Does the ensemble that shapes the function of the neuronal circuit or the individual features shapes the ensemble? Part I: Migration, Laminar allocation, synapse formation of projection and interneurons.

Slide1 (2)


One of the important questions in the field of neurosciences is to understand how the neuronal ensembles acquire physiological properties during development. In other words, how molecular and electrical properties shape the neuronal ensembles to acquire certain physiological responses during development and how they will be refined in the adult stages. It is important to know how the developing circuits undergo different structural changes such as pruning, remodelling of the synapses to weigh the on-going activity impinged upon those circuits during the process of learning and acquisition of memory during the learning. A circuit undergoes structural remodelling to form a specific connectivity during development is as significant to know their convergence and divergence of connectivity later in the development. Series of recent work has explicitly pointed out that neurons have a high probability of firing synchronously, which are involved in common function. They organize into sub-networks with recurrent connections so that they can fire together for a particular modality of function. This holds true in the case of excitatory pyramidal neurons while there is a substantial lack of evidence for interneurons whether they can dynamically organize into functional assemblies. Lineage tracing studies has elegantly demonstrated that sister excitatory neurons are synaptically coupled electrically through gap junctions initially which render them to fire together thereby forming functionally connected neurons and later they have been refined to form chemical synapses. For instance, sister neurons those possess same receptive field for an orientation preference are synaptically connected to form microcircuits in columns in the cortex of the mammalian brain. This has led to the postulation that the intralaminar connectivity between sister neurons in ontogenic columns can make functional columns in the cortex. This is in agreement with the famous “ radial unit hypothesis”(1) put forward by Rakic, which states that sister neurons produced from the same proliferative regions migrate along the radial glial fibers to form ontogenic columns. However in the case of interneurons, experiments labelling clonally related neurons using low titre of viruses to learn whether the clonally related neurons assemble to form functional networks yields inconsistent results. Two recent studies combining the powerful method of unique barcoding and genetic labelling (2,3) has revealed that the clonally related interneurons are widely dispersed unlike that of excitatory neurons and neurons from different clones can cluster together. The majority of the MGE (Medial ganglionic eminence) derived interneurons spread across large regions of the brain. One of the interpretation arises from this studies is that it is this wide dispersion of interneurons is the optimal way to form functional circuits across the cortex given that interneuron comprises only 20% of the neurons. In this way, limited number of interneurons optimally can make multiple connections with excitatory neurons spanning across the cortex. Recently it has been shown that the laminar allocation of interneurons is regulated by the cues from the excitatory neuron, thereby the latter building the allocation of the former.

The fundamental question emanates from this study is that when does the excitatory and inhibitory neuron shapes the ensemble to modulate a specific function during development and how? How does structural features of neuronal wiring influence or determine the emerging neuronal function, both at the level of single – neurons and of microcircuits? Quantifying the convergence and divergence of connectivity between PY to PY and PY to IN and between IN and IN (PY, Pyramidal neurons, IN, Interneurons) within the ensemble will tell us how the information is transferred as an optimal neural code. Circuit pruning, structural remodelling, local activity may change the optimality of the neural coding in different developmental stages in this ensembles. Computational models can be building from the structural connectivity and the neuronal properties can be studied across developmental stages, which might operate with specific neural codes in these ensembles. In essence, how network activity and network structure in the ensemble are shaped by activity dependent rules is one of the important questions to be addressed in neurosciences. Again another emerging theme comes from this study is also that are there a subset of neurons within the ensemble acts as a driver neurons in the sense of control sense? 

I will discuss in the next series of blogs about how this functionally assembled ensembles undergoes synapse maturation and how we can perturb the circuits genetically to understand the physiological properties of these neurons in the ensemble. I will outline how these perturbations can affect network properties especially how we can model the development of networks within the ensembles and how the functional properties are changed. In this blog, I will discuss on how both the excitatory and inhibitory interneurons migrate to the cortical plate and decorate the specific layers in the cortex. The molecular mechanisms guiding the migration of the excitatory neuron is well known nonetheless we still lack information on how inhibitory interneurons reach the cortical place. It will not be surprising when we learn in the future that the molecules that are responsible for the migration of interneurons into the cortex will be the same array of molecules (receptors/ligands/ extracellular matrix/adhesive proteins), which are involved in cell migration in development, axon guidance so on and so forth. I will detail describe how we can approach the questions. Before that I give a brief outline of excitatory and inhibitory neurons migrate to the cortical plate.

A defect in the correct lamination of cortical structures or the mis – positioning of neurons will have a profound effect on the normal functioning of the brain. Research has shown this to be true in the case of autism, epilepsy, schizophrenia, Fragile X – Syndrome, and other diseases. In humans, abnormal migration of neurons leads to lissencephaly. The cortex development of mammals is initiated with the migration of neurons formed in the ventricular or sub -ventricular zone. The first step of cortical development is the formation of a preplate above the ventricular zone. The preplate is decorated with early born neurons, the sub -plate neurons and the Cajal – Retzius cells. This is ensued by the splitting of the pre-plate into a superficial layer having the Cajal – Retzius cells in the marginal zone (MZ) and a deeper subplate. The late generated neurons move past the sub -plate and are embedded in the cortical plate (CP) between the marginal zone and the sub plate. It has been proposed that reelin secreted by the Cajal – Retzius cells in the marginal zone acts as a stop signal for the late generated neurons so that they can circumvent the MZ. This orientation of alignment of neurons in the cortex is called inside – out lamination, where the late born neurons pass through the earlier generated neurons and aligns in a superficial position beneath the MZ in the cortical plate, where the early born neurons reside in the deep layers. The neurons either take a mode of radial migration or somal translocation to reach the pial surface. The neurons which enroute the pial surface along the radial fiber of the radial glial cells in the case of radial migration have a bipolar morphology, with a thick leading process migrating to the pia, and a thin trailing process. In the intermediate zone beneath the sub – plate, the neurons either exist as bipolar or multipolar cells and they traverse along the radial fiber through the subplate to the cortical plate with a bipolar morphology. Interestingly, in the case of somal translocation, the long radial process emanating from the cell attaches to the pia and acts as a leading process pushing the cell soma towards the pia through this elongated process. Conversely, the reeler mutant, possess an outside –in alignment of neurons in the cortical plate because of the inability of the preplate to split in to two, which eventually leads to the formation of a super- plate. In the reeler mutant, the earlier generated neurons stay close to the superficial layer as reeler is characterized by the absence of a well – defined marginal zone causing the late generated neurons reside in the deep layers. It has been shown that reelin acts as a permissive signal in the early stages of development to split the preplate so that the early-generated neurons can plate. Once the migrating neurons reach the marginal zone, the Cajal – Retzius cells, residing there secrete reelin as a stop signal for the migrating neurons. This has been demonstrated in the reeler mutant, where the marginal zone is populated by the early-generated neurons in contrast to the absence of any cells in the marginal zone in the wild type. The stopping function of reelin exerted on the migratory neurons reaching the marginal zone is achieved by its binding to the Vldlr receptors expressed on the incoming neurons. Reelin that acts on the actin cytoskeleton eventually stabilizes the migratory neurons, which reach the marginal zone. This is done by phosphorylating cofilin at serine 3, rendering it to an inactive state, culminating in having stabilized polymerized actin. Moreover, reelin acts on its receptor, α3β1 integrins to detach the neurons from the radial glial scaffold Interestingly, this type of outside – in orientation of cortical neurons in reeler has been mirrored in other phenotypes such as in yotari, scrambler and in apoer2/vldlr double knockout mice. Subsequently it turns out that all the phenotypes mentioned above are in the same genetic or molecular pathway. Reelin turns out to be the ligand for both ApoEr2 and Vldlr receptors. Moreover battery of transcription factors plays an instructive role in the laminar allocation of excitatory neurons in the upper and deep layers of the cortex. They also play a role in the assignment of the positioning of the cortical areas. For instance, ctip1 regulates the identity of sensory areas and represses motor areas. Knockout of these mutants causes the motorization of the sensory output connectivity. Moreover over –expression of Fez2, a transcription factor specifying the deep layer excitatory neurons in E14.5 embryos can turn the callosal projection neurons in the upper layer to corticofugal neurons present in the deep layers of the cortex.

Similar to excitatory neurons, MGE (Medial ganglionic eminence, express the transcription factor Nkx2.1) derived interneurons decorate the cortical plate in an inside-out fashion; whilst on contrary, CGE-derived interneurons do not exhibit their birth date and their final destination in the cortical plates, as they inhabit the superficial layers of the cortex. Temporal cohorts of interneuron produced interneuron diversity. While excitatory pyramidal neurons take a beeline to reach the cortex, interneurons produced in the sub-pallium takes a circuitous route. They migrate tangentially over long-range distance and take either the marginal zone or sub ventricular zone before making a radial turn how to reach the cortical plate. It has been speculated that excitatory neurons which already present in the different layers of cortical plate produces molecules which acts as a chemo attractant/repellant for the invading interneurons as they migrate radially only after the projection neurons to reach their final destination. For instance, it has been shown that the corticofugal axons present in the MZ and VZ will play as a scaffold for the migrating interneuron into the cortex. I will outline a few experimental plans to learn about how the molecular cues produced by the interneurons and projection neurons to form an ensemble.

I) Methodological approach to find the candidate genes:

Given both the projection neurons and interneurons invade the cortical plate in an inside-out pattern; one of the obvious things will be isolating the timely matched neurons, both the excitatory and inhibitory using genetic tricks. For instance inducing tamoxifen in Nkx2.1CRERT2 mice crossed with RosaRtdTomato/tdTomato in E11.5 to E13.5 only labels interneurons destined to deep layers (mostly layer 5 and layer 6) in red while simultaneously performing in utero electroporation in this stage label projection neurons in green. Along the line, inducing Tamoxifen in E14.5 to E17.5 shall labels upper layers (of note: E17.5 labels a specific set of interneurons called chandelier cells). In postnatal stage, P0, most of the interneurons are in the marginal zone or in the sub ventricular zone ready to invade the cortical plate, while the projection neurons are already placed in their appropriate layers in the cortical plate. Dissecting the whole cortical plate from the marginal zone to sub ventricular zone and further performing FACS, one can isolate the interneurons labeled in red and the projection neurons in green. Besides it is also possible to micro dissect the distinct layers of the cortex (upper versus against deep layer) and performing FACS to isolate the labeled neurons. Method A) RNA is isolated from the differentially labeled neurons and transcriptome analysis can be performed. The GO of the cellular component and the biological process gives the fold enrichment of the genes in distinct compartments in the cell. For instance, if one is specifically interested in looking at the membrane bound ligands/ receptors; the fold enrichment of this sets of genes can be analyzed. One can also try to match what membrane proteins (ligands and receptors) can acts as an instructive cues from the list. Method B) This is of my favorite approach: The membranes of the FACS isolated neurons (red and green) will be stripped and multiplexed mass spectrometry (CSC technology) can be performed by covalently labeling the extracellular glycan motives. This would allow the relative quantification of N-linked cell surface glycoproteins, which includes all the membrane-tethered receptors, a comprehensive view of the surface protein landscape. Novel receptor –ligand interactions can be also performed using a chemoproteomic approach called TRICEPS, a modified version of CSC and then applying ligand based receptor technology (LRC) enabling to isolate the receptor-ligand interaction without any genetic modifications (5). Method C) Adhesive molecular match codes flagging the juxtaposed pre and post-synaptic partners in assembling the excitatory and inhibitory circuits can be curated by visualizing the expression of trans membrane proteins from public databases such as Allen brain atlas map, which in turn provides a plethora of information of expression patterns of genes across different regions of the brain during development as well as in adult stages. In essence, a list of transmembrane domain from mouse proteins from Ensembl Biomart (Ensembl Genes 75) can be obtained.  The list of 11368 transcripts/proteins (many were isoforms) which refers to the subset of the Mouse proteome (genome version GRcm38.P2 & Ensembl annotation release 75) can be obtained. The transmembrane receptors representing ion channels, solute carriers, G-protein coupled receptors and their isoforms can be omitted by manual curating. The rest of them were verified against the expression pattern in the Allen brain atlas and can be color-coded according to their expression in the different layers in the developing brain. Potentially, one can color code the expression pattern of any genes in the different layers in the developing as well as adult brain comparing with the insitu expression pattern found in the public database, Allen brain atlas ( for easing the visualization.

One can potentially isolate the time matched interneuron population using FACS and micro dissection at different stages in postnatal development and can perform the method A, B or C to address the adhesive codes used by the excitatory and inhibitory neurons. P0 (Postnatal day 0) represents, the migratory phase and one can presume that they were expressing the genes important for migration, P1, P2, P3, P4 represents the axon guidance phase and synapse formation, P7 to P21 represents circuit pruning, synapse stabilization and synapse maturation. In the case of excitatory neurons, it would be expressing molecules that would guide the axon of the interneurons, synapse formation, maturation and maintenance. This above-mentioned methodology can be applied to CGE derived interneurons as well. The only exception is that the CGE derived interneurons do not exhibit a tight nexus between their birth rate and final laminar allocation, as they inhabit only the superficial layers of the cortex. However, the guidance, synapse formation, maturation and circuit pruning molecules can be addressed as the same way described above.

II) How can one validate the molecules involved in laminar allocation, migration and synapse formation?

Conventional approaches such as knockout models will give information on the laminar allocation of interneurons in the different layers in the cortex. For instance, we know that the early generated interneurons goes to the deep layers, while the late generated interneurons goes to the upper layer in the wild type. Besides depending on the time of birth, there is a distinct allocation of sub-types of interneurons in the cortex. For instance, in the early stages of development both the somatostatin (SST) and Parvalbumin (PV) expressing interneurons are produced moreover in a equal proportion, while in the later stages of development, the SST production declines while the PV generation is continuous. Comparing the laminar allocation in wild-type animals (Control) and in the mutants will helps in understanding the genes involved in that step. Whole cell recordings, Paired recordings and photo stimulation experiments tell whether the connectivity, synaptic strength is compromised in inter as well as intralaminar connections in the mutant. This would also reflect in the behavioral paradigms such as the animal suffering from epilepsy so on and so forth because of the perturbation in the connectivity. However to narrow down those genes involved from a screen, we need finding an alternative way to quickly screen the set of genes before going for the conventional knock-out animals. Gene editing method such as the CRISPR against the gene of interest is a powerful approach however the caveat of this approach is that it will be difficult to distinguish whether the phenotype is due to the defect in the initial tangential migration of the interneurons from the sub-pallium to the SVZ or MZ or rather a defect in the radial migration into the cortical plate. The best way to accomplish a fast paced screen will be performing the knock down of the candidate genes in P0, where they already migrating into the MZ or SVZ. Here, I present two approaches of how this can be achieved with the current art of techniques.

Approach 1:

Here, the MGE from different age (E10.5 to E16.5) is cut off from the brain slices and they were further dissociated. These cells were transduced with Dlx12b-GFP lentivirus harboring the sh-RNA of the candidate genes or with a sham. These cells transfected with the Sh-RNA or the vehicle will be injected into the P0 cortices. At P21, these green cells will be scored whether they reach the final allocation or not. For instance, the early born cells shall go to the deeper while the late born shall go to the upper layer. One could examine whether this has been compromised in neurons infected with Sh-RNA from the list of candidate genes in comparison to that of the wild type. The methodology is adapted from Vogt D et al; Neuron 2014 (5).


Figure 1:

Figure 1: MGE from wild type cells are dissociated and infected with Dlx2b – GFP lentivirus carrying the ShRNA against the candidate gene or sham. These neurons are further transplanted to P0 cortices and they are scored for upper and deeper bound neurons in comparison to the wild type. If the gene is important for the laminar allocation, the Sh-RNA against the gene will cause an ectopic mislocalization.

Approach 2:

Alternative way to validate will be gain of function experiments. The rationale is that the matching adhesive code expressed by the pyramidal neurons that are already present in the cortical plate instructs the invading interneurons to finally allocate its position in different layers. Isolating the birth and time matched GFP expressing pyramidal neurons in upper or deep layers in the cortex and performing either method A and B could isolate those trans membrane proteins instructs the migration of interneurons. This receptor-receptor or receptor –ligand interaction can be checked with the counterpart, i.e. expressed by the interneurons whether they match with the genes expressed by pyramidal neuron. The rationale for the gain of function experiment will be that if you forcibly over-express the match code normally expressed by the pyramidal neurons in the upper layer to the pyramidal neurons that are destined to go to the deeper layer, then you would expect the time matched interneurons that recognize and matched to the code will ends up in an ectopic location, in this case they would go to the deeper layer instead. Conversely you can change the allocation of those interneurons, which are destined to go to deeper layer to that of the upper layer. In this way, one can quickly screen for list of candidate genes expressed by interneurons as well as the pyramidal neurons using gain of function as well as loss of function approaches.

Figure 2:


Figure2: Using in utero electroporation approach, one can forcibly overexpress any candidate gene of interest (green) in pyramidal neurons. Depending on the birth date, early generated goes to the deep layer, while the late generated goes to the upper layer. One can express the trans membrane receptor normally expressed by the upper layer projection neurons into those neurons that are destined to deeper layers. Using a Cre-inducible system one can mark the MGE derived interneuron population (Red) depending on the date of birth as they follow the inside-out lamination pattern like that of the projection neurons. In this case, the interneurons end up in an ectopic location, deeper layer. Red circle marks, the interneurons destined to upper layer, while Blue represents the pyramidal cells and the green attached to blue represents the trans membrane receptors. In wild type animals, the interneuron (red) occupies the normal laminar allocation (upper) while in the case of mutant, the red occupies in the deep layers.

Synapse formation between PY and IN or between either PY and PY or IN and IN can be performed by either mixing co-cultures of neurons expressing the trans membrane receptors on this cultures. Besides, one can also culture neurons with HEK cells expressing any of the candidate genes to see whether the expression of the receptors in the cell line cluster the scaffolding proteins such as PSD-95, AMPA/NMDA/GABA receptors on the post-synaptic part and active zone proteins such as piccolo and bassoon and other synaptic proteins such as synaptotagmins, vGLUTs, VGATs in the pre-synaptic part.

One important thing we need taking into consideration is that it is not only the molecules cues but also the emerging electrical activity between projection neurons and interneurons connect networks depending on the on-going activity, thereby causing the final laminar allocation of neurons within the ensemble. A thorough understanding is needed of how the balanced excitatory and inhibitory activity as well as the recurrent connections that strengthens the synapses by hebbian plasticity that builds the ensemble. The spontaneous and evoked responses in network activity can further modulate the final allocation of interneurons in the cortex thereby shaping the ensemble suit for a particular function. Physiological manipulation and computational modelling will shed more light on the functional properties of this ensembles and how any perturbation can change this ensemble, which can be translated to a phenotypic output will be discussed in my next blog.

Due to the paucity of time, I will not detail the comprehensive list of references except the ones I mentioned in the text.

1.Rakic, P (1988). Specification of cortical areas. Science, 241, 170 – 176.

2.Harwell CC et al; (2015). Wide dispersion and diversity of clonally related inhibitory interneurons. Neuron, 87, 999 – 1007.

3.Mayer C et al; (2015). Clonally related forebrain interneurons disperse broadly across both functional areas and structural boundaries. Neuron, 87, 989 – 998.

4.Frei AP et al; (2012). Direct identification of ligand-receptor interactions on living cells and tissues. Nat Biotechnology, 30, 997 – 1001.

5.Vogt D et al; (2014). Lhx directly regulates Arx and CXCR7 to determine cortical interneuron laminar fate and laminar position. Neuron, 82, 350 – 364.

Leave a Comment