UTA Biology A Jump Start for Electroceuticals & Restoring Vision Summaries

I just need a one page summary for each paper. Where you include your opinion on the paper itself as well.

COMMENT
BIOGEOGRAPHY David Quammen
reappraises The Malay
Archipelago p.165
MUSEUM San Francisco
Exploratorium has a bright
new home p.167
TEXTILES Knitted robots,
mobile-charging shirts and
wearable sensors p.168
PESKIMO/SYNERGY ART
HISTORY In praise of Alfred
Russel Wallace, visionary
scientist, daring explorer p.162
A jump-start for
electroceuticals
Kristoffer Famm and colleagues unveil a multidisciplinary initiative to develop
medicines that use electrical impulses to modulate the body’s neural circuits.
I
magine a day when electrical impulses
are a mainstay of medical treatment.
Your clinician will administer ‘electroceuticals’ that target individual nerve fibres
or specific brain circuits to treat an array of
conditions. These treatments will modulate
the neural impulses controlling the body,
repair lost function and restore health. They
could, for example, coax insulin from cells to
treat diabetes, regulate food intake to treat
obesity and correct balances in smoothmuscle tone to treat hypertension and
pulmonary diseases.
All this is within reach if researchers
from disparate disciplines in academia and
ind­u stry work together. Here, we outline what needs to be done to bring about
electro­ceuticals and unveil a public–private
research initiative and an award that we hope
will catalyse the field.
Electrical impulses — action potentials
— are the language of the body’s nervous
system. Virtually all organs and functions
are regulated through circuits of neurons
communicating through such impulses1.
Two features make these circuits excellent
targets for therapeutic intervention. First,
they comprise discrete components —
interconnected cells, fibre tracts and
nerve bundles — allowing for pinpoint
1 1 A P R I L 2 0 1 3 | VO L 4 9 6 | NAT U R E | 1 5 9
© 2013 Macmillan Publishers Limited. All rights reserved
COMMENT
intervention. Second, they are controlled
by patterns of action potentials, which can be
altered for treatment.
Already, devices that harness electrical
impulses are used to treat disease. Pace­
makers and defibrillators save millions
of lives each year; deep-brain stimulation dramatically improves the quality of
life for people with Parkinson’s disease
and depress­ion; sacral-nerve stimulation
restores some bladder control in people
with paraplegia, and vagus-nerve stimulation shows clinical benefits in diseases
ranging from epilepsy to rheumatoid
arthritis2. But these devices do not target
specific cells within circuits.
Neural tissue is compact: unrelated
circuits often run close together through
brain regions and in peripheral nerves.
At present, electrical devices activate
or inhibit cells in an area of tissue indiscriminately, muddying clinical effects.
For example, electrodes that stimulate
the vagus nerve enclose approximately
100,000 fibres, which innervate many
different internal organs. Similarly, deepbrain stimulation for Parkinson’s disease
affects many cells other than those that
control movement, leading to emotional
and cognitive side effects. In natural urinary control, opposite signals in adjacent
nerve fibres simultaneously contract the
bladder and relax the urethral sphincter
— an elegant process that is poorly
mimicked by today’s devices.
Neither do neurostimulation devices
yet generate naturalistic patterns of action
potentials. Typically, devices block or stimulate with simple waveforms, rather than
modulate dynamically on the milli­second
scale. Precise modulation is important:
in mice, stimulation of cells in the neural
circuit for hunger with a simple 20-Hertz
waveform causes voracious eating within
minutes, and ablation of these cells causes
anorexia; but food intake can be more
finely modulated by the number and frequency of action potentials in specific
cells3. Similarly, single action potentials
in small sets of cortical neurons have been
shown to encode sensory input or perception in mice4. In other words, neural
circuits act through sets of precise electrical
impulses generated in specific sets of cells.
PATH TO PRECISION
We believe that it is now possible to create
medicines that control action potentials in
individual neurons and in functional groups
of them.
Many of the stepping stones are already in
place, thanks to recent advances in a variety
of disciplines. For example, disease-specific
neural circuits, such as the reflex that controls levels of inflammatory mediators5, are
starting to be anatomically and functionally
traced. Tools, such as optogenetics, that
enable cellular-level control have improved
researchers’ ability to analyse the signals
in circuits, and they provide a mechanism
by which future electroceuticals could
elicit action potentials6. Efforts to control
prosthetic limbs and generate brain–machine
interfaces are giving rise to architectures
for electrodes that can interact with individual neurons. Researchers are designing
microchips that mimic brain processing to
facilitate local and low-power computation7.
The development of cochlear and retinal
implants has led to advances in neural signal
processing. Nanotechnology has delivered
approaches for harvesting energy to power
micro­devices8. And neuro­surgery can now
be done through small holes in the skull
and body with the use of needles and
scopes, as in preci“Researchers
sion procedures to
will need to
remove herniated
disc material from
embrace the
the spine or open
tools of other
new fluid channels
fields, and
from blocked brain
even dream
ventricles.
differently.”
The first logical
step towards electroceuticals is to better
map the neural circuits associated with disease and treatment. This needs to happen
on two levels. On the anatomical level,
researchers need to map disease-associated
nerves and brain areas and identify the best
points for intervention. On the signalling
level, the neural language at these intervention points must be decoded, so that
researchers can develop a ‘dictionary’ of
patterns associated with health and disease states — a project synergistic with
international drives to map the human
brain9. In circuits altered by disease, it will
be important to establish how introduced
electrical impulses affect the disease and
which patterns yield the most effective
therapeutic responses. Developing the
technology to record from and stimulate
a larger set of central and peripheral
neurons will be crucial to this pursuit.
This type of research is analogous to the
target-identification and validation steps
at the core of modern molecular-drug discovery. The circuit maps that emerge will
provide the design specifications for future
treatment devices. Early prototypes might
use microchip-controlled electrode arrays
similar to those used today in interfaces for
prosthetic limbs to modulate neural signals
(see ‘It’s electric’). Second-generation
micro- and nanoscale devices may instead
leverage light, mechanical or magnetic
energy to achieve such modulation in
specific cells within targeted circuits.
How will all this come about? Disease biologists will need to work with neuroscientists
to map circuits and with bioinformaticians
1 6 0 | NAT U R E | VO L 4 9 6 | 1 1 A P R I L 2 0 1 3
© 2013 Macmillan Publishers Limited. All rights reserved
to identify the action-potential signatures
of diseases. To develop treatment devices,
bioengineers designing biocompatible
interfaces will need to collaborate with electrical engineers to develop microchips for
real-time signal processing; with nanotechnologists to create energy sources; and with
neurosurgeons to ensure that these designs
can be implanted and connected. Researchers will need to embrace the languages and
tools of other fields, and perhaps even dream
differently: much of the challenge lies in
translating biological understanding into
engineering specifications.
MULTIDISCIPLINARY JOURNEY
We think that initial progress will come from
targeting circuits that have accessible and
peripheral intervention points. For example, it
has been shown that hypertension can be controlled through signals in carotid-sinus and
renal nerves, and the production of certain
inflammatory molecules in rheumatoid
arthritis can be modified through the splenic
nerve. A range of conditions — cardiovascular, metabolic, respiratory, inflammatory and
autoimmune — are likely to have similarly
accessible intervention points, given that they
involve organs and functions that are under
neural control.
We envision adaptive or ‘closed-loop’
electroceuticals that can record incoming
action potentials and physiological parameters, analyse these data in real time and
modulate neural signalling accordingly.
This capability, together with that of spatially
targeting a specific set of neurons, will
underpin the selective therapeutic effect that
we expect from electroceuticals. But these
closed-loop therapies can be realized only
if the required disciplines come together
early on.
We also hope that this effort will result
in interdisciplinary advances that can be
brought to bear on disorders of the brain
itself. Correcting such disorders with
treatment in their own electrical language
and by individually addressing a larger set
of neurons in brain circuits could be the
approach that proves commensurate with
the body’s most complex of organs. In the
long run, it could be the most revolutionary
aspect of electroceuticals.
Critics will argue that we underestimate
the complexity of the nervous system; the
challenges in reliably, durably and non-disruptively manipulating groups of individual
neurons and the sheer volume of neural
information flowing through these circuits.
We would argue that miniaturiz­ation and
big-data handling have been among the most
rapidly advancing areas of scientific research
in the past decade. Starting off with peripheral intervention points and simpler circuits
should also help.
There are a few noteworthy unknowns,
COMMENT
PAUL JACKMAN/NATURE
IT’S ELECTRIC
Electroceuticals deliver electrical impulses targeting the
neural circuits that regulate the body’s organs and functions.
To treat disease, an electroceutical homes in on discrete
components of the nervous system, such as individual
neurons in a specific circuit.
The electroceutical restores health by
modulating the action potentials that
flow through these neurons.
but these will be resolved only when the
approach is put to test. To what extent does
mapping of the neural language in animal
models translate to the human setting?
In which diseases will modulation of the
relevant neural circuits suffice to reverse
or control disease progression? Could the
degree of circuit redundancy or plasticity
limit the efficacy of treatment?
CATALYSING THE FIELD
At GlaxoSmithKline (GSK) and in academia,
we are confident that this field will deliver
real medicines, and we are mobilizing
resources for this journey. This summer,
the University of Pennsylvania will open
its Center for Neuro­e ngineering and
Therapeutics, which will bring together
researchers in medicine, engineering and
business. University investigators (including B.L.) are already mapping neural circuits
in humans and in cats, dogs, rodents and
other models of disease. They are also building and deploying devices that modulate
circuits at the neuronal level, using cloud
computing to mine ‘big’ neural data and
translating these technologies for use in tools
such as anti­seizure devices.
At the Feinstein Institute for Medical
Research (where K.J.T. is president), scientists are trying to establish the neural codes
that underlie diseases of immunity and
inflammation, identify intervention points
and conduct exploratory clinical work.
Results so far indicate that it is feasible to
identify and manipulate neural signals
specific to different inflammatory mediators in standard laboratory models.
At the Massachusetts Institute of
Technology, researchers (including E.S.B.)
are collaborating to map and modulate
neural circuits using technologies that range
from optogenetics6 to scalable, automated
electrophysiology10 ­— and they are distributing the genetic codes, hardware and
software necessary to put these inventions
into practice.
At GSK, we (K.F. and M.S.) are committed
to acting as a catalyst for this emerging field,
through three imm­ediate steps. The first, a
programme that will
fully fund up to 40 “Could
researchers in up to 20 the degree
external labs conduct- of circuit
ing exploratory work redundancy or
mapping disease- plasticity limit
associated neural the efficacy of
circuits, launches this treatment?”
week (www.gsk.com/
bioelectronics). Funding for the first year will
be awarded after a rapid review and approval
process that should take roughly one
month. Early findings will be shared among
researchers in this network, and intellectualproperty rights will remain with the inventors. Throughout this exploratory phase, the
network will be encouraged to shape longerterm efforts in research and development.
In December, as a second step, GSK will
hold a global forum for research leaders to
chart an integrated path forward and to collectively identify a key hurdle in the field.
After the forum, the company will launch
the third step: a US$1-million prize for
innovation, to be awarded to the group that
overcomes this hurdle.
Clearly, open innovation and flexibility
in dealing with intellectual property will be
important. As the poet Cesare Pavese said:
“If you wish to travel far and fast, travel light.
Take off all your envies, jealousies, unforgiveness, selfishness and fears.” Together we
can bring about the era of electroceuticals. ■
Kristoffer Famm is vice-president of
bioelectronics research and development at
GlaxoSmithKline, Brentford TW8 9GS, UK.
Brian Litt, Kevin J. Tracey,
Edward S. Boyden, Moncef Slaoui.
e-mail: kristoffer.h.famm@gsk.com
1. Kandel, E. R., Schwartz, J. S., Jessell, T. M.,
Siegelbaum, S. A. & Hudspeth, A. J. Principles
of Neural Science 5th edn (McGraw Hill
Professional, 2012).
2. Koopman, F. A. et al. Arthritis Rheum. 64 (suppl.),
abstr. 451 (2012).
3. Aponte, Y., Atasoy, D. & Sternson, S. M. Nature
Neurosci. 14, 351–355 (2011).
4. Huber, D. et al. Nature 451, 61–64 (2008).
5. Andersson, U. & Tracey, K. J. J. Exp. Med. 209,
1057–1068 (2012).
6. Chow, B. Y. & Boyden, E. S. Sci. Transl. Med. 5,
177ps5 (2013).
7. Rapoport, B. I., Turicchia, L., Wattanapanitch,
W., Davidson, T. J. & Sarpeshkar, R. PLoS ONE 7,
34292 (2012).
8. Wang, Z. L. & Wu, W. Angew. Chem. Int. Ed. 51,
11700–11721 (2012).
9. Alivisatos, A. P. et al. ACS Nano 7, 1850–1866
(2013).
10. Kodandaramaiah, S. B., Franzesi, G. T., Chow,
B. Y., Boyden, E. S. & Forest, C. R. Nature Meth.
9, 585–587 (2012).
1 1 A P R I L 2 0 1 3 | VO L 4 9 6 | NAT U R E | 1 6 1
© 2013 Macmillan Publishers Limited. All rights reserved
npg
Svisio/ iStock / Getty Images Plus
© 2016 Nature America, Inc. All rights reserved.
N E W S F E AT U R E
A spark at the periphery
Electroceutical devices that stimulate the peripheral nervous
system come to the fore. Emily Waltz reports.
In August, GlaxoSmithKline and Verily Life
Sciences (formerly Google Life Sciences)
announced a new venture, a company called
Galvani Bioelectronics, which, with up to more
than $700 million in funding from the two
partners, will develop miniaturized electronic
devices for peripheral nerve stimulation. This
is the latest in a recent string of successes that
the fledgling electroceuticals field has racked
up, which include a breakthrough first-inhuman clinical success in rheumatoid arthritis
by SetPoint Medical of Valencia, California,
and an approval by the UK’s National Health
Service of a handheld vagus nerve stimulator
for migraine and cluster headache, developed
by ElectroCore of Basking Ridge, New Jersey.
Unlike electrical therapies of the past, these
new commercial efforts aim to harness knowledge of neurophysiology and molecular mechanisms to guide the design of neuromodulation
devices. Together with an ongoing influx of
funding from several large organizations
(Box 1), biomedical engineers are generating
the tools and maps they need to build better,
more targeted neurostimulation devices.
Striking a nerve
Devices that harness electrical impulses have
904
been used for decades to improve health and
save lives—the most common being pacemakers and defibrillators. Deep-brain stimulation is fairly widely used to treat people with
Parkinson’s disease, sacral nerve stimulation
is used to restore bladder control and spinal
cord stimulation has been applied to treat pain.
Most work in peripheral nerve stimulation
(PNS) has focused largely on the voluntary or
somatic nervous system in an effort to restore
movement in people who have paralysis due
to spinal cord injury or stroke. But that field
has seen little commercial activity due to its
relatively small market size.
What’s got entrepreneurs excited about
PNS for the autonomic nervous system is that
it has the potential to treat a wide range of diseases that are currently underserved by oral
or injected drugs. But instead of circulating
throughout the body and causing side effects
as drugs do, PNS can send a message straight
to—and only to—the target. That’s led some in
the field to dub such therapies ‘electroceuticals’.
In fact, some PNS researchers are targeting the
same mechanisms as those targeted by blockbuster commercial drugs.
The peripheral nervous system’s involuntary, or autonomic, nerves play a large role in
organ function and immune responses and in
the body’s inflammatory, respiratory, cardiovascular and urinary systems. Just how central a role it plays in these areas continues to
become more apparent. “It’s starting to be very
well established that the immune system can
be controlled through the nervous system. Ten
years ago, people may have said that was crazy,”
says Kris Famm, vice president of bioelectronics R&D at GlaxoSmithKline in London.
The peripheral nervous system carries
electrical impulses to and from the brain and
spinal cord via action potentials. The signals
travel along neural networks in different
temporal patterns, like a drum beat or Morse
code. These patterns dictate chemical and biological changes throughout the body. When
those communication signals are disrupted
or don’t fire properly, a myriad of things can
go wrong.
Electrical stimulation enables researchers to
hack into the nervous system, and potentially
restore or correct communication. In many
such devices, a pulse generator sends electrical impulses through a lead to electrodes that
are placed on or near a nerve. The pulse generator ramps up intensity of the stimulation to
a threshold that causes neurons to fire. Those
artificially induced action potentials are indistinguishable from those produced by the body.
Compared with the pharmaceutical drug
screening process, developing a stimulation
therapy can be done much faster. Theoretically,
once the mechanisms are identified and signal patterns of the healthy target nerves are
recorded, “the therapeutic intervention is
VOLUME 34 NUMBER 9 SEPTEMBER 2016 NATURE BIOTECHNOLOGY
N E W S F E AT U R E
npg
© 2016 Nature America, Inc. All rights reserved.
Box 1 Funders touch a nerve
The NIH, DARPA and GlaxoSmithKline (GSK) are all funding projects that seek to identify
the functions of neural circuits, decode the signals traveling through nerve fibers and
develop new tools to interface with the nervous system. “When you invest…in the science
and the mechanistic underpinnings, it opens a lot of doors,” says Juan Pablo Mas, a
partner at GSK’s bioelectronics venture capital (VC) arm. “It opens doors to financing.
It helps brings in money when you can answer questions like, Why should this work?
How does it work? And it opens the door to easier conversations with the FDA and payers
because you can provide them a scientific rationale [in addition to] clinical data.”
On top of investments from the NIH, DARPA and GSK, two large centers devoted to
bioelectronics are being built—one at the Karolinska Institute in Stockholm, Sweden,
and the other at the Feinstein Institute in Manhasset, New York. The state of New York
in March committed $50 million to its center, and the investment was matched by $600
million in private funding.
Through its SPARC program, the NIH has committed $248 million to the field over
seven years and plans to announce by October the recipients of the core of that funding.
The bulk will go toward anatomical and functional mapping of organs and associated
nerves. The agency will also fund projects aimed at improving tools such as electrodes
that better interface with nerves, and ways of recording how end organs change when
associated nerves are stimulated. A substantial portion will also go toward projects with
industry to study new uses for existing neuromodulation devices.
DARPA’s nearly $60-million program called ElectRx is similarly focused on the
physiology of nerve circuits and unconventional technologies that record and interact with
peripheral nerve targets. The agency wants to apply these efforts toward inflammatory
diseases and mental health, particularly post-traumatic stress disorder, says Weber,
at DARPA. The agency in March announced another PNS program—the ‘Targeted
Neuroplasticity Training’ program—to explore how stimulation of peripheral nerves can
enhance learning processes in the brain. It plans to announce the winning research
proposals by November.
immediately possible,” says Famm, at GSK.
“The neural signal you record is in essence
also the treatment you want to put back in. In
contrast, if you identify a target for molecular
drug discovery, that’s only the beginning of
a long journey to then find your molecule to
affect that target.”
Focusing on mechanisms
That’s the vision, and a handful of companies have brought PNS devices to market
(Table 1). But most of those don’t begin
to hack the nervous system to the level of
specificity that researchers would like to see.
“Stimulation today is greatly simplified,” says
Milton Morris, CEO of the Santa Clara-based
neuromodulation startup NeuSpera Medical.
For most devices, it’s not clear which of the
tens of thousands of fibers within a nerve are
being stimulated, nor exactly what chemical
or molecular effect the stimulation is having. And the patterns of action potentials
produced by commercial devices likely don’t
come close to mimicking those of a healthy
nerve fiber. “The nerve is not homogeneous.
It has a substructure that is incredibly complicated and it’s largely being ignored in the
design of these therapeutic strategies,” says
Gene Civillico, who runs the US National
Institute of Health’s (NIH) peripheral
nerve stimulation funding program called
Stimulating Peripheral Activity to Relieve
Conditions, or SPARC.
Stimulation devices lack this specificity
largely because of the dearth of knowledge
of the physiology and function of peripheral
nerve circuits. “You have a little glimmer of
understanding here and there. But in general,
we don’t have a map,” says Famm. Rather than
understanding the role of individual fiber
tracks and neural circuits, we’re relying on
gross descriptors of the peripheral nervous system, such as sympathetic and parasympathetic,
adds Kevin Tracey, director of the Feinstein
Institute for Medical Research in Manhasset,
New York and co-founder of SetPoint Medical.
That’s not the way to build a stimulation device,
he says.
There are some hard limits to how selectively
investigators can stimulate within a nerve. In
the vagus nerve, for example, there are about
100,000 fibers at the level of the neck. Fibers
large in diameter, such as A and B fibers, can
be ten to twenty times larger than C fibers, and
fire at lower amplitudes because their thresholds are lower. Stimulating C fibers without
first activating the large-diameter fibers is
practically impossible with current technology, says Doug Weber, of the US Department
of Defense’s research arm—the Defense
NATURE BIOTECHNOLOGY VOLUME 34 NUMBER 9 SEPTEMBER 2016
Advanced Research Projects Agency (DARPA)
in Washington, DC—who runs the agency’s
PNS projects.
The unknowns and technological shortcomings have forced companies to shoot rather
randomly and broadly at a nerve and hope
the therapy reduces symptoms. That’s a far cry
from the drug industry’s approach of finding
and then targeting a known mechanism underlying a disease or disorder. “The pharma industry begins with a molecular target and moves to
screening for drugs,” says Tracey. Bioelectronic
medicine should begin the same way, he says.
Then, instead of screening drug candidates,
companies essentially screen stimulation
parameters that will recruit the groups of nerve
fibers associated with that mechanism. “That’s
the challenge to how the device industry has to
move forward,” he says. “You can’t just build a
device and stick it in a bunch of people and see
what happens when you turn it on.”
The push for using mechanisms as a guide
to building stimulation devices has been
echoed among researchers both in academia
and industry. Most of these new resources are
being focused on peripheral nerve stimulation,
rather than brain or spinal cord, partly because
the opportunity to understand the mechanisms
underlying PNS are more accessible than those
of the central nervous system. The degree of
complexity and integration of the cells of the
brain—thousands of connections for each cell
and billions of cells—is mind blowing, says
Moncef Slaoui, GSK’s chairman of global vaccines, who was instrumental in establishing
GSK’s investments in the field of bioelectronics.
“And honestly, it’s a miracle that when you do
[deep brain stimulation] people have a benefit
on the other side of it. Because you have no
idea what you’re doing,” he says. Deciphering
those connections “is going to take an incredible amount of time” and will probably have to
be hammered out in academic settings, he says.
Nerves of the periphery, on the other hand,
are far less complex, making the decision to
tackle those first a pragmatic one for companies. “Where the nerve hits the organ, you’re
talking about several hundred or thousand
nerve fibers…as compared to billions in the
brain,” says Slaoui. Working in the periphery
also allows researchers to target organs in a
way that’s more specific than can be accessed
through the central nervous system, adds
Famm at GSK.
Getting down to size
It helps that a large pharma company has publicly announced its interest in the field1. Over
the last three years, GSK has set up a 30-person R&D unit, nearly 50 external research collaborations and a $50-million venture capital
905
N E W S F E AT U R E
npg
© 2016 Nature America, Inc. All rights reserved.
Table 1 Selected companies developing electrical stimulation therapies for the peripheral nervous system
Company
Therapeutic targets
Technology
LivaNova, merger
of Cyberonics and
Sorin
Epilepsy, depression
Implanted, cervical vagus nerve stimulator. Stimulation runs at preset Epilepsy: EU 1994; US 1997
intervals, and can be turned on at the onset of seizure by the user or
Depression: US 2005; EU 2001
automatically when a sensor detects signs of a seizure. Pulse generator implanted in the chest is connected by a lead to electrodes cuffed
to the vagus nerve in the neck.
EnteroMedics (St.
Paul, Minnesota,
USA)
Obesity, type 2 diabetes
Implanted device that aims to give the user a feeling of fullness by
blocking the signaling that travels from the brain to the gut along
the vagus nerve. Electrical current is applied to the two trunks of
the vagus nerve below the diaphragm where the esophagus joins the
stomach.
US 2015, EU 2011
ElectroCore
Migraine, cluster headache
Handheld, non-invasive vagus nerve stimulator. Device is held to
neck and stimulation is applied by user. Believed to enhance production of inhibitory neurotransmitters and reduce excitatory neurotransmitters, which has been implicated in primary headaches.
EU 2011, UK 2016
SetPoint Medical
Crohn’s disease, rheumatoid
arthritis and other immunemediated inflammatory
conditions
Inhibits TNF and other inflammatory cytokines by stimulating a
subset of vagus nerve fibers at the neck. Pulse generator, lead and
electrodes are condensed into one small device that sits at the cervical level of the vagus nerve.
None
MicroTransponder
(Austin, TX, USA)
Tinnitus, stroke
Implanted cervical vagus nerve stimulator paired with external training stimuli. Aims to release neurotransmitters, promoting plasticity
in the brain that allows enhanced learning during training. In tinnitus
patients, stimulation is timed with auditory signals in an attempt
to decrease hyperactivity in the auditory cortex that causes ringing
in the ears. In stroke patients, stimulation is timed with physical
therapy movements to treat upper limb deficits.
None
NeuSpera Medical
Undisclosed
Powering technology that enables placement of ultra-miniaturized
stimulation devices through an injection at deep tissue targets.
None
CVRx
Hypertension and heart
failure
Activation of carotid baroreceptors using an implanted device. The
stimulation technique generates afferent signals that travel via the
carotid sinus nerve to the brainstem, modulating the autonomic
nervous system in ways that improve heart and vascular function and
reduce excessive blood pressure
EU: heart failure, 2014, hypertension, 2011
GlaxoSmithKline
Targeting auto-immune,
endocrine and metabolic systems through neuromodulation; not limited to particular
disease areas
(VC) arm, all devoted to bioelectronics, and
specifically PNS. In June, the company notified
three independent research groups that they
were finalists for a $1-million prize GSK had
first dangled in front of the field in 2013. The
first group to show that it can record, stimulate
and block neural signals in one of four organ
systems will win.
GSK’s announcement August 1 that it is
partnering with Verily to establish a new bioelectronics company is a big deal for the field
of PNS, in that it brings together two pharma
and tech powerhouses. Verily, an Alphabet
(formerly Google) company, will contribute
its technical expertise in the miniaturization
of low-power electronics, device development,
big data analytics and software development.
“It became clear to us we needed a strategic
partner that would bring the engineering,”
says Slaoui. In looking for that partner, GSK
had engaged most of the big players in the tech
space and found that many were intrigued but
were also very concerned by the regulatory
requirements of putting their devices in the
human body. With Verily, “there was a com906
Regulatory approvals
US: Received from FDA in 2014 a
humanitarian device exemption for
hypertension, in certain qualifying
patients
None
Conducting basic research to understand physiology and neural circuits responsible for disease states. Developing engineering tools and
devices to interface with the peripheral nervous system. Established
a joint venture called Galvani Bioelectronics that will focus on miniaturization of electronics.
plete alignment of vision for bioelectronic
medicine and appetite for risk and willingness
to give this a chance,” he says.
GSK’s VC arm, called Action Potential
Venture Capital, is also investing in companies
with enabling technologies, such as micro-electronics and wireless powering. Such advancements will be key to interfacing PNS systems
with the tiny nerves that adjoin end organs,
and to appeasing patients. Most stimulation
devices on the market require the implantation of a pacemaker-sized pulse generator, plus
a lead and an electrode cuff that goes around
the entire nerve.
Recruiting fewer, more targeted fibers
could reduce side effects, such as vomiting
and hoarseness—common with vagus nerve
stimulation—and prevent neural circuits
from becoming desensitized from overstimulation. Such improvements could also reduce
the power drawn from the pulse generator,
which enables such devices to be smaller. In
terms of neurostimulation, “there is a general
sense that we are probably overdoing it,” says
Morris at NeuSpera. The moment one learns
which fibers are the ones that need to be stimulated to drive a particular effect, “everything
else is an overdose,” he says. NeuSpera has
licensed from Stanford University a powering
technology that enables placement of ultraminiaturized stimulation devices at deep tissue targets in the body through an injection,
rather than a surgical incision. That technology, called midfield powering, can wirelessly
power injected stimulation devices by propagating electromagnetic waves within tissue.
The company plans to license the powering technology and also apply it to its own
ultra-miniaturized stimulation device for
an undisclosed therapeutic indication. The
company’s investors include GSK’s Action
Potential Venture Capital and New Yorkbased Windham Venture Partners.
Some of the newer, smaller players in the
neuromodulation field are putting more time
and resources into basic science. That’s the
strategy being taken by SetPoint Medical,
which is developing an implantable vagus
nerve stimulator to treat inflammatory diseases such as Crohn’s disease and rheumatoid
VOLUME 34 NUMBER 9 SEPTEMBER 2016 NATURE BIOTECHNOLOGY
npg
© 2016 Nature America, Inc. All rights reserved.
N E W S F E AT U R E
arthritis. The company was co-founded by
Kevin Tracey at Feinstein and is also backed
by Action Potential.
SetPoint’s technology is built on the grounds
that the nervous system regulates immune
responses and can inhibit inflammation2.
SetPoint, Tracey and other academic collaborators reported in July that they had demonstrated in humans that they could inhibit the
production of tumor necrosis factor (TNF) and
other inflammatory cytokines by stimulating
a subset of 4,000–5,000 vagus nerve B fibers3.
TNF is the target of several rheumatoid arthritis drugs on the market, and cytokines are the
target of a $50-billion industry.
The experiment was novel in that the underlying neurophysiology and mechanism of
action of the neural circuit were characterized
first, and that guided the researchers where and
how to stimulate. The results are “very strong,”
says Warren Grill, a biomedical engineer at
Duke University in Durham, North Carolina,
who did not participate in the study, and has
been a consultant for SetPoint.
Tracey’s group is now headed back to the
laboratory. He wants to see the number of target nerve fibers narrowed from B fibers to a
smaller subgroup. He also wants to know what
the patterns of action potentials in healthy
nerve fibers look like so he can replicate them
with electrical stimulation in the nerves of people with rheumatoid arthritis. “That’s what I
have ten people working on right now,” he says.
Commercial limits on R&D
That kind of splicing is severely limited by
anatomical access. And many of the tools
that could less invasively identify the role of
individual populations of fibers are still in
their infancy, and limited to animal models.
Optogenetics, in which light is used to control
neurons that have been genetically modified
to express light-sensitive ion channels, is one
option. Chemogenetics and magnetogenetics,
in which a custom small molecule or a magnetic field, respectively, are used to control
engineered receptors expressed in subsets of
cells are additional options.
Devoting the resources needed to get that
level of detail may be beyond what many companies are able or willing to bankroll. “There
really isn’t a clear way to stimulate only a subset of the neurons without doing a tremendous
amount of splicing out nerves and testing each
one,” says J.P. Errico, chief science and strategy
officer at ElectroCore, which is developing a
handheld, non-invasive vagus nerve stimulation device to treat migraines, cluster headaches and other conditions. “I don’t know how
practical that will be in the commercial environment or real-world clinical practice.”
ElectroCore has focused much of its R&D
dollars on understanding how stimulation
of the vagus nerve plays a role in modulating
neurotransmitters, how that affects inflammation and what the clinical effects of that are.
Stimulation of the vagus nerve seems to enhance
production of inhibitory neurotransmitters,
such as serotonin, acetylcholine, norepinephrine
and GABA, helping bring them back to normal
levels, says Errico. That helps balance the levels
of excitatory neurotransmitters, such as glutamate, which has been implicated in primary
headaches and other disorders.
“You have a little glimmer of
understanding here and there.
But in general, we don’t have a
map,” Kris Famm, GSK.
ElectroCore and its academic research partners have yet to pinpoint the mechanisms by
which non-invasive vagus nerve stimulation
alleviates migraines and other conditions, but
is “absolutely committed” to finding out, says
Errico. But the company has stopped short
of teasing out the role and signal patterns of
individual fiber groups. “I am not convinced
that this line of research will be fruitful,” says
Errico. “Such optimization work will have to
be left to others as we focus our resources on
clinical outcomes and a better understanding
of the mechanism of the signal we and others
are using.”
The stimulation from ElectroCore’s device,
called GammaCore, is applied externally,
through the skin at the neck using a handheld device. It received Europe’s CE mark of
approval in 2011, and a green light in April
from the UK for the device to be prescribed
within the country’s National Health Service.
ElectroCore awaits approval in the US. The
US Food and Drug Administration (FDA)
considers the device class III, which gets the
greatest scrutiny, but is reviewing it through
its de novo device pathway, which could result
in future applications being reviewed as class
II, Ericco says.
It is possible to see success in the clinic with
a fairly primitive understanding of what the
stimulation is actually doing. Cyberonics,
which in 2015 merged with Sorin Group to
form London-based LivaNova, developed one
of the earliest implantable PNS devices, and
received approvals in Europe and the US in
the mid-1990s for treatment of epilepsy. The
company’s device has been implanted in more
than 85,000 people, most of them with epilepsy,
and has substantially reduced seizures and
improved quality of life for many of them. In
NATURE BIOTECHNOLOGY VOLUME 34 NUMBER 9 SEPTEMBER 2016
one study of 436 patients, two-thirds saw their
seizures reduced by half or more4.
Yet the company does not know with any certainty which groups of fibers within the vagus its
device stimulates or how that affects seizures.
“It is something we will continue to explore, but
[it] has not slowed our efforts as the therapy is
clearly performing well with the vast majority of
the patients,” says Eric Relkin, director of global
marketing at LivaNova. “Our focus is working with the clinician community to improve
the quality of care as well as committing R&D
resources to continue to improve patient results.”
Traditional device makers working in the
neuromodulation space, such as Minneapolisbased Medtronic and Boston Scientific in
Marlborough, Massachusetts, similarly have
built devices based on symptomatic endpoints5. “The work to establish the mechanisms and to validate the approach hasn’t been
appreciated so much by medical device companies,” says Mas at Action Potential. “They’ve
been focused more on empirical evidence:
does it work, is it safe, if so, we run the trial, we
get approved, we commercialize and develop
a market around it.” Adds Civillico, “Their
very nicely engineered solutions have gotten
them pretty far without a heavily built-up basic
research effort.”
From empiricism to mechanism
Success will continue to be limited, however,
without digging deeper into the mechanisms.
If a device isn’t working in a third of patients
and researchers have no idea why, it’s hard to
improve on that therapy. “For every successful
trial you have a number of unsuccessful trials
where something was tried and it did not meet
its primary endpoint,” says Civillico. “In many
cases, it didn’t meet its primary endpoint in
a very scientifically tantalizing way, in that it
worked for some of the people really well, and
didn’t work for a lot of the people. So that creates the possibility that there’s a huge room for
improvement if we understood why it’s working sometimes and not other times.”
That point was illustrated in a series of
recent clinical trials that put one company out
of business. The trials tested vagus nerve stimulation as a treatment for heart failure. Animal
studies conducted by different research groups
looked promising. Although the mechanisms
were unclear, vagal nerve stimulation seemed
to have positive effects on the heart, such as
reduced ventricular arrhythmias, anti-inflammatory effects, a normalizing of nitric oxide
synthase levels and general improvement of
heart failure symptoms6,7.
Encouraged by the preclinical results,
BioControl Medical, based in Yehud, Israel,
developed a vagus nerve stimulator called
907
npg
© 2016 Nature America, Inc. All rights reserved.
N E W S F E AT U R E
CardioFit. The device aimed to stimulate B
fibers in the vagus nerve that travel to the
heart. In a couple of pilot studies involving
small numbers of patients, the CardioFit
device was found to be safe and improved
patients’ quality of life, exercise capacity and
their hearts’ blood pumping efficiency8.“The
clinical endpoints and heart failure barometers were all going in the right direction,”
says Ehud Cohen, former CEO of BioControl
Medical.
BioControl ramped up its next trial to
over 700 patients. Called INOVATE-HF, the
trial compared the efficacy of the CardioFit
device to standard care in people with heart
failure. But four months after investigators
completed enrollment, in December 2015,
the company had to stop the trial. There were
no reductions in hospitalizations or deaths—
the primary endpoints—compared with the
control group. BioControl Medical shut down
operations in July.
Boston Scientific’s vagal nerve stimulator
trial for heart failure, called NECTAR-HF, did
not translate well from preclinical experiments
either. In the 96-patient study, effects on cardiac function were not significant compared
to the control group, although some patients
reported better quality of life and exercise
capacity9.
What went wrong in the INOVATE and
NECTAR studies is unclear. Each study used
different stimulation parameters. And in none
of the clinical studies did investigators know
exactly which fibers they were stimulating. “I
think it’s a matter of not using the right levels of
stimulation,” says Benjamin Scherlag, a cardiac
physiologist at University of Oklahoma College
of Medicine in Oklahoma City.
908
Cohen at BioControl Medical says
INOVATE-HF’s failure was a result of the
design of the study. Partway through, the
company realized that about half of the
patients—those who hadn’t responded to a
first-line electrical therapy for heart failure
called Cardiac Resynchronization Therapy
(CRT)—were also not responding to vagus
nerve stimulation. That’s supported by research
by Minneapolis-based CVRx, which is developing a vagal nerve stimulator device for
hypertension and heart failure. The company
in 2015 reported that results were particularly
positive in heart failure patients who had not
been treated with CRT.
But in BioControl’s case, the company had
not built into the trial the flexibility to steer
away from this subpopulation of patients midstudy. “In retrospect, we could have done a
clinical design that is more adaptive,” Cohen
says. Device companies need to take note of the
more sophisticated clinical studies common in
the pharma industry, he says.
For LivaNova, a missed endpoint in a clinical trial may have set back not just the company, but possibly the field of vagus nerve
stimulation. In 2005, the company (then called
Cyberonics), received FDA approval for its
vagal nerve stimulator device in people with
treatment-refractory depression, giving the
company a large potential market outside of
epilepsy. Several payers initially reimbursed
for the device, and it was implanted in several
thousand people. But almost two years later,
the US Centers for Medicare and Medicaid
Services (CMS) announced it would deny coverage of the device.
The decision was based on a primary endpoint that the company had missed in a clini-
cal trial. After CMS’s decision, private payers
stopped covering the device for depression,
leaving LivaNova without that market. “I think
you would have seen a more rapid growth in
PNS if the company were able to convince
CMS to render a favorable coverage decision,”
says Morris at NeuSpera, who was previously
senior vice president of R&D at Cyberonics.
“The company had a couple of false starts and
had to redefine itself ” by turning to the epilepsy market, he says.
LivaNova says it continues to work with
investigators and with CMS to find a pathway
for access to its vagal nerve stimulation therapy
in people with depression. The market is large
and many people with depression don’t like
or don’t respond to commercial drugs for the
condition. PNS offers an alternative for those
people, and LivaNova says it knows much more
about the technology today than when its pivotal
trial was designed. “The results [in depression]
I’ve seen are quite compelling—particularly the
long-term ones,” says Morris. Presenting better
studies and more knowledge of the mechanisms
may help get these payers on board.
Emily Waltz, Nashville, Tennessee
1. Famm, K., Litt, B., Tracey, K.J., Boyden, E.S. & Slaoui,
M. Nature 496, 159–161 (2013).
2. Borovikova, L.V. et al. Nature 405, 458–462 (2000).
3. Koopman, F.A. et al. Proc. Natl. Acad. Sci. USA 113,
8284–8289 (2016).
4. Elliott, R.E. et al. Epilepsy Behav. 20, 57–63 (2011).
5. Slavin, K.V. (ed.). Peripheral Nerve Stimulation
(Karger, 2011).
6. Schwartz, P.J., La Rovere, M.T., De Ferrari, G.M. &
Mann, D.L. Circ. Heart Fail. 8, 619–628 (2015).
7. Murray, A.R., Atkinson, L., Mahadi, M.K., Deuchars,
S.A. & Deuchars, J. Auton. Neurosci. http://dx.doi.
org/10.1016/j.autneu.2016.06.004 (2016).
8. De Ferrari, G.M. et al. Eur. Heart J. 32, 847–855
(2011).
9. Zannad, F. et al. Eur. Heart J. 36, 425–433 (2015).
VOLUME 34 NUMBER 9 SEPTEMBER 2016 NATURE BIOTECHNOLOGY
Review
https://doi.org/10.1038/s41586-018-0076-4
Restoring vision
Botond Roska1,2,3* & José-Alain Sahel4,5,6,7*
Restoring vision to the blind by retinal repair has been a dream of medicine for centuries, and the first successful
procedures have recently been performed. Although we are still far from the restoration of high-resolution vision, stepby-step developments are overcoming crucial bottlenecks in therapy development and have enabled the restoration of
some visual function in patients with specific blindness-causing diseases. Here, we discuss the current state of vision
restoration and the problems related to retinal repair. We describe new model systems and translational technologies,
as well as the clinical conditions in which new methods may help to combat blindness.
R
eversing or slowing loss of vision has been an aim of ophthalmology
since this discipline emerged1 (Fig. 1). Surgical replacement of
the lens has restored visual acuity in patients with cataracts2 and
the control of vascular leakage can stabilize or improve vision in those
with wet age-related macular degeneration3. These treatments have had
major social and economic impacts4–7. Despite these successes, there are
still an estimated 285 million vision-impaired and 39 million blind people
around the world8. Many conditions that cause blindness in under­
developed countries are treatable because they involve refractive errors,
cataracts, infections, or nutritional states, and the main problem is the
delivery of existing care. The delivery of care in developed countries is not
a major issue, but there are still no treatments for most blinding diseases,
including advanced glaucoma, atrophic macular degeneration, advanced
diabetic retinopathy, myopia, and monogenic retinal degeneration.
Here, we discuss the current state of vision restoration in blinding retinal
diseases and the general problems related to retinal repair. We describe
new model systems and translational technologies that have opened up
new opportunities and given momentum to combating blindness—with
human-derived retinal organoids and gene therapy representing the key
advances. Finally, we discuss specific clinical conditions in which these
methods may help to restore vision or slow retinal degeneration in the
next decade.
in blind patients if the optic nerve is intact. A third example is the
transplantation of retinal pigmented epithelial cells behind the retina,
which has produced possible visual improvements or stable vision in
a few patients with age-related macular degeneration21,22 or Stargardt
disease21, a monogenic form of macular degeneration23. Controlled
studies with larger groups of patients are needed to verify the visual
benefits, but the safety of such transplants, when done properly24, is
notable.
In recent years, we have not only seen the emergence of new therapies
but have also learned new lessons about the state of visual pathways
during human visual development. In a healthy visual system, lightdriven developmental events occur during the first years of life, the
so-called critical period25. Whether restoration of retinal output in
congenitally blind patients after the critical period could ever benefit
their vision has been investigated. A set of patients with congenital
cataracts that prevented the formation of a visual image on the retina
from birth received lens replacement surgery at or after eight years of
age. This resulted in the acquisition of pattern vision26–28 and demonstrated that a lack of high-resolution vision during the first few years
of life does not necessarily prevent the acquisition of useful vision after
treatment.
Where we are today
Apart from these major developments in therapies and in our understanding, advances in clinical care to treat blinding diseases have perhaps
been slow from the point of view of the patient. Why is it difficult to
slow down the progression of human vision loss or to restore vision
once it has been lost? There are five central problems.
First, without intervention, intrinsic regeneration of the mammalian
retina is weak or entirely absent. This is in contrast to some other animals,
such as fishes, in which the generation of new retinal cells and regeneration of portions of a damaged retina occur after injury even in
adulthood29.
Second, the retina is a biological computer made up of about 100
different cell types that form specific synaptic connections and reside in
functionally distinct microcircuits30,31 (Fig. 2). Retinal cell types differ
in their physiology, morphology, and patterns of gene expression32,33.
As a consequence of this cell-type diversity, many diseases of vision are
broadly or narrowly cell-type-specific because the disease-causing gene
is expressed in a subset of cell types or because vulnerability is cell-type
specific. To reduce side effects, the ideal treatment should also be
Lens replacement after cataract and the control of fluid leakage from
retinal blood vessels in wet age-related macular degeneration are
now part of standard clinical care2,3. Over the past decade, we have
witnessed major new developments in medical technology that have
led to vision restoration or the slowing of vision loss in blinding
diseases originating from retinal cells. A key example is gene therapy
for a form of Leber congenital amaurosis (LCA)9–11 that produced
significant improvements in performance in a visually guided task in
young adult patients11,12, as well as increased activation13 and long-term
structural plasticity14 of the visual cortex. This demonstrated the safety
of ocular gene therapy and the potential benefit to patients15. Initial
results from other gene therapy trials, involving conditions such as the
hereditary blinding disease choroideraemia, have also indicated possible benefits16,17. A second example is electric stimulation of the retina in
adult patients with photoreceptor degeneration18, which evokes visual
percepts and in some cases has provided form vision19,20. These results
imply that information can flow from the retina to higher visual areas
The problem of retinal repair
1
Institute of Molecular and Clinical Ophthalmology Basel, Basel, Switzerland. 2Neural Circuit Laboratories, Friedrich Miescher Institute, Basel, Switzerland. 3Department of Ophthalmology,
University of Basel, Basel, Switzerland. 4Department of Ophthalmology, The University of Pittsburgh School of Medicine, Pittsburgh, PA, USA. 5Institut de la Vision, Sorbonne Université, INSERM,
CNRS, Paris, France. 6CHNO des Quinze-Vingts, CIC INSERM-DGOS 1423, DHU Sight Restore, Paris, France. 7Department of Ophthalmology, Fondation Ophtalmologique Rothschild, Paris, France.
*e-mail: botond.roska@iob.ch; j.sahel@gmail.com
1 7 M AY 2 0 1 8 | VO L 5 5 7 | NAT U R E | 3 5 9
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
insight
Review
a
b
Optic nerve head
(glaucoma, LHON)
Macula
(AMD, Stargardt)
Choroid
Fovea
Periphery
(LCA, RP)
RPE
POS
PIS
ELM
ONL
OPL
INL
IPL
GCL
RNFL
ILM
Vitreous
Vasculature
(DR)
Periphery
Fig. 1 | The human retina in vivo. a, The retina as seen at the back of
eye by an ophthalmologist using an ophthalmoscope. Four objects can
be detected: the vasculature, where diabetic retinopathy (DR) arises;
the macula, the centre of which is called the fovea (age-related macular
degeneration (AMD) and Stargardt disease affect this area); the optic
nerve head, which is affected by glaucoma and Leber hereditary optic
neuropathy (LHON); and the retinal periphery, which denotes the entire
retina outside the macula, where Leber congenital amaurosis (LCA) and
retinitis pigmentosa (RP) start. b, Cross sections of the retina (along the
line shown in green in a) visualized using optical coherence tomography
(OCT). Top: area of the macula. The fovea is marked by thinning of the
retina. Bottom: retinal periphery. Different layers can be identified in
an OCT image around the fovea. In general, cell body-rich layers and
layers filled with liquid are darker. Layers from the vitreous towards the
choroid (the layer behind the retina that is filled with capillaries) are:
internal limiting membrane (ILM, the end feet of Müller cells on the
vitreal side), retinal nerve fibre layer (RNFL, ganglion cell axons), ganglion
cell layer (GCL, cell bodies of ganglion cells), inner plexiform layer (IPL,
connections between ganglion cells and other cells in the inner retina,
such as amacrine and bipolar cells), inner nuclear layer (INL, cell bodies of
inner retinal neurons such as amacrine and bipolar cells), outer plexiform
layer (OPL, connections between bipolar cells and photoreceptors),
outer nuclear layer (ONL, photoreceptor cell bodies), external limiting
membrane (ELM, the end feet of Müller cells on the choroidal side),
photoreceptor inner segments rich in mitochondria (PIS), photoreceptor
outer segments (POS), retinal pigmented epithelial cells (RPE), choroid.
Distinct but fewer layers can also be distinguished in an OCT image of
the retinal periphery, where the retina is thinner. White arrows show the
direction of light.
cell-type specific, but it is difficult to achieve such treatment even
in non-human models. A second consequence of cell-type diversity
is that both the ability of gene therapy vectors to enter cells and the
efficiency of transcription from vectors are cell-type-specific. Thus,
vectors need to be tailor-made for different cell types34. Furthermore,
specific synaptic connectivity in the retina has consequences for cell
therapy. Repair of diseased retinas by introducing additional cells of a
given type requires that the added cells connect to their natural partners,
which is not efficient in adults35.
Third, it is often assumed that findings in mice, the most common
mammalian disease model, can be directly translated to humans.
Although the overall cellular architecture of the retina is similar in
mice and humans, the two species exhibit certain differences36 that can
complicate translation. There are differences in retinal cell-type composition, in cell-type-specific gene expression, and in the organization of
the retina at the macroscopic level. For example, the human retina contains midget ganglion cells and their related retinal circuitry37, but the
mouse retina apparently does not. Midget cells are thought to mediate
high-resolution image-forming vision38, and preventing the degeneration of midget cells in patients with glaucoma or diabetic retinopathy
would have a major effect on retarding disease progression—but there
is no mouse model. Usher I genes are examples of differences in celltype-specific gene expression between mice and humans. Mutant Usher
I genes lead to Usher I syndrome, an inherited form of blindness paired
with deafness. Mouse photoreceptors express Usher I proteins at much
lower levels39,40 than human photoreceptors, and lack the cellular
compartment in which the proteins are localized in human photoreceptors39. Not surprisingly, mice bearing the same mutations as patients
with Usher I syndrome have normal vision39. Finally, a striking difference in the organization of the retina at the macroscopic level between
mice and humans is the presence in humans of a small compartment
at the centre of the retina called the fovea. This is absent in mice41; in
fact, primates are the only mammals with a fovea. The human fovea
makes up only 0.2% of the retina42 but is necessary for high-resolution
colour vision38, and is essential for important everyday tasks such as
recognizing faces and reading43. Age-related macular degeneration3
and Stargardt disease23 both initially affect the fovea.
Fourth, the eyes of primates have a thick membrane, the so-called
inner limiting membrane, between the retina and the inner compartment of the eye (the vitreous). This membrane limits diffusion and thus
restricts the efficacy of intravitreally delivered gene therapy vectors,
such as adeno-associated viral vectors (AAVs)44. Newly constructed
AAVs show slightly increased peripheral expression, but access is still
limited45. However, the limitation of diffusion across the inner limiting
membrane is lower in the foveal region of the retina and gene transfer
by intravitreal AAV injections around the fovea in primates has been
successful46. The difficulty in transducing the peripheral retina with
gene therapy vectors is important, as intravitreal injections are easier
for ophthalmologists and safer to perform than sub-retinal injections45.
Fifth, the surface area of the human retina is very large (about
1,000 mm2). It is seventy times larger than that of the mouse (about
15 mm2)47,48. In contrast to treatments with small molecules, which
distribute across the retina when injected into the eye or delivered via
the bloodstream, scaling up gene delivery to reach all cells of a given
type across the human retina is a daunting task.
New model systems
Given these difficulties related to retinal repair and the differences
between the retinas of humans and other species, how can we understand human retinal diseases and develop new therapies? Historically,
researchers have used genetically tractable small animal models and
cell lines to study retinal diseases and to test treatment strategies. While
work on these systems will continue, three new retina models have
become available in recent years that may increase our understanding
of human retinal disease and benefit the search for new therapies
(Table 1).
3 6 0 | NAT U R E | VO L 5 5 7 | 1 7 M AY 2 0 1 8
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Review
a
b
insight
Input:
Photoreceptors
(AMD, Stargardt,
LCA, RP)
Horizontal cells
Bipolar cells
(CSNB)
Amacrine cells
(CN)
Ganglion cells
(glaucoma, LHON)



Output:
30
4
1
2
3
Fig. 2 | Cell types, circuits and computations performed by the
vertebrate retina. a, Retinal cell types (about 100 types) can be divided
into a few major classes. The characteristic locations of the cell bodies,
gross morphology and connectivity of some of these classes in the
retinal periphery are shown. Photoreceptors (rods in cyan, and cones in
light blue) capture light. Photoreceptors pass information to inhibitory
horizontal cells (orange) and excitatory bipolar cells (dark blue).
Horizontal cells feed back to photoreceptors, while bipolar cells pass
information further to inhibitory amacrine cells (red) and excitatory
ganglion cells (purple). Amacrine cells either feed back to bipolar cells or
pass information to ganglion cells. Ganglion cells are the output neurons
of the retina, sending information to the rest of the brain via their axons,
which come together to form the optic nerve. This circuit layout is highly
conserved across vertebrates. The retinal circuit is similar in the fovea,
but the locations of the cell bodies are different: cell bodies of all cell
classes except cones are displaced to the side, which allows light to hit
the cones directly. The black arrow shows the direction of light. Different
diseases (in parentheses) affect different cell classes. AMD, age-related
macular degeneration; LCA, Leber congenital amaurosis; RP, retinitis
pigmentosa; CSNB, congenital stationary night blindness; CN, congenital
nystagmus; LHON, Leber hereditary optic neuropathy. b, Retinal cell
types are organized into about 30 circuits, each of which receives input
from the mosaic of photoreceptors and ends with a mosaic of ganglion
cells of a given type that forms the retinal output. There are about 30 types
of ganglion cell and the retina therefore creates about 30 different image
representations (three examples are shown at the bottom) of the original
image that enters the eye (the face at the top). The retina is thus a parallel
image processor that describes to the brain the image that falls on the
retina using its own ‘language’.
Human retinal organoids
A simple skin biopsy from a patient with a hereditary retinal disease
can be induced within one year to generate layered retina-like neuronal structures, termed retinal organoids. These consist of different
cell types and appear to bear genetic markings similar to those of retinal
neurons49–51. Alternatively, retinal organoids grown from a healthy
human biopsy can be engineered to carry relevant mutations 52.
Such organoids could enable unprecedented insights into the development of retinal diseases and provide a controlled, biologically
realistic system in which to test new treatment strategies. Disease-
model retinal organoids could prove useful in several ways. With
diseases that cause symptoms early in life, a disease phenotype may
develop in the organoids, thus allowing investigation of the mechanistic basis of the disease and the development of repair strategies. The extent to which retinal organoids can develop into adult
tissue53 is currently unclear and thus disease-model retinal organoids for hereditary diseases in which patients develop symptoms
later in life may not present relevant phenotypes. However, such
retinal organoids could still be useful for testing cell-type-specific
gene delivery or the efficacy of gene editing in selected cell types.
Finally, organoid culture allows control of the growth environment of
human retina-like tissue and can thus be used for experiments previously impossible with human retinas, such as the precise investigation
of physical and chemical stresses (for example, pressure, levels of oxygen
or glucose) that may underlie disease progression.
Table 1 | The advantages and limitations of different model systems
Fovea
In vivo
In vitro
In vivo screen
In vitro screen
Adult
Genetic engineering
Availability
Cost
Zebrafish
Mouse
Marmoset
Human
Organoid
No
Yes
Yes
Yes
No
Yes
Yes
+  +  +
+
No
Yes
Yes
No
No
Yes
Yes
+  +  +
++
Yes
Yes
Yes
No
No
Yes
Yes
+
+++
Yes
No
Yes
No
No
Yes
No
+  +
+
No
No
Yes
No
Yes
No
Yes
+  + +
++
Research combined in a variety of model systems allows a better understanding of the pathomechanism of disease and the development of therapy.
Post-mortem human retinas
Retinas from human donors can be kept alive in culture for at least three
weeks post mortem54,55, and recent improvements in culture conditions
may substantially extend their longevity. This is a key model system for
understanding the cell types and circuits of the adult human retina.
The recording of functional light responses at cellular resolution from
the retinas of animal models in vitro has been possible for years56 and
1 7 M AY 2 0 1 8 | VO L 5 5 7 | NAT U R E | 3 6 1
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
insight
Review
Small molecule
Antibody
Viral vector
Cell
Electric impla
implant
Size (m)
10–9
Number
of units
–
10–8
6 × 1015
2 × 10–8
1011
Stage Early
10–5
3 × 10–3
6 × 104
1
Late
Diffusion
Fig. 3 | Current and planned tools for treating blinding diseases.
The approximate physical sizes span six log units, and the numbers of
tools delivered to the eye span 15 log units. The approximate sizes and
numbers of tools delivered are shown below each tool. The complexity
and sophistication of the tools increase with size, as does the difficulty
of delivering treatment across a large retinal surface in vivo. The
characteristics of the different tools are shown below the images.
further technical advances will allow this to also be done in human
retinas. This will improve the comparison of retinal responses with
visual behaviour in humans and facilitate direct correlation of gene
expression and physiological responses for human retinal cell types.
Furthermore, direct access to human retinas will allow the testing of
gene therapy vectors, cell transplantation, and neuroprotection strategies in a more controlled manner and with a higher throughput than
is currently possible: a single human retina can be dissected into many
smaller pieces that can be examined individually.
and in vivo in primates eyes65 could identify vectors that are relevant for
humans. Post mortem retinas can be used to identify the cell types that are
targeted in adult human retinas; primate retinas can reveal the efficacy of
infection in living foveated eyes. However, given an effective AAV for delivering a gene of interest to a desired cell type in the human retina, the challenge
remains of reliably and safely delivering AAVs to the entire human retina.
As mentioned above, shielding of the human retina by the inner limiting
membrane restricts infection to the foveal area. Sub-retinal injections may
provide access to a larger retinal area but this is technically more complicated and separates the retina from the retinal pigmented epithelium, with
the danger of photoreceptor damage66. Furthermore, the human retina is
so large that a single AAV injection site is likely to be insufficient for full
coverage. Despite these limitations, gene therapy via AAVs is likely to
remain a leading strategy for the treatment of vision loss in the coming
years. Non-viral gene delivery approaches67–69 are also being pursued, such
as laser, ultrasound, and electrical discharges, which create momentary
pores in cell membranes and allow DNA or RNA fragments to enter the
cell. Unlike AAV-based delivery, such non-viral gene delivery techno­logies
produce transient expression of the target protein.
Genetic disease models in marmosets
Non-human primates have been used for a number of years to study
vision and vision loss, with most of the classic work done in macaques.
However, it is difficult to apply transgenic methods to this species and
the marmoset has emerged recently as a much easier non-human
primate model for transgenic work57. Importantly, marmoset retinas
have a fovea58, making the marmoset more relevant for developing
treatment strategies than non-primate (that is, non-foveated) models
such as mice. Thus, the advancement of transgenic applications in the
marmoset as an in vivo, foveated model of human visual diseases is of
crucial importance for future vision research.
New translational technologies
In order to treat blindness, we need not only new model systems
but also new translational technologies. While treatments based on
small molecules, antibodies, and surgery remain the main therapeutic
approaches used today, new technologies for repairing damaged retinas
and restoring light sensitivity to impaired retinas are also being developed (Fig. 3). We will discuss four new technologies here.
Gene therapy
Gene therapy59 has gained momentum with the introduction of AAVs60
as vectors for delivering genes of interest to retinal cells61,62. AAVs diffuse
easily in tissue, have little toxicity, and bring about sustained gene expression, although their integration into the human genome is minimal63. AAVs
have three important components: the capsid for cell entry64, the promoter
driving transgene expression and cell-type-specific expression62,65, and the
therapeutic gene as the viral payload. A weakness of AAVs is their relatively
small carrying capacity, which limits their application to small genes. As
the efficacies of both capsid and promoter depend on cell type and species,
AAVs optimized in mice are unlikely to be efficient and cell-type-specific
in humans. Screening of AAVs in vitro in post mortem human retinas54,55
Cell therapy
Cell therapy provides retinal repair via ectopic cell transplantation to
replace damaged or dead cells. Cells can be obtained from cell lines
derived from embryonic or induced pluripotent stem cells, or alternatively from retinal organoids in which the developmental stage of the
transplanted cells can be controlled. Cell therapy via retinal pigmented
epithelial cells21 appears to be simpler than via retinal neurons such
as photoreceptors, because pigmented epithelial cells do not need to
form neuronal signal-carrying contacts such as synapses. Nevertheless,
transplanted epithelial cells need to interact with photoreceptors to
be useful for retinal function. Cell therapy for retinal neurons is more
complex because transplanted cells need to migrate to the correct location, make the proper synaptic connections, and exhibit cell-type-specific physiological responses. While there has been some success in
transplanting photoreceptors into mouse retinas70,71, it is not clear that
current strategies will be efficient and effective in a clinical setting72.
Only a small proportion of transplanted cells appears to survive in mice
and make the necessary synaptic connections71, and the transfer of
material between transplanted and native photoreceptors complicates
interpretation of the results73–75. Recently, transplanted ganglion cells
were shown to integrate into retinal circuits, providing yet another cellular source for repair76,77; however, it has not yet been shown that the
3 6 2 | NAT U R E | VO L 5 5 7 | 1 7 M AY 2 0 1 8
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Review
transplanted ganglion cells grow axons that reach their targets in the
brain. Future experiments with retinal pigmented epithelial cells, photoreceptors and ganglion cells, in both animal models and in patients,
will further define the feasibility of this therapeutic approach.
Induced retinal regeneration
Induced retinal regeneration attempts to apply lessons learned from
animals, such as fishes, in which regeneration of the injured retina
occurs. In response to damage to the retina, Müller glial cells in fish
return to a stem-cell-like state, divide, and develop into different cell
types that integrate into the retinal circuitry29. Mammals do not have
this innate capability, but research on experimentally driven regeneration is in progress. For example, mouse Müller cells activated by specific
small molecules or genes start to divide78 and the newly generated
cells integrate into the retinal circuitry79. Induced retinal regeneration
has also been used to induce the regrowth of ganglion cell axons
towards their brain targets80,81. The cut axons of retinal ganglion cells
regrow axonal projections into the brain when provided with appropriate genes, small molecules, or neuronal activity81–83. The axons even
reach their correct cell-type-specific targets and establish functional
connections81. While these methods are currently performed only in
animal models and their efficiency is relatively low, future work is likely
to improve performance and add induced retinal regeneration as an
important tool to the translational repertoire.
Artificial retinal stimulation
Artificial retinal stimulation using electronic implants, optogenetics or
photoswitches is designed to restore the image-acquiring properties of
the retina84,85. The patient wears a camera mounted, for example, on a
pair of spectacles. The video signal is converted into an electronic pattern or a pattern of light, depending on the vision restoration strategy.
In electronic implant therapy, the implant is placed in physical contact
with the retina and an electronic pattern created by implant electrodes
stimulates the remaining light-insensitive retinal cells. The current
generation of retinal implants provides relatively modest visual improvement, owing partly to the low density and low number of electrodes in
the implants20,86, as well as to the large distances between implanted
electrodes and target cells87. Cortical implants are an alternative to retinal implants and a cortical implant that provides surface stimulation to
the primary visual cortex is in clinical trials88. In optogenetic therapy,
monochromatic light activates light-gated ion channels delivered by
gene therapy to remaining specific cell types in the blind retina54,89,90.
In photoswitch therapy91, monochromatic light activates light-sensitive
small molecules delivered to retinal cells that couple to intrinsic channels
on the surface of retinal cells. Theoretically, these latter approaches act
at a higher resolution than electronic implants and can activate specific
retinal cell types, giving increased control of specific retinal micro­
circuits. Limitations of current optogenetic and photoswitch techno­
logies are that the activation of cells requires bright visible light and that
the dynamic range of 2–3 logarithmic units is narrow92. In a number of
blinding diseases, including macular degeneration, most visual sensitivity is lost in a specific region of the retina, while other regions remain
light sensitive. The remaining light-sensitive retinal regions rule out or
impede the use of bright visible light for activation. Although it is possible to increase the sensitivity of optogenetic sensors, this often slows
the response93, and thus rules out restoration of useful motion vision.
Furthermore, since the dynamic range remains narrow, this approach
still requires an external camera. The development of near-infrared
optogenetic sensors or photoswitches could mitigate the problem of
the remaining light sensitivity. Optogenetic approaches are currently in
early clinical development and higher resolution electronic implants are
being developed. In the coming years, different forms of artificial retinal
stimulation are likely to be important strategies in vision-loss therapy.
Implementing technologies for therapy
The various translational technologies outlined above could be used
either alone or in combination to treat a variety of diseases. However,
insight
the decision about which technology is suitable for which disease and at
which stage requires a detailed understanding of the different blinding
diseases and their natural history. For example, it is important to discriminate loss of function, which is potentially reversible, from loss of
cells, which is irreversible. In general, the therapeutic strategy depends
on the stage of the disease. If vision has severely deteriorated or is completely lost, the aim has to be to restore vision; if significant vision is
still present, the aim may be to prevent or slow further loss. As is typical
in medicine, the border between these stages is not sharp but the two
ends of the spectrum do allow us to highlight differences in the ways
in which the available technologies can be used.
Vision is lost
If the patient has lost all useful vision, the only therapeutic option is
vision restoration. The available technological repertoire is wide, since
only general safety considerations are important, such as avoidance
of an immune response or the malignant transformation of cells.
Importantly, restoring vision even in a small patch of the retina may
lead to useful vision and, thus, the new technologies listed above that
lead to regained light sensitivity would be promising approaches. Most
vision restoration strategies that are considered today in clinical settings
aim to target diseases stemming from photoreceptor degeneration.
Cones are the photoreceptors that are responsible for high-resolution
colour vision under daylight conditions, and rods are responsible
for vision under dim light conditions. Many monogenic diseases
primarily affect rods. Late-onset forms of these monogenic diseases are
called retinitis pigmentosa94; Leber congenital amaurosis95 describes
the condition in which symptoms of retinitis pigmentosa are present
during infancy. Although rods in humans outnumber cones 20 to 1,
we artificially illuminate our environment today to such an extent that
individuals without functional rods can lead their lives relatively unaffected. However, for reasons not yet fully understood, rod degeneration
leads to secondary cone degeneration, even when the gene associated
with the disease is not expressed in cones59. This means that the most
severe effect of rod loss is the subsequent loss of cones and cone-derived
vision. Repair strategies at later stages of retinitis pigmentosa aim to
connect the retina to the visual world by providing artificial retinal
stimulation using electronic implants, optogenetics, or photoswitches.
An alternative is the transplantation of photoreceptors59,84,85.
Vision restoration strategies are also being considered that target
diseases in which ganglion cells degenerate. After ganglion cells have
died, such as in advanced stages of glaucoma or diabetic retinopathy,
there is currently little that can be done. But there are at least three ways
to develop potential therapy. First, to restore vision at higher levels in
the visual system such as the thalamus or visual cortex via electronic
implants96 or optogenetic stimulation. Second, to transplant ganglion
cells77. Third, to perform whole eye97 or organoid transplantation.
With the exception of a cortical implant now in clinical trials88, these
approaches are in their early stages or are simply theoretical options
with no proof of concept as yet.
Useful vision remains
If the patient still has some useful vision, the best approach for curing
blindness is the prevention, or at least slowing, of vision loss. Here, a
major factor when designing a new therapy is to guarantee that the
treatment does not interfere in any way with the remaining natural
vision. One key point is to protect foveal vision. A second factor concerns the distribution of the therapeutic agents. If a blinding disease
is progressive and affects the entire retina, the molecules or genes to
slow visual loss need to be delivered to the disease-affected retinal cell
types across the whole retina. With the exception of the fovea, deliver­
ing thera­peutic agents to a small patch of the retina will be likely to
have little effect on overall degeneration and loss of vision. Therefore,
approaches to slow vision loss differ between fovea-affecting diseases
and diseases that affect the whole retina. Since we are far from being
able to deliver genes across the whole human retina, retina-wide
delivery is most likely to succeed for small-molecule or antibody-based
1 7 M AY 2 0 1 8 | VO L 5 5 7 | NAT U R E | 3 6 3
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
insight
Review
treatments, or by using local gene therapy that produces secreted factors
that spread over a wide area.
A successful example of slowing down vision loss is the intravitreal
injection of anti-VEGF antibodies in wet age-related macular degeneration; this decreases the leakage of fluid from abnormal retinal
blood vessels and its destructive accumulation (oedema). However,
while this approach substantially delays vision loss, anti-VEGF therapies target only the vascular component of the disease and mostly
reduce blood leakage. The causative and amplifying processes (for
example, genetic variants, inflammatory and immune cell activation
and photoreceptor damage) are not treated98. In atrophic age-related
macular degeneration, cell therapy involving transplantation of retinal
pigmented epithelial cells is the approach that is gaining the most
traction99. Larger clinical trials will show whether this approach
increases vision in patients.
Our most precious photoreceptors are the foveal cones, the loss of
which leads to severe visual handicap. Juvenile macular degeneration,
including Stargardt disease, and atrophic age-related macular degeneration mostly affect foveal cones. Although Stargardt disease is a target
for gene therapy23, the most frequently mutated gene in this disease,
ABCA4, is too large to fit into an AAV vector. The options to be considered here are the use of alternative viral vectors, trimming the length of
the gene, or its assembly from shorter pieces within target cells. Gene
therapy using ABCA4 in larger-capacity lentiviral vectors is now in
clinical trials100.
Recent promising developments have reduced secondary degeneration of cones in animal models of retinitis pigmentosa. The delivery of
rod-derived cone viability factor101–103 or activation of its receptor104,
or the delivery of genes countering energy starvation105,106 or oxidative
stress107, have been shown to decrease cone degeneration in animal
models and are being considered for human therapy. Starvation108,109
and oxidative stress110,111 appear to be major contributors to cone
degeneration after rod death; therefore, feeding cones and preventing oxidative damage emerge as central themes in their protection.
Methods that may increase nutrient availability by increasing blood
supply112, such as microcurrent stimulation113,114, are in clinical trials,
while methods to reduce oxidative stress via antioxidant supplementation are already in clinical practice115.
There have also been developments in the reduction of vision loss in
a form of Leber hereditary optic neuropathy (LHON)116. Mutations in
different genes encoded by the mitochondrial genome cause different
forms of LHON. This disease is a clear example of cell-type-specific
vulnerability; despite the presence of mitochondria in every cell in the
body, only or mostly retinal ganglion cells degenerate. Gene therapy and
small molecule-based therapies have both shown promising slowing
of vision loss in LHON.
Everything we have described so far relates to retinal vision loss;
however, in a minority of cases vision loss is due not to retinal problems but rather to defects in other regions of the visual system, such as
the visual cortex. In these conditions, influencing the plasticity of the
remaining visual cortex has been invoked as a way to improve visual
function117.
Evaluating the outcome of therapy
The emergence of new therapies to slow loss of vision or to restore
vision creates the need for major innovation and international collaboration on ways to measure the outcome of therapies in patients.
One key challenge is variability in the results of clinical trials. For
example, one group reported increased visual function12 after LCA gene
therapy, while other groups found initial improvement that declined
over time118,119, or even an overall decline in visual function compared
to baseline in some patients118. Variability has also been observed in the
visual benefit to patients of electronic implants. Variability can originate
in many ways: different ways of evaluating visual function, measuring
visual function at different times after treatment, differences in the
details of gene therapy vectors and their delivery66, the details of how
visual stimulation devices are implanted, the ability of patients to learn
the meaning of altered peripheral input, and differences in the disease
stages at which the therapy was provided.
Another challenge is that retinal degenerative diseases are progressive and it is therefore essential to study progression of the disease
before enrolling patients in clinical trials, and to characterize changes
in progression during and following therapy85. Moreover, new and
existing diagnostic tools should be incorporated into the development
of a standardized and quantitative evaluation scheme to assess retinal
and cortical117 structure and function as well as visual function in both
control patients and patients receiving therapy.
New and improved in vivo imaging methods (including adaptive
optics, optical coherence tomography, ultrasound and autofluorescence
imaging) and psychophysical methods (for example, microperimetry)
can assist such evaluations and comparisons by allowing determination of the density of photoreceptors, the existence and length of inner
and outer segments, the status of ganglion cells, interactions between
photoreceptors and retinal pigmented epithelial cells, retinal blood
flow112, and any remaining function in a small patch of the retina85.
However, demonstration of a clinically relevant benefit extends beyond
such quantitative measures. Regulatory agencies and funding bodies
will require real-life performance-based tests, such as mobility testing15,
as well as patient-reported outcomes of treatment. Measures of outcome
must be robust, reproducible, relevant and easy to duplicate across
various centres120–122.
Rehabilitation is another key issue, since the outcome of therapy
can improve with time and a well-designed rehabilitation program123.
Improvement may be gradual because the connectivity of the visual
cortex may have reconfigured during the state of blindness to support
other cognitive functions124. Therefore, it may require visual training and time to readjust the visual cortex to its original function.
Furthermore, vision restoration approaches in which technologies
converge will require new rehabilitation strategies. As an example, optogenetic therapy combines gene therapy with a medical device, a goggle.
This technology convergence will require different types of expertise
and strong patient engagement during rehabilitation.
Technologies that complement vision restoration
Aside from strategies that explicitly set out to restore vision or slow
vision loss, there are many alternative approaches currently available or
being developed for conveying visual information to visually impaired
patients through one of their remaining sensory modalities. Classic
strategies include the use of guide dogs and canes. However, the digital
revolution is also adding to the capabilities of blind patients: smartphone applications provide services as varied as using the phone as a
simple light detector, or as a GPS-linked talking map, or to perform
voice-written emailing, word processing and web browsing125,126. It
has recently become possible to describe particular images to blind
people using artificial intelligence127. As the power of artificial intelligence increases, more and more digital tools that are useful for blind
individuals will be generated.
Other emerging technologies focus on the use of more complex sensory
substitution, where the remaining senses are trained to process information normally processed by the visual system. One approach that has
brought the highest recorded ‘visual’ acuity for any vision rehabilitation
strategy applied to vision-impaired patients is visual-to-auditory sensory
substitution128. Here a head-mounted camera records the visual environment and the video signal is then converted into a series of sounds.
Remarkably, visual-to-auditory sensory substitution activates ‘visual’
brain areas with auditory stimulation129. Another approach is the use
in humans of echolocation similar to that of bats and dolphins130.
Although these advances may make life easier for the blind, the goal
of medicine remains the prevention of vision loss in the first place and
the cure of vision loss that has already occurred.
Outlook
Blinding diseases of the eye, which are increasing in step with increasing human longevity, cause major social and economic burdens131. The
3 6 4 | NAT U R E | VO L 5 5 7 | 1 7 M AY 2 0 1 8
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Review
recent developments outlined in this paper provide the hope that some
of these diseases can be effectively addressed in the coming decade.
Nevertheless, the challenge is severe because the retina is a biological
computer and not just a simple collection of photosensitive cells.
Therefore, we argue that both the diagnosis and the treatment of retinal
diseases should be cell-type focused: disease by disease, we need to
understand the causes and mechanisms of the death of disease-affected
human retinal cell types. We must also learn to interact with retinal cell
types inside the patient’s eye, specifically and efficiently.
Finally, the retina, as a biological computer comprising about 100
cell types with specific rules of connectivity, is not unique. The cerebral cortex32 and some deep brain nuclei132 are both cell-type rich and
have specific connectivity. The retina is in fact an outpost of the brain
sitting in the eye56 and following rules similar to the rest of the brain.
This is important for two reasons. First, key pathological phenomena,
including cell-type specific degeneration, are shared by the retina
and other brain regions. Understanding why retinal cells die and why
this process is cell-type specific could enhance our understanding of
neurodegeneration in general. Second, fine perturbations of the activity
of specific retinal cell types are likely to occur in a variety of neurological and psychiatric diseases with a genetic or environmental origin,
because the disease-associated genes are expressed in specific retinal
cell types or because the environmental insult affects specific retinal
cell types. The fact that the retina is visible through the lens in humans
makes it the part of the brain that is simplest to study in patients in vivo.
The development of new imaging techniques to record the activity of
retinal cells at cellular resolution in vivo in humans would make the
retina the ideal location for following the progress of brain disease and
reaction to therapy.
Received: 1 October 2017; Accepted: 16 February 2018;
Published online 16 May 2018.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
The Lasker Foundation. Restoring Vision to the Blind http://www.
laskerfoundation.org/new-noteworthy/articles/restoring-vision-blind/ (2015).
Olson, R. J. Cataract surgery from 1918 to the present and future—just
imagine! Am. J. Ophthalmol. 185, 10–13 (2017).
Jager, R. D., Mieler, W. F. & Miller, J. W. Age-related macular degeneration.
N. Engl. J. Med. 358, 2606–2617 (2008).
Hodgson, N. et al. Economic and quality of life benefits of anti-VEGF therapy.
Mol. Pharm. 13, 2877–2880 (2016).
Finger, R. P., Guymer, R. H., Gillies, M. C. & Keeffe, J. E. The impact of
anti-vascular endothelial growth factor treatment on quality of life in
neovascular age-related macular degeneration. Ophthalmology 121,
1246–1251 (2014).
Essue, B. M. et al. A multicenter prospective cohort study of quality of life and
economic outcomes after cataract surgery in Vietnam: the VISIONARY study.
Ophthalmology 121, 2138–2146 (2014).
Lamoureux, E. L., Fenwick, E., Pesudovs, K. & Tan, D. The impact of cataract
surgery on quality of life. Curr. Opin. Ophthalmol. 22, 19–27 (2011).
WHO. Visual Impairment and Blindness http://www.who.int/mediacentre/
factsheets/fs282/en/ (2017).
Bainbridge, J. W. B. et al. Effect of gene therapy on visual function in Leber’s
congenital amaurosis. N. Engl. J. Med. 358, 2231–2239 (2008).
Hauswirth, W. W. et al. Treatment of Leber congenital amaurosis due to RPE65
mutations by ocular subretinal injection of adeno-associated virus gene
vector: short-term results of a phase I trial. Hum. Gene Ther. 19, 979–990
(2008).
Maguire, A. M. et al. Safety and efficacy of gene transfer for Leber’s congenital
amaurosis. N. Engl. J. Med. 358, 2240–2248 (2008).
This study led to the first FDA-approved AAV gene therapy.
Russell, S. et al. Efficacy and safety of voretigene neparvovec (AAV2hRPE65v2) in patients with RPE65-mediated inherited retinal dystrophy: a
randomised, controlled, open-label, phase 3 trial. Lancet 390, 849–860
(2017).
The first FDA-approved AAV gene therapy.
Ashtari, M. et al. The human visual cortex responds to gene therapy-mediated
recovery of retinal function. J. Clin. Invest. 121, 2160–2168 (2011).
Ashtari, M. et al. Plasticity of the human visual system after retinal gene
therapy in patients with Leber’s congenital amaurosis. Sci. Transl. Med. 7,
296ra110 (2015).
Ledford, H. FDA advisers back gene therapy for rare form of blindness. Nature
550, 314 (2017).
MacLaren, R. E. et al. Retinal gene therapy in patients with choroideremia:
initial findings from a phase 1/2 clinical trial. Lancet 383, 1129–1137 (2014).
Edwards, T. L. et al. Visual acuity after retinal gene therapy for choroideremia.
N. Engl. J. Med. 374, 1996–1998 (2016).
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
insight
Humayun, M. S. et al. Visual perception elicited by electrical stimulation of
retina in blind humans. Arch. Ophthalmol. 114, 40–46 (1996).
This work led to the development of epiretinal implants.
Mills, J. O., Jalil, A. & Stanga, P. E. Electronic retinal implants and artificial
vision: journey and present. Eye 31, 1383–1398 (2017).
da Cruz, L. et al. Five-year safety and performance results from the Argus II
retinal prosthesis system clinical trial. Ophthalmology 123, 2248–2254 (2016).
Schwartz, S. D. et al. Human embryonic stem cell-derived retinal pigment
epithelium in patients with age-related macular degeneration and Stargardt’s
macular dystrophy: follow-up of two open-label phase 1/2 studies. Lancet
385, 509–516 (2015).
Mandai, M. et al. Autologous induced stem-cell-derived retinal cells for
macular degeneration. N. Engl. J. Med. 376, 1038–1046 (2017).
Tanna, P., Strauss, R. W., Fujinami, K. & Michaelides, M. Stargardt disease:
clinical features, molecular genetics, animal models and therapeutic options.
Br. J. Ophthalmol. 101, 25–30 (2017).
Daley, G. Q. Polar extremes in the clinical use of stem cells. N. Engl. J. Med. 376,
1075–1077 (2017).
Espinosa, J. S. & Stryker, M. P. Development and plasticity of the primary visual
cortex. Neuron 75, 230–249 (2012).
Ganesh, S. et al. Results of late surgical intervention in children with
early-onset bilateral cataracts. Br. J. Ophthalmol. 98, 1424–1428 (2014).
Kalia, A. et al. Development of pattern vision following early and extended
blindness. Proc. Natl Acad. Sci. USA 111, 2035–2039 (2014).
Sinha, P., Chatterjee, G., Gandhi, T. & Kalia, A. Restoring vision through “Project
Prakash”: the opportunities for merging science and service. PLoS Biol. 11,
e1001741 (2013).
Wan, J. & Goldman, D. Retina regeneration in zebrafish. Curr. Opin. Genet. Dev.
40, 41–47 (2016).
Azeredo da Silveira, R. & Roska, B. Cell types, circuits, computation. Curr. Opin.
Neurobiol. 21, 664–671 (2011).
Masland, R. H. The neuronal organization of the retina. Neuron 76, 266–280
(2012).
Zeng, H. & Sanes, J. R. Neuronal cell-type classification: challenges,
opportunities and the path forward. Nat. Rev. Neurosci. 18, 530–546 (2017).
Seung, H. S. & Sümbül, U. Neuronal cell types and connectivity: lessons from
the retina. Neuron 83, 1262–1272 (2014).
Weinmann, J. & Grimm, D. Next-generation AAV vectors for clinical use: an
ever-accelerating race. Virus Genes 53, 707–713 (2017).
Santos-Ferreira, T. F., Borsch, O. & Ader, M. Rebuilding the missing part—a
review on photoreceptor transplantation. Front. Syst. Neurosci. 10, 105 (2017).
Chalupa, L. M. & Williams, R. W. (eds) Eye, Retina, and Visual System of the
Mouse (MIT Press, Cambridge, 2008).
Dacey, D. M. The mosaic of midget ganglion cells in the human retina.
J. Neurosci. 13, 5334–5355 (1993).
Kolb, H. in Webvision: The Organization of the Retina and Visual System (eds
Kolb, H. et al.) (Univ. Utah Health Sciences Center, Salt Lake City, 1995).
Sahly, I. et al. Localization of Usher 1 proteins to the photoreceptor calyceal
processes, which are absent from mice. J. Cell Biol. 199, 381–399 (2012).
Example of the limitations of mice as a disease model.
Siegert, S. et al. Transcriptional code and disease map for adult retinal cell
types. Nat. Neurosci. 15, 487–495 (2012).
Kostic, C. & Arsenijevic, Y. Animal modelling for inherited central vision loss.
J. Pathol. 238, 300–310 (2016).
Kolb, H., Fernandez, E. & Nelson, R. (eds) Webvision: The Organization of the Retina
and Visual System (Univ. Utah Health Sciences Center, Salt Lake City, 1995).
Seiple, W., Rosen, R. B. & Garcia, P. M. T. Abnormal fixation in individuals with
age-related macular degeneration when viewing an image of a face. Optom.
Vis. Sci. 90, 45–56 (2013).
Dalkara, D. et al. Inner limiting membrane barriers to AAV-mediated retinal
transduction from the vitreous. Mol. Ther. 17, 2096–2102 (2009).
Dalkara, D. et al. In vivo-directed evolution of a new adeno-associated virus for
therapeutic outer retinal gene delivery from the vitreous. Sci. Transl. Med. 5,
189ra76 (2013).
Yin, L. et al. Intravitreal injection of AAV2 transduces macaque inner retina.
Invest. Ophthalmol. Vis. Sci. 52, 2775–2783 (2011).
Panda-Jonas, S., Jonas, J. B., Jakobczyk, M. & Schneider, U. Retinal
photoreceptor count, retinal surface area, and optic disc size in normal human
eyes. Ophthalmology 101, 519–523 (1994).
Remtulla, S. & Hallett, P. E. A schematic eye for the mouse, and comparisons
with the rat. Vision Res. 25, 21–31 (1985).
Eiraku, M. et al. Self-organizing optic-cup morphogenesis in three-dimensional
culture. Nature 472, 51–56 (2011).
This study led to the development of human retinal organoids.
Nakano, T. et al. Self-formation of optic cups and storable stratified neural
retina from human ESCs. Cell Stem Cell 10, 771–785 (2012).
Sasai, Y. Grow your own eye: biologists have coaxed cells to form a retina, a
step toward growing replacement organs outside the body. Sci. Am. 307,
44–49 (2012).
Ovando-Roche, P., Georgiadis, A., Smith, A. J., Pearson, R. A. & Ali, R. R.
Harnessing the potential of human pluripotent stem cells and gene editing for
the treatment of retinal degeneration. Curr. Stem Cell R…

Order an Essay Now & Get These Features For Free:

Turnitin Report

Formatting

Title Page

Citation

Outline

Place an Order