Commentary on Some Recent Theses Relevant to Combating Aging: October 2019

Cancer and Stem Cell Extravasation through Angiopellosis

Tyler Allen, PhD, North Carolina State University

Cardiovascular disease, including myocardial infarction (MI), remains one of the most common causes of death in western society. Stem cell therapies have shown immense potential for the treatment of MI. However, there are currently limited methods for introducing stem cells to damaged heart tissue, with one promising method involves infusing stem cells directly into cardiac vessels allowing the cells to extravasate/transmigrate across the blood vessel wall into the target tissue for repair. Although such transmigration does occur, an understanding of this process remains elusive. Here, we characterize the method of extravasation these cells utilize to exit blood vessels.

First, we used a transgenic zebrafish embryo model and intravital microscopy to elucidate how stem cells exit vessels when injected into the circulation. The stem cells, once injected, underwent a distinct method of extravasation that was markedly different from native cells found in the blood stream, such as leukocytes. In this unreported mechanism, the vascular wall undergoes an extensive remodeling to allow the cell to exit the lumen, while the cell itself remains distinctively passive in activity. We termed this new cell extravasation process as Angio-pello-sis (Angio = related to blood vessels; pello = expel). Both individual cells as well as multicellular clusters were able to utilize this process to extravasate, pointing to a potential means for circulating tumor clusters to extravasate.

Next, we focused on the leading cause of cancer-related deaths worldwide—metastasis. A key step in the metastasis process involves circulating tumor cells exiting the circulation and forming secondary tumors. The exact process tumor cells use to extravasate vessels, remains poorly understood, and it is currently unknown if tumor cells possess the ability to utilize angiopellosis to extravasate. Here, we discovered tumor cells possess the ability to utilize the angiopellosis method to exit vessels and form tumors.

Accumulating data suggests metastatic primary tumor cells disseminate through the circulation as multicellular clusters. Although tumor clusters correlate to increased secondary tumor formation, the mechanism clusters use to exit circulation remains largely unknown. We hypothesized that tumor cell clusters were able to exit blood vessels without first dissociating. We discovered circulating tumor cells possess the ability to exit blood vessels while maintaining a multicellular phenotype, through angiopellosis, challenging the hypothesis that circulating tumor cell clusters must first disassociate to exit blood vessels.

Additionally, we found that cervical (HeLa) and melanoma (B16F10) tumor cells which extravasate as multicellular clusters through angiopellosis exhibit an augmented ability to proliferate while individually-extravasating cells remain dormant. Furthermore, we established that melanoma tumor cell clusters tail-vein injected into murine circulation, induced increased metastatic lung foci, suggesting this method of extravasation and proliferation behavior is conserved in mammals. The results suggest angiopellosis as the prevailing method of extravasation used by both infused stem cells and metastatic circulating tumor cells, and the results of these studies have implications for both stem cell and cancer metastasis therapies.

Comment: Although some tumors can pose a life-threatening risk even while confined to their site of origin, the majority of the death toll associated with cancer is due to metastasis—the escape of malignant cells to other parts of the body—since this complication effectively rules out the use of the arsenal of ablative techniques (surgery, radiotherapy, etc.), which can already be employed to destroy localized cell masses. Inhibiting metastasis has therefore been a major goal of oncological research for several decades; however, attempts to do so by blocking steps of the classical extravasation process (diapedesis) have led to confusing and disappointing results, with negligible reduction of metastatic growth even when diapedesis is robustly impaired. This failure remained unexplained until the identification, in the course of this exceptional doctoral project, of angiopellosis—an entirely novel mechanism of vascular escape utilized by stem cells, and co-opted by transformed cells (most likely as a side effect of the acquisition of other stem-like properties early in their development; it is somewhat difficult to see how the phenotype could otherwise be so consistently selected for within the dense and poorly vascularized tumor microenvironment). Like diapedesis, angiopellosis is triggered by interactions between the migrating cell's outer membrane and the endothelial wall—indeed, rather remarkably, inert polymer microspheres can undergo angiopellosis if coated with membrane derived from cardiac stem cells. However, the process is otherwise quite different, involving engulfment and translocation by the endothelium instead of adhesion and invasion by the extravasating cell, supporting the transport of intact cell clusters as well as individual cells, and taking place over a much longer time course (hours rather than minutes). Crucially, while a proportion of cancerous cells do also exit the vasculature by diapedesis, such emigrants appear to have an extremely limited capacity for survival and proliferation. Thus, blocking angiopellosis—a prospect that would seem to require only the modulation of endothelial cell behavior, rather than the infamously challenging task of targeting cancerous cells themselves—may be clinically equivalent to blocking metastasis entirely! Further work is needed to verify these findings in a broader range of cancers, to identify the specific molecular signals involved in inducing angiopellosis, and to ascertain the side effects of inhibiting angiopellosis on healthy physiology; yet even at this early stage it seems certain that this discovery will have profound implications.

Human Chimeric Antigen Receptor Macrophages for Cancer Immunotherapy

Michael Klichinsky, PhD, University of Pennsylvania

Despite recent landmark advances in chimeric antigen receptor (CAR) T cell immunotherapy for the treatment of human cancer, metastatic solid tumors remain an intractable challenge. Myeloid cells are actively recruited to the tumor microenvironment (TME), where tumor associated macrophages (TAMs) are often the most abundant infiltrating immune cell. Currently, macrophage orientated immunotherapeutic approaches under clinical development in oncology seek to reduce TAM infiltration or enhance TAM phagocytosis. We hypothesized that genetically engineering human macrophages with CARs against tumor-associated antigens could redirect their phagocytic activity and lead to therapeutic efficacy with the potential for the induction of an anti-tumor T cell response.

In this thesis, we demonstrate that CD3-zeta based CARs are capable of inducing phagocytosis by human macrophages. Notably, an active intracellular CAR signaling domain was required for activity. Targeted phagocytosis and clearance of CD19+, mesothelin+, and HER2+ cells by CARs targeted against each respective antigen was significantly superior to that by control untransduced (UTD) macrophages. Importantly, CAR macrophages were capable of polyphagocytosis and serial phagocytosis of tumor cells.

We demonstrate that primary human monocyte derived macrophages, which are resistant to most viral vectors, are efficiently transduced by the chimeric-fiber adenoviral vector Ad5f35. Ad5f35 transduced primary human CAR macrophages demonstrated targeted phagocytosis, with phagocytic activity dependent on both the CAR and antigen densities. CAR, but not UTD, macrophages led to potent dose-dependent killing of tumor cells in vitro and led to tumor regression and improved overall survival in murine xenograft models of human cancer.

Macrophage transduction with Ad5f35 leads to a broad gene expression change, an interferon signaling signature, and induction of a classically activated M1 phenotype. CAR macrophages upregulated co-stimulatory ligand and antigen processing/presentation genes and led to enhanced T cell stimulation in vitro and in vivo. Lastly, CAR, but not UTD, macrophages showed a broad resistance for M2 conversion in response to immunosuppressive cytokines.

In conclusion, human CAR macrophages display targeted tumor phagocytosis, lead to improved overall survival in xenograft models, and demonstrate enhanced T cell stimulation. Taken together, these data show that CAR macrophages are a novel cell therapy platform for the treatment of human cancer.

Comment: The remarkable effectiveness of CAR-T cells against hematological cancers has led to a correspondingly remarkable improvement in remission rates and survival time for many such malignancies. Unfortunately, attempts to employ them against solid tumors have been much less successful; the density and poor vascularization of such masses inhibit T cell entry, while the hostile microenvironment potently suppresses the effector functions of those cells that do infiltrate. Although efforts to overcome those barriers through further refinement of the CAR framework are ongoing, there is also growing interest in combination immunotherapies that integrate elements of the innate immune system to induce a more comprehensive antitumor response. Many such projects have focused on the modulation of tumor-associated macrophages, either by depleting such cells or by driving their “repolarization” from the protumoral M2 phenotype to the antitumoral M1 phenotype, thereby reducing obstacles to the function of coadministered CAR-T cells. This dissertation, however, explores a much closer integration of the two arms of the immune system: engineering CARs directly into macrophages! Such expression is not entirely without precedent—the existence of small subpopulations of TCR⁺ cells within the innate immune system has been known since the mid-2000s, overturning a previous dogma that such expression was exclusive to T cells per se—yet has been little studied. The benefits conferred by employing CARs in this context are demonstrated herein to be well beyond what could have been anticipated—and are of especially great interest from a translational perspective, since the clinical-scale production of macrophages ex vivo is markedly easier than that of T cells.

Insights on Alzheimer's Disease Etiology from Network Approaches in Healthy Aging

Katelyn Arnemann, PhD, University of California, Berkeley

The etiology of Alzheimer's disease involves the presymptomatic development and progression of amyloid-β and tau in healthy aging. Amyloid-β and tau are naturally occurring proteins that can form abnormal aggregates—amyloid-β plaques and neurofibrillary tangles—which constitute the pathological hallmarks of Alzheimer's disease. The initial formation of these aggregates occurs decades before the onset of cognitive symptoms, in individuals otherwise considered to be healthy and unimpaired. This dissertation hinges on in-vivo PET imaging of amyloid-β and tau in humans using PIB-PET and AV1451-PET to explore this presymptomatic phase of Alzheimer's disease—when pathology is present without detectable symptoms. I place particular emphasis on amyloid-β pathology—understanding the factors that underlie vulnerability to amyloid-β as well as identifying the initial sources and progressive spread of amyloid-β pathology in healthy aging. My focus on amyloid-β is consistent with the predominant framework for Alzheimer's disease, the amyloid cascade hypothesis, which contends that amyloid-β initiates a slow and ultimately deadly chain of events that results, decades later, in deteriorating memory and breakdown of cognition. In recognition that Alzheimer's disease does not reflect a focal disorder, but rather network failure of large-scale brain systems, I adapt a network-based framework to account for the role of the complex interdependencies between distributed brain regions—of glucose metabolism from FDG-PET, of brain activity from resting-state functional MRI, and of amyloid-β from PIB-PET. Examining metabolic brain networks, I reveal widespread, highly systematic reorganization of glucose metabolism in old age—well beyond what has been revealed using other methods—that is more heterogeneous in those possessing both substantial amyloid-β and genetic risk for Alzheimer's disease. Further, I demonstrate that the topology of early-life “metabolic inefficiency”—a novel metric that removes the potential association of glucose metabolism with highly connected hubs—explains the topology of amyloid-β in healthy aging. Finally, I provide evidence that very early amyloid-β accumulation, in those without substantial amyloid-β pathology, is multifocal and broadly distributed across brain networks—consistent with shared tissue vulnerability, not transneuronal spread, being the driving force of accumulation of amyloid-β pathology. These findings support the notion that shared tissue vulnerability of a metabolic origin drives widespread, systematic accumulation of amyloid-β in healthy aging. Future work should uncover the nature and origin of metabolic tissue vulnerability to amyloid-β, exploring the complex chain of events that drive widespread age-related reorganization—especially of cerebral glucose metabolism—and its links other age-related changes and the onset of pathological accumulation of amyloid-β.

Comment: Alzheimer's disease is the most common form of dementia, affecting tens of millions of people worldwide and resulting in millions of deaths per year. Despite extensive research, contemporary therapies are still unable to do more than modestly slow the rate of clinical decline. This dire situation is now widely suspected to be the consequence of a very prolonged presymptomatic phase, during which molecular damage to the brain gradually accumulates but does not reach a threshold sufficient to detectably disrupt normal mental function. By the time cognitive symptoms manifest—decades later—a significant fraction of that damage has already become irreversible (especially due to the loss of neurons), limiting the potential efficacy of medical interventions. Earlier treatment is increasingly held to be necessary to meaningfully improve patient outcomes, yet has proven challenging precisely because the preclinical phase of the disease is so difficult to detect in cross-sectional analysis, especially in the presence of the confounding effects of “healthy aging” (i.e., chronic degenerative processes not yet accorded the status of a recognized disease, such as the accumulation of senescent cells, ¹¹ or changes to the systemic milieu secondary to damage elsewhere in the body ¹² ). Meanwhile, longitudinal studies face the daunting prospect of decades of follow-up before progression to Alzheimer's disease can be confirmed. One metric that distinguishes Alzheimer's disease in particular is a unique spatial pattern of reduced metabolism (hypometabolism) that associates strongly with the pattern of amyloid-β deposition, and which crucially appears to precede both plaque formation and tissue atrophy. This thesis investigates the relationship between hypometabolism and amyloid-β in considerably greater detail than has previously been achieved, utilizing mathematical tools from graph theory to characterize the extent to which metabolic rates correlate between regions of the brain, and observes marked and distinctive alterations of those correlations during both aging and Alzheimer's. Such findings have the potential to be developed into a diagnostic test for presymptomatic disease—which will be essential for the clinical evaluation of therapies designed to target that phase of pathology—and may also lead directly to an understanding of the origins of the metabolic dysregulation that increasingly appears to be culpable for the earliest stages of neurodegeneration.

Myelin and Glial Pathology in Aging and Cognitive Decline: Evidence for Faulty Myelin Clearance in the Rhesus Monkey

Eli Townsend-Shobin, PhD, Boston University

Aging is associated with a loss of cognitive function related to learning, memory, and executive function with varying severity. Although there is no age-related loss of neurons in healthy aging, myelin damage accumulates and is associated with cognitive decline. The brain's resident macrophages, microglia, are responsible for clearing damaged myelin and promoting subsequent oligodendrocyte-mediated remyelination. To test the hypothesis that age-related dysfunction of microglial phagocytosis and oligodendrocyte remyelination capacity contributes to myelin pathology and cognitive impairment. To test this, rhesus monkeys from across the lifespan (7–30 years of age) were tested in three specific aims. 1) To characterize gene expression of myelin basic protein (MBP) in the brain and clearance of MBP to the cerebrospinal fluid (CSF) in relation to age-related myelin pathology. The density of myelinated axons visualized using label-free spectral confocal reflectance imaging did not correlate with age, but was significantly lower in aged animals with cognitive impairment. Next, MBP gene expression was measured using qPCR in the dorsal prefrontal cortex along with quantification of MBP protein levels in the CSF using ELISA. Age-dependent increases of MBP gene expression in the brain and MBP protein levels in the CSF were observed. Interestingly, MBP levels in the CSF were lower in animals with cognitive impairment. 2) To test the hypothesis that microglia would become increasingly primed for phagocytosis with age-related myelin pathology. The number of microglia immunostained with galectin-3, a marker for phagocytic activation, was quantified in the frontal white matter and increases in both aging and cognitive decline were detected. 3) To evaluate the hypothesis that lipofuscin, an age-related accumulation indicative of autophagic dysfunction, would accumulate and impair glial cells of the white matter in aged animals. Lipofuscin accumulation was increased with age in the frontal white matter and the size of lipofuscin clusters was associated with cognitive impairment. Lipofuscin was found primarily in microglia and oligodendrocytes, but not in astrocytes. These data suggest that lipofuscin burden in microglia and oligodendrocytes inhibits their homeostatic functions resulting in improper myelin clearance and turnover, leading to a devastating feed-forward cycle of myelin damage that contributes to age-related cognitive impairment.

Comment: The function of the vertebrate central nervous system (CNS) depends critically on myelin, an insulating sheath formed from as many as a hundred consecutive layers of lipid-rich membrane bound tightly by members of the myelin basic protein (MBP) family. This coating both protects the delicate axon from damage and dramatically increases the speed and reliability of transmission of electrochemical action potentials. Myelin sheaths are produced by oligodendrocytes in a process that largely takes place during infancy and adolescence, reaching a peak in early adulthood after which total myelin coverage declines gradually with increasing age (and more rapidly in association with neurodegenerative disease). Although oligodendrocyte dysfunction is clearly a partial contributor to this age-related demyelination, some capacity for remyelination exists even in very elderly brains; furthermore, that capacity can be substantially enhanced by stimulating the expansion of oligodendrocyte progenitor cells resident in the subventricular zone, the progeny of which spontaneously migrates to myelin-depleted lesions and attempts repair. Importantly, however, oligodendrocytes can only ensheath naked axons; the elaborate cell–cell communication that underpins the wrapping process depends initially on direct membrane contact, and in particular is entirely blocked by pre-existing myelin. Thus, damage that does not result in complete loss of the sheath cannot generally be repaired until the former coating has been removed by the action of microglia. Even under ideal conditions, that clearance faces a nontrivial challenge; myelin is highly enriched in cholesterol (volumetrically; the fraction of cholesterol present in the bilayer is only modestly elevated compared with other cell membranes, but of course most cells are not encased in a hundred layers of membrane) and there is good evidence that this alone can overwhelm the efflux capacities of microglia, resulting in cholesterol crystal formation and phagolysosome rupture. In middle-aged mice, stimulation of reverse cholesterol transport is actually sufficient to restore remyelination; we strongly suspect, however, that the situation in very elderly nervous tissue will prove less favorable. The high energetic demand of axons, coupled with their limited antioxidant capacity, leaves the sheath intrinsically vulnerable to the accumulation of oxidative modifications. (Consistent with this, myelin is remarkably enriched in a class of antioxidant phospholipids called plasmalogens, depletion of which significantly accelerates the rate of loss; and it is well established that a variety of antioxidants demonstrate neuroprotective effects, at least in vitro. ^{13 –17} ) A tiny fraction of those modifications create atypical oxidized species—especially cholesterol derivatives—which are resistant to further metabolism, and which disproportionately impair the function of the phagocytic cells that consume them. This process is thought to be a critical driver of atherosclerosis and seems very likely to also be involved in the microglial lipofuscin accumulation reported in this thesis. Thankfully, work is already under way to develop agents to eliminate oxysterols from the vasculature, providing a foundation for similar therapy in the CNS.

Phosphatidylserine-Based Nanoparticles for Tolerance Induction Towards Therapeutic Proteins

Fiona Glassman, PhD, State University of New York at Buffalo

Phosphatidylserine (PS) is the most abundant negatively charged phospholipid in the cell membrane. Despite comprising 12% of total phospholipid and being restricted primarily in the inner leaflet of the membrane bilayer, the consequences of PS externalization to the outer leaflet is biologically significant, especially in hemostasis and apoptosis. The exposure of PS on the surface of apoptotic cells functions as an “eat me” signal that initiates macrophage recognition, uptake, and removal of the apoptotic debris. The apoptotic process is an immunologically silent event, as these apoptotic cells are cleared without activation of the immune system. There is considerable literature evidence that support the fact that PS is a critical mediator in apoptosis, facilitating the efficient removal of cell debris in order to maintain tolerance to self-proteins by immunosuppressive mechanisms and tissue homeostasis.

The key function of PS exposure in apoptosis could be exploited for therapeutic implications, such as mitigation of immunogenicity towards therapeutic proteins. The focus of our laboratory is the development of tolerogenic liposomal formulations to reduce immunogenicity and induce immunological hypo-responsiveness towards therapeutic proteins. Factor VIII (FVIII) served as the model protein, as the dysfunction or deficiency of FVIII manifests in Hemophilia A (HA). Despite enzyme replacement therapy as the first line of therapy, about 30% of severe HA patients develop neutralizing (NAbs) or inhibitory titers, severely complicating efficacy of therapy. As a result, any approach designed to reduce immunogenicity and induce tolerance would address an unmet clinical need. Previous studies have shown that subcutaneous pre-treatment of FVIII in the presence of PS liposomes induced hypo-responsiveness after subsequent rechallenge to free FVIII in HA mice. The PS-mediated mechanism involves induction of tolerogenic dendritic cells, secretion of regulatory cytokine TGF-β, generation of regulatory T cells, inhibition of memory B cell development, and reduction of antibody production, resulting in the property of PS to convert an immunogen to a tolerogen. The overall goal of this dissertation is to extend the application of our PS tolerance strategy into multiple sclerosis, a model disease for autoimmunity, as well as to develop and optimize this strategy to enhance the tolerogenic effects of PS tolerance induction.

Chapter 1 provides an overview of the biological functions of PS. The involvement of PS in hemostasis and apoptosis are discussed along with the contribution of receptors and structural modifications that may influence the biological outcome of PS exposure. We then highlight how the understanding of PS mechanism can be exploited for therapeutic applications, such as in viral infection, cancer, autoimmunity, and mitigation of immunogenicity. Further, we highlight the property of PS conversion of an immunogen to a tolerogen through our previous work and how that property could be exploited to induce immunological tolerance towards a protein.

In Chapter 2, we investigated whether PS liposomes can utilized with multiple antigens other than FVIII. We found that pre-exposure of acid alpha glucosidase (GAA) in the presence of PS can induce a durable and long-lasting hypo-responsiveness in Pompe Disease mice. In addition, we demonstrated that the PS receptor, T-cell immunoglobulin and mucin 4 (TIM-4) receptor, is involved in PS-mediated hypo-responsiveness. Finally, we showed that pre-exposure of myelin oligodendrocyte glycoprotein (MOG) peptide, a known auto-antigen for multiple sclerosis (MS), in the presence of PS can alter the disease onset and disease severity of experimental autoimmune encephalomyelitis (EAE), a murine model for human MS. We further determined that an increase in CD4⁺FoxP3⁺ regulatory T cell expression was exhibited in mice that were pre-treated with PS MOG_35–55, suggesting the involvement of regulatory T cells in PS-mediated effects in EAE.

After demonstrating that PS can be utilized to induce tolerance towards therapeutic proteins and can be extended to re-tolerize the immune system to a self-protein in autoimmune conditions, we sought to optimize the PS particle in Chapter 3 in order to improve the translatability of this strategy into the clinic and enhance its tolerogenic properties. The identification of a specific structural PS species that could be responsible for the PS ability to convert an immunogen to a tolerogen is essential in order to develop a homogeneous formulation for successful translation.

Comment: A very general obstacle to the practical effectiveness of many emerging biological therapeutics is the propensity of the immune system to (accurately, if inconveniently for physicians) recognize a novel molecular structure as foreign and act to neutralize and remove it. Such reactions are particularly likely, rapid, and potent in cases of repeat exposure to the antigen in question and thus are most problematic for therapies that are intended to be reapplied periodically or which permanently introduce a new protein or cell population to the body—including many of the proposed damage/repair tools of rejuvenation biotechnology. ¹⁸ The capacity to (re)induce immunological tolerance to a given substance has of course been sought for many decades as a cure for autoimmune disease, with an increasing number of options available to modulate the function of relevant immune cells or to mimic an immune-privileged site. Still, few such techniques have progressed beyond the laboratory, and there remains an urgent need for a widely applicable method of tolerogenesis suitable for clinical translation. One of the more obvious candidates for such an agent is phosphatidylserine (PS), with its well-established role in preventing immunological reactions during the routine clearance of dying cells; the fact that PS is co-opted by an astonishing number of major pathogens to evade immune responses (and is almost universally dysregulated in cancers for the same reason) can only further support that candidacy. This dissertation provides very promising proofs-of-principle for the use of PS nanoparticles to induce lasting tolerance in multiple preclinical model systems, while also laying explicit groundwork for their clinical translation. We are somewhat hopeful that the said translation will be more expeditious than is typical, since phosphatidylserine is so ubiquitous as to be “Generally Recognized as Safe” as a food additive.

Role of Endothelial Cell Stiffening in Choroidal Atrophy Associated with Dry Age-Related Macular Degeneration

Andrea Cabrera, PhD, University of California, Riverside

Age-related macular degeneration (AMD) is the leading cause of blindness in the aging population. Recent studies have implicated choriocapillaris (CC) dropout and choroidal thinning in AMD pathogenesis, potentially mediated by hypoxia-induced retinal pigment epithelium dysfunction. However, the precise mechanism underlying choroidal atrophy remains unknown. The goal of this research was to address this vital gap in our mechanistic understanding of choroidal atrophy associated with dry AMD. Since complement activation, a major risk factor for dry AMD, leads to the deposition of membrane attack complex (MAC; C5b-9n) on choroidal vessels, it may lead to choroidal endothelial cell (EC) atrophy and the observed loss of CC in AMD eyes. Interestingly, MAC deposition also occurs in young non-AMD eyes, thus indicating that specific age-related factors may contribute to MAC-induced choroidal degeneration in AMD eyes.

Since aging, a major nonmodifiable risk factor for AMD, has been shown to increase stiffness of retinal vessels and enhance pro-inflammatory cues in non-opthalmic vessels such as aorta and arteries, the central hypothesis of this research was that aging increases CC stiffening that, in turn, enhances EC susceptibility to MAC injury.

Findings from the current research revealed for the first time that aging leads to choroidal EC stiffening that in turn, contributes significantly to the increased susceptibility to MAC injury. Further, these studies showed that the stiffening of aged choroidal ECs is cytoskeletal-mediated. Remarkably, reducing age-induced EC stiffness prevented MAC injury. Taken together, these novel findings not only elucidate a key role of EC stiffness but also identify potentially new targets (e.g., Rho-mediated cytoskeletal stiffening) for more effective therapies in the future.

Comment: The choroid is a highly vascularized connective tissue located between the outermost layer of the eye—the protective sclera—and the innermost layer, the retina. Its primary function is to provide oxygen and nutrients to, and drain waste products from, the outer and middle layers of the retina (comprising the photoreceptor cells and intermediate neurons), from which it is separated by Bruch's membrane. The vast majority of the inner layer of the retina (comprising the ganglion cells that integrate visual information and feed into the optic nerve) is supplied instead by branches of the central retinal artery traversing the inner surface of the retina. Notably, however, this supply is diminished across the macular region and is entirely absent at the fovea, within which all layers of the retina are dependent on the choroidal circulation. This most likely reflects the intense trade-off between the requirement for vascular supply and the requirement to maximize light absorption and thus visual acuity—needless to say, blood vessels overlaying the photoreceptors are not conducive to the optimal transmission of light. Despite that anatomical observation, the choroid was historically thought to play a crucial role only in the rarer (∼10%–20% of cases) “wet” form of AMD, where abnormal growth of capillaries into the macula causes a relatively sudden loss of central vision. Meanwhile, “dry” AMD was considered largely an intrinsic pathology of the retina associated with the deposition of drusen—a zinc-enriched aggregate of proteins and lipids—on the retinal side of Bruch's membrane, accompanied by abnormalities of the overlying retinal pigmented epithelium (RPE). Drusen's composition and location suggest that it originates from the incomplete degradation of photoreceptor outer segments, a task primarily performed by the RPE, yet it correlates with AMD symptomatology only when present in high concentrations (and not at all at the levels found in a majority of middle-aged and older adults). Particularly damning, laser therapy has been employed within the last few years to destroy drusen with apparently robust success—yet this treatment has been shown to have no impact on the risk of further macular degeneration! With drusen thus largely excluded from consideration as the instigator of pathology, and work already well under way to address other intrinsic toxicity in the RPE, ¹⁹ changes to the choroidal vasculature more subtle than those observed in wet AMD have come under scrutiny as potential unaddressed drivers of the dry form. The vasculature in general becomes stiffer with age, and this thesis demonstrates that such stiffness contributes directly to the susceptibility of choroidal capillaries to attack by the complement system (also providing a novel explanation as to why polymorphisms in complement proteins are among the strongest genetic risk factors for AMD, independent of the established yet now apparently incidental presence of such proteins in drusen deposits). Intriguingly, while modification of the extracellular matrix is the most well-known contributor to vascular stiffening and its sequelae, ^{20 –22} the phenomenon described here appears to be mediated instead by cellular biomechanics. From a therapeutic perspective, this finding may be a blessing; such cytoskeletal stiffness is strongly associated with cellular senescence, and of course senolytics is among the fastest growing areas of contemporary medicine.