grades within each level (23 g for cervical, 35 g for thoracic) that reflected a moderate-severe traumatic SCI. Future studies, should attempt to compare between the molecular

116

pathology of a 23 g thoracic injury to that of a 23 g cervical injury to more clearly understand whether there are significant changes in the spatiotemporal expression of SCI-associated genes.

7.3.6 Sex considerations

The female gender bias within the SCI field is well-documented and is primarily driven (like the use of thoracic model) by ease-of-use [320,321]. In both cSCI and tSCI, bladder function is lost after injury. As such, researchers must manually expel the bladder at least once a day and up to thrice a day—as in our study [322]. In male rodents, the enlarged prostate makes it difficult to expel the bladder without causing considerable discomfort for the animals. The study of sex as a variable in SCI is controversial with studies that show no significant difference to ones that demonstrate that female sex hormones can significant influence immunological response to SCI. In summary, while it should be recognized that there are limitations to using only female animals in this study, there are no well-controlled studies that definitively outline the impact of sex as a biological variable after SCI [323,324].

7.4 Future Directions 7.4.1 Overexpression of basement membrane ECM molecules

In my thesis work, it was shown that baseline levels of key basement membrane ECM molecules collagen, fibronectin and laminin were less in cervical than thoracic cords. To-date, no studies have ever examined the impact of the overexpression of these basement membrane molecules in the setting of CNS disease. While it would be interesting to individually design inducible overexpression of Col4a1, Lama1, and Fn1. It would be more feasible to induce transient overexpression of Tgfb1 a few weeks prior to injury as it is well known to induce the expression of collagen, fibronectin and laminin molecules [325–327]. Using this model, we can access the impact of a thicker basement membrane in the cervical cord on vascular disruption and injury progression following SCI.

117

7.4.2 Inhibition of thrombin targets to reduce intraspinal hemorrhage

In my thesis work, cSCI induced a greater hemorrhage area, syrinx and large vessel disruption relative to tSCI. This suggests that the level-dependent differences exacerbate vascular pathology after cSCI. Baseline thrombin transcripts were shown to be lower in baseline cervical relative to thoracic cords. However, after injury, the magnitude of Thbs upregulation was several folds higher than that of tSCI. It has been shown that thrombin directly contributes to BBB disruption through protease-activated receptor-1 (PAR-1). In the absence of PAR-1 thrombin failed to induce severe BBB disruption [328]. It would be interesting to assess the BSCB disruption after cSCI in PAR-1 knockouts and determine whether thrombin significantly contributes to BSCB disruption in the cervical cord.

7.4.3 Acute infusion of mesenchymal stromal cells to attenuate vascular disruption

MSCs are an intensely studied population of multipotent mesodermal progenitors, which are relatively easy to isolate, and display homing properties to the site of injury.

They differentiate quickly into chondrocytes, adipocytes, myocytes, osteoblasts [23,329]. MSCs can be derived from various sources including bone-marrow (BM MSC), umbilical cord (UC- MSC), adipose-derived MSC (AD-MSC). These display both anti-inflammatory properties, notably through macrophage phenotype modulation, and provide trophic support for neuroprotection and regeneration [330,331]. Interestingly, in both human and animal models they also decrease peripheral inflammatory cell infiltration through protection of the BSCB [23]. Various factors have been reported to contribute to this function, notably IL-10, TSG-6 and TIMP- 6 [332–334]. Recently, MSC’s have been reported to reduce intraparenchymal hemorrhage and increase systemic levels of IL-10 in a spleen-dependent manner, which was further supported by altered splenic cytokine levels after MSC infusion [335]. Vawda et. al. reported that infusion of BM MSC reduced glial scarring and increased vascular density without decreasing cavity volume

118

significantly [336]. El-keir et. al report that in a phase I/II controlled single-blind trial (ID: NCT00816803) where chronic cervical and thoracic SCI patients were administered autologous adherent BM-BMC, smaller lesions and greater functional improvements were seen in thoracic SCI in comparison with cervical injuries [337]. There are several clinical trials currently assessing MSC’s. Notably, a phase II/II trial for BM-MSCs in chronic SCI (NCT01676441) with results expected in 2020, a phase I trial for AD-MSC’s (NCT03308565). UC-MSC’s are also being tested for use in subacute and chronic injury in phase I/II trials (NCT03521323, NCT03505034).

This topic was examined during my thesis, and I co-authored the following related manuscripts: 1. Badner, A., et al. (2018). "Splenic involvement in umbilical cord matrix-derived mesenchymal stromal cell-mediated effects following traumatic spinal cord injury." J Neuroinflammation 15(1): 219. 2. Badner, A., et al. (2019). "Endogenous Interleukin-10 Deficiency Exacerbates Vascular Pathology in Traumatic Cervical Spinal Cord Injury." J Neurotrauma 36(15): 2298-2307. 3. Vawda, R., et al. (2019). "Early Intravenous Infusion of Mesenchymal Stromal Cells Exerts a Tissue Source Age-Dependent Beneficial Effect on Neurovascular Integrity and Neurobehavioral Recovery After Traumatic Cervical Spinal Cord Injury." Stem Cells Transl Med 8(7): 639-649.

7.4.4 PKC-β inhibition as a target for attenuating vascular disruption

In my thesis work, I found a level-dependent increase in PKC isoforms after cSCI. Protein Kinase C- (PKC-) is a member of the conventional (cPKC) group of the PKC serine/threonine kinase family, which are regulated by diacylglycerol, phospholipids and Ca2+. PKC regulates multiple physiological processes in the CNS, including signaling cascades for actin and microtubule dynamics and NMDA pathways [338]. Activation of PKC’s increases the microvascular permeability (Tang 2018, Qi 2008, Pastore 2019:17-20), and therefore PKC inhibitors may

119

counteract BSCB compromise induced by neutrophils in SCI, among other neurovascular conditions [339–344]. Additionally, PKC inhibition may attenuate the inhibitory effects of chondroitin sulfate proteoglycans (CSPG) and CNS myelin on neurite outgrowth [345]. This is one of several previously reported potential effects on neural regeneration [346–348].

Midostaurin was originally developed as a PKC inhibitor and was also found to also inhibit VEGF receptors. Its VEGF-related inhibition of angiogenesis led to trials in diabetic retinopathy, although limited efficacy and gastrointestinal toxicity suppressed further development. Based on its inhibition of the fms-like tyrosine kinase 3 (FTL3) and the KIT proto-oncogene receptor tyrosine kinase, it is approved by the FDA and EMA for FTL3-Mutated acute myeloid leukemia (AML) and systemic mastocytosis (SM) respectively [349]. Interestingly, no clinical trials using Midostaurin in SCI are currently listed in the clinicaltrials.gov database. Based on the preclinical evidence of neuroprotection discussed here, this may represent an opportunity for further exploration of Midostaurin’s potential BSCB protection in SCI.

Enzastaurin is a PKC- inhibitor which has showed anti-angiogenesis activity, as well as preferred cytotoxicity towards subsets of tumor cells [350]. Notably, at specific concentrations, administration of Enzastaurin increased platelet aggregation, which may be relevant to thrombosis- derived ischemia and microvascular protection after SCI [351]. Stranahan et. al report that Enzastaurin restored BBB integrity in a leptin-receptor deficient mouse model, significantly reducing macrophage infiltration and activation [352]. Carducci et al. reported toxicity at doses above 700 mg daily and observed only mild neurologic effects (sensory neuropathy) [353]. A subsequent phase II trial reported a relatively favorable safety profile, with no grade 4 toxic effects, at doses of 500 mg daily [354]. In this study, fatigue (grade 2) and syncope (grade 3) were observed in two patients and one patient, respectively. Interestingly, in their work with rats, Willeman et al report significant memory impairment after Enzastaurin administration [355]. This should be taken into consideration in its future applications to treat vascular dysfunction in SCI.

120

7.4.5 Acute IgG infusion for immunomodulation and rescue of BSCB disruption

Intravenous immunoglobulin G (IVIG) consists of pooled serum immunoglobulin G (IgG) from healthy donors. Despite a relatively uncertain mechanism of action, it is currently approved by the FDA for treatment of autoimmune and immunodeficiency disorders, namely primary humoral deficiency idiopathic, thrombocytopenic purpura and chronic inflammatory demyelinating polyneuropathy [151,356]. Brennan et. al report IVIG decreased lesion enlargement and axonal degeneration after contusion SCI in a mouse model. This study also suggests that IVIG may achieve these effects through attenuation of the complement response and a decrease in the phagocytic activity of macrophages [357]. Nguyen et al. reported that in a rat model of cervical compression injury, IVIG administration reduced pro-inflammatory signaling (IL-1 and IL-6). This may have had neuroprotective effects via the BSCB, as demonstrated by the reduction in neutrophil infiltration and reductions in the level of BSCB-damaging metalloproteinase-9 (MMP- 9) [358]. This is further supported by reports of neurovascular protection through increases in tight junction protein expression and significant decreases in MMP-9. The improved BSCB integrity was associated with increased serum levels of inflammatory cytokines (IL-8, MIP-1α, CCL- 2/MCP-1, and IL-5). IgG also co-localized with vascular cell adhesion molecule 1 (VCAM-1) without reducing its expression. VCAM-1 is used by neutrophils in extravasation [356]. Notably, IgG co-localized with astrocytes and pericytes, which suggests a role in modulating immune cell infiltration through the BSCB. Administration of 2g/kg resulted in increased tissue preservation, blood flow and behavioral recovery, effects comparable with methylprednisolone (MPPS), the former standard of care for acute SCI.

This topic was examined during my thesis, and I co-authored the following manuscript:

121

1. Chio, J. C. T., et al. (2019). "The effects of human immunoglobulin G on enhancing tissue protection and neurobehavioral recovery after traumatic cervical spinal cord injury are mediated through the neurovascular unit." J Neuroinflammation 16(1): 141.

7.4.6 Remote ischemic preconditioning prior to surgical decompression

In remote ischemic preconditioning (RIPC), cycles of ischemia and reperfusion are administered to a remote organ, which protects the injured organ from ischemic/reperfusion injury. This has been applied in the protection of several types of tissues in animal models, including the brain, kidney and heart, among others [359]. Using an aortic occlusion model of spinal ischemia in rabbits, Dong et. al. reported improved neurological outcomes after RIPC. This study also suggests that the ROS produced in the remote organ during RIPC might increase the effects of endogenous antioxidant enzymes, which impart a neuroprotective effect in spinal cord ischemia [360]. In a porcine model of segmental artery occlusion, Haapanen et al. show that RIPC increased motor- evoked potentials in comparison to the control group, suggesting neuroprotection from non- traumatic ischemia of the spinal cord. Interestingly however, the histopathological scores of edema, hemorrhage, infarction, and neuronal degeneration did not differ significantly between groups [361]. The same group report that RIPC increased protection against oxidative stress as demonstrated by increased nuclear factor erythroid 2-related factor (Nrf2), which regulates the cellular antioxidant response. Interestingly, histopathological findings of edema, neuron degeneration, hemorrhage and infarction were again not significantly different between groups [362]. Interestingly Fukui et al. report that direct ischemic preconditioning (DIPC) of the rabbit spinal cord before occlusion-based ischemia improved neurological and histopathological outcomes, while kidney and limb RIPC resulted in no similarly protective effects. Notably, the number of morphologically normal neurons in the kidney and limb RIPC did not differ significantly from the control and pre-treatment with the ROS scavenger dimethylthiourea (DMTU) did not attenuate the neuroprotective effects of DIPC [363]. This is in contrast with the

122

previously reported effects of oxidative stress by Dong et. al. [360]. Importantly however, Dong et. al. showed that ROS were implicated in a systemic response to RIPC and did not explore oxidative stress in DIPC. RIPC may also exert its neuroprotective effects through maintenance of BSCB integrity after SCI. Jing et. al. showed that it preserved the BSCB and improved motor function in rats after spinal cord occlusion ischemia. The effects on the BSCB were related to increased occludin expression, dependent on increased expression of cannabinoid 1 and 2 receptors (CB1/CB2) [364]. Ueno (2016) report that in a mouse model of occlusive ischemia RIPC’s neuroprotective effects are associated with an increased concentration of plasma VEGF through downregulation of miRNA-762 and miR-3072-5 [365]. The effects of VEGF in neuroprotection are widely reported, notably by Jin et al. in 2002 [366]. In a systematic review, Dezfulian (2013) suggests that pre- and post-conditioning may provide similar neuroprotection in cerebral ischemia. These authors also highlight the differentiation between early and late effects of preconditioning, associated with post-translational regulation and gene expression modulation, respectively [367].

While the studies discussed here have implications for purely ischemic insults to the spinal cord, these results are not necessarily translatable to acute or chronic mechanical SCI. Currently, RIPC is easily translatable into the clinical setting through the use of a forelimb as the remote tissue. Five-minute alternating cycles of ischemia and reperfusion can safely and effectively be achieved using a blood pressure cuff, which is non-invasive, simple and cost-effective.

123

References

1. Lee, B.B.; Cripps, R.A.; Fitzharris, M.; Wing, P.C. The global map for traumatic spinal cord injury epidemiology: Update 2011, global incidence rate. Spinal Cord 2014, 52, 110– 116.

2. Fitzharris, M.; Cripps, R.A.; Lee, B.B. Estimating the global incidence of traumatic spinal cord injury. Spinal Cord 2014, 52, 117–122.

3. Singh, A.; Tetreault, L.; Kalsi-Ryan, S.; Nouri, A.; Fehlings, M.G. Global Prevalence and incidence of traumatic spinal cord injury. Clin. Epidemiol. 2014, 6, 309–331.

4. James, S.L.; Theadom, A.; Ellenbogen, R.G.; Bannick, M.S.; Montjoy-Venning, W.; Lucchesi, L.R.; Abbasi, N.; Abdulkader, R.; Abraha, H.N.; Adsuar, J.C.; et al. Global, regional, and national burden of traumatic brain injury and spinal cord injury, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 2019, 18, 56–87.

5. Jain, N.B.; Ayers, G.D.; Peterson, E.N.; Harris, M.B.; Morse, L.; O’Connor, K.C.; Garshick, E. Traumatic spinal cord injury in the United States, 1993-2012. JAMA - J. Am. Med. Assoc. 2015, 313, 2236–2243.

6. Stover, S.L.; Fine, P.R. The epidemiology and economics of spinal cord injury. Paraplegia 1987, 25, 225–228.

7. Center, N.S.C.I.S. Spinal cord injury facts and figures at a glance. J. Spinal Cord Med. 2013, 36, 1–2.

8. Poorman, G.W.; Segreto, F.A.; Beaubrun, B.M.; Jalai, C.M.; Horn, S.R.; Bortz, C.A.; Diebo, B.G.; Vira, S.; Bono, O.J.; de la Garza-Ramos, R.; et al. Traumatic fracture of the

124

pediatric cervical spine: Etiology, epidemiology, concurrent injuries, and an analysis of perioperative outcomes using the kids’ inpatient database. Int. J. Spine Surg. 2019, 13, 68–78.

9. Krueger, H.; Noonan, V.K.; Trenaman, L.M.; Joshi, P.; Rivers, C.S. The economic burden of traumatic spinal cord injury in Canada. Chronic Dis. Inj. Can. 2013, 33, 113–122.

10. Noonan, V.K.; Fingas, M.; Farry, A.; Baxter, D.; Singh, A.; Fehlings, M.G.; Dvorak, M.F. Incidence and prevalence of spinal cord injury in Canada: A national perspective. Neuroepidemiology 2012, 38, 219–226.

11. Ahuja, C.S.; Wilson, J.R.; Nori, S.; Kotter, M.R.N.; Curt, A.; Fehlings, M.G. Traumatic Spinal Cord Injury. Nat. Rev. Dis. Prim. 2017.

12. Cripps, R.A.; Lee, B.B.; Wing, P.; Weerts, E.; MacKay, J.; Brown, D. A global map for traumatic spinal cord injury epidemiology: Towards a living data repository for injury prevention. Spinal Cord 2011, 49, 493–501.

13. Montoto-Marqués, A.; Ferreiro-Velasco, M.E.; Salvador-De La Barrera, S.; Balboa- Barreiro, V.; Rodriguez-Sotillo, A.; Meijide-Failde, R. Epidemiology of traumatic spinal cord injury in Galicia, Spain: Trends over a 20-year period. Spinal Cord 2017, 55, 588– 594.

14. Katoh, S.; Enishi, T.; Sato, N.; Sairyo, K. High incidence of acute traumatic spinal cord injury in a rural population in Japan in 2011 and 2012: An epidemiological study. Spinal Cord 2014, 52, 264–267.

15. Kudo, D.; Miyakoshi, N.; Hongo, M.; Kasukawa, Y.; Ishikawa, Y.; Ishikawa, N.; Shimada, Y. An epidemiological study of traumatic spinal cord injuries in the fastest

125

aging area in Japan. Spinal Cord 2019, 57, 509–515.

16. Tator, C.H.; Fehlings, M.G. Review of the secondary injur theory of acute spinal cord trauma with emphasis on vascular mechanisms. J. Neurosurg. 1991, 75, 15–26.

17. Alizadeh, A.; Dyck, S.M.; Karimi-Abdolrezaee, S. Traumatic Spinal Cord Injury: An Overview of Pathophysiology, Models and Acute Injury Mechanisms. Front. Neurol. 2019, 10.

18. Sekhon, L.H.S.; Fehlings, M.G. Epidemiology, demographics, and pathophysiology of acute spinal cord injury. Spine (Phila. Pa. 1976). 2001, 26, S2-12.

19. Alshareef, M.; Krishna, V.; Ferdous, J.; Alshareef, A.; Kindy, M.; Kolachalama, V.B.; Shazly, T. Effect of spinal cord compression on local vascular blood flow and perfusion capacity. PLoS One 2014, 9, e108820.

20. Stokes, B.T.; Jakeman, L.B. Experimental modelling of human spinal cord injury: A model that crosses the species barrier and mimics the spectrum of human cytopathology. Spinal Cord 2002, 40, 101–109.

21. DeVivo, M.J.; Rutt, R.D.; Black, K.J.; Go, B.K.; Stover, S.L. Trends in Spinal Cord Injury demographics and treatment outcomes between 1973 and 1986. Arch. Phys. Med. Rehabil. 1992, 73, 424–430.

22. Oyinbo, C.A. Secondary injury mechanisms in traumatic spinal cord injury: A nugget of this multiply cascade. Acta Neurobiol. Exp. (Wars). 2011, 71, 281–299.

23. Ahuja, C.S.; Nori, S.; Tetreault, L.; Wilson, J.; Kwon, B.; Harrop, J.; Choi, D.; Fehlings, M.G. Traumatic spinal cord injury - Repair and regeneration. Clin. Neurosurg. 2017, 80, S22–S90.

126

24. Tanhoffer, R.A.; Yamazaki, R.K.; Nunes, E.A.; Pchevozniki, A.I.; Pchevozniki, A.M.; Nogata, C.; Aikawa, J.; Bonatto, S.J.; Brito, G.; Lissa, M.D.; et al. Glutamine concentration and immune response of spinal cord-injured rats. J. Spinal Cord Med. 2007, 30, 140–146.

25. Göritz, C.; Dias, D.O.; Tomilin, N.; Barbacid, M.; Shupliakov, O.; Frisén, J. A pericyte origin of spinal cord scar tissue. Science (80-. ). 2011, 333, 238–242.

26. Tator, C.H.; Koyanagi, I. Vascular mechanisms in the pathophysiology of human spinal cord injury. J. Neurosurg. 1997, 86, 483–492.

27. Wolman, L. The disturbance of circulation in traumatic paraplegia in acute and late stages: A pathological study. Paraplegia 1965, 2, 213–226.

28. Figley, S.A.; Khosravi, R.; Legasto, J.M.; Tseng, Y.-F.F.; Fehlings, M.G. Characterization of vascular disruption and blood-spinal cord barrier permeability following traumatic spinal cord injury. J. Neurotrauma 2014, 31, 541–552.

29. Tomko, P.; Farkaš, D.; Čížková, D.; Vanický, I. Longitudinal enlargement of the lesion after spinal cord injury in the rat: A consequence of malignant edema? Spinal Cord 2017, 55, 255–263.

30. Nelson, E.; Gertz, S.D.; Rennels, M.L.; Ducker, T.B.; Blaumanis, O.R. Spinal Cord Injury: The Role of Vascular Damage in the Pathogenesis of Central Hemorrhagic Necrosis. Arch. Neurol. 1977, 34, 332–333.

31. Gerzanich, V.; Woo, S.K.; Vennekens, R.; Tsymbalyuk, O.; Ivanova, S.; Ivanov, A.; Geng, Z.; Chen, Z.; Nilius, B.; Flockerzi, V.; et al. De novo expression of Trpm4 initiates secondary hemorrhage in spinal cord injury. Nat. Med. 2009, 15, 185–191.

127

32. Henke, D.; Gorgas, D.; Doherr, M.G.; Howard, J.; Forterre, F.; Vandevelde, M. Longitudinal extension of myelomalacia by intramedullary and subdural hemorrhage in a canine model of spinal cord injury. Spine J. 2016, 16, 82–90.

33. Fehlings, M.G.; Tator, C.H.; Linden, R.D. The relationships among the severity of spinal cord injury, motor and somatosensory evoked potentials and spinal cord blood flow. Electroencephalogr. Clin. Neurophysiol. Evoked Potentials 1989, 74, 241–259.

34. Anthes, D.L.; Theriault, E.; Tator, C.H. Ultrastructural evidence for arteriolar vasospasm after spinal cord trauma. Neurosurgery 1996, 39, 804–814.

35. Furlan, J.C.; Fehlings, M.G. Cardiovascular complications after acute spinal cord injury: Pathophysiology, diagnosis, and management. Neurosurg. Focus 2008, 25, E13.

36. Jacobs, T.P.; Kempski, O.; McKinley, D.; Dutka, A.J.; Hallenbeck, J.M.; Feuerstein, G. Blood Flow and Vascular Permeability During Motor Dysfunction in a Rabbit Model of Spinal Cord Ischemia. Stroke 1992, 23, 367–373.

37. Faden, A.I.; Chan, P.H.; Longar, S. Alterations in Lipid Metabolism, Na+,K+‐ATPase Activity, and Tissue Water Content of Spinal Cord Following Experimental Traumatic Injury. J. Neurochem. 1987, 48, 1809–1816.

38. Ley, K. The Microcirculation in Inflammation. In Microcirculation; Elsevier Inc., 2008; pp. 387–448 ISBN 9780123745309.

39. Hsu, C.Y.; Halushka, P. V.; Hogan, E.L.; Banik, N.L.; Lee, W.A.; Perot, P.L. Alteration of thromboxane and prostacyclin levels in experimental spinal cord injury. Neurology 1985, 35, 1003–1009.

40. Sharma, H.S.S.; Olsson, Y.; Nyberg, F.; Dey, P.K.K. Prostaglandins modulate alterations

128

of microvascular permeability, blood flow, edema and serotonin levels following spinal cord injury: An experimental study in the rat. Neuroscience 1993, 57, 443–449.

41. Hu, Z.; Tu, J. The roads to mitochondrial dysfunction in a rat model of posttraumatic syringomyelia. Biomed Res. Int. 2015, 2015.

42. Stys, P.K.; Lopachin, R.M. Mechanisms of calcium and sodium fluxes in anoxic myelinated central nervous system axons. Neuroscience 1997, 82, 21–32.

43. Liu, D.; Xu, G.Y.; Pan, E.; McAdoo, D.J. Neurotoxicity of glutamate at the concentration released upon spinal cord injury. Neuroscience 1999, 93, 1383–1389.

44. Xu, G.Y.; Hughes, M.G.; Ye, Z.; Hulsebosch, C.E.; McAdoo, D.J. Concentrations of glutamate released following spinal cord injury kill oligodendrocytes in the spinal cord. Exp. Neurol. 2004, 187, 329–336.

45. Rothstein, J.D.; Dykes-Hoberg, M.; Pardo, C.A.; Bristol, L.A.; Jin, L.; Kuncl, R.W.; Kanai, Y.; Hediger, M.A.; Wang, Y.; Schielke, J.P.; et al. Knockout of glutamate transporters reveals a major role for astroglial transport in excitotoxicity and clearance of glutamate. Neuron 1996, 16, 675–686.

46. Kolodziejczyk, K.; Saab, A.S.; Nave, K.A.; Attwell, D. Why do oligodendrocyte lineage cells express glutamate receptors? F1000 Biol. Rep. 2010, 2.

47. Imaizumi, T.; Kocsis, J.D.; Waxman, S.G. Anoxic Injury in the Rat Spinal Cord: Pharmacological Evidence for Multiple Steps in Ca2+-Dependent Injury of the Dorsal Columns; Mary Ann Lieber!, Inc, 1997; Vol. 14, pp. 299–311;.

48. Agrawal, S.K.; Fehlings, M.G. Role of NMDA and non-NMDA ionotropic glutamate receptors in traumatic spinal cord axonal injury. J. Neurosci. 1997, 17, 1055–1063.

129

49. Faden, A.I.; Simon, R.P. A potential role for excitotoxins in the pathophysiology of spinal cord injury. Ann. Neurol. 1988, 23, 623–626.

50. Choi, D.W. Ionic dependence of glutamate neurotoxicity. J. Neurosci. 1987, 7, 369–379.

51. Fehlings, M.G.; Agrawal, S. Role of sodium in the pathophysiology of secondary spinal cord injury. Spine (Phila. Pa. 1976). 1995, 20, 2187–2191.

52. Ates, O.; Cayli, S.R.; Gurses, I.; Turkoz, Y.; Tarim, O.; Cakir, C.O.; Kocak, A. Comparative neuroprotective effect of sodium channel blockers after experimental spinal cord injury. J. Clin. Neurosci. 2007, 14, 658–665.

53. Agrawal, S.K.; Fehlings, M.G. The Effect of the Sodium Channel Blocker QX-314 on Recovery after Acute Spinal Cord Injury. J. Neurotrauma 1997, 14, 81–88.

54. Li, S.; Jiang, Q.; Stys, P.K. Important role of reverse Na+-Ca2+ exchange in spinal cord white matter injury at physiological temperature. J. Neurophysiol. 2000, 84, 1116–1119.

55. Stout, A.K.; Raphael, H.M.; Kanterewicz, B.I.; Klann, E.; Reynolds, I.J. Glutamate- induced neuron death requires mitochondrial calcium uptake. Nat. Neurosci. 1998, 1, 366– 373.

56. Peng, T.I.; Jou, M.J.; Sheu, S.S.; Greenamyre, J.T. Visualization of NMDA Receptor- Induced Mitochondrial Calcium Accumulation in Striatal Neurons Tsung-I. Exp. Neurol. 1998, 149, 1–12.

57. Epstein, F.H.; Cheung, J.Y.; Bonventre, J. V; Malis, C.D.; Leaf, A. Calcium and Ischemic Injury. N. Engl. J. Med. 1986, 314, 1670–1676.

58. Khayrullina, G.; Bermudez, S.; Byrnes, K.R. Inhibition of NOX2 reduces locomotor

130

impairment, inflammation, and oxidative stress after spinal cord injury. J. Neuroinflammation 2015, 12.

59. Poh Loh, K.; Hong Huang, S.; De Silva, R.; H. Tan, B.; Zhun Zhu, Y. Oxidative Stress: Apoptosis in Neuronal Injury. Curr. Alzheimer Res. 2006, 3, 327–337.

60. Donnelly, D.J.; Popovich, P.G. Inflammation and its role in neuroprotection, axonal regeneration and functional recovery after spinal cord injury. Exp. Neurol. 2008, 209, 378–388.

61. Jia, Z.; Zhu, H.; Li, J.; Wang, X.; Misra, H.; Li, Y. Oxidative stress in spinal cord injury and antioxidant-based intervention. Spinal Cord 2012, 50, 264–274.

62. Robinson, J.M. Reactive oxygen species in phagocytic leukocytes. Histochem. Cell Biol. 2008, 130, 281–297.

63. Kamencic, H.; Griebel, R.W.; Lyon, A.W.; Paterson, P.G.; Juurlink, B.H.J. Promoting glutathione synthesis after spinal cord trauma decreases secondary damage and promotes retention of function. FASEB J. 2001, 15, 243–250.

64. Christie, S.D.; Comeau, B.; Myers, T.; Sadi, D.; Purdy, M.; Mendez, I. Duration of lipid peroxidation after acute spinal cord injury in rats and the effect of methylprednisolone: Laboratory investigation. Neurosurg. Focus 2008, 25.

65. Okamoto, T.; Akaike, T.; Sawa, T.; Miyamoto, Y.; Van der Vliet, A.; Maeda, H. Activation of Matrix Metalloproteinases by Peroxynitrite-induced Protein S- Glutathiolation via Disulfide S-Oxide Formation. J. Biol. Chem. 2001, 276, 29596–29602.

66. Lochhead, J.J.; McCaffrey, G.; Quigley, C.E.; Finch, J.; Demarco, K.M.; Nametz, N.; Davis, T.P. Oxidative stress increases blood-brain barrier permeability and induces

131

alterations in occludin during hypoxia-reoxygenation. J. Cereb. Blood Flow Metab. 2010, 30, 1625–1636.

67. Fehlings, M.G.; Nguyen, D.H. Immunoglobulin G: A potential treatment to attenuate neuroinflammation following spinal cord injury. J. Clin. Immunol. 2010, 30, S109-12.

68. Maikos, J.T.; Shreiber, D.I. Immediate damage to the blood-spinal cord barrier due to mechanical trauma. J. Neurotrauma 2007, 24, 492–507.

69. Reyes-Alva, H.J.; Franco-Bourland, R.E.; Martinez-Cruz, A.; Grijalva, I.; Madrazo, I.; Guizar-Sahagun, G. Characterization of spinal subarachnoid bleeding associated to graded traumatic spinal cord injury in the rat. Spinal Cord 2014, 52.

70. Muldoon, L.L.; Alvarez, J.I.; Begley, D.J.; Boado, R.J.; Del Zoppo, G.J.; Doolittle, N.D.; Engelhardt, B.; Hallenbeck, J.M.; Lonser, R.R.; Ohlfest, J.R.; et al. Immunologic privilege in the central nervous system and the blood-brain barrier. J. Cereb. Blood Flow Metab. 2013, 33, 13–21.

71. Engelhardt, B.; Sorokin, L. The blood-brain and the blood-cerebrospinal fluid barriers: Function and dysfunction. Semin. Immunopathol. 2009, 31, 497–511.

72. Larochelle, C.; Alvarez, J.I.; Prat, A. How do immune cells overcome the blood-brain barrier in multiple sclerosis? FEBS Lett. 2011, 585, 3770–3780.

73. Bolton, S.J.; Anthony, D.C.; Perry, V.H. Loss of the tight junction proteins occludin and zonula occludens-1 from cerebral vascular endothelium during neutrophil-induced blood- brain barrier breakdown in vivo. Neuroscience 1998, 86, 1245–1257.

74. Holtz, A.; Nyström, B.; Gerdin, B. Effect of methylprednisolone on motor function and spinal cord blood flow after spinal cord compression in rats. Acta Neurol. Scand. 1990,

132

82, 68–73.

75. Gensel, J.C.; Zhang, B. Macrophage activation and its role in repair and pathology after spinal cord injury. Brain Res. 2015, 1619, 1–11.

76. Schnell, L. Acute inflammatory responses to mechanical lesions in the CNS: Differences between brain and spinal cord. Eur. J. Neurosci. 1999, 11, 3648–3658.

77. Tzekou, A.; Fehlings, M.G. Treatment of spinal cord injury with intravenous immunoglobulin G: Preliminary evidence and future perspectives. J. Clin. Immunol. 2014, 34.

78. Gaudet, A.D.; Popovich, P.G. Extracellular matrix regulation of inflammation in the healthy and injured spinal cord. Exp. Neurol. 2014, 258, 24–34.

79. Losey, P.; Young, C.; Krimholtz, E.; Bordet, R.; Anthony, D.C. The role of hemorrhage following spinal-cord injury. Brain Res. 2014, 1569, 9–18.

80. Bradbury, E.J.; Burnside, E.R. Moving beyond the glial scar for spinal cord repair. Nat. Commun. 2019, 10.

81. Titova, E.; Ostrowski, R.P.; Kevil, C.G.; Tong, W.; Rojas, H.; Sowers, L.C.; Zhang, J.H.; Tang, J. Reduced brain injury in CD18-deficient mice after experimental intracerebral hemorrhage. J. Neurosci. Res. 2008, 86, 3240–3245.

82. Dusart, I.; Schwab, M.E. Secondary Cell Death and the Inflammatory Reaction After Dorsal Hemisection of the Rat Spinal Cord. Eur. J. Neurosci. 1994, 6, 712–724.

83. Kong, X.; Gao, J. Macrophage polarization: a key event in the secondary phase of acute spinal cord injury. J. Cell. Mol. Med. 2017, 21, 941–954.

133

84. Katoh, H.; Yokota, K.; Fehlings, M.G. Regeneration of Spinal Cord Connectivity Through Stem Cell Transplantation and Biomaterial Scaffolds. Front. Cell. Neurosci. 2019, 13.

85. Cherry, J.D.; Olschowka, J.A.; O’Banion, M.K. Neuroinflammation and M2 microglia: The good, the bad, and the inflamed. J. Neuroinflammation 2014, 11.

86. Gao, Z.; Wang, J.; Thiex, R.; Rogove, A.D.; Heppner, F.L.; Tsirka, S.E. Microglial activation and intracerebral hemorrhage. Acta Neurochir. Suppl. 2008, 51–53.

87. Mautes, A.E.; Weinzierl, M.R.; Donovan, F.; Noble, L.J. Vascular Events After Spinal Cord Injury: Contribution to Secondary Pathogenesis. Phys. Ther. 2000, 80, 673.

88. Anderson, M.A.; Burda, J.E.; Ren, Y.; Ao, Y.; O’Shea, T.M.; Kawaguchi, R.; Coppola, G.; Khakh, B.S.; Deming, T.J.; Sofroniew, M. V Astrocyte scar formation aids central nervous system axon regeneration. Nature 2016, 532, 195–200.

89. Standring, S.; Borley, N.; Collins, P.; Crossman, A.; Gatzoulis, M.; Healy, J.; Johnson, D.; Mahadevan, V.; Newell, R.; Wigley, C. Gray’s Anatomy; 40th ed.; Elsevier, 2008;

90. Siclari, F.; Burger, I.M.; Fasel, J.H.D.; Gailloud, P. Developmental anatomy of the distal vertebral artery in relationship to variants of the posterior and lateral spinal arterial systems. Am. J. Neuroradiol. 2007, 28, 1185–1190.

91. Alleyne, C.H.; Cawley, C.M.; Shengelaia, G.G.; Barrow, D.L. Microsurgical anatomy of the artery of Adamkiewicz and its segmental artery. J. Neurosurg. 1998, 89, 791–795.

92. Santillan, A.; Nacarino, V.; Greenberg, E.; Riina, H.A.; Gobin, P.; Patsalides, A.; Gobin, Y.P.; Patsalides, A. Vascular anatomy of the spinal cord. J. Neurointerv. Surg. 2012, 4, 67–74.

134

93. Turnbull, I.M. Microvasculature of the human spinal cord. J. Neurosurg. 1971, 35, 141– 147.

94. Miyasaka, K.; Asano, T.; Ushikoshi, S.; Hida, K.; Koyanagi, I. Vascular anatomy of the spinal cord and classification of spinal arteriovenous malformations. In Proceedings of the Interventional Neuroradiology; 2000; Vol. 6, pp. 195–198.

95. Lasjaunias, P.; Vallee, B.; Person, H.; Ter Brugge, K.; Chiu, M. The lateral spinal artery of the upper cervical spinal cord. Anatomy, normal variations, and angiographic aspects. J. Neurosurg. 1985, 63, 235–241.

96. Kroszczynski, A.C.; Kohan, K.; Kurowski, M.; Olson, T.R.; Downie, S.A. Intraforaminal location of thoracolumbar anterior medullary arteries. Pain Med. (United States) 2013, 14, 808–812.

97. Dommisse, G.F. The blood supply of the spinal cord. A critical vascular zone in spinal surgery. J. Bone Jt. Surg. - Ser. B 1974, 56, 225–235.

98. Amato, A.C.M.; Stolf, N.A.G. Anatomia da circulação medular. J. Vasc. Bras. 2015, 14, 248–252.

99. Bosmia, A.N.; Hogan, E.; Loukas, M.; Tubbs, R.S.; Cohen-Gadol, A.A. Blood supply to the human spinal cord: Part I. Anatomy and hemodynamics. Clin. Anat. 2015, 28, 52–64.

100. Gailloud, P. The artery of von Haller: A constant anterior radiculomedullary artery at the upper thoracic level. Neurosurgery 2013, 73, 1034–1043.

101. Yoshioka, K.; Niinuma, H.; Ehara, S.; Nakajima, T.; Nakamura, M.; Kawazoe, K. MR angiography and CT angiography of the artery of Adamkiewicz: State of the art. Radiographics 2006, 26.

135

102. Biglioli, P.; Spirito, R.; Roberto, M.; Grillo, F.; Cannata, A.; Parolari, A.; Maggioni, M.; Coggi, G. The anterior spinal artery: The main arterial supply of the human spinal cord - A preliminary anatomic study. J. Thorac. Cardiovasc. Surg. 2000, 119, 376–379.

103. Tveten, L. Spinal cord vascularity. I. Extraspinal sources of spinal cord arteries in man. Acta Radiol. Diagn. (Stockh). 1976, 17, 1–16.

104. Sukeeyamanon, W.; Sirsiapisith, T.; Wasinrat, J. Preoperative localization of adamkiewicz arteries and their origins by using MDCT angiography. J. Med. Assoc. Thail. 2010, 93, 1430–1436.

105. Rodriguez-Baeza, A.; Muset-Lara, A.; Rodriguez-Pazos, M.; Domenech-Mateu, J.M. The arterial supply of the human spinal cord: A new approach to the arteria radicularis magna of Adamkiewicz. Acta Neurochir. (Wien). 1991, 109, 57–62.

106. Guziński, M.; Bryl, M.; Ziemińska, K.; Wolny, K.; Saiadek, M.; Garcarek, J.S. Detection of the Adamkiewicz artery in computed tomography of the thorax & abdomen. Adv. Clin. Exp. Med. 2017, 26, 31–37.

107. Bley, T.A.; Duffek, C.C.; François, C.J.; Schiebler, M.L.; Acher, C.W.; Mell, M.; Grist, T.M.; Reeder, S.B. Presurgical localization of the artery of Adamkiewicz with time- resolved 3.0-T MR angiography. Radiology 2010, 255, 873–881.

108. Taterra, D.; Skinningsrud, B.; Pękala, P.A.; Hsieh, W.C.; Cirocchi, R.; Walocha, J.A.; Tubbs, R.S.; Tomaszewski, K.A.; Henry, B.M. Artery of Adamkiewicz: a meta-analysis of anatomical characteristics. Neuroradiology 2019, 61, 869–880.

109. Lazorthes, G.; Gouaze, A.; Zadeh, J.O.; Santini, J.J.; Lazorthes, Y.; Burdin, P. Arterial vascularization of the spinal cord. Recent studies of the anastomotic substitution

136

pathways. J. Neurosurg. 1971, 35, 253–262.

110. Griepp, E.B.; Di Luozzo, G.; Schray, D.; Stefanovic, A.; Geisbüsch, S.; Griepp, R.B. The anatomy of the spinal cord collateral circulation. Ann. Cardiothorac. Surg. 2012, 1, 350– 357.

111. Murakami, H.; Kawahara, N.; Tomita, K.; Demura, S.; Kato, S.; Yoshioka, K. Does interruption of the artery of adamkiewicz during total en bloc spondylectomy affect neurologic function? Spine (Phila. Pa. 1976). 2010, 35.

112. Etz, C.D.; Kari, F.A.; Mueller, C.S.; Silovitz, D.; Brenner, R.M.; Lin, H.M.; Griepp, R.B. The collateral network concept: A reassessment of the anatomy of spinal cord perfusion. J. Thorac. Cardiovasc. Surg. 2011, 141, 1020–1028.

113. Rojas, S.; Ortega, M.; Rodríguez-Baeza, A. Vascular configurations of anastomotic basket of conus medullaris in human spinal cord. Clin. Anat. 2018, 31, 441–448.

114. Crosby; Humphrey; Lauer Correlative Anatomy of the Nervous System; 1962;

115. El-Toraei, I.; Juler, G. Ischemic Myelopathy. Angiology 1979, 30, 81–94.

116. Di Chiro, G. Spinal cord angiography. Proc. R. Soc. Med. 1970, 63, 184.

117. Gailloud, P. The artery of von Haller: A constant anterior radiculomedullary artery at the upper thoracic level. Neurosurgery 2013, 73, 1034–1043.

118. Backes, W.H.; Nijenhuis, R.J. Advances in spinal cord MR angiography. Am. J. Neuroradiol. 2008, 29, 619–631.

119. Bernacki, J.; Dobrowołska, A.; Nerwińska, K.; Małecki, A. Physiology and pharmacological role of the blood-brain barrier. Pharmacol. Reports 2008, 60, 600–622.

137

120. Bartanusz, V.; Jezova, D.; Alajajian, B.; Digicaylioglu, M. The blood-spinal cord barrier: Morphology and clinical implications. Ann. Neurol. 2011, 70, 194–206.

121. Sims, D.E. Recent advances in pericyte biology--implications for health and disease. Can. J. Cardiol. 1991, 7, 431–443.

122. Frank, R.N.; Dutta, S.; Mancini, M.A. Pericyte coverage is greater in the retinal than in the cerebral capillaries of the rat. Investig. Ophthalmol. Vis. Sci. 1987, 28, 1086–1091.

123. Daneman, R.; Zhou, L.; Kebede, A.A.; Barres, B.A. Pericytes are required for blood-brain barrier integrity during embryogenesis. Nature 2010, 468, 562–566.

124. Armulik, A.; Genové, G.; Mäe, M.; Nisancioglu, M.H.; Wallgard, E.; Niaudet, C.; He, L.; Norlin, J.; Lindblom, P.; Strittmatter, K.; et al. Pericytes regulate the blood-brain barrier. Nature 2010, 468, 557–561.

125. Ramsauer, M.; Krause, D.; Dermietzel, R. Cerebral Pericytes and Angiogenesis in vitro. FASEB J. 2002, 16, 1274–1276.

126. Bergers, G.; Song, S. The role of pericytes in blood-vessel formation and maintenance. Neuro. Oncol. 2005, 7, 452–464.

127. Kelley, C.; D’Amore, P.; Hechtman, H.B.; Shepro, D. Vasoactive hormones and cAMP affect pericyte contraction and stress fibres in vitro. J. Muscle Res. Cell Motil. 1988, 9, 184–194.

128. Hamilton, N.B. Pericyte-mediated regulation of capillary diameter: a component of neurovascular coupling in health and disease. Front. Neuroenergetics 2010, 2.

129. Abbott, N.J.; Rönnbäck, L.; Hansson, E. Astrocyte-endothelial interactions at the blood-

138

brain barrier. Nat. Rev. Neurosci. 2006, 7, 41–53.

130. Thomsen, M.S.; Routhe, L.J.; Moos, T. The vascular basement membrane in the healthy and pathological brain. J. Cereb. Blood Flow Metab. 2017, 37, 3300–3317.

131. Thomsen, M.S.; Birkelund, S.; Burkhart, A.; Stensballe, A.; Moos, T. Synthesis and deposition of basement membrane proteins by primary brain capillary endothelial cells in a murine model of the blood–brain barrier. J. Neurochem. 2017, 140, 741–754.

132. ALCOLADO, R.; WELLER, R.O.; PARRISH, E.P.; GARROD, D. The Cranial Arachnoid and Pia Mater in Man: Anatomical and Ultrastructural Observations. Neuropathol. Appl. Neurobiol. 1988, 14, 1–17.

133. Zhang, E.T.; Inman, C.B.; Weller, R.O. Interrelationships of the pia mater and the perivascular (Virchow-Robin) spaces in the human cerebrum. J. Anat. 1990, 170, 111– 123.

134. Morris, A.W.J.; Sharp, M.M.G.; Albargothy, N.J.; Fernandes, R.; Hawkes, C.A.; Verma, A.; Weller, R.O.; Carare, R.O. Vascular basement membranes as pathways for the passage of fluid into and out of the brain. Acta Neuropathol. 2016, 131, 725–736.

135. Gautam, J.; Zhang, X.; Yao, Y. The role of pericytic laminin in blood brain barrier integrity maintenance. Sci. Rep. 2016, 6, 1–13.

136. Aumailley, M. The laminin family. Cell Adhes. Migr. 2013, 7, 48–55.

137. Aumailley, M.; Bruckner-Tuderman, L.; Carter, W.G.; Deutzmann, R.; Edgar, D.; Ekblom, P.; Engel, J.; Engvall, E.; Hohenester, E.; Jones, J.C.R.; et al. A simplified laminin nomenclature. Matrix Biol. 2005, 24, 326–332.

139

138. Nirwane, A.; Yao, Y. Laminins and their receptors in the CNS. Biol. Rev. 2019, 94, 283– 306.

139. Yao, Y.; Chen, Z.L.; Norris, E.H.; Strickland, S. Astrocytic laminin regulates pericyte differentiation and maintains blood brain barrier integrity. Nat. Commun. 2014, 5.

140. Brown, L.S.; Foster, C.G.; Courtney, J.M.; King, N.E.; Howells, D.W.; Sutherland, B.A. Pericytes and neurovascular function in the healthy and diseased brain. Front. Cell. Neurosci. 2019, 13.

141. Engelhardt, B.; Carare, R.O.; Bechmann, I.; Flügel, A.; Laman, J.D.; Weller, R.O. Vascular, glial, and lymphatic immune gateways of the central nervous system. Acta Neuropathol. 2016, 132, 317–338.

142. Ma, Q.; Ineichen, B. V; Detmar, M.; Proulx, S.T. Outflow of cerebrospinal fluid is predominantly through lymphatic vessels and is reduced in aged mice. Nat. Commun. 2017, 8, 1434.

143. Dupont, G.; Schmidt, C.; Yilmaz, E.; Oskouian, R.J.; Macchi, V.; de Caro, R.; Tubbs, R.S. Our current understanding of the lymphatics of the brain and spinal cord. Clin. Anat. 2019, 32, 117–121.

144. Louveau, A.; Smirnov, I.; Keyes, T.J.; Eccles, J.D.; Rouhani, S.J.; Peske, J.D.; Derecki, N.C.; Castle, D.; Mandell, J.W.; Lee, K.S.; et al. Structural and functional features of central nervous system lymphatic vessels. Nature 2015, 523, 337–341.

145. Antila, S.; Karaman, S.; Nurmi, H.; Airavaara, M.; Voutilainen, M.H.; Mathivet, T.; Chilov, D.; Li, Z.; Koppinen, T.; Park, J.-H.; et al. Development and plasticity of meningeal lymphatic vessels. J. Exp. Med. 2017, 214, 3645–3667.

140

146. Jacob, L.; Boisserand, L.S.B.; Geraldo, L.H.M.; de Brito Neto, J.; Mathivet, T.; Antila, S.; Barka, B.; Xu, Y.; Thomas, J.-M.; Pestel, J.; et al. Anatomy and function of the vertebral column lymphatic network in mice. Nat. Commun. 2019, 10, 4594.

147. Van Zwam, M.; Huizinga, R.; Heijmans, N.; Van Meurs, M.; Wierenga-Wolf, A.F.; Melief, M.J.; Hintzen, R.Q.; Hart, B.A.T.; Amor, S.; Boven, L.A.; et al. Surgical excision of CNS-draining lymph nodes reduces relapse severity in chronic-relapsing experimental autoimmune encephalomyelitis. J. Pathol. 2009, 217, 543–551.

148. Muradov, J.M.; Ewan, E.E.; Hagg, T. Dorsal column sensory axons degenerate due to impaired microvascular perfusion after spinal cord injury in rats. Exp. Neurol. 2013, 249, 59–73.

149. Wilcox, J.T.; Satkunendrarajah, K.; Nasirzadeh, Y.; Laliberte, A.M.; Lip, A.; Cadotte, D.W.; Foltz, W.D.; Fehlings, M.G. Generating level-dependent models of cervical and thoracic spinal cord injury: Exploring the interplay of neuroanatomy, physiology, and function. Neurobiol. Dis. 2017, 105, 194–212.

150. Winkler, E.A.; Sengillo, J.D.; Bell, R.D.; Wang, J.; Zlokovic, B. V. Blood-spinal cord barrier pericyte reductions contribute to increased capillary permeability. J. Cereb. Blood Flow Metab. 2012, 32, 1841–1852.

151. Fehlings, M.G.; Ulndreaj, A.; Badner, A. Promising neuroprotective strategies for traumatic spinal cord injury with a focus on the differential effects among anatomical levels of injury. F1000Research 2017, 6.

152. Liu, S.J.; Wang, Q.; Tang, H.H.; Bai, J.Z.; Wang, F.Y.; Lv, Z.; Chen, S.Z.; Liu, J.S.; Hong, Y.; Zhang, J.W. Heterogeneity among traumatic spinal cord injuries at the thoracolumbar junction: helping select patients for clinical trials. Spinal Cord 2019.

141

153. Vaccaro, A.R.; Nachwalter, R.S.; Klein, G.R.; Sewards, J.M.; Albert, T.J.; Garfin, S.R. The Significance of Thoracolumbar Spinal Canal Size in Spinal Cord Injury Patients. Spine (Phila. Pa. 1976). 2001, 26, 371–376.

154. Brommer, B.; Engel, O.; Kopp, M.A.; Watzlawick, R.; Müller, S.; Prüss, H.; Chen, Y.; DeVivo, M.J.; Finkenstaedt, F.W.; Dirnagl, U.; et al. Spinal cord injury-induced immune deficiency syndrome enhances infection susceptibility dependent on lesion level. Brain 2016, 139, 692–707.

155. Ibarra, A.; Jiménez, A.; Cortes, C.; Correa, D. Influence of the intensity, level and phase of spinal cord injury on the proliferation of T cells and T-cell-dependent antibody reactions in rats. Spinal Cord 2007, 45, 380–386.

156. Hong, J.; Chang, A.; Liu, Y.; Wang, J.; Fehlings, M.G. Incomplete Spinal Cord Injury Reverses the Level-Dependence of Spinal Cord Injury Immune Deficiency Syndrome. Int. J. Mol. Sci. 2019, 20, 3762.

157. Stiell, I.G.; Clement, C.M.; Grimshaw, J.; Brison, R.J.; Rowe, B.H.; Schull, M.J.; Lee, J.S.; Brehaut, J.; McKnight, R.D.; Eisenhauer, M.A.; et al. Implementation of the Canadian C-spine rule: Prospective 12 centre cluster randomised trial. BMJ 2009, 339, 1071.

158. Stiell, I.G.; Clement, C.M.; McKnight, R.D.; Brison, R.; Schull, M.J.; Rowe, B.H.; Worthington, J.R.; Eisenhauer, M.A.; Cass, D.; Greenberg, G.; et al. The Canadian C- Spine Rule versus the NEXUS Low-Risk Criteria in Patients with Trauma. N. Engl. J. Med. 2003, 349, 2510–2518.

159. O’Toole, J.E.; Kaiser, M.G.; Anderson, P.A.; Arnold, P.M.; Chi, J.H.; Dailey, A.T.; Dhall, S.S.; Eichholz, K.M.; Harrop, J.S.; Hoh, D.J.; et al. Congress of neurological surgeons

142

systematic review and evidence-based guidelines on the evaluation and treatment of patients with thoracolumbar spine trauma: Executive summary. Clin. Neurosurg. 2019, 84, 2–6.

160. Ryken, T.C.; Hurlbert, R.J.; Hadley, M.N.; Aarabi, B.; Dhall, S.S.; Gelb, D.E.; Rozzelle, C.J.; Theodore, N.; Walters, B.C. The acute cardiopulmonary management of patients with cervical spinal cord injuries. Neurosurgery 2013, 72, 84–92.

161. Consortium for Spinal Cord Medicine Early Acute Management in Adults with Spinal Cord Injury; 2008; Vol. 31; ISBN 0929819209.

162. Wing, P.C. Early acute management in adults with spinal cord injury: A clinical practice guideline for health-care providers. Who should read it?; 2008; Vol. 31; ISBN 0929819209.

163. Aarabi, B.; Hadley, M.N.; Dhall, S.S.; Gelb, D.E.; John Hurlbert, R.; Rozzelle, C.J.; Ryken, T.C.; Theodore, N.; Walters, B.C. Management of acute traumatic central cord syndrome (ATCCS). Neurosurgery 2013, 72, 195–204.

164. Saadeh, Y.S.; Smith, B.W.; Joseph, J.R.; Jaffer, S.Y.; Buckingham, M.J.; Oppenlander, M.E.; Szerlip, N.J.; Park, P. The impact of blood pressure management after spinal cord injury: A systematic review of the literature. Neurosurg. Focus 2017, 43, E20.

165. Yue, J.K.; Tsolinas, R.E.; Burke, J.F.; Deng, H.; Upadhyayula, P.S.; Robinson, C.K.; Lee, Y.M.; Chan, A.K.; Winkler, E.A.; Dhall, S.S. Vasopressor support in managing acute spinal cord injury: current knowledge. J. Neurosurg. Sci. 2019, 63, 308–317.

166. Inoue, T.; Manley, G.T.; Patel, N.; Whetstone, W.D. Medical and surgical management after spinal cord injury: Vasopressor usage, early surgerys, and complications. J.

143

Neurotrauma 2014, 31, 284–291.

167. Readdy, W.J.; Whetstone, W.D.; Ferguson, A.R.; Talbott, J.F.; Inoue, T.; Saigal, R.; Bresnahan, J.C.; Beattie, M.S.; Pan, J.Z.; Manley, G.T.; et al. Complications and outcomes of vasopressor usage in acute traumatic central cord syndrome. J. Neurosurg. Spine 2015, 23, 574–580.

168. O’Toole, J.E.; Kaiser, M.G.; Anderson, P.A.; Arnold, P.M.; Chi, J.H.; Dailey, A.T.; Dhall, S.S.; Eichholz, K.M.; Harrop, J.S.; Hoh, D.J.; et al. Congress of neurological surgeons systematic review and evidence-based guidelines on the evaluation and treatment of patients with thoracolumbar spine trauma: Executive summary. Clin. Neurosurg. 2019, 84, 2–6.

169. Hadley, M.N.; Walters, B.C.; Aarabi, B.; Dhall, S.S.; Gelb, D.E.; Hurlbert, R.J.; Rozzelle, C.J.; Ryken, T.C.; Theodore, N. Clinical assessment following acute cervical spinal cord injury. Neurosurgery 2013, 72, 40–53.

170. Kirshblum, S.C.; Waring, W.; Biering-Sorensen, F.; Burns, S.P.; Johansen, M.; Schmidt- Read, M.; Donovan, W.; Graves, D.; Jha, A.; Jones, L.; et al. Reference for the 2011 revision of the International Standards for Neurological Classification of Spinal Cord Injury. J. Spinal Cord Med. 2011, 34, 547–554.

171. Aarabi, B.; Walters, B.C.; Dhall, S.S.; Gelb, D.E.; John Hurlbert, R.; Rozzelle, C.J.; Ryken, T.C.; Theodore, N.; Hadley, M.N. Subaxial cervical spine injury classification systems. Neurosurgery 2013, 72, 170–186.

172. Dailey, A.T.; Arnold, P.M.; Anderson, P.A.; Chi, J.H.; Dhall, S.S.; Eichholz, K.M.; Harrop, J.S.; Hoh, D.J.; Qureshi, S.; Rabb, C.H.; et al. Congress of neurological surgeons systematic review and evidence-based guidelines on the evaluation and treatment of

144

patients with thoracolumbar spine trauma: Classification of injury. Clin. Neurosurg. 2019, 84, E24–E27.

173. Lee, J.Y.; Vaccaro, A.R.; Lim, M.R.; Öner, F.C.; Hulbert, R.J.; Hedlund, R.; Fehlings, M.G.; Arnold, P.; Harrop, J.; Bono, C.M.; et al. Thoracolumbar injury classification and severity score: A new paradigm for the treatment of thoracolumbar spine trauma. J. Orthop. Sci. 2005, 10, 671–675.

174. Van Middendorp, J.J. Letter to the editor regarding: “Early versus delayed decompression for traumatic cervical spinal cord injury: Results of the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS).” Spine J. 2012, 12, 540.

175. Wilson, J.R.; Tetreault, L.A.; Kwon, B.K.; Arnold, P.M.; Mroz, T.E.; Shaffrey, C.; Harrop, J.S.; Chapman, J.R.; Casha, S.; Skelly, A.C.; et al. Timing of Decompression in Patients With Acute Spinal Cord Injury: A Systematic Review. Glob. Spine J. 2017, 7, 95S-115S.

176. Du, J.P.; Fan, Y.; Zhang, J.N.; Liu, J.J.; Meng, Y. Bin; Hao, D.J. Early versus delayed decompression for traumatic cervical spinal cord injury: application of the AOSpine subaxial cervical spinal injury classification system to guide surgical timing. Eur. Spine J. 2019.

177. Medicine, C. for S.C. Early Acute Management in Adults with Spinal Cord Injury. J. Spinal Cord Med. 2008, 31, 403–479.

178. Streijger, F.; So, K.; Manouchehri, N.; Gheorghe, A.; Okon, E.B.; Chan, R.M.; Ng, B.; Shortt, K.; Sekhon, M.S.; Griesdale, D.E.; et al. A Direct Comparison between Norepinephrine and Phenylephrine for Augmenting Spinal Cord Perfusion in a Porcine Model of Spinal Cord Injury. J. Neurotrauma 2018, 35, 1345–1357.

145

179. Simard, J.M.; Tsymbalyuk, O.; Ivanov, A.; Ivanova, S.; Bhatta, S.; Geng, Z.; Woo, S.K.; Gerzanich, V. Endothelial sulfonylurea receptor 1-regulated NC Ca-ATP channels mediate progressive hemorrhagic necrosis following spinal cord injury. J. Clin. Invest. 2007, 117, 2105–2113.

180. Coderre, T.J.; Kumar, N.; Lefebvre, C.D.; Yu, J.S.C. A comparison of the glutamate release inhibition and anti-allodynic effects of gabapentin, lamotrigine, and riluzole in a model of neuropathic pain. J. Neurochem. 2007, 100, 1289–1299.

181. Badhiwala, J.H.; Ahuja, C.S.; Fehlings, M.G. Time is spine: A review of translational advances in spinal cord injury. J. Neurosurg. Spine 2019, 30, 1–18.

182. Sámano, C.; Nasrabady, S.E.; Nistri, A. A study of the potential neuroprotective effect of riluzole on locomotor networks of the neonatal rat spinal cord in vitro damaged by excitotoxicity. Neuroscience 2012, 222, 356–365.

183. Grossman, R.G.; Fehlings, M.G.; Frankowski, R.F.; Burau, K.D.; Chow, D.S.L.; Tator, C.; Teng, A.; Toups, E.G.; Harrop, J.S.; Aarabi, B.; et al. A Prospective, Multicenter, Phase I Matched-Comparison Group Trial of Safety, Pharmacokinetics, and Preliminary Efficacy of Riluzole in Patients with Traumatic Spinal Cord Injury. J. Neurotrauma 2014, 31, 239–255.

184. Fournier, A.E.; Takizawa, B.T.; Strittmatter, S.M. Rho kinase inhibition enhances axonal regeneration in the injured CNS. J. Neurosci. 2003.

185. Dergham, P.; Ellezam, B.; Essagian, C.; Avedissian, H.; Lubell, W.D.; McKerracher, L. Rho signaling pathway targeted to promote spinal cord repair. J. Neurosci. 2002.

186. Fehlings, M.G.; Theodore, N.; Harrop, J.; Maurais, G.; Kuntz, C.; Shaffrey, C.I.; Kwon,

146

B.K.; Chapman, J.; Yee, A.; Tighe, A.; et al. A phase I/IIa clinical trial of a recombinant Rho protein antagonist in acute spinal cord injury. J. Neurotrauma 2011, 28, 787–796.

187. Ahuja, C.S.; Wilson, J.R.; Nori, S.; Kotter, M.R.N.; Druschel, C.; Curt, A.; Fehlings, M.G. Traumatic spinal cord injury. Nat Rev Dis Prim. 2017, 3, 17018.

188. Casha, S.; Zygun, D.; McGowan, M.D.; Bains, I.; Yong, V.W.; John Hurlbert, R. Results of a phase II placebo-controlled randomized trial of minocycline in acute spinal cord injury. Brain 2012, 135, 1224–1236.

189. Kitamura, K.; Fujiyoshi, K.; Yamane, J. ichi; Toyota, F.; Hikishima, K.; Nomura, T.; Funakoshi, H.; Nakamura, T.; Aoki, M.; Toyama, Y.; et al. Human hepatocyte growth factor promotes functional recovery in primates after spinal cord injury. PLoS One 2011.

190. Yamane, K.; Mazaki, T.; Shiozaki, Y.; Yoshida, A.; Shinohara, K.; Nakamura, M.; Yoshida, Y.; Zhou, D.; Kitajima, T.; Tanaka, M.; et al. Collagen-Binding Hepatocyte Growth Factor (HGF) alone or with a Gelatin-furfurylamine Hydrogel Enhances Functional Recovery in Mice after Spinal Cord Injury. Sci. Rep. 2018.

191. Yamane, K.; Misawa, H.; Takigawa, T.; Ito, Y.; Ozaki, T.; Matsukawa, A. Multipotent Neurotrophic Effects of Hepatocyte Growth Factor in Spinal Cord Injury. Int. J. Mol. Sci. 2019.

192. Date, I.; Takagi, N.; Takagi, K.; Kago, T.; Matsumoto, K.; Nakamura, T.; Takeo, S. Hepatocyte growth factor attenuates cerebral ischemia-induced learning dysfunction. Biochem. Biophys. Res. Commun. 2004.

193. Date, I.; Takagi, N.; Takagi, K.; Kago, T.; Matsumoto, K.; Nakamura, T.; Takeo, S. Hepatocyte growth factor improved learning and memory dysfunction of microsphere-

147

embolized rats. J. Neurosci. Res. 2004.

194. Tzekou, A.; Fehlings, M.G. Treatment of spinal cord injury with intravenous immunoglobulin G: Preliminary evidence and future perspectives. J. Clin. Immunol. 2014, 34.

195. Leypold, B.G.; Flanders, A.E.; Schwartz, E.D.; Burns, A.S. The impact of methylprednisolone on lesion severity following spinal cord injury. Spine (Phila. Pa. 1976). 2007, 32, 373–378.

196. Crockard, A.D.; Boylan, M.T.; Droogan, A.G.; McMillan, S.A.; Hawkins, S.A. Methylprednisolone-induced neutrophil leukocytosis--down-modulation of neutrophil L- selectin and Mac-1 expression and induction of granulocyte-colony stimulating factor. Int. J. Clin. Lab. Res. 1998, 28, 110–115.

197. Hall, E.D.; Springer, J.E. Neuroprotection and Acute Spinal Cord Injury: A Reappraisal. NeuroRx 2004, 1, 80–100.

198. Bracken, M.B.; Shepard, M.J.; Holford, T.R.; Leo-Summers, L.; Aldrich, E.F.; Fazl, M.; Fehlings, M.; Herr, D.L.; Hitchon, P.W.; Marshall, L.F.; et al. Administration of methylprednisolone for 24 or 48 hours or tirilazad mesylate for 48 hours in the treatment of acute spinal cord injury: Results of the Third National Acute Spinal Cord Injury randomized controlled trial. J. Am. Med. Assoc. 1997, 277, 1597–1604.

199. Bracken, M.B.; Collins, W.F.; Freeman, D.F.; Shepard, M.J.; Wagner, F.W.; Silten, R.M.; Hellenbrand, K.G.; Ransohoff, J.; Hunt, W.E.; Perot, P.L.; et al. Efficacy of Methylprednisolone in Acute Spinal Cord Injury. JAMA J. Am. Med. Assoc. 1984, 251, 45–52.

148

200. Bracken, M.B.; Shepard, M. jo; Collins, W.F.; Holford, T.R.; Young, W.; Baskin, D.S.; Eisenberg, H.M.; Flamm, E.; Leo-Summers, L.; Maroon, J.; et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury: Results of the second national acute spinal cord injury study. N. Engl. J. Med. 1990, 322, 1405–1411.

201. Hextrum, S.; Bennett, S. A critical examination of subgroup analyses: The national acute spinal cord injury studies and beyond. Front. Neurol. 2018, 9.

202. Sayer, F.T.; Kronvall, E.; Nilsson, O.G. Methylprednisolone treatment in acute spinal cord injury: the myth challenged through a structured analysis of published literature. Spine J. 2006, 6, 335–343.

203. Bracken, M.B. Cochrane review NASCIS (Steroids for acute spinal cord injury) by same author as nascis.

204. Fehlings, M.G.; Wilson, J.R.; Tetreault, L.A.; Aarabi, B.; Anderson, P.; Arnold, P.M.; Brodke, D.S.; Burns, A.S.; Chiba, K.; Dettori, J.R.; et al. A Clinical Practice Guideline for the Management of Patients With Acute Spinal Cord Injury: Recommendations on the Use of Methylprednisolone Sodium Succinate. Glob. Spine J. 2017, 7, 203S-211S.

205. Evaniew, N.; Noonan, V.K.; Fallah, N.; Kwon, B.K.; Rivers, C.S.; Ahn, H.; Bailey, C.S.; Christie, S.D.; Fourney, D.R.; Hurlbert, R.J.; et al. Methylprednisolone for the Treatment of Patients with Acute Spinal Cord Injuries: A Propensity Score-Matched Cohort Study from a Canadian Multi-Center Spinal Cord Injury Registry. J. Neurotrauma 2015, 32, 1674–1683.

206. Rahimi-Movaghar, V. Clinical trials for the treatment of spinal cord injury: Cervical and lumbar enlargements versus thoracic area. Brain 2009, 132.

149

207. Ankeny, D.P.; Guan, Z.; Popovich, P.G. B cells produce pathogenic antibodies and impair recovery after spinal cord injury in mice. J. Clin. Invest. 2009, 119, 2990–2999.

208. Ueno, M.; Ueno-Nakamura, Y.; Niehaus, J.; Popovich, P.G.; Yoshida, Y. Silencing spinal interneurons inhibits immune suppressive autonomic reflexes caused by spinal cord injury. Nat Neurosci 2016, 19, 784–787.

209. Riegger, T.; Conrad, S.; Liu, K.; Schluesener, H.J.; Adibzahdeh, M.; Schwab, J.M. Spinal cord injury-induced immune depression syndrome (SCI-IDS). Eur J Neurosci 2007, 25, 1743–1747.

210. Riegger, T.; Conrad, S.; Schluesener, H.J.; Kaps, H.P.; Badke, A.; Baron, C.; Gerstein, J.; Dietz, K.; Abdizahdeh, M.; Schwab, J.M. Immune depression syndrome following human spinal cord injury (SCI): a pilot study. Neuroscience 2009, 158, 1194–1199.

211. Kadoya, K.; Lu, P.; Nguyen, K.; Lee-Kubli, C.; Kumamaru, H.; Yao, L.; Knackert, J.; Poplawski, G.; Dulin, J.N.; Strobl, H.; et al. Spinal cord reconstitution with homologous neural grafts enables robust corticospinal regeneration. Nat. Med. 2016.

212. Andrews, S. FASTQC A Quality Control tool for High Throughput Sequence Data. Babraham Inst. 2015.

213. Marcel, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 2011.

214. Trapnell, C.; Pachter, L.; Salzberg, S.L. TopHat: Discovering splice junctions with RNA- Seq. Bioinformatics 2009.

215. Liao, Y.; Smyth, G.K.; Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 2014, 30, 923–930.

150

216. Zerbino, D.R.; Achuthan, P.; Akanni, W.; Amode, M.R.; Barrell, D.; Bhai, J.; Billis, K.; Cummins, C.; Gall, A.; Girón, C.G.; et al. Ensembl 2018. Nucleic Acids Res. 2018.

217. Tarazona, S.; García, F.; Ferrer, A.; Dopazo, J.; Conesa, A. NOIseq: a RNA-seq differential expression method robust for sequencing depth biases. EMBnet.journal 2012.

218. Risso, D.; Ngai, J.; Speed, T.P.; Dudoit, S. Normalization of RNA-seq data using factor analysis of control genes or samples. Nat Biotechnol 2014, 32, 896–902.

219. Love, M.I.; Huber, W.; Anders, S. DESeq2; 2014; ISBN 1474-760X (Electronic)1474- 7596 (Linking).

220. Reimand, J.; Arak, T.; Adler, P.; Kolberg, L.; Reisberg, S.; Peterson, H.; Vilo, J. g:Profiler-a web server for functional interpretation of gene lists (2016 update). Nucleic Acids Res 2016, 44, W83-9.

221. He, L.; Vanlandewijck, M.; Mäe, M.A.; Andrae, J.; Ando, K.; Gaudio, F. Del; Nahar, K.; Lebouvier, T.; Laviña, B.; Gouveia, L.; et al. Data descriptor: Single-cell RNA sequencing of mouse brain and lung vascular and vessel-associated cell types. Sci. Data 2018.

222. Newman, A.M.; Steen, C.B.; Liu, C.L.; Gentles, A.J.; Chaudhuri, A.A.; Scherer, F.; Khodadoust, M.S.; Esfahani, M.S.; Luca, B.A.; Steiner, D.; et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat. Biotechnol. 2019.

223. Rusk, N. Expanded CIBERSORTx. Nat. Methods 2019.

224. Soubeyrand, M.; Badner, A.; Vawda, R.; Chung, Y.S.; Fehlings, M.G. Very high resolution ultrasound imaging for real-time quantitative visualization of vascular disruption after spinal cord injury. J Neurotrauma 2014, 31, 1767–1775.

151

225. Schindelin, J.; Arganda-Carreras, I.; Frise, E.; Kaynig, V.; Longair, M.; Pietzsch, T.; Preibisch, S.; Rueden, C.; Saalfeld, S.; Schmid, B.; et al. Fiji: An open-source platform for biological-image analysis. Nat. Methods 2012.

226. Cardona, A.; Saalfeld, S.; Schindelin, J.; Arganda-Carreras, I.; Preibisch, S.; Longair, M.; Tomancak, P.; Hartenstein, V.; Douglas, R.J. TrakEM2 software for neural circuit reconstruction. PLoS One 2012.

227. SOLNITZKY, O. The vascular supply of the spinal cord: anatomical and clinical considerations. Bull. Georgetown Univ. Med. Cent. 1959.

228. Seki, T.; Fehlings, M.G. Mechanistic insights into posttraumatic syringomyelia based on a novel in vivo animal model: Laboratory investigation. J. Neurosurg. Spine 2008.

229. Armulik, A.; Genové, G.; Betsholtz, C. Pericytes: Developmental, Physiological, and Pathological Perspectives, Problems, and Promises. Dev. Cell 2011.

230. Sa-Pereira, I.; Brites, D.; Brito, M.A. Neurovascular unit: a focus on pericytes. Mol Neurobiol 2012, 45, 327–347.

231. Goritz, C.; Dias, D.O.; Tomilin, N.; Barbacid, M.; Shupliakov, O.; Frisen, J. A pericyte origin of spinal cord scar tissue. Science (80-. ). 2011, 333, 238–242.

232. Dias, D.O.; Kim, H.; Holl, D.; Werne Solnestam, B.; Lundeberg, J.; Carlén, M.; Göritz, C.; Frisén, J. Reducing Pericyte-Derived Scarring Promotes Recovery after Spinal Cord Injury. Cell 2018.

233. Vanlandewijck, M.; Betsholtz, C. Single-cell mRNA sequencing of the mouse brain vasculature. In Methods in Molecular Biology; 2018.

152

234. Zamanian, J.L.; Xu, L.; Foo, L.C.; Nouri, N.; Zhou, L.; Giffard, R.G.; Barres, B.A. Genomic analysis of reactive astrogliosis. J. Neurosci. 2012.

235. Braughler, J.M.; Duncan, L.A.; Chase, R.L. Interaction of lipid peroxidation and calcium in the pathogenesis of neuronal injury. Cent Nerv Syst Trauma 1985, 2, 269–283.

236. Xiong, Y.; Rabchevsky, A.G.; Hall, E.D. Role of peroxynitrite in secondary oxidative damage after spinal cord injury. J. Neurochem. 2007, 100, 639–649.

237. Whetstone, W.D.; Hsu, J.-Y.C.; Eisenberg, M.; Werb, Z.; Noble-Haeusslein, L.J. Blood- spinal cord barrier after spinal cord injury: Relation to revascularization and wound healing. J. Neurosci. Res. 2003, 74, 227–239.

238. Noble, L.J.; Donovan, F.; Igarashi, T.; Goussev, S.; Werb, Z. Matrix metalloproteinases limit functional recovery after spinal cord injury by modulation of early vascular events. J. Neurosci. 2002, 22, 7526–7535.

239. Zhang, Y.; Guan, Z.; Reader, B.; Shawler, T.; Mandrekar-Colucci, S.; Huang, K.; Weil, Z.; Bratasz, A.; Wells, J.; Powell, N.D.; et al. Autonomic Dysreflexia Causes Chronic Immune Suppression after Spinal Cord Injury. J. Neurosci. 2013, 33, 12970–12981.

240. Schwab, J.M.; Zhang, Y.; Kopp, M.A.; Brommer, B.; Popovich, P.G. The paradox of chronic neuroinflammation, systemic immune suppression, autoimmunity after traumatic chronic spinal cord injury. Exp. Neurol. 2014, 258, 121–129.

241. Hagen, E.M. Acute complications of spinal cord injuries. World J Orthop 2015, 6, 17–23.

242. Levi, L.; Wolf, A.; Belzberg, H. Hemodynamic parameters in patients with acute cervical cord trauma: description, intervention, and prediction of outcome. Neurosurgery 33, 1007.

153

243. Barros Filho, T.E.; Molina, A.E. Analysis of the sensitivity and reproducibility of the Basso, Beattie, Bresnahan (BBB) scale in Wistar rats. Clin. (Sao Paulo) 2008, 63, 103– 108.

244. Basso, D.M.; Beattie, M.S.; Bresnahan, J.C. Graded histological and locomotor outcomes after spinal cord contusion using the NYU weight-drop device versus transection. Exp. Neurol.

245. Waykole, Y.; Doiphode, S.; Rakhewar, P.; Mhaske, M. Anticytokine therapy for periodontal diseases: Where are we now? J. Indian Soc. Periodontol. 2009.

246. Audet, M.C.; Anisman, H. Interplay between pro-inflammatory cytokines and growth factors in depressive illnesses. Front. Cell. Neurosci. 2013.

247. Kofler, S.; Nickel, T.; Weis, M. Role of cytokines in cardiovascular diseases: A focus on endothelial responses to inflammation. Clin. Sci. 2005.

248. Mukhamedshina, Y.O.; Akhmetzyanova, E.R.; Martynova, E. V.; Khaiboullina, S.F.; Galieva, L.R.; Rizvanov, A.A. Systemic and local cytokine profile following spinal cord injury in rats: A multiplex analysis. Front. Neurol. 2017.

249. Bartholdi, D.; Schwab, M.E. Expression of pro-inflammatory cytokine and chemokine mRNA upon experimental spinal cord injury in mouse: an in situ hybridization study. Eur. J. Neurosci. 1997, 9, 1422–1438.

250. Klusman, I.; Schwab, M.E. Effects of pro-inflammatory cytokines in experimental spinal cord injury. Brain Res. 1997.

251. Schnell, L.; Fearn, S.; Schwab, M.E.; Perry, V.H.; Anthony, D.C. Cytokine-induced acute inflammation in the brain and spinal cord. J. Neuropathol. Exp. Neurol. 58, 245–254.

154

252. Francos-Quijorna, I.; Santos-Nogueira, E.; Gronert, K.; Sullivan, A.B.; Kopp, M.A.; Brommer, B.; David, S.; Schwab, J.M.; López-Vales, R. Maresin 1 promotes inflammatory resolution, neuroprotection, and functional neurological recovery after spinal cord injury. J. Neurosci. 2017.

253. Peterson, S.L.; Anderson, A.J. Complement and spinal cord injury: traditional and non- traditional aspects of complement cascade function in the injured spinal cord microenvironment. Exp Neurol 2014, 258, 35–47.

254. Anderson, A.J.; Robert, S.; Huang, W.; Young, W.; Cotman, C.W. Activation of complement pathways after contusion-induced spinal cord injury. J. Neurotrauma 2004.

255. Kasinathan, N.; Vanathi, M.; Subrahmanyam, V.; Rao, J. A Review on Response of Immune System in Spinal Cord Injury and Therapeutic Agents useful in Treatment. Curr. Pharm. Biotechnol. 2015.

256. Liu, Y.; Figley, S.; Spratt, S.K.; Lee, G.; Ando, D.; Surosky, R.; Fehlings, M.G. An engineered transcription factor which activates VEGF-A enhances recovery after spinal cord injury. Neurobiol. Dis. 2010.

257. Figley, S.A.; Liu, Y.; Karadimas, S.K.; Satkunendrarajah, K.; Fettes, P.; Kaye Spratt, S.; Lee, G.; Ando, D.; Surosky, R.; Giedlin, M.; et al. Delayed administration of a bio- engineered zinc-finger VEGF-A gene therapy is neuroprotective and attenuates allodynia following traumatic spinal cord injury. PLoS One 2014.

258. Wang, H.; Wang, Y.; Li, D.; Liu, Z.; Zhao, Z.; Han, D.; Yuan, Y.; Bi, J.; Mei, X. VEGF inhibits the inflammation in spinal cord injury through activation of autophagy. Biochem. Biophys. Res. Commun. 2015.

155

259. Wang, L.; Tang, X.; Zhang, H.; Yuan, J.; Ding, H.; Wei, Y. Elevated leptin expression in rat model of traumatic spinal cord injury and femoral fracture. J. Spinal Cord Med. 2011.

260. Gezici, A.R.; Ergun, R.; Karakas, A.; Gunduz, B. Serum Leptin Levels Following Acute Experimental Spinal Cord Injury. J Spinal Cord Med 2009, 32, 416–421.

261. Garshick, E.; Walia, P.; Goldstein, R.L.; Teylan, M.; Lazzari, A.A.; Tun, C.G.; Hart, J.E. Plasma Leptin and Reduced FEV1 and FVC in Chronic Spinal Cord Injury. PM R 2018.

262. Jeon, J.Y.; Steadward, R.D.; Wheeler, G.D.; Bell, G.; McCargar, L.; Harber, V. Intact sympathetic nervous system is required for leptin effects on resting metabolic rate in people with spinal cord injury. J. Clin. Endocrinol. Metab. 2003.

263. Fernández-Martos, C.M.; González, P.; Rodriguez, F.J. Acute leptin treatment enhances functional recovery after spinal cord injury. PLoS One 2012.

264. Lubieniecka, J.M.; Streijger, F.; Lee, J.H.T.; Stoynov, N.; Liu, J.; Mottus, R.; Pfeifer, T.; Kwon, B.K.; Coorssen, J.R.; Foster, L.J.; et al. Biomarkers for severity of spinal cord injury in the cerebrospinal fluid of rats. PLoS One 2011.

265. Kwon, B.K.; Stammers, A.M.T.; Belanger, L.M.; Bernardo, A.; Chan, D.; Bishop, C.M.; Slobogean, G.P.; Zhang, H.; Umedaly, H.; Giffin, M.; et al. Cerebrospinal fluid inflammatory cytokines and biomarkers of injury severity in acute human spinal cord injury. J. Neurotrauma 2010.

266. Glaser, J.; Gonzalez, R.; Perreau, V.M.; Cotman, C.W.; Keirstead, H.S. Neutralization of the chemokine CXCL10 enhances tissue sparing and angiogenesis following spinal cord injury. J. Neurosci. Res. 2004.

267. Gonzalez, R.; Hickey, M.J.; Espinosa, J.M.; Nistor, G.; Lane, T.E.; Keirstead, H.S.

156

Therapeutic neutralization of CXCL10 decreases secondary degeneration and functional deficit after spinal cord injury in mice. Regen. Med. 2007, 2, 771–783.

268. Min, J.; Kim, J.H.; Choi, K.H.; Yoon, H.H.; Jeon, S.R. Is there additive therapeutic effect when GCSF combined with adipose-derived stem cell in a rat model of acute spinal cord injury? J. Korean Neurosurg. Soc. 2017.

269. Koda, M.; Nishio, Y.; Kamada, T.; Someya, Y.; Okawa, A. … stimulating factor (G-CSF) mobilizes bone marrow-derived cells into injured spinal cord and promotes functional recovery after compression-induced spinal cord injury …. Brain Res.

270. Takahashi, H.; Yamazaki, M.; Okawa, A.; Sakuma, T.; Kato, K.; Hashimoto, M.; Hayashi, K.; Furuya, T.; Fujiyoshi, T.; Kawabe, J.; et al. Neuroprotective therapy using granulocyte colony-stimulating factor for acute spinal cord injury: a phase I/IIa clinical trial. Eur. Spine J. 2012, 21, 2580–2587.

271. Pitzer, C.; Kruger, C.; Plaas, C.; Kirsch, F.; Dittgen, T.; Muller, R.; Laage, R.; Kastner, S.; Suess, S.; Spoelgen, R.; et al. Granulocyte-colony stimulating factor improves outcome in a mouse model of amyotrophic lateral sclerosis. Brain 2008, 131, 3335–3347.

272. Guo, Y.; Liu, S.; Wang, P.; Zhang, H.; Wang, F.; Bing, L.; Gao, J.; Yang, J.; Hao, A. Granulocyte colony-stimulating factor improves neuron survival in experimental spinal cord injury by regulating nucleophosmin-1 expression. J. Neurosci. Res. 2014, 92, 751– 760.

273. Koda, M.; Nishio, Y.; Kamada, T.; Someya, Y.; Okawa, A.; Mori, C.; Yoshinaga, K.; Okada, S.; Moriya, H.; Yamazaki, M. Granulocyte colony-stimulating factor (G-CSF) mobilizes bone marrow-derived cells into injured spinal cord and promotes functional recovery after compression-induced spinal cord injury in mice. Brain Res. 2007, 1149,

157

223–231.

274. Lee, J.S.; Yang, C.C.; Kuo, Y.M.; Sze, C.I.; Hsu, J.Y.C.; Huang, Y.H.; Tzeng, S.F.; Tsai, C.L.; Chen, H.H.; Jou, I.M. Delayed granulocyte colony-stimulating factor treatment promotes functional recovery in rats with severe contusive spinal cord injury. Spine (Phila. Pa. 1976). 2012.

275. Yoon, S.H.; Shim, Y.S.; Park, Y.H.; Chung, J.K.; Nam, J.H.; Kim, M.O.; Park, H.C.; Park, S.R.; Min, B.-H.; Kim, E.Y.; et al. Complete Spinal Cord Injury Treatment Using Autologous Bone Marrow Cell Transplantation and Bone Marrow Stimulation with Granulocyte Macrophage-Colony Stimulating Factor: Phase I/II Clinical Trial. Stem Cells 2007, 25, 2066–2073.

276. Poniatowski, Ł.A.; Wojdasiewicz, P.; Krawczyk, M.; Szukiewicz, D.; Gasik, R.; Kubaszewski, Ł.; Kurkowska-Jastrzębska, I. Analysis of the Role of CX3CL1 (Fractalkine) and Its Receptor CX3CR1 in Traumatic Brain and Spinal Cord Injury: Insight into Recent Advances in Actions of Neurochemokine Agents. Mol. Neurobiol. 2017.

277. Donnelly, D.J.; Longbrake, E.E.; Shawler, T.M.; Kigerl, K.A.; Lai, W.; Amy Tovar, C.; Ransohoff, R.M.; Popovich, P.G. Deficient CX3CR1 signaling promotes recovery after mouse spinal cord injury by limiting the recruitment andactivation of Ly6Clo/iNOS+ macrophages. J. Neurosci. 2011.

278. Freria, C.M.; Hall, J.C.E.; Wei, P.; Guan, Z.; McTigue, D.M.; Popovich, P.G. Deletion of the fractalkine receptor, CX3CR1, improves endogenous repair, axon sprouting, and synaptogenesis after spinal cord injury in mice. J. Neurosci. 2017.

279. Pruss, H.; Tedeschi, A.; Thiriot, A.; Lynch, L.; Loughhead, S.M.; Stutte, S.; Mazo, I.B.;

158

Kopp, M.A.; Brommer, B.; Blex, C.; et al. Spinal cord injury-induced immunodeficiency is mediated by a sympathetic-neuroendocrine adrenal reflex. Nat Neurosci 2017, 20, 1549–1559.

280. Schwab, J.M.; Zhang, Y.; Kopp, M.A.; Brommer, B.; Popovich, P.G. The paradox of chronic neuroinflammation, systemic immune suppression, autoimmunity after traumatic chronic spinal cord injury. Exp Neurol 2014, 258, 121–129.

281. Lee, Y.L.; Shih, K.; Bao, P.; Ghirnikar, R.S.; Eng, L.F. Cytokine chemokine expression in contused rat spinal cord. In Proceedings of the Neurochemistry International; 2000.

282. Kopp, M.A.; Druschel, C.; Meisel, C.; Liebscher, T.; Prilipp, E.; Watzlawick, R.; Cinelli, P.; Niedeggen, A.; Schaser, K.D.; Wanner, G.A.; et al. The SCIentinel study--prospective multicenter study to define the spinal cord injury-induced immune depression syndrome (SCI-IDS)--study protocol and interim feasibility data. BMC Neurol 2013, 13, 168.

283. Noble, B.T.; Brennan, F.H.; Popovich, P.G. The spleen as a neuroimmune interface after spinal cord injury. J Neuroimmunol 2018, 321, 1–11.

284. Lucin, K.M.; Sanders, V.M.; Jones, T.B.; Malarkey, W.B.; Popovich, P.G. Impaired antibody synthesis after spinal cord injury is level dependent and is due to sympathetic nervous system dysregulation. Exp Neurol 2007, 207, 75–84.

285. Lucin, K.M.; Sanders, V.M.; Popovich, P.G. Stress hormones collaborate to induce lymphocyte apoptosis after high level spinal cord injury. J Neurochem 2009, 110, 1409– 1421.

286. Ahuja, C.S.; Wilson, J.R.; Nori, S.; Kotter, M.R.N.; Druschel, C.; Curt, A.; Fehlings, M.G. Traumatic spinal cord injury. Nat Rev Dis Prim. 2017, 3, 17018.

159

287. Hong, J.; Chang, A.; Zavvarian, M.M.; Wang, J.; Liu, Y.; Fehlings, M.G. Level-Specific Differences in Systemic Expression of Pro- and Anti-Inflammatory Cytokines and Chemokines after Spinal Cord Injury. Int J Mol Sci 2018, 19.

288. Zha, J.; Smith, A.; Andreansky, S.; Bracchi-Ricard, V.; Bethea, J.R. Chronic thoracic spinal cord injury impairs CD8+ T-cell function by up-regulating programmed cell death- 1 expression. J Neuroinflammation 2014, 11, 65.

289. Norden, D.M.; Bethea, J.R.; Jiang, J. Impaired CD8 T cell antiviral immunity following acute spinal cord injury. J Neuroinflammation 2018, 15, 149.

290. Jahnsen, F.L.; Haye, R.; Gran, E.; Brandtzaeg, P.; Johansen, F.E. Glucocorticosteroids inhibit mRNA expression for eotaxin, eotaxin-2, and monocyte-chemotactic protein-4 in human airway inflammation with eosinophilia. J Immunol 1999, 163, 1545–1551.

291. Dufour, J.H.; Dziejman, M.; Liu, M.T.; Leung, J.H.; Lane, T.E.; Luster, A.D. IFN- gamma-inducible protein 10 (IP-10; CXCL10)-deficient mice reveal a role for IP-10 in effector T cell generation and trafficking. J Immunol 2002, 168, 3195–3204.

292. Latifi, S.; Koushki, D.; Norouzi Javidan, A.; Matin, M.; Sabour, H. Changes of leptin concentration in plasma in patients with spinal cord injury: a meta-analysis. Spinal Cord 2013, 51, 728–731.

293. Ramirez, O.; Garza, K.M. Leptin deficiency in vivo enhances the ability of splenic dendritic cells to activate T cells. Int Immunol 2014, 26, 627–636.

294. Kristo, C.; Godang, K.; Ueland, T.; Lien, E.; Aukrust, P.; Froland, S.S.; Bollerslev, J. Raised serum levels of interleukin-8 and interleukin-18 in relation to bone metabolism in endogenous Cushing’s syndrome. Eur J Endocrinol 2002, 146, 389–395.

160

295. Takahashi, H.K.; Iwagaki, H.; Mori, S.; Yoshino, T.; Tanaka, N.; Nishibori, M. Beta 2- adrenergic receptor agonist induces IL-18 production without IL-12 production. J Neuroimmunol 2004, 151, 137–147.

296. Seagard, J.L.; Hopp, F.A.; Bosnjak, Z.J.; Osborn, J.L.; Kampine, J.P. Sympathetic efferent nerve activity in conscious and isoflurane-anesthetized dogs. Anesthesiology 1984, 61, 266–270.

297. Soderblom, C.; Luo, X.; Blumenthal, E.; Bray, E.; Lyapichev, K.; Ramos, J.; Krishnan, V.; Lai-Hsu, C.; Park, K.K.; Tsoulfas, P.; et al. Perivascular fibroblasts form the fibrotic scar after contusive spinal cord injury. J. Neurosci. 2013.

298. Li, J.; Lepski, G. Cell Transplantation for Spinal Cord Injury: A Systematic Review. Biomed Res. Int. 2013, 2013, 1–32.

299. Wilcox, J.T.; Satkunendrarajah, K.; Zuccato, J.A.; Nassiri, F.; Fehlings, M.G. Neural precursor cell transplantation enhances functional recovery and reduces astrogliosis in bilateral compressive/contusive cervical spinal cord injury. Stem Cells Transl. Med. 2014, 3, 1148–1159.

300. Cummings, B.J.; Uchida, N.; Tamaki, S.J. Human neural stem cell differentiation following transplantation into spinal cord injured mice: association with recovery of locomotor function. Neurol. Res. 2006, 28, 474–481.

301. Karimi-Abdolrezaee, S.; Schut, D.; Wang, J.; Fehlings, M.G. Chondroitinase and Growth Factors Enhance Activation and Oligodendrocyte Differentiation of Endogenous Neural Precursor Cells after Spinal Cord Injury. PLoS One 2012, 7, e37589.

302. Karimi-Abdolrezaee, S. Delayed Transplantation of Adult Neural Precursor Cells

161

Promotes Remyelination and Functional Neurological Recovery after Spinal Cord Injury. J. Neurosci. 2006, 26, 3377–3389.

303. Ulndreaj, A.; Badner, A.; Fehlings, M.G. Promising neuroprotective strategies for traumatic spinal cord injury with a focus on the differential effects among anatomical levels of injury. F1000Res 2017, 6, 1907.

304. Thuret, S.; Moon, L.D.F.; Gage, F.H. Therapeutic interventions after spinal cord injury. Nat. Rev. Neurosci. 2006.

305. Center, N.S.C.I.S. Spinal Cord Injury Facts and Figures at a Glance; 2018;

306. Tetzlaff, W.; Okon, E.B.; Karimi-Abdolrezaee, S.; Hill, C.E.; Sparling, J.S.; Plemel, J.R.; Plunet, W.T.; Tsai, E.C.; Baptiste, D.; Smithson, L.J.; et al. A Systematic Review of Cellular Transplantation Therapies for Spinal Cord Injury. J. Neurotrauma 2011, 28, 1611–1682.

307. Lemon, R.N.; Griffiths, J. Comparing the function of the corticospinal system in different species: Organizational differences for motor specialization? Muscle and Nerve 2005.

308. Terashima, T. Anatomy, development and lesion-induced plasticity of rodent corticospinal tract. Neurosci. Res. 1995.

309. Welniarz, Q.; Dusart, I.; Roze, E. The corticospinal tract: Evolution, development, and human disorders. Dev. Neurobiol. 2017.

310. Mestas, J.; Hughes, C.C.W. Of Mice and Not Men: Differences between Mouse and Human Immunology. J. Immunol. 2004.

311. Mestas, J.; Hughes, C.C.W. Mouse and Human Immunology Of Mice and Not Men:

162

Differences between Of Mice and Not Men: Differences between Mouse and Human Immunology. J Immunol Ref. J. Immunol. Mirand Libr. Roswell Park Cancer Inst 2015.

312. Schneemann, M.; Schoedon, G. Species differences in macrophage NO production are important (multiple letters). Nat. Immunol. 2002.

313. Schaffner, A.; Schneemann, M.; Schoedon, G.; Hofer, S.; Blau, N.; Guerrero, L. Nitric Oxide Synthase Is Not a Constituent of the Antimicrobial Armature of Human Mononuclear Phagocytes. J. Infect. Dis. 1993.

314. Carpenter, R.S.; Kigerl, K.A.; Marbourg, J.M.; Gaudet, A.D.; Huey, D.; Niewiesk, S.; Popovich, P.G. Traumatic spinal cord injury in mice with human immune systems. Exp. Neurol. 2015.

315. Akkina, R. Human immune responses and potential for vaccine assessment in humanized mice. Curr. Opin. Immunol. 2013.

316. Forgione, N.; Karadimas, S.K.; Foltz, W.D.; Satkunendrarajah, K.; Lip, A.; Fehlings, M.G. Bilateral contusion-compression model of incomplete traumatic cervical spinal cord injury. 2014, 31, 1776–1788.

317. Wilcox, J.T.; Fehlings, M.G. Acute clip compression model of SCI. In Neuromethods; 2016.

318. Fehlings, M.G. The relationships among the severity of spinal cord injury, residual neurological function, axon counts, and counts of retrogradely labeled neurons after experimental spinal cord injury. Exp. Neurol. 132, 220–228.

319. Fehlings, M.G.; Tator, C.H. The relationships among the severity of spinal cord injury, residual neurological function, axon counts, and counts of retrogradely labeled neurons

163

after experimental spinal cord injury. Exp. Neurol. 1995.

320. Sipski, M.L.; Jackson, A.B.; Gómez-Marín, O.; Estores, I.; Stein, A. Effects of gender on neurologic and functional recovery after spinal cord injury. Arch. Phys. Med. Rehabil. 2004.

321. Farooque, M.; Suo, Z.; Arnold, P.M.; Wulser, M.J.; Chou, C.T.; Vancura, R.W.; Fowler, S.; Festoff, B.W. Gender-related differences in recovery of locomotor function after spinal cord injury in mice. Spinal Cord 2006.

322. Chan, W.M.; Mohammed, Y.; Lee, I.; Pearse, D.D. Effect of Gender on Recovery After Spinal Cord Injury. Transl. Stroke Res. 2013.

323. Furlan, J.C.; Krassioukov, A. V.; Fehlings, M.G. The effects of gender on clinical and neurological outcomes after acute cervical spinal cord injury. J. Neurotrauma 2005.

324. Greenwald, B.D.; Seel, R.T.; Cifu, D.X.; Shah, A.N. Gender-related differences in acute rehabilitation lengths of stay, charges, and functional outcomes for a matched sample with spinal cord injury: A multicenter investigation. Arch. Phys. Med. Rehabil. 2001.

325. Ignotz, R.A.; Endo, T.; Massagué, J. Regulation of fibronectin and type I collagen mRNA levels by transforming growth factor-beta. J. Biol. Chem. 1987.

326. Varga, J.; Rosenbloom, J.; Jimenez, S.A. Transforming growth factor β (TGFβ) causes a persistent increase in steady-state amounts of type I and type III collagen and fibronectin mRNAs in normal human dermal fibroblasts. Biochem. J. 1987.

327. Paralkar, V.M.; Vukicevic, S.; Reddi, A.H. Transforming growth factor β type 1 binds to collagen IV of basement membrane matrix: Implications for development. Dev. Biol. 1991.

164

328. Cheng, Y.; Xi, G.; Jin, H.; Keep, R.F.; Feng, J.; Hua, Y. Thrombin-Induced Cerebral Hemorrhage: Role of Protease-Activated Receptor-1. Transl. Stroke Res. 2014.

329. Cofano, F.; Boido, M.; Monticelli, M.; Zenga, F.; Ducati, A.; Vercelli, A.; Garbossa, D. Mesenchymal Stem Cells for Spinal Cord Injury: Current Options, Limitations, and Future of Cell Therapy. Int J Mol Sci 2019, 20.

330. Spejo, A.B.; Chiarotto, G.B.; Ferreira, A.D.F.; Gomes, D.A.; Ferreira, R.S.; Barraviera, B.; Oliveira, A.L.R. Neuroprotection and immunomodulation following intraspinal axotomy of motoneurons by treatment with adult mesenchymal stem cells. J. Neuroinflammation 2018, 15, 230.

331. Nakajima, H.; Uchida, K.; Guerrero, A.R.; Watanabe, S.; Sugita, D.; Takeura, N.; Yoshida, A.; Long, G.; Wright, K.T.; Johnson, W.E.B.; et al. Transplantation of mesenchymal stem cells promotes an alternative pathway of macrophage activation and functional recovery after spinal cord injury. J. Neurotrauma 2012, 29, 1614–1625.

332. Badner, A.; Vawda, R.; Laliberte, A.; Hong, J.; Mikhail, M.; Jose, A.; Dragas, R.; Fehlings, M. Early Intravenous Delivery of Human Brain Stromal Cells Modulates Systemic Inflammation and Leads to Vasoprotection in Traumatic Spinal Cord Injury. Stem Cells Transl. Med. 2016, 5, 991–1003.

333. Watanabe, J.; Shetty, A.K.; Hattiangady, B.; Kim, D.K.; Foraker, J.E.; Nishida, H.; Prockop, D.J. Administration of TSG-6 improves memory after traumatic brain injury in mice. Neurobiol. Dis. 2013, 59, 86–99.

334. Li, R.; Liu, W.; Yin, J.; Chen, Y.; Guo, S.; Fan, H.; Li, X.; Zhang, X.; He, X.; Duan, C. TSG-6 attenuates inflammation-induced brain injury via modulation of microglial polarization in SAH rats through the SOCS3/STAT3 pathway. J. Neuroinflammation

165

2018, 15.

335. Badner, A.; Hacker, J.; Hong, J.; Mikhail, M.; Vawda, R.; Fehlings, M.G. Splenic involvement in umbilical cord matrix-derived mesenchymal stromal cell-mediated effects following traumatic spinal cord injury. J. Neuroinflammation 2018, 15.

336. Vawda, R.; Badner, A.; Hong, J.; Mikhail, M.; Lakhani, A.; Dragas, R.; Xhima, K.; Barretto, T.; Librach, C.L.; Fehlings, M.G. Early Intravenous Infusion of Mesenchymal Stromal Cells Exerts a Tissue Source Age-Dependent Beneficial Effect on Neurovascular Integrity and Neurobehavioral Recovery After Traumatic Cervical Spinal Cord Injury. Stem Cells Transl. Med. 2019, 8, 639–649.

337. El-Kheir, W.A.; Gabr, H.; Awad, M.R.; Ghannam, O.; Barakat, Y.; Farghali, H.A.M.A.; El Maadawi, Z.M.; Ewes, I.; Sabaawy, H.E. Autologous bone marrow-derived cell therapy combined with physical therapy induces functional improvement in chronic spinal cord injury patients. Cell Transplant. 2014, 23, 729–745.

338. Pastore, D.; Pacifici, F.; Dave, K.R.; Palmirotta, R.; Bellia, A.; Pasquantonio, G.; Guadagni, F.; Donadel, G.; Di Daniele, N.; Abete, P.; et al. Age-Dependent Levels of Protein Kinase Cs in Brain: Reduction of Endogenous Mechanisms of Neuroprotection. Int. J. Mol. Sci. 2019, 20, 3544.

339. Lanz, T. V; Becker, S.; Osswald, M.; Bittner, S.; Schuhmann, M.K.; Opitz, C.A.; Gaikwad, S.; Wiestler, B.; Litzenburger, U.M.; Sahm, F.; et al. Protein kinase Cβ as a therapeutic target stabilizing blood-brain barrier disruption in experimental autoimmune encephalomyelitis. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 14735–14740.

340. Aiello, L.P.; Bursell, S.E.; Clermont, A.; Duh, E.; Ishii, H.; Takagi, C.; Mori, F.; Ciulla, T.A.; Ways, K.; Jirousek, M.; et al. Vascular endothelial growth factor-induced retinal

166

permeability is mediated by protein kinase C in vivo and suppressed by an orally effective β-isoform-selective inhibitor. Diabetes 1997, 46, 1473–1480.

341. Habib, A.; Karmali, V.; Polavarapu, R.; Akahori, H.; Cheng, Q.; Pachura, K.; Kolodgie, F.D.; Finn, A. V. Sirolimus-FKBP12.6 impairs endothelial barrier function through protein kinase C-α activation and disruption of the p120-vascular endothelial cadherin interaction. Arterioscler. Thromb. Vasc. Biol. 2013, 33, 2425–2431.

342. Qi, X.; Inagaki, K.; Sobel, R.A.; Mochly-Rosen, D. Sustained pharmacological inhibition of δPKC protects against hypertensive encephalopathy through prevention of blood-brain barrier breakdown in rats. J. Clin. Invest. 2008, 118, 173–182.

343. Tang, Y.; Soroush, F.; Sun, S.; Liverani, E.; Langston, J.C.; Yang, Q.; Kilpatrick, L.E.; Kiani, M.F. Protein kinase C-delta inhibition protects blood-brain barrier from sepsis- induced vascular damage. J. Neuroinflammation 2018, 15.

344. Lucke-Wold, B.P.; Turner, R.C.; Logsdon, A.F.; Simpkins, J.W.; Alkon, D.L.; Smith, K.E.; Chen, Y.W.; Tan, Z.; Huber, J.D.; Rosen, C.L. Common mechanisms of Alzheimer’s disease and ischemic stroke: The role of protein kinase C in the progression of age-related neurodegeneration. J. Alzheimer’s Dis. 2015, 43, 711–724.

345. Sivasankaran, R.; Pei, J.; Wang, K.C.; Zhang, Y.P.; Shields, C.B.; Xu, X.M.; He, Z. PKC mediates inhibitory effects of myelin and chondroitin sulfate proteoglycans on axonal regeneration. Nat. Neurosci. 2004, 7, 261–268.

346. Yang, Q.; Zhang, X.F.; Van Goor, D.; Dunn, A.P.; Hyland, C.; Medeiros, N.; Forscher, P. Protein kinase C activation decreases peripheral actin network density and increases central nonmuscle myosin II contractility in neuronal growth cones. Mol. Biol. Cell 2013, 24, 3097–3114.

167

347. Yang, P.; Li, Z.Q.; Song, L.; Yin, Y.Q. Protein kinase C regulates neurite outgrowth in spinal cord neurons. Neurosci. Bull. 2010, 26, 117–125.

348. Wang, X.; Hu, J.; She, Y.; Smith, G.M.; Xu, X.M. Cortical PKC inhibition promotes axonal regeneration of the corticospinal tract and forelimb functional recovery after cervical dorsal spinal hemisection in adult rats. Cereb. Cortex 2014, 24, 3069–3079.

349. Stone, R.M.; Manley, P.W.; Larson, R.A.; Capdeville, R. Midostaurin: its odyssey from discovery to approval for treating acute myeloid leukemia and advanced systemic mastocytosis. Blood Adv. 2018, 2, 444–453.

350. Ouaret, D.; Larsen, A.K. Protein kinase C β inhibition by enzastaurin leads to mitotic missegregation and preferential cytotoxicity toward colorectal cancer cells with chromosomal instability (CIN). Cell Cycle 2014, 13, 2697–2706.

351. Lesyk, G.; Fong, T.; Ruvolo, P.P.; Jurasz, P. The potential of enzastaurin to enhance platelet aggregation and growth factor secretion: implications for cancer cell survival. J. Thromb. Haemost. 2015, 13, 1514–1520.

352. Stranahan, A.M.; Hao, S.; Dey, A.; Yu, X.; Baban, B. Blood-brain barrier breakdown promotes macrophage infiltration and cognitive impairment in leptin receptor-deficient mice. J. Cereb. Blood Flow Metab. 2016, 36, 2108–2121.

353. Carducci, M.A.; Musib, L.; Kies, M.S.; Pili, R.; Truong, M.; Brahmer, J.R.; Cole, P.; Sullivan, R.; Riddle, J.; Schmidt, J.; et al. Phase I dose escalation and pharmacokinetic study of enzastaurin, an oral protein kinase C beta inhibitor, in patients with advanced cancer. J. Clin. Oncol. 2006, 24, 4092–4099.

354. Morschhauser, F.; Seymour, J.F.; Kluin-Nelemans, H.C.; Grigg, A.; Wolf, M.;

168

Pfreundschuh, M.; Tilly, H.; Raemaekers, J.; van’t Veer, M.B.; Milpied, N.; et al. A phase II study of enzastaurin, a protein kinase C beta inhibitor, in patients with relapsed or refractory mantle cell lymphoma. Ann. Oncol. 2008, 19, 247–253.

355. Willeman, M.N.; Mennenga, S.E.; Siniard, A.L.; Corneveaux, J.J.; De Both, M.; Hewitt, L.T.; Tsang, C.W.S.; Caselli, J.; Braden, B.B.; Bimonte-Nelson, H.A.; et al. The PKC-β selective inhibitor, Enzastaurin, impairs memory in middle-aged rats. PLoS One 2018, 13.

356. Chio, J.C.T.; Wang, J.; Badner, A.; Hong, J.; Surendran, V.; Fehlings, M.G. The effects of human immunoglobulin G on enhancing tissue protection and neurobehavioral recovery after traumatic cervical spinal cord injury are mediated through the neurovascular unit. J. Neuroinflammation 2019, 16.

357. Brennan, F.H.; Kurniawan, N.D.; Vukovic, J.; Bartlett, P.F.; Käsermann, F.; Arumugam, T. V; Basta, M.; Ruitenberg, M.J. IVIg attenuates complement and improves spinal cord injury outcomes in mice. Ann. Clin. Transl. Neurol. 2016, 3, 495–511.

358. Nguyen, D.H.; Cho, N.; Satkunendrarajah, K.; Austin, J.W.; Wang, J.; Fehlings, M.G. Immunoglobulin G (IgG) attenuates neuroinflammation and improves neurobehavioral recovery after cervical spinal cord injury. J. Neuroinflammation 2012, 9.

359. Yu, Q.; Huang, J.; Hu, J.; Zhu, H. Advance in spinal cord ischemia reperfusion injury: Blood–spinal cord barrier and remote ischemic preconditioning. Life Sci. 2016, 154, 34– 38.

360. Dong, H.L.; Zhang, Y.; Su, B.X.; Zhu, Z.H.; Gu, Q.H.; Sang, H.F.; Xiong, L. Limb remote ischemic preconditioning protects the spinal cord from ischemia-reperfusion injury: A newly identified nonneuronal but reactive oxygen species-dependent pathway. Anesthesiology 2010, 112, 881–891.

169

361. Haapanen, H.; Herajärvi, J.; Arvola, O.; Anttila, T.; Starck, T.; Kallio, M.; Anttila, V.; Tuominen, H.; Kiviluoma, K.; Juvonen, T. Remote ischemic preconditioning protects the spinal cord against ischemic insult: An experimental study in a porcine model. J. Thorac. Cardiovasc. Surg. 2016, 151, 777–785.

362. Herajärvi, J.; Anttila, T.; Sarja, H.; Mustonen, C.; Haapanen, H.; Mäkelä, T.; Yannopoulos, F.; Starck, T.; Kallio, M.; Tuominen, H.; et al. Exploring Spinal Cord Protection by Remote Ischemic Preconditioning: An Experimental Study. Ann. Thorac. Surg. 2017, 103, 804–811.

363. Fukui, T.; Ishida, K.; Mizukami, Y.; Shiramoto, K.; Harada, H.; Yamashita, A.; Yamashita, S.; Matsumoto, M. Comparison of the protective effects of direct ischemic preconditioning and remote ischemic preconditioning in a rabbit model of transient spinal cord ischemia. J. Anesth. 2018, 32, 3–14.

364. Jing, N.; Fang, B.; Wang, Z.L.; Ma, H. Remote Ischemia Preconditioning Attenuates Blood-Spinal Cord Barrier Breakdown in Rats Undergoing Spinal Cord Ischemia Reperfusion Injury: Associated with Activation and Upregulation of CB1 and CB2 Receptors. Cell. Physiol. Biochem. 2017, 43, 2516–2524.

365. Ueno, K.; Samura, M.; Nakamura, T.; Tanaka, Y.; Takeuchi, Y.; Kawamura, D.; Takahashi, M.; Hosoyama, T.; Morikage, N.; Hamano, K. Increased plasma VEGF levels following ischemic preconditioning are associated with downregulation of miRNA-762 and miR-3072-5p. Sci. Rep. 2016, 6.

366. Jin, K.L.; Mao, X.O.; Greenberg, D.A. Vascular endothelial growth factor: Direct neuroprotective effect in in vitro ischemia. Proc. Natl. Acad. Sci. 2000, 97, 10242–10247.

367. Dezfulian, C.; Garrett, M.; Gonzalez, N.R. Clinical Application of Preconditioning and

170

Postconditioning to Achieve Neuroprotection. Transl. Stroke Res. 2013, 4, 19–24.

171

172