So to start off by explaining what is ONC and GCAMP6?

Optic nerve crush is used an experimental model of glaucoma because it

produces rapid and reproducible degeneration of retinal ganglion cells and their
axons which mimicks the neurodegenerative outcome of glaucoma which is
progressive RGC loss and optic nerve damage. (doesn’t replicate the chronic,
pressure-dependent nature of the disease). Optic nerve crush was performed
under ketamine anesthesia by exposing the optic nerve through a small
conjunctival incision and applying a controlled 3-second clamp with cross-action
forceps 1–3 mm behind the globe. After closure of the conjunctiva and
administration of reversal and analgesia, mice were recovered on a heated pad
and monitored for postoperative wellbeing.
This is essentially the two part system design to allow cell type specific
expression. CaMKII (calcium-calmodulin protein kinase 2) is a promoter that is
mainly active in excitatory pyramidal neurons (ensures that our system only
targets those). TTA (tetracycline controlled transactivator) is a regulatory protein
that acts like a genetic switch. When expressed, it binds to specific DNA
sequences and turns on the expression of downstream genes. GCaMP6 is a
genetically encoded calcium indicator which fluorescens when calcium levels rise
inside the cell and is used as a proxy for neural activity. TRE (tetracycline
response element) is the DNA sequence that TTA binds to, when TTA binds to the
TRE, it activates the expression of GCaMP6.
This second construct is a viral tool used to manipulate neuronal activity. AAV9/2
refers to the viral vector. AAV (adeno-associated virus which is safe and efficient
in neurons). ChRmine is a red shifted opsin which is a light sensitive ion channel,
when illuminated with red light, it opens and depolarises the neuron allowing us
to activate those neurons with precise timing. mScarlet is a red fluorescent
protein fused to ChRmine which allows us to visually confirm which cells are
expressing the opsin under a microscope. Kv2.1 is a targeting sequence that
directs the protein to the somatic membrane of the neuron which helps localise
expression to the cell body and reduces unwanted activation of axons or
dendrites. And it is injected into the primary visual cortex.
And this is the timeline where testing of the mice will occur 4 days before, 2 days
after, 7 days after and 14 days after ONC.

Pattern ERG recordings were performed under ketamine anesthesia using the
Celeris system, with pupils dilated and the corneal electrodes positioned in
contact with the eye using Viscotears. After confirming low and balanced
electrode impedance, monocular PERG stimuli were presented to measure retinal
ganglion cell-driven responses.
Two photon calcium imaging allows us to monitor the activity of individual
neurons in vivo by exciting GCaMP fluorescence with deep-penetrating infrared
laser pulses. This technique provides high resolution, real time recordings of
neural population responses in the intact cortex.
The mice are placed on a treadmill so they can behave naturally while we record
neural activity. Running and stationary states produce very different cortical
responses, so keeping the animal awake and able to move is essential for
interpreting the calcium-imaging data correctly.
Mice are presented drifiting gratins of 8 orientations, each shown for 300s
followed by a 300s inter-stimulus interval.
So first to explain the basic idea of how to measure visual cortex activity. We
have mouse with a cranial window. These mouse express GCaMP6 so when a
neuron becomes actives, it briefly gets brighter. So we can record these
fluorescence changes.
We show the mouse simple a simple visual pattern of black and white gratings at
different orientations.
This is an example two-photon image from visual cortex. Two infrared photons hit
the same molecule at the same and together they have enough energy to excite
it and this allows you to see individual neurons in the mouse visual cortex to
allow for in vivo calcium imaging. If you selected one cell and recorded its
activity it would look like this. The red trace is the change in fluorescence over
time and each spike in the trace corresponds to the neuron responding strongly
to a particular stimulus orientation demonstrating orientation selectivity.
To give further background, this is a study that looks at the way locomotion
causes a change in the cortical state. It is thought to be via two processes: the
first being that when the animal is moving, neurons in V1 fire more which is
important and the second being that the noise between the neurons becomes
less synchronised so the group of neurons as a whole represents visual input
more cleanly and distinctly.
The mice were presented with moving grating in the visual field. Mice were free
to run or stand stationary on a spherical treadmill floating on an air stream, while
their movements were monitored. They counted the number of spikes evoked in
the first 500 ms after the stimulus onset and then labelled each trial as a ‘run
trial’ or a ‘still trial’ based on the mouse’s average running speed during that
period. These were not GCamp6 mice but instead they used a high density
silicon microelectrode which picks up action potentials from nearby neurons.
It was found that locomotion strengthened average single-cell responses to the
stimuli. The change in mutual information of cells was predicted by the
multiplicative component of modulation and was weakly inversely related to the
additive component of modulation. Mutual information essentially quantifies how
much information one variable contains about another and calculates the
average reduction in uncertainty. This is important because the fact that it is
weakly additive suggests that the neurons are retaining their orientation
selectivity when in locomotion.
This image compares the functional organisation of primary visual cortex in cats
versus mice. Each little line represents the preferred orientation of a single
neuron and its colour shows which eye it responds to. The mouse visual cortex
does not have a highly ordered map and the orientation preferences are mixed
together in a ‘salt and pepper’ pattern.
In primates they have a highly organised visual cortex which orientation columns
and ocular dominance columns (alternating stripes for left-eye vs right-eye
dominance) which is not seen in mice.
This means that large-scale map organisation cannot be directly extrapolated to
mice which means the approach must focus on individual neurons and their
tuning properties rather than relying on macroscale maps.
In this study they are looking at how adult mouse visual cortex responds to optic
nerve injury. They used GCaMP6 mice.
This shows the visual responses of 50 neurons before and after ONC. This panel
illustrates how optic nerve injury disrupts and then reverse ocular dominance in
adult mouse visual cortex. The heatmaps show calcium responses from
individual layer 2/3 neurons to stimulation of either the contralateral eye (red) or
ipsilateral eye (blue), comparing control mice with mice 7 days after optic nerve
crush (right). In controls, neurons respond strongly to contralateral eye input and
only weakly to ipsilateral stimulation, reflects the normal contralateral
dominance of mouse V1. After ONC, however responses to the crushed
contralateral eye are markedly reduced, while ipsilateral eye responses remain
intact and can even become more prominent. As seen in the individual traces. Its
positive and then negative ODI indicating contralateral preference.
Again +1 neurons responds to contralateral -1 responds to the ipsilateral.
In humans, both hemispheres receive balanced input from both eyes because of
extensive binocular overlap, allowing strong integration of visual information for
depth perception. In contrast, the mouse visual system is highly contralaterally
biased, with most retinal fibres crossing at the optic chiasm, so each hemisphere
predominantly processes input from the opposite eye. This anatomical difference
is important when interpreting monocular manipulations like optic nerve crush in
mice, as cortical changes reflect a much more eye-specific organisation than in
humans.
This code loads the Suite2p calcium-imaging outputs (F, Fneu, iscell), corrects
the fluorescence for neuropil contamination, and extracts only the ROIs that were
classified as real neurons.
It then reshapes the data into days and blocks, selects the specific imaging days
defined for that mouse, applies the correct day labels, and prints a summary of
the dataset being analyzed.
This code picks 15 random neurons and plots their full calcium-imaging traces
across all recorded days, stacking each trace vertically for clarity. The vertical red
lines mark the boundaries between days, so each segment between red lines
represent one full imaging session.
This plot shows the calcium activity of many neurons concatenated across days.
Activity becomes synchronised on day+2 and later days some neurons show
increased individual fluorescence again.
This code plots a heatmap for each day, showing how strongly pairs of neurons
are correlated on that day. The neurons are sorted so that recovering cells are at
the top, and the code draws red lines to mark the boundary between recovering
and non-recovering neurons.
On day +2, the large yellow block sits entirely within the deteriorating group,
showing abnormally high synchrony among neurons that later lose function. This
reflects an injury-induced network collapse, where deteriorating neurons become
strongly co-active because they lose normal retinal drive. When we compare
correlation to spatial distance, these neurons also form “local clusters”—
physically nearby cells become more correlated—indicating that the pathological
synchrony is spatially organized rather than random.
This code creates a 3D scatter plot using Principal Component Analysis (PCA) to
compare neurons that are classified as recovering versus deteriorating after
optic nerve crush (ONC). First, PCA is applied to the neuronal activity data, which
converts each neuron’s complex activity pattern (across time and days) into
three new values: PC1, PC2 and PC3. These represent the main axes of variation
in the data, with PC1 explaining the greatest differences between neurons, PC2
the second most, and PC3 the third. The code then plots each neuron in this
reduced 3D space: recovering neurons are shown in magenta and deteriorating
neurons in grey. The viewing angle is optimised for each mouse to best visualise
clustering.
The resulting graph shows that recovering neurons tend to cluster in a distinct
region of PCA space compared to deteriorating neurons, indicating that their
overall activity patterns are different in a consistent way. This means that
recovery is not random, but associated with a specific neural activity signature.
In the context of this project, this suggests that neurons capable of recovery
after ONC have a characteristic functional profile, potentially reflecting greater
plasticity or adaptive network integration. PCA therefore allows us to visualise
and quantify how functional organisation of neural populations relates to
resilience and degeneration, supporting the idea that recovery following optic
nerve injury is driven by structured, biologically meaningful circuit dynamics
rather than noise.
In terms of potential implications, this highlights that glaucoma is not purely a
retinal disease but also involves significant pathological changes at the cortical
level. Similar to Alzheimer’s disease, where accumulation of amyloid beta and
tau disrupts excitatory glutamatergic and inhibitory GABAergic signalling through
increased extracellular glutamate and loss of GABAergic function, evidence
suggests that glaucoma may follow a parallel neurodegenerative pathway within
the visual system. This study investigated this by measuring GABA and
glutamate levels in the visual cortex of glaucoma patients and healthy controls
using magnetic resonance spectroscopy alongside detailed ophthalmic
assessments. They demonstrated that GABA levels were significantly reduced in
advanced glaucoma compared to healthy controls, and that structural retinal
damage (summarised as a retinal structural index derived via PCA from OCT
measures) strongly predicted both reduced GABA and glutamate levels,
indicating a link between peripheral retinal degeneration and central cortical
neurochemistry. Furthermore, neural specificity — reflecting how distinctly the
visual cortex encoded horizontal versus vertical stimuli — was shown to be
significantly associated with GABA levels, with reduced GABA leading to
degraded neural specificity even after accounting for age, retinal damage and
grey matter volume. This identifies GABA as a key modulator of cortical visual
processing in glaucoma rather than glutamate. These findings directly align with
my project, which examines visual cortical activity and network reorganisation
following optic nerve injury, demonstrating how disruption of inhibitory-
excitatory balance contributes to cortical plasticity changes. Together, this
supports the concept that glaucoma accelerates cortical ageing and alters
inhibitory control mechanisms, reinforcing the importance of targeting cortical
neurobiology when developing therapeutic approaches and expanding glaucoma
management beyond the eye alone to include central visual processing and
rehabilitation strategies.