We met with hundreds of neuroscientists to find out. What we learned is that there is a clear path to safer AI through insights from the brain, but it requires an unconventional bet: we must actively accelerate neuroscience. This is the core of defensive accelerationism, a strategy to reduce systemic risks by selectively accelerating beneficial innovations to outpace harmful ones. Our position is that neuroAI is among the most important investments for AI safety, and that we can extract these insights fast enough to matter, even on relatively short AGI timelines.

Closing the Loop: Aligning Neuroscience with AI Timelines

A common objection to a brain-inspired path to AI safety is that neuroscience moves too slowly to keep pace with rapid AI development. However, our work in frontier biology has taught us that you can dramatically increase the speed of discovery by building well-specified virtual models that allow for high-throughput testing of hypotheses. In the context of neuroAI, there are effectively two brain modeling strategies: emulation and distillation. Emulation aims to copy the brain’s precise biological structure, while distillation extracts its underlying rules and algorithms. These two approaches have fundamentally different development trajectories.

The Emulation Path: High Reward, Long Horizon

The emulation path is a mainstay of classic AI safety literature (e.g. Sandberg & Bostrom 2008, Bostrom 2014). The basic idea is straightforward: we scan the human brain with high-resolution microscopes to obtain a map of the brain – a connectome. Then, we emulate this scan in silico to create a biophysical simulation of a brain. This would allow us to run infinite experiments to understand human intelligence and our prosocial instincts. It would also allow us to spin up huge numbers of virtual researchers to solve aging, ASI alignment, and beyond.

We evaluated this idea rigorously. Alongside the Foresight Institute, we sponsored a workshop at Oxford, convening dozens of neuroscientists and tool developers. We spoke with leading AI safety researchers, and many of them view brain emulation, if realized, as the most powerful way to reduce risk.

We found a field experiencing a dramatic drop in data acquisition costs. In 2023, it was estimated that it would cost $15B for a single mouse connectome, but we’ve already seen rapid progress. Disruptive technologies, like E11 Bio’s protein labeling and Google Connectomics’s computer vision advances, could dramatically decrease this cost to below $1B over the next decade. This would be a remarkable achievement and accelerate neuroscience broadly. We’re glad to support E11’s work on this.

Yet, getting the connectome is only half the battle. The human brain is the most complex system humanity has ever attempted to simulate, and even a perfect structural map would not suffice for faithful emulation. We would still lack critical information about synaptic strengths, kinetics, and neuromodulatory states that actually drive the system’s behavior. Acquiring this calibration data will likely take over a decade. Barring a Manhattan-Project-style mobilization, the emulation path is unlikely to yield results in time for short AGI timelines.

The Distillation Path: Fast, Functional, and Attainable

The other path to leveraging the brain for AGI safety is distillation. To keep human societies stable and collaborative, specific aspects of cognition, decision-making, and computation are essential. The challenge is identifying them.

As researchers like Ilya Sutskever and Doris Tsao have warned, we should not merely emulate the brain, but rather extract the correct features from it. That is, distill its distributed representations and underlying algorithms from functional brain models.

There is much to learn from the brain’s representational structure: it is incredibly robust, generalizing and learning effectively with minimal data exposure. It demonstrates how sparse coding manifests as compressed, highly interpretable representations. Studying this could lead to new architectures that are more interpretable, which is one of the most important goals of AI safety.

Instead of simulating every neuron and synapse, we can train neural networks to replicate recorded neural responses, building highly accurate functional models of the brain (Tolias Lab, 2025). We can then dissect these systems in silico (Ganguli Lab, 2023) to deconstruct how the brain’s representations are structured and how they change. These functional simulations provide a sandbox for virtual experiments at scale, allowing us to test and generate hypotheses in silico before verifying them back in vivo. Extending this approach to circuits underlying emotion and social behavior could reveal the precise mechanics of how the brain encodes empathy, cooperation, and moral reasoning.

This is why we funded the Enigma Project at Stanford University and Metamorphic: an integrated, frontier research effort that co-designs experiments and modeling. Through Enigma, we built conviction that the data required to scale functional modeling is highly attainable on short timelines, based on two key insights:

First, we don’t need to record every neuron! Sparse sampling theory (the JL lemma) proves that the geometry of neural representations is preserved even when subsampled. Because the brain is highly interconnected, meaningful low-dimensional structures can be recovered from surprisingly sparse recordings.

Second, the hardware is ready to scale. Neuropixels provide a well-laid technological path to dramatically increase representation fidelity. Our funding for Neuropixels will deliver significantly improved density of the previous generations and stands on the shoulders of a decade of previous work at IMEC and funding from their supporters.

Building a Deep Bench for AI Safety

Finally, why not just fund conventional AI safety research? While we strongly support traditional approaches, we believe unconventional research offers uniquely high leverage at this critical juncture.

Historically, safety research has mirrored the prevailing AI paradigms of the day. In 2014, the field implicitly assumed Reinforcement Learning was the sole path to AGI. Today, safety efforts are heavily indexed on LLMs and AI agents built on top of them. However, even if LLMs cross the threshold to AGI, they will not be the final stop in the evolutionary lineage of AI systems. Detailed AI projection scenarios (e.g. AI 2027) assume that the first AGI systems will be used to create the next generation of more efficient AGI, likely based on entirely different principles and architectures. The first general intelligence will neither be the last, nor the best.

To secure these new AI systems with unforeseen capabilities, the brain offers foundational safety principles that can guide their architectural design. In any conceivable scenario, the human brain remains the ultimate alignment target. Put simply: you cannot align a system to human values without fundamentally understanding how those values are implemented.

Subscribe now

Join the Frontier

The window to align AGI is closing, but the tools to map and understand the only working model of safe general intelligence are scaling rapidly. We are actively deploying capital and building teams to accelerate this future.

• Metamorphic is developing large-scale foundation models trained on rich, continuous neural data — high-resolution brain models at a scale never before possible. They are looking for research engineers and scientists to join their growing AI research team.

• E11 Bio is recruiting at multiple levels, including for an experienced operations lead – if you’re excited about connectomics as a high-leverage path to AI safety and possess exceptional talent, integrity, and innate motivation, check out their open positions.

• Constellation is looking for passionate operations, ML engineering, and post-training experts to develop their foundation model of human state.

Many thanks to Sophia Sanborn, Andreas Tolias, Adam Marblestone, Andrew Payne, and Grant Hummer for feedback and review

Over the past several years, our team has consulted with hundreds of leading scientists to identify the most transformative ideas in their fields. Often, these high-impact concepts fall outside the scope of R01 federal grants and therefore remain unrealized. We are dedicated to sponsoring and accelerating these overlooked breakthroughs, and we are incredibly proud of the pioneering teams we have backed so far. By sharing our work, we hope to inspire others to join us in tackling these civilizational challenges.

We take an unconventional approach that blends philanthropy with venture capital. By operating at the earliest stages of development, we act as a bridge across the “valley of death” that separates academic research from commercial venture funding. To date, we have deployed nearly $100M in philanthropic grants and $250M+ in early-stage investments through the Amaranth Foundation and Starbloom Capital.

The brain as the crux for AGI safety and longevity

Our top priorities are ensuring a safe transition to Artificial General Intelligence (AGI) and extending the human lifespan so as many people as possible can participate in the next phase of civilization.

We’ve converged on neuroscience as a primary focus because the brain is both the only relatively safe general intelligence that we know of and the most vital organ for longevity. As AI becomes more capable, we believe it is essential to study the brain to understand:

• Prosocial Behavior: The neural mechanisms behind traits like empathy, maternal instincts, and moral reasoning to build more robust moral agents

• Biological Network Architecture: Biological neural networks exhibit modularity, hierarchy, and sparse coding that may guide the design of AI systems that are more amenable to interpretation and control

No matter how AGI emerges, understanding human intelligence remains essential. It is both the alignment target and the only known example of a general intelligence that is broadly cooperative.

We care about longevity because aging is the top cause of death and suffering in the world. 50 million deaths per year are from age-related disease. By 2029, the U.S. will spend $3T+ annually on adults 65 and older. Research into the underlying biology of aging remains drastically underfunded relative to the scale of the problem.

Our existing portfolio

I believe the endeavors we are funding today will become some of the most important platforms of tomorrow. We will go into more detail on the areas we care about in subsequent posts, but here’s some of what we’ve funded:

• Bexorg, a next-generation platform dramatically improving brain drug development

• Cognito Therapeutics, a promising progression-halting therapy for Alzheimer’s Disease and potentially Parkinson’s

• The Enigma Project → Metamorphic, a Stanford project to build comprehensive models of brain structure and neural firing that led to a company scaling these models for better, safer AI

• Neuropixels, advancing the leading neural probe for large-scale recordings of individual neurons

• Becoming, engineering replacement tissues — and eventually organs — at scale

• Loyal, bringing the first FDA-approved longevity drug to market — starting with dogs

• Magic Lifescience, a lab-on-chip platform to cheaply and accurately detect pathogens and biomarkers in any biofluid

• Swift Solar, low-cost perovskite solar cells that are 40%+ more efficient than silicon cells

• Etherealize, building the settlement rails for Wall Street on Ethereum

• Forest Neurotech → Merge, a non-profit de-risking ultrasound neuromodulation technology that led to a company developing an ultrasound BCI

• E11 Bio, technology to map brain wiring at 100x lower cost

Building roadmaps to the future

If you want a bird’s eye view of the fields we care about, we’ve openly written comprehensive field maps to help you make sense of them. These maps guide our own allocation decisions and have been widely acclaimed and referenced in their fields:

• Bottlenecks of Aging (2022): Our advisory board helped identify 12 key constraints slowing progress in extending healthy lifespan, from regulatory reform to brain aging research and talent development

• Brain Aging White Paper (2023): We gathered some of the top minds in neuroscience to outline their top moonshot ideas for combating brain aging

• NeuroAI for AI Safety Roadmap (2024): A 152-page technical document identifying 8 paths where neuroscience can inform safer AI development, synthesizing over 700 references and coordinating input from researchers across multiple institutions

• Silver Linings (2025): An interactive economic model quantifying the ROI of longevity breakthroughs, allowing users to simulate impacts on GDP and aging trajectories

We’re currently concentrating our efforts on neuroAI and AI safety, decoding intelligence to ensure its safe and aligned development. If you’re building in this direction, we’d like to hear from you.

Email us at: info@amaranth.foundation

1 – Neuroscience ↔ AGI: Open‑Ended Relevance

How can your area of neuroscience meaningfully inform, constrain, or inspire the design and governance of future AGI systems? We welcome perspectives spanning cellular and circuit mechanisms, cognitive and social neuroscience, computational modelling, neuro‑inspired architectures, brain‑data‑driven training regimes, and ethical or societal insights grounded in the study of natural intelligence. You may wish to reflect on themes surfaced in the “NeuroAI for AI Safety” roadmap (Mineault et al., 2024), which outlines eight ways brain science could mitigate key AI‑safety failure modes—from robustness and out‑of‑distribution generalisation to the cultivation of pro‑social agency.

2 — Short‑Timeline Scenario (~2025‑2028)

Now, assume that frontier AI labs achieve highly capable, largely autonomous AI R&D systems by late 2027 (Koktajlo et al. 2025, Aschenbrenner 2024), and that this rapidly leads to AGI. In that world:

What near‑term research and development opportunities exist for the neuroscience community—e.g., data generation, tool development, fundamental theory, interdisciplinary partnerships, etc.—to most effectively shape the safety, alignment, and societal integration of frontier AI systems?

Concrete proposals might touch on rapid‑cycle brain‑data benchmarks for agentic models, causal intervention toolkits inspired by modern neurotech, biologically grounded metrics of value formation, or multi-agent governance frameworks informed by comparative cognition. Feel free to challenge these examples.

Practicalities

• Response format: ≤ 2 pages (PDF or equivalent), free‑form prose

• Deadline: 23:59 ET, 29 Aug 2025

• Submission: neuroaisafety@amaranth.foundation

• Review: Internal; selected authors may be invited to a virtual salon or to submit full proposals. No material will be shared externally without permission.

For clarifications, email neuroaisafety@amaranth.foundation with the subject “Neuro × AGI RFI Query”.

Thank you for helping map how neuroscience can guide the safe evolution of artificial general intelligence.

A lot of things can go wrong with cells in the brain as we age, and the same is true for the environment that these cells exist in. Extracellular space accounts for a whopping 20% of the volume in the adult brain [1] and this space is filled with the extracellular matrix (ECM), a rich meshwork of proteins, sugars, and other macromolecules [2]. These molecules surround, influence, and are in turn shaped by the cells of the nervous system in highly dynamic and highly underappreciated ways. 

Despite the considerable space it takes up, the brain ECM is often ignored. Researchers instead focus on the cells in the brain as if they existed in a vacuum. For the scientists out there, this is perhaps best illustrated with a thought experiment. Imagine you want to study your favorite cells in a dish, whether they are neurons or glia. How often do you consider culturing them on a halfway decent extracellular matrix, let alone any at all? 

Ironically, “glia” is derived from the Greek word for glue, as such cells were mistakenly thought to serve as a connective structure in the brain [3]. This term would perhaps more aptly describe the ECM, although it does far more than simply provide structural support. In fact, the ECM interfaces with multiple pathways critical to brain aging. As a result, any attempts to intervene in this process will have to contend with the matrix.

At the time of writing, there is only ~$90M in active funding combined from the two primary government sponsors of brain and aging research for projects containing the key words “extracellular matrix” in the National Institutes of Health’s RePORTER database. A crude measure to be sure, but comparing this to the ~$643M in “microglia” projects by these agencies reveals a pretty apparent lack of interest in the matrix relative to glial cells, let alone neurons.

Using the framework provided by age1 to classify aging interventions, I’ll illustrate how the ECM will be critical to three of the major approaches as they apply to the brain. These strategies include delaying brain aging and replacing or restoring aged cells and tissues. Then I’ll get into what we can do about it.

In some cases, the relevance of the ECM is intuitive - if you were going to replace your old cells with shiny new ones, would you want to put them in the same old, junk-filled environment? 

In other cases, a deeper understanding of the brain ECM will be necessary to fully grasp its importance.

Delaying aging: The matrix in neuroplasticity

The brain extracellular matrix can be divided into three primary compartments:

1. The diffuse ECM, located in the interstitial space and around synapses

2. Perineuronal nets, dense ECM coats that form around certain neurons

3. The basement membrane of the blood-brain barrier

Each of these compartments has distinct structural and functional characteristics [4].

The diffuse ECM is a loose network of proteins and sugars that surrounds cells and synapses throughout the brain, where in the latter case it is referred to as the perisynaptic matrix. It’s distinct in its molecular makeup from ECM found in other parts of the body, reflecting its specialized role in brain tissue. 

In contrast, perineuronal nets (PNNs) are more structurally compact and surround only certain neuronal subtypes. Although PNNs share many of the same matrix components as the diffuse ECM, these components differ in their concentration and organization. 

The basement membrane, with its own unique repertoire of matrix molecules, lines the blood vessels of the brain and supports the integrity of the blood-brain barrier.

With this in mind, there are two major changes to the brain ECM that occur with age and associated pathologies that can be targeted to delay brain aging, and both of these directly impact synaptic plasticity:

1. A build-up of diffuse matrix material around synapses

2. A breakdown of perineuronal nets

Diffuse ECM builds up

Brain plasticity, a frequent target of therapeutic efforts, is critically regulated by the matrix. Matrix material builds up in the brain interstitial space with age [5], in the so-called diffuse ECM, where it accumulates around synapses and restricts synaptic plasticity [6]. Similar diffuse ECM builds up in neurodegeneration, as in Alzheimer’s [7] and Huntington’s disease [8], where deficits in synaptic plasticity are also rampant.

The maintenance of healthy synapses is critical to cognitive health in aging [9], and it may be possible to attenuate age-related synaptic deficits by reducing or preventing the accumulation of ECM around synapses. This can be done by directing the brain’s resident immune cells, microglia, to clear perisynaptic ECM via cytokine delivery, as has been shown with IL-33 [6]. Alternatively, enzymes could be delivered to break down matrix material directly [10]. Both approaches enhance synaptic plasticity.

Perineuronal nets break down

At the same time that the diffuse interstitial matrix is building up with age, the ECM is breaking down in other parts of the brain. In contrast to the diffuse ECM, extracellular proteins and polysaccharides can also condense around specific subsets of neurons in a lattice-like structure known as the perineuronal net, which serves to stabilize and support synaptic architecture [11].  

The formation of PNNs is a critical event that brings about the closure of heightened periods of plasticity in the developing brain. By stabilizing synapses in certain neurons, PNNs fundamentally control and restrict synaptic plasticity. In line with this, their removal in the adult brain is able to reinstate plasticity to an extent seen in early brain development [12, 13].

In my graduate work with Dr. Kim Green at UC Irvine, we found that PNNs are disrupted during the course of normal aging as well as in Alzheimer’s disease [7], and that this process is also regulated by microglia [14]. In Alzheimer’s disease patients, the loss of nets correlates with amyloid plaque load, and ECM components are found deposited in plaques [7]. 

But the loss of PNNs would mean more synaptic plasticity, which is a good thing – right? It’s complicated. Particularly when it comes to a hallmark deficit of aging: memory. 

Many subtypes of memory - from spatial to social to fear memories - are critically regulated by PNNs [15]. Take episodic memory, through which we remember our personal experiences as we go about our daily lives. The ability to form precise memories of events develops after birth, with young children only able to form imprecise, fuzzy memories until this system comes online at ~5-8 years of age [16].

In mice, this process is controlled by the maturation of hippocampal PNNs, and adult episodic-like memory precision can be produced by bringing about PNN formation in the brains of younger mice [16]. Memory imprecision characteristic of the immature brain occurs in adults if PNNs are removed.

The Nobel Prize-winning chemist Roger Tsien proposed that PNNs may serve as the physical basis for long-term memory storage [17]. While this is a pretty grandiose claim, it is clear that PNNs and memory are closely intertwined, albeit often in ways that remain somewhat opaque. 

The increased plasticity required to encode new memories has to be balanced with sufficient synaptic stability to retain the old ones [11]. As such, PNN removal may enhance plasticity at the cost of long-term memory stability, due to interference with new memories [18]. So by disrupting PNNs, we might be sacrificing the fidelity of our memories for this boost in plasticity.

For more information on PNNs in aging and AD, and their role in plasticity/memory, see:

• A review on the role of microglia in ECM and PNN regulation by yours truly, which more thoroughly covers the literature on ECM changes in aging and neurodegeneration

• A primer on PNNs in memory by Fawcett et al., a leading PNN expert

• A review by Reichelt et al. on PNNs in plasticity and the “stability-plasticity tradeoff”

A dual approach: Compartment-specific matrix regulation for brain health

In order to delay brain aging, efforts may be best aimed at reducing the build-up of perisynaptic ECM while simultaneously preventing the breakdown or loss of perineuronal nets. With this approach, we could promote general synaptic plasticity and integrity while still maintaining the stability of long-term memory storage.

As a bonus, PNNs are inherently protective against age-related pathologies, protecting the neurons they enwrap from toxicity associated with amyloid beta [19], oxidative stress [20], and tau [21]. They also keep neurotransmitters from spilling over outside of their synapses [22]. It’s not a stretch to imagine then that PNN maintenance could provide protection for the aging brain across multiple dimensions.

Replacing the aging brain: The matrix in cell transplantation approaches

Replacing cells

Beyond plasticity, the brain ECM actively regulates functions of local brain cells including survival, proliferation, differentiation and migration [4]. Approaches to replace aging or diseased cells in the brain via transplantation of healthy cells will have to factor in the ECM, which accumulates its own age-related damage. 

Aged ECM is stiffer, for instance, at least in the rat brain (things are less clear in humans - see the Appendix). This mechanical change impairs the ability of oligodendrocyte precursor cells to proliferate and differentiate. In line with this, precursor cells from neonatal rats lose their proliferative capabilities when seeded on aged ECM [23]. 

Putting new cells into the brain without considering the extracellular environment in which they exist would be like trying to put a human on Mars without planning for the unique atmosphere, and yet transplantation efforts seldom consider it.

Brain cell replacement therapies should therefore develop and incorporate auxiliary strategies to make the aged ECM more hospitable. One potential approach is to deliver enzymes that temporarily break down specific ECM components alongside transplanted cells. This method could temporarily modify tissue stiffness and reduce synaptic ECM accumulation, facilitating integration of donor cells and promoting therapeutic efficacy. 

Alternatively, transplanted cells could be engineered to express their own ECM-degrading enzymes, or a unique repertoire of integrins, which are receptors by which cells bind to and interact with the matrix. These could be fine-tuned to maximize integration and function of transplanted cells, much like the spacesuit we would give our astronaut to ensure survival on another planet.

Replacing ECM?

It would be interesting to consider whether replacement of the ECM itself could serve as a viable therapeutic strategy to combat brain aging. Could old, beaten-up ECM that likely carries similar age-related damage as reported for the matrix in other tissues [24] be broken down and replaced by the delivery of young ECM? Certainly for the diffuse matrix this seems possible. 

In PNNs, on the other hand, matrix components are tightly packed together by what are known as link proteins. Overexpression of these link proteins can induce PNN formation where PNNs are lacking [16]. One possibility might be to deliver young ECM around neurons with disrupted PNNs and then promote link protein expression, with the hope that the donor ECM would be incorporated into new nets. 

Given the importance of the ECM in cellular health, plasticity, and synaptic function, these replacement strategies could be undertaken as primary therapies or as adjuncts to promote the efficacy of cell transplantations.

Restoring a youthful brain: The matrix in neurogenesis

Strategies to counter brain aging may also look to exploit natural mechanisms that promote neuronal rejuvenation and that already exist in nervous tissue, as in neurogenesis. Here, too, we have to contend with the matrix. 

Regions of the adult brain where new neurons are formed - neurogenic niches - have their own specialized ECM that is stiffer than non-neurogenic regions [25]. Further age-related alterations in niche stiffness might directly impact neurogenesis. In fact, neural stem cells are more adhesive and have difficulty migrating out of certain neurogenic niches with age [26], a deficit that may be impacted by local changes in ECM stiffness.

By targeting the ECM, we may be able to create environments that better support the brain’s own repair mechanisms, potentially slowing or even reversing age-related decline.

Interestingly, the enzymatic removal of ECM in neurogenic niches can impair neurogenesis [27]. This might be related to the ability of the ECM to serve as a reservoir for important cellular growth factors. Increasing [28] or decreasing [29] specific ECM components can also boost or inhibit neurogenesis, respectively. A key challenge in developing therapies will be to selectively target the detrimental age-related accumulation of ECM without impairing the essential pro-neurogenic functions of the healthy, homeostatic matrix - and ideally, even enhancing them. 

There’s far more to talk about on the role of the ECM in neurogenesis than we have room for here, and work is still being done. But it’s clear that there’s a lot we can do to influence these endogenous restorative pathways in aging by tinkering with the matrix.

For more information:

• A review on the ECM-neurogenesis interface by Elise Cope and Elizabeth Gould

• A discussion on the matrix in the neurogenic niche in a review by Dityatev et al.

Okay, the ECM is important in aging. So what can we do about it?

We can:

1. Look into the matrix

2. Manipulate the matrix

Looking into the matrix

In a proteomic study of blood plasma from over 40,000 individuals, higher levels of specific brain-derived ECM proteins were associated with a more youthful brain age. Even more striking, perineuronal nets and synapse-associated ECM were the top two brain protein pathways associated with human longevity [30]. 

By characterizing the brain ECM and how it changes with age and age-related disease, we can determine the best way to factor it into brain aging interventions. 

There remain gaps in our understanding of the precise compositional differences between the ECM compartments of the brain, as well as how these differences are regulated. Proteomic screens to define the ‘matrisome’ [31] have been and will continue to be critical, with a strong emphasis on human data. While transcriptomic analysis is important, the secretory and long-lived nature of ECM components to some extent uncouples them from transcriptional dynamics.

One particular challenge is that the different brain ECM compartments - PNNs, diffuse ECM, and basement membranes - overlap in the proteins and sugars they contain. Changes at these smaller scales may be missed with proteomic analysis of bulk tissue. Enrichment approaches have been developed to address this issue, as in the isolation of cerebrovascular- [32] and PNN-associated [33] matrices, and should be applied to studies on the aging brain.

To further complicate things, ECM molecules are heavily post-translationally modified, and these modifications impart critical functional properties [34]. This often occurs through the attachment of sugar side chains that, in turn, have their own age-related chemical modifications. Combining proteomics with glycomics to capture these modifications will be most fruitful [35]. 

At the moment, different approaches are taken by researchers to characterize the various ECM compartments and components, making integration and comparison of data across studies difficult. Ultimately, a consensus will need to be reached in the brain ECM community about the methodology most suitable for screening the matrix and its changes so that we know exactly what we’re trying to fix.

To really understand the biology of matrix constituents, these approaches need to be complemented with thorough functional experimentation, and our ability to do so depends on the methodological tools at hand. However, the availability of tools to manipulate the brain ECM in vivo is fundamentally limited, and progress in studying and consequently targeting the brain ECM therapeutically will depend on addressing this roadblock.

Manipulating the matrix

The most commonly used method in studying matrix function is to take the molecular equivalent of a sledgehammer to it and see how this affects your favorite biological process. Chondroitinase ABC, a bacterial enzyme that cleaves the side chains of certain brain ECM-enriched molecules, has been delivered to nervous tissue in vivo as an investigational technique for over twenty years [36]. 

As someone who spent his time in graduate school studying microglia and their influence on the ECM by pharmacologically removing them from the brain, I’m the first to acknowledge the utility of such heavy-handed approaches in revealing fundamental biology [37]. 

However, chondroitinase has many drawbacks: 

• It generally requires direct, invasive delivery to the brain

• Because it quickly loses activity at 37°C, ECM effects are temporary

• It’s nonspecific, acting on a family of ECM molecules (CSPGs)

• At the same time, it only acts on one subset of ECM molecules

• The degradation products it generates from cleaving ECM can directly influence inflammation [38, 39] and confound experiments via off-target effects

This tool is a mainstay of ECM research and is constantly being upgraded. However, we need to acknowledge its limitations. Enzymes that act on other matrix components [40, 41] can also be used but may run into similar problems related to invasiveness of delivery and the generation of byproducts with their own biological activities, resulting in off-target effects [42].

So what else is there? Mice with ECM-related genes knocked out are a good bet. A number of mouse lines have been created that lack expression of individual ECM molecules [43-46] or the enzymes that synthesize [47] or modify them [34]. More advanced genetic models are now available that enable conditional manipulation of some of these proteins in particular cell types [13]. 

ECM substructure specificity is also possible, to a degree, as indicated earlier: by genetically removing the link proteins that hold the components of PNNs together, you can disrupt PNNs with minimal effects on the total levels of other core ECM components [48, 49]. Viruses driving the expression of a link protein or its mutant variant can be delivered to promote PNN formation and disassembly, respectively [16]. 

To the point made in the beginning of this article about in vitro experiments often neglecting the ECM: culturing methods have been developed to grow neurons and glia on decellularized brain tissue [23, 50], so the ECM can be manipulated both in an organism and in a dish. The development of ex vivo post-mortem brain perfusion systems [51] will further advance the translational potential of ECM screening approaches and functional experimentation.

Existing approaches to manipulating the brain ECM:

• Chondroitinase and other ECM-cleaving enzymes

• Constitutive and inducible cell type-specific ECM knockout mice or tissue

• Mouse lines lacking enzymes that synthesize or chemically modify the ECM

• Viral delivery of link protein HAPLN1 or a mutant variant to induce PNN formation or loss

• In vitro systems using decellularized brain tissue

A major roadblock in the field is the lack of small molecules that can regulate ECM synthesis, modification, and clearance. This relegates experimental studies to genetic, viral, or enzymatic approaches, which, while useful, are cumbersome and/or often nonspecific. And they are much more difficult to get off the ground translationally. 

The tools we need for the human matrix

What we really need are better pharmacological agents that can act on the ECM, either directly or indirectly. 

Important examples currently in use include:

• A peptide that interrupts a specific ECM-receptor interaction [52] 

• An antibody that binds to and blocks an inhibitory ECM component [53]

• Several small molecules that act as substrate analogues in ECM synthesis, which can reduce levels of core ECM proteins and the elongation of their sugar side chains [54, 55]. 

We need more of these. Ideally, we need small molecules that are ECM compartment-specific and that can shape the diffuse ECM, perineuronal nets, and/or vascular basement membrane ECM, either individually or in combination.

The development of blood-brain barrier-penetrant small molecules that activate or inhibit enzymes controlling ECM synthesis or degradation would unlock the potential for precise control over the ECM to an extent that is not currently possible. These could include drug candidates that act on proteolytic enzymes [56] or the molecules that regulate them [5], or compounds that target the enzymes directly involved in ECM synthesis and chemical modification. 

Alternatively, one could hijack the brain’s natural ECM clean-up crew by targeting the cytokines that control microglial processing of the ECM, as with IL-33 [6], or even microglial activation more broadly as a condition associated with ECM remodeling [7]. Whatever the approach, we need better tools to manipulate the brain matrix that can be more readily applied to the clinic.

Conclusion

Returning to the question I asked in the beginning, who cares about the brain ECM? The sad news is not too many people, at least as far as funding is concerned. The graph of active research spend shown at the start was meant to be as inclusive as possible for projects related to the matrix on RePORTER by the National Institute on Aging and the National Institute of Neurological Disorders and Stroke, but funding from the former likely also includes projects related to body-wide ECM.

Searching for active projects specifically related to the brain extracellular matrix returned just ~$62M in ongoing funding. This needs to change.

We have only very briefly covered the biology of the brain ECM in aging and age-related disorders here. The hope is that those working on interventions for brain aging will reconsider the role of the ECM in their approach - but I’ll settle for it being considered at all. Researchers who may want to look into the matrix further in their own research now have a place to begin, and an idea of the problems that need to be addressed for therapeutic advancement. 

Until we face the matrix, the possibility of restoring the aged brain may remain fundamentally out of reach.

Appendix

NIH funding breakdown

Data from RePORTER used to generate the funding chart. The funding is broken down into combined active funding by the National Institute on Aging and the National Institute of Neurological Disorders and Stroke as well as the total funding reported across all institutes and agencies. The total numbers of active projects that are currently funded are also displayed.

It’s interesting to note that there is a sizable investment in the extracellular matrix outside of the brain, with ~$875M total funding across institutes. Over a third of this comes from active projects sponsored by the National Heart, Lung, and Blood Institute (~$225M) and the National Cancer Institute (~$120M). Researchers studying the brain ECM may therefore benefit from the insights gained in these fields, such as the finding that local ECM stiffness hinders therapeutic delivery and immune cell infiltration into tumors [57]. Similar factors may affect therapeutic delivery to the aging brain.

Still, the numbers for total ECM funding - accounting for the matrix across all organs and tissues - barely hold up to the funding directed to individual glial cell types.

The search conditions used to generate these numbers are found in the table below. Searches were intended to be as inclusive as possible while still retaining sufficient specificity. 

Additional notes on the aging brain matrix

There are other aspects of brain ECM biology that are implicated in aging but that lie outside of the scope of this article: 

• ECM constituents are chemically modified during aging in a way that contributes to plasticity and memory impairments [34]

• ECM molecules deposit in amyloid plaques [58, 59] and clearing ECM molecules can reduce amyloid plaque pathology [40, 58, 59]

• Genetic variants that can stave off cognitive decline in familial Alzheimer’s disease patients by decades are related to their effects on ECM molecules [60]

• In addition to being regulated by immune cells, the ECM directly regulates inflammation and serves as a reservoir of immune-relevant cytokines

Interestingly, whereas the rat brain stiffens, human brains display an overall softening with age [61]. It is unclear the extent to which this is due to cell loss vs. changes in the mechanical properties of the ECM itself.

It may be the case that ECM in the human brain stiffens during the course of aging, as in the aging rat brain, while age-related cell loss and atrophy leads to a general softening of the human brain overall. Amyloid fibrils that accumulate in the brain extracellular space in Alzheimer’s disease are exceptionally stiff [62], for instance, and may augment the stiffness of the local ECM while the disease-related loss of cells and synapses in humans results in softening at the macro scale.

Further research is needed to close this translational gap. In any case, given what we know about the effects of the ECM on the functions of endogenous and transplanted cells - which extend beyond just its mechanical properties - it will remain prudent to keep the matrix in mind when developing strategies to treat brain aging.

Literally looking into the matrix

Recent developments in live imaging over the last year have enabled the use of dyes and fluorescently-labeled compounds to visualize the major brain ECM subcompartments in vivo in real-time [63, 64]. Live imaging of brain ECM and PNNs in vivo is also now possible with the use of a fluorescent probe that is genetically fused to an important ECM protein (the link protein HAPLN1) [65], and the mice are available on JAX. Another fusion-labeling system based on this same protein has already been developed to both birthdate brain ECM and to reveal novel matrix architecture [66]. Applying these tools to the aging brain will allow us to see how the ECM changes at a resolution never before possible.

Acknowledgements

As stated earlier, the framework used to address the importance of the ECM in brain aging was inspired by and based on age1’s categories of interventions for aging. Inspiration was also drawn from our team’s own Bottlenecks of Aging and research roadmap for combatting brain aging. Infographics were generated with BioRender unless otherwise indicated.

Huge thanks to everyone at the Amaranth Foundation for helping this come together - Joanne Peng for the constant feedback, Jarod Rutledge, Rohan Krajeski, and Raiany Romanni (several times over) for their thoughts and comments, and Alex Colville for input and for suggesting I do this in the first place. And of course thank you to James Fickel for making this possible. 

1. Nicholson, C., and Syková, E. (1998). Extracellular space structure revealed by diffusion analysis. Trends Neurosci. 21, 207–215. https://doi.org/10.1016/S0166-2236(98)01261-2.

2. Dityatev, A., Seidenbecher, C.I., and Schachner, M. (2010). Compartmentalization from the outside: the extracellular matrix and functional microdomains in the brain. Trends Neurosci. 33, 503–512. https://doi.org/10.1016/j.tins.2010.08.003.

3. Allen, N.J., and Lyons, D.A. (2018). Glia as Architects of Central Nervous System Formation and Function. Science 362, 181–185. https://doi.org/10.1126/science.aat0473.

4. Lau, L.W., Cua, R., Keough, M.B., Haylock-Jacobs, S., and Yong, V.W. (2013). Pathophysiology of the brain extracellular matrix: a new target for remyelination. Nat. Rev. Neurosci. 14, 722–729. https://doi.org/10.1038/nrn3550.

5. Ferreira, A.C., Hemmer, B.M., Philippi, S.M., Grau-Perales, A.B., Rosenstadt, J.L., Liu, H., Zhu, J.D., Kareva, T., Ahfeldt, T., Varghese, M., et al. (2023). Neuronal TIMP2 regulates hippocampus-dependent plasticity and extracellular matrix complexity. Mol. Psychiatry 28, 3943–3954. https://doi.org/10.1038/s41380-023-02296-5.

6. Nguyen, P.T., Dorman, L.C., Pan, S., Vainchtein, I.D., Han, R.T., Nakao-Inoue, H., Taloma, S.E., Barron, J.J., Molofsky, A.B., Kheirbek, M.A., et al. (2020). Microglial Remodeling of the Extracellular Matrix Promotes Synapse Plasticity. Cell 182, 388-403.e15. https://doi.org/10.1016/j.cell.2020.05.050.

7. Crapser, J.D., Spangenberg, E.E., Barahona, R.A., Arreola, M.A., Hohsfield, L.A., and Green, K.N. (2020). Microglia facilitate loss of perineuronal nets in the Alzheimer’s disease brain. EBioMedicine 58, 102919. https://doi.org/10.1016/j.ebiom.2020.102919.

8. Crapser, J.D., Ochaba, J., Soni, N., Reidling, J.C., Thompson, L.M., and Green, K.N. (2020). Microglial depletion prevents extracellular matrix changes and striatal volume reduction in a model of Huntington’s disease. Brain 143, 266–288. https://doi.org/10.1093/brain/awz363.

9. Morrison, J.H., and Baxter, M.G. (2012). The Aging Cortical Synapse: Hallmarks and Implications for Cognitive Decline. Nat. Rev. Neurosci. 13, 240–250. https://doi.org/10.1038/nrn3200.

10. Orlando, C., Ster, J., Gerber, U., Fawcett, J.W., and Raineteau, O. (2012). Perisynaptic Chondroitin Sulfate Proteoglycans Restrict Structural Plasticity in an Integrin-Dependent Manner. J. Neurosci. 32, 18009–18017. https://doi.org/10.1523/JNEUROSCI.2406-12.2012.

11. Reichelt, A.C., Hare, D.J., Bussey, T.J., and Saksida, L.M. (2019). Perineuronal Nets: Plasticity, Protection, and Therapeutic Potential. Trends Neurosci. 42, 458–470. https://doi.org/10.1016/j.tins.2019.04.003.

12. Pizzorusso, T., Medini, P., Berardi, N., Chierzi, S., Fawcett, J.W., and Maffei, L. (2002). Reactivation of Ocular Dominance Plasticity in the Adult Visual Cortex. Science 298, 1248–1251. https://doi.org/10.1126/science.1072699.

13. Rowlands, D., Lensjø, K.K., Dinh, T., Yang, S., Andrews, M.R., Hafting, T., Fyhn, M., Fawcett, J.W., and Dick, G. (2018). Aggrecan Directs Extracellular Matrix-Mediated Neuronal Plasticity. J. Neurosci. 38, 10102–10113. https://doi.org/10.1523/JNEUROSCI.1122-18.2018.

14. Crapser, J.D., Arreola, M.A., Tsourmas, K.I., and Green, K.N. (2021). Microglia as hackers of the matrix: sculpting synapses and the extracellular space. Cell. Mol. Immunol. 18, 2472–2488. https://doi.org/10.1038/s41423-021-00751-3.

15. Fawcett, J.W., Fyhn, M., Jendelova, P., Kwok, J.C.F., Ruzicka, J., and Sorg, B.A. (2022). The extracellular matrix and perineuronal nets in memory. Mol. Psychiatry 27, 3192–3203. https://doi.org/10.1038/s41380-022-01634-3.

16. Ramsaran, A.I., Wang, Y., Golbabaei, A., Aleshin, S., de Snoo, M.L., Yeung, B.A., Rashid, A.J., Awasthi, A., Lau, J., Tran, L.M., et al. (2023). A shift in the mechanisms controlling hippocampal engram formation during brain maturation. Science 380, 543–551. https://doi.org/10.1126/science.ade6530.

17. Tsien, R.Y. (2013). Very long-term memories may be stored in the pattern of holes in the perineuronal net. Proc. Natl. Acad. Sci. 110, 12456–12461. https://doi.org/10.1073/pnas.1310158110.

18. Christensen, A.C., Lensjø, K.K., Lepperød, M.E., Dragly, S.-A., Sutterud, H., Blackstad, J.S., Fyhn, M., and Hafting, T. (2021). Perineuronal nets stabilize the grid cell network. Nat. Commun. 12, 253. https://doi.org/10.1038/s41467-020-20241-w.

19. Miyata, S., Nishimura, Y., and Nakashima, T. (2007). Perineuronal nets protect against amyloid beta-protein neurotoxicity in cultured cortical neurons. Brain Res. 1150, 200–206. https://doi.org/10.1016/j.brainres.2007.02.066.

20. Cabungcal, J.-H., Steullet, P., Morishita, H., Kraftsik, R., Cuenod, M., Hensch, T.K., and Do, K.Q. (2013). Perineuronal nets protect fast-spiking interneurons against oxidative stress. Proc. Natl. Acad. Sci. U. S. A. 110, 9130–9135. https://doi.org/10.1073/pnas.1300454110.

21. Suttkus, A., Holzer, M., Morawski, M., and Arendt, T. (2016). The neuronal extracellular matrix restricts distribution and internalization of aggregated Tau-protein. Neuroscience 313, 225–235. https://doi.org/10.1016/j.neuroscience.2015.11.040.

22. Tewari, B.P., Woo, A.M., Prim, C.E., Chaunsali, L., Patel, D.C., Kimbrough, I.F., Engel, K., Browning, J.L., Campbell, S.L., and Sontheimer, H. (2024). Astrocytes require perineuronal nets to maintain synaptic homeostasis in mice. Nat. Neurosci., 1–14. https://doi.org/10.1038/s41593-024-01714-3.

23. Segel, M., Neumann, B., Hill, M.F., Weber, I., Viscomi, C., Zhao, C., Young, A., Agley, C.C., Thompson, A.J., Gonzalez, G., et al. (2019). Niche stiffness underlies the aging of central nervous system progenitor cells. Nature 573, 130–134. https://doi.org/10.1038/s41586-019-1484-9.

24. Fournet, M., Bonté, F., and Desmoulière, A. (2018). Glycation Damage: A Possible Hub for Major Pathophysiological Disorders and Aging. Aging Dis. 9, 880–900. https://doi.org/10.14336/AD.2017.1121.

25. Kjell, J., Fischer-Sternjak, J., Thompson, A.J., Friess, C., Sticco, M.J., Salinas, F., Cox, J., Martinelli, D.C., Ninkovic, J., Franze, K., et al. (2020). Defining the Adult Neural Stem Cell Niche Proteome Identifies Key Regulators of Adult Neurogenesis. Cell Stem Cell 26, 277-293.e8. https://doi.org/10.1016/j.stem.2020.01.002.

26. Yeo, R.W., Zhou, O.Y., Zhong, B.L., Sun, E.D., Navarro Negredo, P., Nair, S., Sharmin, M., Ruetz, T.J., Wilson, M., Kundaje, A., et al. (2023). Chromatin accessibility dynamics of neurogenic niche cells reveal defects in neural stem cell adhesion and migration during aging. Nat. Aging 3, 866–893. https://doi.org/10.1038/s43587-023-00449-3.

27. Yamada, J., Nadanaka, S., Kitagawa, H., Takeuchi, K., and Jinno, S. (2018). Increased Synthesis of Chondroitin Sulfate Proteoglycan Promotes Adult Hippocampal Neurogenesis in Response to Enriched Environment. J. Neurosci. 38, 8496–8513. https://doi.org/10.1523/JNEUROSCI.0632-18.2018.

28. Pujadas, L., Gruart, A., Bosch, C., Delgado, L., Teixeira, C.M., Rossi, D., Lecea, L. de, Martínez, A., Delgado-García, J.M., and Soriano, E. (2010). Reelin Regulates Postnatal Neurogenesis and Enhances Spine Hypertrophy and Long-Term Potentiation. J. Neurosci. 30, 4636–4649. https://doi.org/10.1523/JNEUROSCI.5284-09.2010.

29. Won, S.J., Kim, S.H., Xie, L., Wang, Y., Mao, X.O., Jin, K., and Greenberg, D.A. (2006). Reelin-deficient mice show impaired neurogenesis and increased stroke size. Exp. Neurol. 198, 250–259. https://doi.org/10.1016/j.expneurol.2005.12.008.

30. Oh, H.S.-H., Le Guen, Y., Rappoport, N., Urey, D.Y., Rutledge, J., Brunet, A., Greicius, M.D., and Wyss-Coray, T. (2024). Plasma proteomics in the UK Biobank reveals youthful brains and immune systems promote healthspan and longevity. Preprint, https://doi.org/10.1101/2024.06.07.597771 https://doi.org/10.1101/2024.06.07.597771.

31. Naba, A., Clauser, K.R., Hoersch, S., Liu, H., Carr, S.A., and Hynes, R.O. (2012). The Matrisome: In Silico Definition and In Vivo Characterization by Proteomics of Normal and Tumor Extracellular Matrices. Mol. Cell. Proteomics MCP 11, M111.014647. https://doi.org/10.1074/mcp.M111.014647.

32. Pokhilko, A., Brezzo, G., Handunnetthi, L., Heilig, R., Lennon, R., Smith, C., Allan, S.M., Granata, A., Sinha, S., Wang, T., et al. (2021). Global proteomic analysis of extracellular matrix in mouse and human brain highlights relevance to cerebrovascular disease. J. Cereb. Blood Flow Metab. 41, 2423–2438. https://doi.org/10.1177/0271678X211004307.

33. Deepa, S.S., Carulli, D., Galtrey, C., Rhodes, K., Fukuda, J., Mikami, T., Sugahara, K., and Fawcett, J.W. (2006). Composition of perineuronal net extracellular matrix in rat brain: a different disaccharide composition for the net-associated proteoglycans. J. Biol. Chem. 281, 17789–17800. https://doi.org/10.1074/jbc.M600544200.

34. Yang, S., Gigout, S., Molinaro, A., Naito-Matsui, Y., Hilton, S., Foscarin, S., Nieuwenhuis, B., Tan, C.L., Verhaagen, J., Pizzorusso, T., et al. (2021). Chondroitin 6-sulphate is required for neuroplasticity and memory in ageing. Mol. Psychiatry 26, 5658–5668. https://doi.org/10.1038/s41380-021-01208-9.

35. Raghunathan, R., Hogan, J.D., Labadorf, A., Myers, R.H., and Zaia, J. (2020). A glycomics and proteomics study of aging and Parkinson’s disease in human brain. Sci. Rep. 10, 12804. https://doi.org/10.1038/s41598-020-69480-3.

36. Silver, J., and Miller, J.H. (2004). Regeneration beyond the glial scar. Nat. Rev. Neurosci. 5, 146–156. https://doi.org/10.1038/nrn1326.

37. Green, K.N., Crapser, J.D., and Hohsfield, L.A. (2020). To Kill a Microglia: A Case for CSF1R Inhibitors. Trends Immunol. 41, 771–784. https://doi.org/10.1016/j.it.2020.07.001.

38. Ebert, S., Schoeberl, T., Walczak, Y., Stoecker, K., Stempfl, T., Moehle, C., Weber, B.H.F., and Langmann, T. (2008). Chondroitin sulfate disaccharide stimulates microglia to adopt a novel regulatory phenotype. J. Leukoc. Biol. 84, 736–740. https://doi.org/10.1189/jlb.0208138.

39. Rolls, A., Cahalon, L., Bakalash, S., Avidan, H., Lider, O., and Schwartz, M. (2006). A sulfated disaccharide derived from chondroitin sulfate proteoglycan protects against inflammation-associated neurodegeneration. FASEB J. Off. Publ. Fed. Am. Soc. Exp. Biol. 20, 547–549. https://doi.org/10.1096/fj.05-4540fje.

40. Jendresen, C.B., Cui, H., Zhang, X., Vlodavsky, I., Nilsson, L.N.G., and Li, J.-P. (2015). Overexpression of Heparanase Lowers the Amyloid Burden in Amyloid-β Precursor Protein Transgenic Mice. J. Biol. Chem. 290, 5053–5064. https://doi.org/10.1074/jbc.M114.600569.

41. Kochlamazashvili, G., Henneberger, C., Bukalo, O., Dvoretskova, E., Senkov, O., Lievens, P.M.-J., Westenbroek, R., Engel, A.K., Catterall, W.A., Rusakov, D.A., et al. (2010). The Extracellular Matrix Molecule Hyaluronic Acid Regulates Hippocampal Synaptic Plasticity by Modulating Postsynaptic L-Type Ca2+ Channels. Neuron 67, 116–128. https://doi.org/10.1016/j.neuron.2010.05.030.

42. Perkins, K.L., Arranz, A.M., Yamaguchi, Y., and Hrabetova, S. (2017). Brain extracellular space, hyaluronan, and the prevention of epileptic seizures. Rev. Neurosci. 28, 869–892. https://doi.org/10.1515/revneuro-2017-0017.

43. Brakebusch, C., Seidenbecher, C.I., Asztely, F., Rauch, U., Matthies, H., Meyer, H., Krug, M., Böckers, T.M., Zhou, X., Kreutz, M.R., et al. (2002). Brevican-Deficient Mice Display Impaired Hippocampal CA1 Long-Term Potentiation but Show No Obvious Deficits in Learning and Memory. Mol. Cell. Biol. 22, 7417–7427. https://doi.org/10.1128/MCB.22.21.7417-7427.2002.

44. Favuzzi, E., Marques-Smith, A., Deogracias, R., Winterflood, C.M., Sánchez-Aguilera, A., Mantoan, L., Maeso, P., Fernandes, C., Ewers, H., and Rico, B. (2017). Activity-Dependent Gating of Parvalbumin Interneuron Function by the Perineuronal Net Protein Brevican. Neuron 95, 639-655.e10. https://doi.org/10.1016/j.neuron.2017.06.028.

45. Zhou, X.H., Brakebusch, C., Matthies, H., Oohashi, T., Hirsch, E., Moser, M., Krug, M., Seidenbecher, C.I., Boeckers, T.M., Rauch, U., et al. (2001). Neurocan is dispensable for brain development. Mol. Cell. Biol. 21, 5970–5978. https://doi.org/10.1128/MCB.21.17.5970-5978.2001.

46. Irala, D., Wang, S., Sakers, K., Nagendren, L., Ulloa Severino, F.P., Bindu, D.S., Savage, J.T., and Eroglu, C. (2024). Astrocyte-secreted neurocan controls inhibitory synapse formation and function. Neuron 112, 1657-1675.e10. https://doi.org/10.1016/j.neuron.2024.03.007.

47. Arranz, A.M., Perkins, K.L., Irie, F., Lewis, D.P., Hrabe, J., Xiao, F., Itano, N., Kimata, K., Hrabetova, S., and Yamaguchi, Y. (2014). Hyaluronan deficiency due to Has3 knock-out causes altered neuronal activity and seizures via reduction in brain extracellular space. J. Neurosci. Off. J. Soc. Neurosci. 34, 6164–6176. https://doi.org/10.1523/JNEUROSCI.3458-13.2014.

48. Carulli, D., Pizzorusso, T., Kwok, J.C.F., Putignano, E., Poli, A., Forostyak, S., Andrews, M.R., Deepa, S.S., Glant, T.T., and Fawcett, J.W. (2010). Animals lacking link protein have attenuated perineuronal nets and persistent plasticity. Brain 133, 2331–2347. https://doi.org/10.1093/brain/awq145.

49. Bekku, Y., Saito, M., Moser, M., Fuchigami, M., Maehara, A., Nakayama, M., Kusachi, S., Ninomiya, Y., and Oohashi, T. (2012). Bral2 is indispensable for the proper localization of brevican and the structural integrity of the perineuronal net in the brainstem and cerebellum. J. Comp. Neurol. 520, 1721–1736. https://doi.org/10.1002/cne.23009.

50. Lam, D., Enright, H.A., Cadena, J., Peters, S.K.G., Sales, A.P., Osburn, J.J., Soscia, D.A., Kulp, K.S., Wheeler, E.K., and Fischer, N.O. (2019). Tissue-specific extracellular matrix accelerates the formation of neural networks and communities in a neuron-glia co-culture on a multi-electrode array. Sci. Rep. 9, 4159. https://doi.org/10.1038/s41598-019-40128-1.

51. Vrselja, Z., Daniele, S.G., Silbereis, J., Talpo, F., Morozov, Y.M., Sousa, A.M.M., Tanaka, B.S., Skarica, M., Pletikos, M., Kaur, N., et al. (2019). Restoration of brain circulation and cellular functions hours post-mortem. Nature 568, 336–343. https://doi.org/10.1038/s41586-019-1099-1.

52. Lang, B.T., Cregg, J.M., DePaul, M.A., Tran, A.P., Xu, K., Dyck, S.M., Madalena, K.M., Brown, B.P., Weng, Y.-L., Li, S., et al. (2015). Modulation of the proteoglycan receptor PTPσ promotes recovery after spinal cord injury. Nature 518, 404–408. https://doi.org/10.1038/nature13974.

53. Yang, S., Hilton, S., Alves, J.N., Saksida, L.M., Bussey, T., Matthews, R.T., Kitagawa, H., Spillantini, M.G., Kwok, J.C.F., and Fawcett, J.W. (2017). Antibody recognizing 4-sulfated chondroitin sulfate proteoglycans restores memory in tauopathy-induced neurodegeneration. Neurobiol. Aging 59, 197–209. https://doi.org/10.1016/j.neurobiolaging.2017.08.002.

54. Keough, M.B., Rogers, J.A., Zhang, P., Jensen, S.K., Stephenson, E.L., Chen, T., Hurlbert, M.G., Lau, L.W., Rawji, K.S., Plemel, J.R., et al. (2016). An inhibitor of chondroitin sulfate proteoglycan synthesis promotes central nervous system remyelination. Nat. Commun. 7, 11312. https://doi.org/10.1038/ncomms11312.

55. Stephenson, E.L., Zhang, P., Ghorbani, S., Wang, A., Gu, J., Keough, M.B., Rawji, K.S., Silva, C., Yong, V.W., and Ling, C.-C. (2019). Targeting the Chondroitin Sulfate Proteoglycans: Evaluating Fluorinated Glucosamines and Xylosides in Screens Pertinent to Multiple Sclerosis. ACS Cent. Sci. 5, 1223–1234. https://doi.org/10.1021/acscentsci.9b00327.

56. Cathomas, F., Lin, H.-Y., Chan, K.L., Li, L., Parise, L.F., Alvarez, J., Durand-de Cuttoli, R., Aubry, A.V., Muhareb, S., Desland, F., et al. (2024). Circulating myeloid-derived MMP8 in stress susceptibility and depression. Nature 626, 1108–1115. https://doi.org/10.1038/s41586-023-07015-2.

57. Mai, Z., Lin, Y., Lin, P., Zhao, X., and Cui, L. (2024). Modulating extracellular matrix stiffness: a strategic approach to boost cancer immunotherapy. Cell Death Dis. 15, 1–16. https://doi.org/10.1038/s41419-024-06697-4.

58. Liu, C.-C., Zhao, N., Yamaguchi, Y., Cirrito, J.R., Kanekiyo, T., Holtzman, D.M., and Bu, G. (2016). Neuronal heparan sulfates promote amyloid pathology by modulating brain amyloid-β clearance and aggregation in Alzheimer’s disease. Sci. Transl. Med. 8, 332ra44. https://doi.org/10.1126/scitranslmed.aad3650.

59. Howell, M.D., Bailey, L.A., Cozart, M.A., Gannon, B.M., and Gottschall, P.E. (2015). Hippocampal administration of chondroitinase ABC increases plaque-adjacent synaptic marker and diminishes amyloid burden in aged APPswe/PS1dE9 mice. Acta Neuropathol. Commun. 3, 54. https://doi.org/10.1186/s40478-015-0233-z.

60. Lopera, F., Marino, C., Chandrahas, A.S., O’Hare, M., Villalba-Moreno, N.D., Aguillon, D., Baena, A., Sanchez, J.S., Vila-Castelar, C., Ramirez Gomez, L., et al. (2023). Resilience to autosomal dominant Alzheimer’s disease in a Reelin-COLBOS heterozygous man. Nat. Med. 29, 1243–1252. https://doi.org/10.1038/s41591-023-02318-3.

61. Hiscox, L.V., Johnson, C.L., McGarry, M.D.J., Perrins, M., Littlejohn, A., van Beek, E.J.R., Roberts, N., and Starr, J.M. (2018). High-resolution magnetic resonance elastography reveals differences in subcortical gray matter viscoelasticity between young and healthy older adults. Neurobiol. Aging 65, 158–167. https://doi.org/10.1016/j.neurobiolaging.2018.01.010.

62. Knowles, T.P.J., and Buehler, M.J. (2011). Nanomechanics of functional and pathological amyloid materials. Nat. Nanotechnol. 6, 469–479. https://doi.org/10.1038/nnano.2011.102.

63. Ge, X., Xu, X., Cai, Q., Xiong, H., Xie, C., Hong, Y., Gao, X., Yao, Y., Bachoo, R., and Qin, Z. (2024). Live Mapping of the Brain Extracellular Matrix and Remodeling in Neurological Disorders. Small Methods 8, 2301117. https://doi.org/10.1002/smtd.202301117.

64. Benbenishty, A., Peled-Hajaj, S., Krishnaswamy, V.R., Har-Gil, H., Havusha-Laufer, S., Ruggiero, A., Slutsky, I., Blinder, P., and Sagi, I. (2023). Longitudinal in vivo imaging of perineuronal nets. Neurophotonics 10, 015008. https://doi.org/10.1117/1.NPh.10.1.015008.

65. Lemieux, S.P., Lev-Ram, V., Tsien, R.Y., and Ellisman, M.H. (2023). Perineuronal nets and the neuronal extracellular matrix can be imaged by genetically encoded labeling of HAPLN1 in vitro and in vivo. Preprint at bioRxiv, https://doi.org/10.1101/2023.11.29.569151 https://doi.org/10.1101/2023.11.29.569151.

66. Sterin, I., Niazi, A., Kim, J., Park, J., and Park, S. (2024). Novel extracellular matrix architecture on excitatory neurons revealed by HaloTag-HAPLN1. bioRxiv, 2024.03.29.587384. https://doi.org/10.1101/2024.03.29.587384.

Introducing The Amaranth Letters

By Raiany Romanni

“Strengthening connections between neurons” - Troy Littleton

“We are going to die, and that makes us the lucky ones. Most people are never going to die because they are never going to be born. Those unborn ghosts include greater poets than Keats, scientists greater than Newton. We know this because the set of possible people allowed by our DNA so massively exceeds the set of actual people. In the teeth of these stupefying odds it is you and I, in our ordinariness, that are here.”

Richard Dawkins, Unweaving the Rainbow

Imagine the year is 1950, early spring. Heart transplants are the stuff of repugnant fiction. Birth control is yet to be tested in humans, and no woman on Earth has the choice of controlling her innate biology.  By the year’s end, smallpox will take some 5 million lives — maybe more.  In developing countries, a tooth infection is deadly. Everywhere, a cancer diagnosis is almost as likely to be managed by a doctor as it is by a priest. 

In 2024, most of us accept some enhancements to our natural life course, like polio vaccines. We do not refuse antibiotics just because they unnaturally add some 23 years to our life expectancy. And heart transplants are no longer heretical, but the stuff of strait-laced, good old medicine.

Still, the idea of extending the human lifespan beyond its unchanged maximum of (allegedly) 122 lies somewhere between avant-garde and fringe. 

Why?

One explanation is that most people, when they hear “life extension,” think of the 20th century — when we doubled our average life years without a corresponding increase in healthy life. Another is that improving the biology of aging — so that we get to live longer without age-related frailty and disease — is technically difficult. 

In a 1993 study, changing one gene (daf-2, a pathway humans share) in C. elegans worms doubled their lifespan. Changing one additional gene (rsks-1) resulted in a five-fold lifespan increase—the equivalent of a 400-year-old human. But technologies like genome engineering for polygenic conditions may be decades away from safety. And promising approaches like replacing aged cells, organs, and tissues will be difficult to deploy at scale.

To date, no single intervention has been proven to reverse biological aging or to extend the maximum lifespan in humans. Clinical trials for multiple diseases—let alone health & lifespan—are costly, lengthy, and difficult to translate between species. As a result, the effects of existing therapeutics on human aging remain largely speculative, and a long list of bottlenecks in the field remains underfunded.

But funding often precedes the technically ambitious results we desire.

This has so far been the case for AI alignment and neurodegenerative diseases, and it was the case for COVID vaccines. 

In 2024, the young science of aging remains the subject of often repulsive fiction — and receives only about 1% of all National Institutes of Health funds: about 8 times less than research on neurodegenerative diseases, even though they’re just one of the several things that go awry with age.

On the private investment front, aging research makes up roughly 3% of all venture capital in biotechnology. And commercial incentives have so far mostly optimized for unproven supplements, imprecise biological-age apps, or unsafe experimental therapies and cosmetics. 

Today, the longevity field largely places its bets on 1) a small roster of scientists achieving a breakthrough, and 2) technological convergence (e.g. AI) spontaneously arriving at a health-&-lifespan wonder drug. 

This lack of funding for serious research on the biology of aging can only be partly explained by the snake-oil “cures” that continue to plague the longevity field. A less circular reason why we haven’t improved how we age over several millennia is that humans just don’t buy the idea that improving the biology of aging is a noble and pressing cause to begin with.

In a world with non-infinite resources, isn’t it more noble and pressing to treat the already sick? Aren’t, say, programmable drugs for future pandemics more pressing? Given the climate crisis, shouldn’t we be reducing the number of humans on Earth? If aging drugs could delay up to 90% of all deaths, is this the world we’d want to engineer? Doesn’t science advance, as perhaps every STEM student on Earth has learned, one funeral at a time?

I’m often asked why, as a bioethicist, I’ve joined a team of people pushing the frontiers of science to compress the timeline for technologies that extend the human lifespan.

My answer is simple: because human life is what I value most — above all other things. 

In today’s letter, I’ll walk you through my reasoning. 

Progress, as I will try to show you, doesn’t magically unfold when funerals take place. Most often, it is architected by the living.

Our goal at Amaranth is to be, fund, and catalyze these living people.

Since it was founded in 2021, Amaranth has donated over $30 million to ambitious research in longevity and neuroscience. 

Underlying our mission is the idea that even today — in an age of human-vilifying headlines and increasingly ingenious machines — our ability to engineer progress and to improve human welfare is more often than not contingent on human brains. 

If we do our jobs right, we’re almost sure to live through generations of poets greater than Keats, scientists greater than Newton. We may unlock the unborn ghosts that lie already within each of us — with lives now too short and overwhelmed with grief and pre-abundance tradeoffs and disease, and bodies too brittle to live up to our infinite dreams as a finite, dainty species.

In the coming months, we’ll publish new data on the economic value of improving biological aging. We’ll look into the ethics and technical viability of accelerating different biotechnologies; and we’ll discuss overlooked approaches to reversing or postponing death — like better cryopreservation methods for different organs, including the brain.

Today, I’ll just begin to answer the one question that rustles from underneath each of our efforts.

Subscribe now

Why prioritize human life?

To start with, because it’s the honest answer.

Modern governments prioritize fire stations over witch trials; they sooner fund vaccines to stop public health crises than funeral homes; and they even build safe roads and mandate building codes and enforce laws like wearing seatbelts because the loss of human lives is socially expensive.

For every human death, an average 9 humans are negatively impacted in profound ways. Grief costs the U.S. alone an estimated $75 billion a year in productivity losses. Unpaid caretakers of older adults in declining health in the U.S.  add up to a combined lost income of $522 billion every year.

Conversely, for every 1% increase in population growth, a 3 -  4% increase in personal abundance can be tracked, measured by access to goods like food or services like transportation.

Economic Growth Over the Very Long Run - Chad Jones

Of course, governments do not fund our wellbeing just because it can improve our federal budget. If that were the case, states may well incentivize for their populations to smoke and die quickly and young, as a short-termist path to saving trillions of dollars on Social Security & Medicare. 

But it’s worth noting that people and governments didn’t always appreciate human life as valuable. 

In Ancient Rome, humans gathered around gleefully to watch as other low-status humans were tortured and killed. Codes like the Hammurabi existed since 1750 BC — but the value of life was far from egalitarian. Most humans, most of the time, were perceived as fungible means.

This can still be the case, but we have made much progress.

The Enlightenment can be credited with highlighting the possibility that all humans, of all classes, could be valuable. As one economist put it:

“It became obvious for the first time that wealth could grow, that classes could be fluid, that life could be extended, that the population could grow and grow, that we could progress together. With that evidence [from the late Middle Ages] came the entrenchment of a commitment to individual rights and a longing for universal emancipation and the good life for all. A related commitment emerged to the cause of medical science not just to help the few but to improve the whole lot of humanity.”

It’s been a lucky truth that human brains have been a major cause of economic growth and progress. This truth led (partly) to the creation of welfare states, and to the once-idealistic notion that all lives should be valued equally, since classes can be fluid.

But this view of people “as human capital fully replaceable first by immigration, and then eventually by technology” has its shortcomings.

The project of ensuring Homo Sapiens remains relevant — in a world filled with sprightly, artsy, and never-tired robots — will demand extreme honesty from us as a species.

The rise of independently intelligent machines may challenge our assumptions on why we ought to value human life, when anything we can do, a robot will likely be able to do better, and when any economic output linked to unaided biological brains may become obsolete.

We can work towards a biosingularity where we evolve together with machines.

But for humans to make it through this first half of the third millennium with a continued sense of value, we’ll also need to intently work towards a new type of self-worth. 

In 2024, many humans continue to derive their worth from their production value. One 21st-century farmer, engineer, or baker can produce more resources than she consumes in her lifetime. Yet in the coming decades, we’ll need to bolster the idea that human life — regardless of its output — has intrinsic worth. 

It may be the case that we can’t rationally explain why we ought to prioritize human life over, say, other conscious animals or machines without veering into what philosophers have called infinite regress. The best explanation available to our primate brains may be  “we ought to value human life because it’s what is best for us.”

For children of the Enlightenment, this may be an unsatisfying answer. 

But if we found ourselves paralyzed by this justification, unable to decide whether our hospital beds should prioritize conscious humans over conscious mice, the loved ones of the humans hypothetically lost in our deliberation would be rightly outraged.

Nearly all noble things a human being can do — from building climate technologies to helping a stranger cross the street to engineering rockets that traverse the stars — are done to celebrate, improve, or protect human life. 

Sometimes, we’ll grant secondary moral status to non-human lives that seem to function like ours.

And I happen to think it’s critical we admit this hierarchy — just as it’s wise to assume that if superintelligent machines aren’t aligned to our brains, to serve our goals, they may well treat us as we do other less intelligent life forms. (Not out of malice, but mere competence.) 

We should — and I trust, will — work towards a future with non-scarce resources, where human and non-human lives can co-exist with equal rights. 

Yet in the near term, even those of us who don’t eat other conscious animals (I don’t) generally support their use in clinical findings to save human lives — even if the long-term goal should be to move towards preclinical development on never-before-conscious subjects through methodologies like organs-on-chips or computer models.

With some isolated exceptions to this rule (e.g. terrorist or ecocentric groups), our chief aim as a civilization — the occupation of some of the brightest people on Earth, who work on goals like nuclear policy and climate innovation and AI alignment — can now be summarized as to nurture and protect human life. 

This is no small achievement. 

But there’s a way in which this is, more than a feat, a feature.  We value human life because it’s what we’re biologically wired to value. Because life is what enables each possible subjective experience we can have — from enjoying a late-summer breeze to building history-changing companies to caressing a beloved pet.

It may be time we reconcile this evolutionary thirst for life with the Enlightenment view of human lives as inherently worthy of investment.

Subscribe now

Why aging research?

There’ll be no single day on the calendar of Homo Sapiens when we finally engineer “longevity.” 

If there were, it would have been when the first hominid attempted to postpone their death or suffering by some non-innate means; or when the first single-cell organism sought collective life in the hopes of improving its odds of survival.

If we look closely, just about all of medicine is designed to extend life or diminish suffering. 

Health is only occasionally the goal of medicine. Palliative care aims to make end-of-life less painful. Resuscitation techniques (like defibrillation or CPR) and a good deal of drugs (like blood thinners or statins) aim to extend life. 

The same is true of aging research, whose promise isn’t to extend life lived with zero disabilities — health and youth are discrete processes, even if they strongly correlate — but to prolong functional, active, livable life. (Some bioethicists have argued all lives are livable; others that we should not live past the age of 75; and the right answer is probably “it’s up to each human to decide.”) 

Our goal, as it has been written before, should be to make aging drugs boring. (Think safer and more effective versions of ozempic, or rapamycin.) To make life-saving interventions available — like neuron transplants or therapies to restructure the extracellular matrix — and then have patients decide whether or not to adhere to them. And to enhance our biopreservation methods so that pausing life for extended periods of time (longer than the current lull between life and death for, say, patients suffering from cardiac arrest) becomes routine medicine.

Aging research is one way to extend and improve human life. But it’s one vastly overlooked way.

NIH funding breakdown

In the past few decades, the allocation of resources to “Global-South diseases” has been a growing moral trend. The Global Fund, for instance, “cut the combined death rate from AIDS, TB, and malaria by more than half, saving 59 million lives [in 20 years].” Space exploration and AI alignment also promise to more effectively extend our longevity as a species.

From a total-utilitarian standpoint, simply creating more working-age humans would solve many of our economic problems.  But this solution, even if its timeline could be compressed — which it cannot, since newborns don’t work, and they one day grow old — overlooks the non-fungible value of existing human lives. 

The world is aging — and this means the disease burden is shifting, in the Global North and South. 

In Brazil, where I come from, non-communicable diseases now make up 75% of all deaths. By 2050, 80% of all older adults on Earth will live in low- to middle-income countries.

Non-communicable diseases in Brazil

Knowing what we do for certain, given no hypotheticals — that some 50 million age-related deaths happen every year, like clockwork, causing incalculable suffering in the Global North and South — drugs and tools designed to improve biological aging could save more lives in the coming decade than any other neglected drug or cosmic expedition. 

Yes, an asteroid could hit the Earth and make a focus on individual longevity seem vain. But if we compress the timeline for aging drugs, we’ll have more resources and talent to engineer solutions that extend our longevity as a species. 

And while the future may be sufficiently abundant that we get to divorce ourselves from economic datasets, the coming years will demand critical tradeoffs on how we allocate scarce resources.

Rounded estimates from the Congressional Budget Office

In just 11 years — between 2018 and 2029 — the U.S. mandatory spending on Social Security and Medicare will more than double, from $1.3 trillion to $2.7 trillion per year.

In the coming decade, safe and effective interventions for biological aging could allow us to reallocate trillions of dollars now projected for the social and medical care of adults aged 65 and older towards other noble causes — like pandemic preparedness or lowering infant mortality.

Indeed, biological aging is a neglected factor in childhood cancers (where patients experience premature aging), and in several communicable diseases.

The COVID death rate for people under 40, for instance, was around 0.2%.

COVID Death Toll - Vox

As Operation Warp Speed has shown, time alone is a deceitful metric when it comes to the advancement of fundamental science.

Market incentives aren’t by default aligned to what is best for us as a species. But with enough public and philanthropic funds, we can align them.

We can develop age-improving therapeutics — eventually, available to all — while building history’s most profitable life sciences companies.

We can unveil the principles of intelligence while solving Alzheimer’s.

And we can profitably accelerate technologies like brain interfaces and modeling to ensure the future of humanity remains contingent on and aligned to human minds.

The unifying theme of The Amaranth Letters will be that human life is rare; mostly evolved to celebrate, improve, and protect itself; and that human brains and aspirations are, more often than not, deeply good.

I hope you’ll join us in brainstorming how we can best use our primate brains to create more, better, and safer life. 

Raiany Romanni is a Policy and Ethics researcher at the Amaranth Foundation. She is writing a book on the moral underpinnings and socio-economic implications of life extension.

Acknowledgments: Joanne Peng, Alex Colville, Rohan Krajeski, Aaron Cravens, James Fickel (& many others!)

We fund ambitious research in longevity and neuroscience. If you support the ideas in this letter, please consider pledging an amount that feels right to you to sustain our mission as a philanthropic organization.