Researchers developed a breakthrough cell therapy to improve memory and prevent seizures


Researchers from the University of California, Irvine developed a breakthrough cell therapy to improve memory and prevent seizures in mice following traumatic brain injury.

The study, titled “Transplanted interneurons improve memory precision after traumatic brain injury,” was published today in Nature Communications.

Traumatic brain injuries affect 2 million Americans each year and cause cell death and inflammation in the brain.

People who experience a head injury often suffer from lifelong memory loss and can develop epilepsy.

In the study, the UCI team transplanted embryonic progenitor cells capable of generating inhibitory interneurons, a specific type of nerve cell that controls the activity of brain circuits, into the brains of mice with traumatic brain injury.

They targeted the hippocampus, a brain region responsible for learning and memory.

The researchers discovered that the transplanted neurons migrated into the injury where they formed new connections with the injured brain cells and thrived long term.

Within a month after treatment, the mice showed signs of memory improvement, such as being able to tell the difference between a box where they had an unpleasant experience from one where they did not.

They were able to do this just as well as mice that never had a brain injury.

The cell transplants also prevented the mice from developing epilepsy, which affected more than half of the mice who were not treated with new interneurons.

“Inhibitory neurons are critically involved in many aspects of memory, and they are extremely vulnerable to dying after a brain injury,” said Robert Hunt, PhD, assistant professor of anatomy and neurobiology at UCI School of Medicine who led the study.

“While we cannot stop interneurons from dying, it was exciting to find that we can replace them and rebuild their circuits.”

This is not the first time Hunt and his team has used interneuron transplantation therapy to restore memory in mice. In 2018, the UCI team used a similar approach, delivered the same way but to newborn mice, to improve memory of mice with a genetic disorder.

Still, this was an exciting advance for the researchers.

“The idea to regrow neurons that die off after a brain injury is something that neuroscientists have been trying to do for a long time,” Hunt said.

“But often, the transplanted cells don’t survive, or they aren’t able to migrate or develop into functional neurons.”

This shows a mouse hippocampus

These are transplanted inhibitory neurons (green) successfully incorporated into the hippocampus of a mouse with traumatic brain injury. The image is credited to UCI School of Medicine.

To further test their observations, Hunt and his team silenced the transplanted neurons with a drug, which caused the memory problems to return.

“It was exciting to see the animals’ memory problems come back after we silenced the transplanted cells, because it showed that the new neurons really were the reason for the memory improvement,” said Bingyao Zhu, a junior specialist and first author of the study.

Currently, there are no treatments for people who experience a head injury.

If the results in mice can be replicated in humans, it could have a tremendous impact for patients. The next step is to create interneurons from human stem cells.

“So far, nobody has been able to convincingly create the same types of interneurons from human pluripotent stem cells,” Hunt said. “But I think we’re close to being able to do this.”

Jisu Eom, an undergraduate researcher, also contributed to this study. Funding was provided by the National Institutes of Health.

Brain injuries represent a large variety of disabling pathologies. They may originate from different causes and affect distinct brain locations, leading to an enormous multiplicity of various symptoms ranging from cognitive deficits to sensorimotor disabilities.

They can also result in secondary disturbances, such as epileptic foci, which occur within the lesioned and perilesional tissues (Herman, 2002). Indeed, frequently a secondary functional damage can take place in a region distant from the first insult (e.g., the hippocampus after traumatic brain injury), providing an explanation for cognitive and memory deficits arising after a brain lesion (Girgis et al., 2016).

Brain injuries can have traumatic or non-traumatic etiologies, including focal brain lesions, anoxia, tumors, aneurysms, vascular malformations, encephalitis, meningitis and stroke (Teasell et al., 2007). In particular, stroke covers a vast majority of acquired brain lesions. Stroke is usually referred to as a sudden cerebrovascular dysfunction leading to focal deficits and/or impairment of global brain functions and lasting more than 24 h (Mackay and Mensah, 2004).

Since brain function is strictly dependent on a constant supply of oxygen and glucose, normally assured by blood circulation, a sudden block of blood supply determines suppression of neural function in less than 1 min, primarily due to interference with synaptic functions (Hofmeijer and Van Putten, 2012). Brief blood deprivations may cause only a reversible damage, which becomes permanent only if the circulation is not promptly restored (Krnjević, 2008Vrselja et al., 2019).

The neocortex represents the highest level of cognitive and sensorimotor integration, and it is therefore not surprising that, independently of different etiologies, lesions occurring in the cerebral cortex are particularly impacting on the clinical phenotype (Delavaran et al., 2013). For example, an insult occurring in the motor cortex results in functional impairment of one or more body parts contralateral to the infarct.

The degree of the motor impairment depends on many factors, such as the extent of the lesion, the identity of the damaged region and the effectiveness of the initial neuroprotective interventions.

Following stroke, there is a window of neuroplasticity during which the greatest gains in recovery occur (Zeiler and Krakauer, 2013). Indeed, in the first weeks after stroke a limited spontaneous restoration of function may be observed, and about 30% of stroke survivors are able to carry on everyday activities (Activity of Daily Living or ADLs, i.e., eating, drinking, walking, etc.) without any help (Mozaffarian et al., 2014). However, other patients do not recover at all (Winters et al., 2018).

In particular, impairments of upper and lower limbs make very hard to retain a sufficient degree of independence in ADLs. These impairments can be ameliorated with a variable degree of success, through rehabilitation of the affected body parts, including several physical activities improving strength and coordination of the affected muscles and promoting recover of motility. Furthermore, combining rehabilitation with treatments that enhance neuroplasticity has been demonstrated to boost recovery (Alia et al., 2017Spalletti et al., 2017) but further steps forward in the field are necessary for clinical translation.

Besides physical rehabilitation and plasticizing treatments, another therapeutic approach is cell-based therapy, which has been pioneered in the therapy of Parkinson Disease (PD). Indeed, initial studies showed that fetal dopaminergic neurons grafted in the striatum ameliorated PD symptoms, both in animal models (Herman and Abrous, 1994) and in patients (Lindvall et al., 1990Kordower et al., 1998). Since fetal transplantation poises both ethical issues and technical challenges (Robertson, 2001), other non-neural cells, such as mesenchymal stem cells (MSCs), may represent a more accessible alternative.

In fact, MSCs can be readily derived from various sources, show a low immunogenic effect and proved to be beneficial in stroke treatment (Eckert et al., 2013Zhang and Chopp, 2013). Notably, MSCs-conditioned medium alone is sufficient for a similar therapeutic effect, suggesting that the beneficial effect is likely due to a bystander effect and trophic support, rather than actual cell replacement (Eckert et al., 2013).

Recent clinical trials proved the safety of immortalized neural cells when stereotaxically injected in the brain of patients with stable paresis of the arm following an ischemic stroke (Kalladka et al., 2016).

Finally, induced pluripotent stem cells (iPSCs) are easily accessible, since they can be reprogrammed starting from somatic cells, can be in vitro differentiated in the desired type of neurons and, being autologously derived and carry low risk of immune rejection (Jung et al., 2012Trounson and DeWitt, 2016).

Due to the high clinical relevance of cerebral lesions, protocols to differentiate neural stem cells in telencephalic neurons are highly valuable. These protocols are frequently inspired by findings in neural development.

Developmentally Inspired Stem Cells Differentiation Paradigms

Attempts to generate identity-specific cortical neurons require exploitation of the right embryonic developmental signals in a culture dish. The six layers of the mammalian neocortex originate from inside-out in a timely regulated manner during embryonic development and are characterized by a specific pattern of gene expression and connectivity (reviewed in: Rubenstein, 2011Bronner and Hatten, 2013Greig et al., 2013Lodato and Arlotta, 2015). Both intrinsic and environmental factors control the progressive restriction of competence of cortical progenitors resulting in the sequential generation of the different cortical neurons, with deeper layer cortical neurons generated earlier than upper layer neurons (Gorski et al., 2002Kriegstein and Noctor, 2004Molyneaux et al., 2007).

The hippocampus arises from the invagination of the dorsal midline of the telencephalon, adjacent to the cortical hem. The boundaries between choroid plexus, cortical hem and hippocampus are defined early in development by the non-overlapping expression of molecular markers signatures. In particular, BMP4/7 are strongly expressed by the choroid plexus (Grove et al., 1998Bulchand et al., 2001), while the cortical hem expresses high level of WNT3a, WNT2, and WNT5, whose signaling is essential for the correct development of the hippocampus (Grove et al., 1998) (Figure 1A).

Consistently, experiments in which the cortical hem is ablated, or hem-dependent WNT expression is abrogated or compromised, showed the crucial role of the cortical hem in the hippocampal development (Grove et al., 1998Grove and Tole, 1999Lee et al., 2000). The Dentate Gyrus (DG), the CA1 and CA3 layers are the different fields forming the hippocampal formation; each one has its own cellular composition, morphology, connectivity and expresses specific molecular markers.

In particular, KA1, a glutamate receptor subunit, is detectable in the CA3, while SCIP, a POU-domain transcription factor, is expressed in the CA1. The pyramidal neurons of the CA1-3 originate in the more dorsal ventricular zone of the hippocampus and then migrate toward the pial surface on a glial scaffold (Nadarajah et al., 2001), while the DG cells arise from a smaller and specialized medial region close to the fimbria, the dentate neuroepithelium (Danglot et al., 2006).

Furthermore, new neurons are generated in the hippocampus of rodents throughout life, while cortical neurogenesis rapidly subsides after birth (Eriksson et al., 1998van Praag et al., 2002). This makes hippocampus the region of choice to assay the ability of in vitro produced PSC-derived hippocampal-like neural cells to make projections and to send them to appropriate targets.

An external file that holds a picture, illustration, etc.
Object name is fnins-13-00684-g001.jpg
FIGURE 1 : (A) Drawing depicting the main morphogens involved in the patterning of the forebrain of a mouse embryo (E10). T, Telencephalon; D, Diencephalon. (B) Cartoon showing main signaling pathways active in differentiating PSC and the inhibitors most commonly used. BMP signaling: Noggin, Bone Morphogenetic Protein Receptor 1A Fc chimera (BMPR1A-Fc), LDN193189, or Dorsomorphin; WNT signaling: CHIR99021, Dickkopf-related protein 1 (Dkk1), IWR-1-endo, 53AH or WNT-C59; TGFβ signaling: Lefty, SB431542; SHH signaling: Cyclopamine. β: β-catenin (CTNNB1).

Pluripotent stem cells cultured in low-density, serum-free and feeder-free conditions, in absence of any exogenous growth factor, readily differentiate into neural progenitor cells (Tropepe et al., 2001Ying et al., 2003).

This process faithfully reproduces in vivo neurogenesis (reviewed in: van den Ameele et al., 2014Brown et al., 2018Varrault et al., 2019), following the same timely regulated sequential neurogenic waves observed in vivo (Eiraku et al., 2008Gaspard et al., 2008Bertacchi et al., 2013Molyneaux et al., 2015), which is therefore conserved in vitro (Shen et al., 2006).

In fact, differentiating PSCs give rise to neural rosettes expressing the early neural markers Sox1, Pax6 and Nestin and closely resembling the early embryonic neural tube (Perrier et al., 2004). When dissociated and replated, PSC-derived neural rosettes give rise to neural progenitor cells (NPCs) which subsequently differentiate into a heterogeneous mixture of different subtypes of neurons (GABAergic, dopaminergic, serotonergic, and cholinergic), astrocytes and oligodendrocytes (Zhang et al., 2001Barberi et al., 2003).

Furthermore, PSCs cultured in this conditions spontaneously acquire an anterior dorsal neural identity (Watanabe et al., 2005Pankratz et al., 2007), while the addition of growth factors to the medium can steer the differentiation of PSCs toward midbrain/hindbrain (Kawasaki et al., 2000Perrier et al., 2004) or spinal cord (Wichterle et al., 2002Li et al., 2009) fates.

Thus, PSCs are first committed toward an anterior neural fate and successively patterned by caudalizing factors to acquire a more posterior neural identity, supporting a mammalian model of default neural induction (Muñoz-Sanjuán and Brivanlou, 2002Smukler et al., 2006Stern, 2006Levine and Brivanlou, 2007Gaulden and Reiter, 2008). Finally, recent evidence suggest that the regionalization of the presumptive neural ectoderm precedes neural commitment; indeed these results further support the notion that FGF, WNT and RA signaling are required for the acquisition of posterior neural identity by PSCs differentiating in vitro (Metzis et al., 2018).

The positional identity of neural cells obtained by pluripotent cell during in vitro differentiation is often established by the simple study of their neurotransmitter phenotype (Eiraku and Sasai, 2011Shi et al., 2012Yu et al., 2014Shiraishi et al., 2017), or by deeper investigation of their molecular nature through methods of global gene expression analysis (Bertacchi et al., 20132014Espuny-Camacho et al., 2013Leemput et al., 2014Edri et al., 2015Yao et al., 2017Terrigno et al., 2018a).

However, specific neural subtypes present in different areas of the adult brain show, in addition to distinct gene expression profiles, also defined connectivity patterns, electrophysiological proprieties and morphologies (Cadwell et al., 2015Fuzik et al., 2015Tiklová et al., 2019). Whereas the molecular signature of neurons can be established prior their transplantation, the evaluation of all the other parameters require in vivo analysis. Understanding how to obtain a precise subtype of neuron, patterned with the proper temporal and positional identity, is pivotal and might have considerable repercussion on possible clinical applications.

In this review, we have considered available protocols to differentiate murine and human pluripotent stem cells into cortical- or hippocampal-like stem cells (Figures 2​,3)3) expressing at least a forebrain marker (FOXG1 or EMX1/2) and an early cortical or hippocampal marker (TBR1/2, BCL11B or PROX1).

An external file that holds a picture, illustration, etc.
Object name is fnins-13-00684-g002.jpg
FIGURE 2 : Table showing original reports of differentiation protocols of murine PSCs in cortical and hippocampal neurons. For each entry we indicated (from left to right): reference, schematic layout of the protocol, neural markers, cell-type markers, electrophysiological proprieties, target of transplantation assay to assess integration, global gene expression analysis to identify cell positional identity. If the information was available, we indicated the time of emergence of main cortical cell types (TBR1, BCL11B, BRN2, SATB2, and CUX1). Due to the sheer number of studies on this topic, derivative reports of minor modifications to previously published protocols are not included in this table. SFEBq, serum-free culture of embryoid body-like aggregates; EB, embryoid body; ML, monolayer; LD, low-density; sPSA, spontaneous post-synaptic activity; ePSA, evoked post-synaptic activity
An external file that holds a picture, illustration, etc.
Object name is fnins-13-00684-g003.jpg
FIGURE 3 : Table showing original reports of differentiation protocols of human PSCs in cortical and hippocampal neurons. With same criteria as those indicated in Figure 2 SFEBq, serum-free culture of embryoid body-like aggregates; EB, embryoid body; ML, monolayer; scRNAseq, single cell RNA sequencing; eAP, evoked action potential; sPSA, spontaneous post-synaptic activity; ePSA, evoked post-synaptic activity.

Derivation of Cortical-Like Neurons From Pluripotent Stem Cells

The derivation of cortical-like neurons from PSCs was first reported a decade ago by the laboratories of Pierre Vanderhaeghen (Gaspard et al., 2008) and Yoshiki Sasai (Eiraku et al., 2008). Interestingly, Sasai’s protocol is also the first report on 3D cortical organoid from both murine and human PSCs obtained by inhibiting both the endogenous WNT and TGFβ signaling. On the contrary, in Vanderhaeghen’s protocol PSCs are cultured as low density adherent monolayer in a minimal, chemically defined medium supplied with a SHH inhibitor, Cyclopamine. Indeed, these authors pioneered the generation of cortical-like cells from PSCs, paving the road for the subsequent studies on human PSCs and induced PSCs (Figures 2​,3).3).

Notably, cortical-like cultures, expressing anterior neural markers (FOXG1 and EMX1/2) and enriched in several cortical cell types (BCL11B, TBR1/2, SATB2, BRN2, and CUX1), can be obtained by simply differentiating PSCs, cultured in suspension as embryoid bodies (Ebs), or as attached monolayer, in a chemically defined minimum medium, devoid of any grow factors (Gaspard et al., 2008Li et al., 2009Zeng et al., 2010Lancaster et al., 2013).

However, most authors block endogenous posteriorizing or ventralizing factors (Figure 1) to promote the commitment toward an anterior dorsal fate and enrich the culture in cortical cell types (Eiraku et al., 2008Gaspard et al., 2008Mariani et al., 2012Shi et al., 2012Bertacchi et al., 20132014Espuny-Camacho et al., 20132018Kadoshima et al., 2013Renner et al., 2013Yu et al., 2014Paşca et al., 2015Sakaguchi et al., 2015Motono, 2016Gunhanlar et al., 2017Sarkar et al., 2018Terrigno et al., 2018a), including BMP inhibitors (Noggin, BMPR1A-Fc, LDN193189, or Dorsomorphin), WNT inhibitors (Dkk1, IWR-1-endo, 53AH, or WNTC59) and TGFβ inhibitors (Lefty, SB431542). In addition, the default fate of PSC-derived neural progenitors appears to be dorsal, thus fewer protocols include the inhibition of SHH signaling (Gaspard et al., 2008Yu et al., 2014Sarkar et al., 2018).

Interestingly, by default PSC-derived cortical neurons appear to acquire mainly limbic and occipital cortical identities (Gaspard et al., 2008Bertacchi et al., 2013Terrigno et al., 2018b). In fact, cortical-like cultures show high levels of posterior cortical markers, namely COUP-TF1 (Gaspard et al., 2008Bertacchi et al., 2013Terrigno et al., 2018b), an orphan nuclear receptor that is expressed in a high posterior-lateral to low anterior-medial expression gradient by both progenitors and cortical neurons, and is required to balance the antero-posterior patterning of the neocortex (Armentano et al., 2007Liu et al., 2010).

In addition, heterotopic transplantation experiments into the adult murine cortex showed patterns of projection strikingly similar to grafted embryonic visual cortical cells (Michelsen et al., 2015Espuny-Camacho et al., 2018), supporting the hypothesis of a default occipital fate for PSC-derived neurons.

Indeed, treatment with the morphogen FGF8 can induce frontal motor cortical fates in PSC-derived cultures (Eiraku et al., 2008Kadoshima et al., 2013Imaizumi et al., 2018Terrigno et al., 2018b). Intriguingly, Motono et al. reported the generation of PSC-derived anterior cortical neurons (low COUP-TF1 / high SP8 expression) by blocking both the canonical and non-canonical WNT signaling with an inhibitor of Porcupine (PORCN), a membrane-bound O-acyltransferase involved in the palmitoylation, and subsequent secretion, of WNT proteins (Motono et al., 2016).

Although these results remain to be confirmed by whole genome expression profiling of the cells and a more extensive analysis of their morphology and connectivity following homotopic and heterotopic transplantation, they are yet another example of how culture conditions profoundly influence the positional identity of PSC-derived cortical neurons. Finally, since different protocol to obtain cells with anterior cortical identity are now available (Eiraku et al., 2008Kadoshima et al., 2013Motono, 2016Imaizumi et al., 2018Terrigno et al., 2018b), further studies should focus on PSC-derived neurons with motor cortical identity transplanted in the damaged motor cortex.

Besides positional identity, the cellular composition of the cell culture has great clinical implications, since the projection pattern of transplanted PSC-derived neurons varies according to their cellular identity.

That is, TBR1- and BCL11B-positive PSC-derived neurons transplanted in the adult brain generate, respectively, cortico-thalamic and corticofugal projections (Espuny-Camacho et al., 2013Michelsen et al., 2015), consistently with the connectivity of normal TBR1- and BCL11B-positive cells. In fact, differentiating PSCs can generate all the cell-types of the six cortical layers (van den Ameele et al., 2014), although culture conditions strongly influence their ratio. Indeed, the production of cortical pyramidal neurons by PSCs cultured in minimal medium as monolayer cultures is strongly skewed toward lower layer identity (Gaspard et al., 2008Espuny-Camacho et al., 2013).

However, the proportion of upper cortical neurons can be increased by supplying retinoids during dual SMAD inhibition (Shi et al., 2012), by transplantation in host brain (Espuny-Camacho et al., 2013) or by culturing PSCs as aggregates (Eiraku et al., 2008Mariani et al., 2012Lancaster et al., 2013). Organoid cortical cultures more faithfully reproduce several features of in vivo corticogenesis (reviewed in Lancaster and Knoblich, 2014Di Lullo and Kriegstein, 2017Giandomenico and Lancaster, 2017), such as migration of newborn neurons on radial glia processes (Bamba et al., 2017), interkinetic nuclear migration (Paşca et al., 2015), and interactions and connections between different regions of the neuroepithelium, such as the interneuron migration and cortical invasion in 3D co-cultures of dorsalized and ventralized PSC-derived neurons (Birey et al., 2017).

Finally, although these methods do not allow the identification of the signals responsible for PSC specification, aggregate cultures display higher axonal outgrowth (Chandrasekaran et al., 2017) and cell survival (Daviaud et al., 2018) following cortical transplantation, holding great promises for in vitro modeling of neurodegenerative diseases (Quadrato et al., 2016).

During telencephalic development, GABAergic interneurons are generated in the subpallial caudal and medial ganglionic eminence (CGE and MGE) and migrate tangentially to the cortex and hippocampus (Marín and Rubenstein, 2001Xu et al., 2004Hansen et al., 2013Ma et al., 2013). Several neurological disorders are caused by dysfunctions of the GABAergic system and in the last decades GABAergic cell-based therapies have emerged as a promising treatment in restoring the lost balance between inhibitory and excitatory circuits (Danjo et al., 2011; reviewed in Alvarez Dolado and Broccoli, 2011Hunt et al., 2013Shetty and Upadhya, 2016).

Furthermore, several groups were able to efficiently generate forebrain GABAergic interneurons from PSCs (Gaspard et al., 2008Li et al., 2009Zeng et al., 2010Falk et al., 2012Germain et al., 2013Liu et al., 2013Maroof et al., 2013Nicholas et al., 2013Tornero et al., 2013Yuan et al., 2015Liao et al., 2016Sun et al., 2016Xu et al., 2016). However, due to the ventral origin of the forebrain interneurons, PSCs committed in culture to dorsal telencephalic fates, such as cortex and hippocampus, generate mainly glutamatergic neurons (Zhang et al., 2001Bibel et al., 2004Eiraku et al., 2008Gaspard et al., 20082009Bertacchi et al., 20132014Espuny-Camacho et al., 2013Yu et al., 2014Michelsen et al., 2015Livesey et al., 2016Motono, 2016Sarkar et al., 2018Terrigno et al., 2018a). Since GABA-ergic inhibition is essential for maintaining the balanced activity of cortical neural circuits and for the development of neural network activity (Ben-Ari et al., 2007), grafting of PSC-derived cortical or hippocampal cultures enriched with interneurons could have significant clinical repercussions and should be further investigated.

Derivation of Hippocampal-Like Neurons From Pluripotent Stem Cells

The first report of PSC-derived neural progenitors committed to a medial pallial identity was by Sasai and colleagues (Eiraku et al., 2008). Since then, several authors described methods to commit PSC to either choroid plexus or hippocampus (Figures 1​,2;2Yu et al., 2014Sakaguchi et al., 2015Sarkar et al., 2018Terrigno et al., 2018a).

Most of these protocols share an initial phase of WNT, BMP and TGFβ inhibition, followed by activation of BMP and WNT signaling, by either suppling WNT3a or a small-molecule inhibitor of GSK3 (CHIR).

In particular, Sakaguchi et al. showed that a short-term activation of WNT and BMP signaling in anteriorized neural progenitors specifies a hippocampal primordium-like tissue in vitro, expressing the hippocampal markers ZBTB20, KA1 and PROX1; while a prolonged treatment results in increased expression of LMX1A and TTR, markers of choroid plexus (Sakaguchi et al., 2015).

Consistently, other authors showed that double inhibition of WNT and BMP endogenous signaling in differentiating PSC followed by timely activation of WNT signaling results in hippocampal specification (Yu et al., 2014Sarkar et al., 2018Terrigno et al., 2018a). In particular, the laboratory of Fred Gage showed that a prolonged treatment with WNT3a results in the commitment of PSC to DG fate, while a shorter treatment and with a lower concentration of WNT3a specifies CA3 identity (Yu et al., 2014Sarkar et al., 2018).

Notably, PSC-derived hippocampal-like neurons, but not cortical-like cells, transplanted into adult hippocampus were capable of long-term survival, contacted relevant targets of the hippocampal formation and formed functional synaptic contacts with the host network, further confirming the importance of the identity match between transplanted cells and surrounding environment (Terrigno et al., 2018a).

UC Irvine
Media Contacts:
Anne Warde – UC Irvine
Image Source:
The image is credited to UCI School of Medicine.

Original Research: Open access
“Transplanted interneurons improve memory precision after traumatic brain injury?”. Bingyao Zhu, Jisu Eom & Robert F. Hunt.
Nature Communications doi:10.1038/s41467-019-13170-w.


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.