Differentiation and functional connectivity of fetal tectal transplants

Abstract

Data from studies analyzing the differentiation and functional connectivity of embryonic neural tissue grafted into the mammalian nervous system has led to the clinical testing of the fetal graft approach in patients with neurodegenerative disease. While some success has been achieved, ethical concerns have led to a search for alternative therapeutic strategies, mostly exploring the use of neural precursors or neurons derived from pluripotent stem cells to replace damaged host neurons and restore lost circuitries. These more recent studies address questions of graft viability, differentiation, and connectivity similar to those posed by researchers in earlier fetal transplant work, thus reviews of the fetal graft literature may inform and help guide ongoing research in the stem cell/organoid field. This brief review describes some key observations from research into the transplantation of neural tissue into the rat visual system, focusing on grafts of the fetal superior colliculus (tectal grafts) into neonatal or adult hosts. In neonate hosts, grafts quickly develop connections with the underlying host midbrain and attain a morphology typical of mature grafts by about 2 weeks. Grafts consistently contain numerous localized regions which, based on neurofibrillar staining, neuronal morphology (Golgi), neurochemistry, receptor expression, and glial architecture, are homologous to the stratum griseum superficiale of normal superior colliculus. These localized “patches” are also seen after explant culture and when donor tectal tissue is dissociated and reaggregated prior to transplantation. In almost all circumstances, host retinal innervation is restricted to these localized patches, but only those that are located adjacent to the graft surface. Synapses are formed and there is evidence of functional drive. The only exception occurs when Schwann cells are added to dissociated tecta prior to reaggregation. In these co-grafts, the peripheral glia appear to compete with local target factors and host retinal ingrowth is more widespread. Other afferent systems (e.g., host cortex, serotonin) show different patterns of innervation. The host cortical input originates more from extrastriate regions and establishes functional excitatory synapses with grafted neurons. Finally, when grafted into optic tract lesions in adult rat hosts, spontaneously regrowing host retinal axons retain the capacity to selectively innervate the localized patches in embryonic tectal grafts, showing that the specific affinities between adult retinal axons and their targets are not lost during regeneration. While the research described here provides some pertinent information about development and plasticity in visual pathways, a more general aim is to highlight how the review of the extensive fetal graft literature may aid in an appreciation of the positive (and negative) factors that influence survival, differentiation, connectivity and functionality of engineered cells and organoids transplanted into the central nervous system.

Introduction

There is a long history of research investigating the possibility of transplanting viable fetal neural tissue from a range of different sources into the immature or mature mammalian central nervous system (CNS). In the last five or so decades in particular, with the advent of markers to (i) identify grafted cells, (ii) monitor cell division and cell phenotype, and (iii) characterize connectivity with host tissue using physiological techniques and neuroanatomical tracers, we have learned a great deal about the adaptability and plasticity of developing neural tissue, and how such tissue interacts with intact, injured or degenerating host brains. It has also become clear that, in some circumstances, transplants of embryonic tissue from appropriate regions of the developing brain can be used to restore functionality, resulting in the use of grafts in the clinical treatment of Parkinson’s disease (Lindvall, 2015) and other neurological conditions (Bachoud-Lévi et al., 2021; Van den Bos et al., 2022). Even though the use of fetal grafts to treat human neurological dysfunction has shown some success, ethical concerns have led to a search for other therapeutic treatments and alternate cell sources, such as neural precursors, neurons derived from pluripotent stem cells or organoids, to replace damaged or diseased cells and eventually repairing compromised circuitries (Ferrari et al., 2018; Toman et al., 2019; Han et al., 2020; Parmar et al., 2020; Limone et al., 2022; Liu et al., 2022; Seng et al., 2022; Xu et al., 2022).

While the possibilities of neural stem cell and organoid therapies (Eichmüller and Knoblich, 2022; Kelly and Paşca, 2022; Penna et al., 2022) are increasingly the subject of pre-clinical research, data from these more recent studies sometimes mirror results from similar types of studies carried out in the 1980s and 1990s using fetal neural tissue grafts. In potentially important translational fields such as this, it is therefore instructive to review earlier data on embryonic neural transplants and the factors that influence graft survival, differentiation, and host-graft connectivity, to inform new researchers in the field of the remarkable range of observations and discoveries made using the fetal graft approach, both in immature and mature host brains. To this end, this review highlights some key observations from several decades of research involving the transplantation of fetal neural tissue into regions of the rat visual system (Lund and Hauschka, 1976; Lund et al., 1982). While the focus is on grafts of fetal superior colliculus (SC) tissue (herein referred to as tectal grafts) and how such studies continue to provide relevant information about development and plasticity in mammalian visual pathways, the research also highlights in more generic terms the factors that can impact the survival, differentiation, connectivity, and functionality of cells or tissues when transplanted into the CNS.

Search Strategy

Studies cited in this narrative review were searched on the PubMed database. These searches were done between 1976 and 2022. Key words included in different searches over that period included: tectal transplants, neural precursors, visual system development, transdifferentiation, axon guidance, superior colliculus, cell death, neural transplantation, Parkinson’s disease, glia and synaptic plasticity, organoids, neocortical grafts, and stem cell therapies.

Techniques

Developing tectal tissue is readily identifiable in rat embryos removed from anesthetized pregnant dams and can be dissected away from more ventral parts of the mesencephalon (Figure 1A and B). The tissue mostly contains the developing SC but caudally may also contain immature inferior colliculus, thus in our studies pieces were always trimmed to ensure grafts comprised only developing SC. In most cases, tissue was derived from rat fetuses at embryonic (E) day 15 (the day after mating = E0), although donor tectal material from embryos aged, 14–17, 18, and 20 has also been used (Lund and Harvey, 1981; Golden et al., 1989; Majda and Harvey, 1989; Thanos and Vanselow, 1990; Girman, 1993). Usually, embryonic tectal tissue was grafted via a glass pipette onto the dorsal surface of the immature or mature host midbrain; however, in some studies tissue was transplanted into the optic tract (Harvey et al., 1987; Harvey and Tan 1992), cerebral cortex (Vanselow and Thanos, 1990) and onto the transected optic nerve (Gravina et al., 1990).

Figure 1.

Embryonic rat tectal tissue before and after dissection.

(A) Rat embryos at embryonic (E) day 15 (the day after mating = E0). The tectal region of the dorsal midbrain containing the developing superior colliculus is arrowed. (B) Tectal pieces after dissection from E15 embryos; the midline is arrowed. Scale bars: 5 mm in A, 2 mm in B. Unpublished data.

Grafts quickly established contact with underlying host SC – either unilaterally or bilaterally (Harvey and Lund, 1984), but as the cerebral cortex matured the grafts were pushed more caudally, often resulting in the de novo formation of fiber tracts between the transplant and host SC. Transplantation onto the midbrain in older hosts sometimes required partial removal of the caudal cortex to expose the underlying SC or optic tract. These studies are described later in this review. In some experiments, to maintain dorsoventral and rostrocaudal orientation, fetal tectal tissue was floated onto a small tungsten wire loop and carefully lowered onto the exposed host midbrain (Lund and Harvey, 1981). Although the success rate of these “flat” grafts was less than after injection, this technique markedly increased donor tissue integrity compared to injected grafts and yielded evidence of intrinsic lamination of donor fetal SC in an ectopic location (Figure 2A).

Figure 2.

Sagittal view (rostral to the left) of a mature E15 tectal graft that was transplanted onto the midbrain of a newborn host rat using a small tungsten wire loop to maintain tissue orientation (“flat” transplant).

(A) Neurofibrillar stained section showing a fiber tract (arrow) running caudally from the host superior colliculus. An almost entirely fiber-free region is visible on the dorsal aspect of this graft (short arrow). (B) Adjacent section processed for autoradiography after an injection of ³H proline into the host eye. Note labeled retinal axons in the tract (arrow) and the highly selective and dense innervation of the dorsal fiber-free layer (short arrow). (C) Fiber-free patches superficial (arrow) and deep (asterisk) in a mature tectal graft. (D) Coronal section of normal SC stained for neurofibrils showing a reduced density of fibers in the stratum griseum superficiale (SGS). Scale bars: 200 µm in A–C, 500 µm in D. Cb: Cerebellum; IC: host inferior colliculus; SO: stratum opticum. Unpublished data.

Viability and Growth of Tectal Grafts

At E15, in rats, 20–25% of tectal cells are still to be born and yet to migrate away from their ventricular origin (Mustari et al., 1979), a proportion of these are destined to form the superficial layers (Altman and Bayer, 1981). E15/16 tectal transplants attained a morphology typical of mature grafts by about 17 days after transplantation into neonate hosts, a time course similar to the maturation profile of normal SC in situ (Harvey and Lund, 1984). Furthermore, in the days following transplantation, grafts increased in volume by 20–40 fold, much of this increase occurring between weeks 1 and 3; given that the day of birth (P0) is at about E = 22.5, this is consistent with the rapid growth of the rat SC that normally occurs postnatally (Harvey and Lund, 1984). Loss of 85–95% of cells in the first few days after transplantation has been reported for suspensions of embryonic rat mesencephalon grafted within the adult rat striatum (e.g., Emgård et al., 1999; Sortwell et al., 2000). Gliotic reaction and lack of outgrowth into the surrounding tissue may contribute to this death (Barker et al., 1996), while viability can be enhanced by combining trophic support with biomimetic matrix scaffolds (Wang et al., 2016; Penna et al., 2022). There is naturally occurring neuronal death in the neonatal rat SC (Giordano et al., 1980); however, at least for E15/16 tectal grafts grafted onto the neonatal midbrain, we saw little evidence of significantly increased cell death in the days post-transplantation. As described later, poorer survival was seen if the tectal tissue was held in culture for too long, or when using tectum derived from older donors.

Prelabeling donor fetuses (at E13/E14) or neonatal hosts (E13 or at birth) with tritiated (³H) thymidine revealed limited intermixing of host and graft tissue, with the exception of meningeal coverings and vascularization by host blood vessels (Lund and Harvey 1981). In silver-stained neurofibrillary preparations of mature grafts, much of the graft neuropil contained a dense network of fibers; however, there were also local areas located both superficially and deep that were relatively free of stained fibers (Figure 2A and C). Similar “fiber-free” regions were also seen in grafts more deeply embedded within the host. Note for comparison that a similar paucity of neurofibrillar staining is typical of the superficial layer of normal rat SC (SGS, stratum griseum superficiale, Figure 2D), the layer that normally receives most of the input from the retina. Within the fiber-free areas, or “patches”, in mature tectal grafts, Nissl staining revealed that cells were generally smaller and relatively tightly packed compared to other regions of the grafted neuropil.

Differentiation of Grafts and Host Retinal Innervation

Staining of tectal grafts with Golgi-Cox revealed a variety of neurons based on soma size and dendritic morphology, including marginal, ganglion type I, stellate and horizontal cells, usually found in superficial SC, as well much larger multipolar cells typical of intermediate and deeper SC layers (Harvey and Warton, 1986). Consistent with Nissl staining, clusters of the smaller cell types were sometimes seen in the Golgi material, supporting the homology between tectal graft “patches” and normal SGS. While a proportion of neurons in all layers of the SC are yet to be born at E15 (Mustari et al., 1979; Altman and Bayer, 1981), the range of neuronal classes consistently seen in mature tectal transplants suggests the survival and differentiation of many post-mitotic cells already committed to a particular neuronal phenotype. In addition, neurons of the superficial layers are reported to be produced last, many after E15 (Altman and Bayer, 1981), suggesting the retention of signals that drive differentiation of these newly born neurons even after transplantation.

In initial tectal graft studies, the host retinal innervation into grafts was visualized by injecting ³H proline into host eyes (Lund and Harvey, 1981). In the example in Figure 2A and B, labeled host retinal axons can be seen in a tract (arrow) that connects the host SC with the graft, and there is selective and dense innervation of the dorsal fiber-free zone (short arrow). Comparison of the laminar-like organization seen in this “flat” graft (Figure 2A and B) and normal SC (Figure 2D) suggests some homology, especially since both are targets for retinal axons. This appropriate and selective retinal axon targeting of the superficial layer in these “flat” grafts is remarkable and consistent with studies that describe the importance of repulsive guidance molecules in the formation of layer-specific retinotectal projections in the chick (Banerjee et al., 2016). The rat tectal graft model thus remains a useful model in which to examine and potentially modify similar targeting mechanisms in mammals.

Selective host retinal innervation of fiber-free regions was also consistently seen in tectal grafts injected via a pipette onto the SC in neonate hosts; however, innervation was only to those regions located adjacent to the graft surface (Lund and Hauschka, 1976; Lund and Harvey 1981). Thus the superficial patch in Figure 2C was innervated by host retinal axons but the deeper patch was not. The reason for this is that, as during normal development (Lund and Bunt, 1976), growing retinal axons tend to grow at tissue surfaces, and the initial ingrowth into developing tectal grafts, beginning at 3 days post-transplantation, was always over a graft surface (Harvey and Lund 1984). Localized, superficial “patches” of innervation were seen at about 6–7 days and the pattern was mature by 12–14 days. There was no evidence of a widespread projection that was later refined. The only exception to this occurred when other cell types were intermixed with tectal cells at the time of transplantation, as discussed later in this review. In transplants of fetal tectal tissue injected onto the neonatal host SC, as many as 8–10 retino-recipient patches were evident within mature grafts, sometimes located towards the caudal pole of the grafts and from the host-graft interface. Clearly then, host retinal axons do not necessarily innervate the first neurons they encounter in the transplanted tissue. Consistent with this selective pattern of innervation, there was no host retinal input into the fetal cerebral cortex transplanted heterotopically onto the neonatal rat midbrain (Lund et al., 1982).

One notable observation was made when the host retinal innervation into grafts was examined using two different methods for tracing left and right eye inputs respectively. One host eye was injected with ³H proline and the other eye was removed to allow the identification of degenerating retinal afferents (Fink-Heimer stain) (Lund and Harvey, 1981). Host retinal axons from the left and right eyes were segregated within tracts newly established between the neonatal host midbrain and tectal graft, and within superficial graft patches, the inputs from left and right eyes again remained essentially separate, with little overlap. This inter-eye recognition and subsequent parcellation of axons has been well-documented in amphibians and has also been reported after experimental intervention in other rodents (e.g., So and Schneider, 1978; So, 1979). Competitive interactions between axons from each eye are typical of what occurs during normal visual system development, influenced by the presence of non-correlated electrical activity within each retina (Zhang et al., 2012; Fassier and Nicol, 2021). While much of this work has examined the mechanisms involved in the separation of terminal arbors in target sites, our graft data show that even in experimentally induced fiber tracts the growing retinal axons from the left and right eyes prefer to keep apart, perhaps involving mechanisms similar to those operative during the formation of binocular visual pathways (Murcia-Belmonte and Erskine, 2019).

Electron microscopy of tectal transplants revealed that degenerating host retinal axons made asymmetric synaptic contacts, mostly with dendritic spines on grafted neurons (Lund and Harvey, 1981). Because, in injected grafts, retino-recipient regions are located superficially, are highly localized, and randomly distributed, attempts to record from them were not successful (Harvey et al., 1982; Golden et al, 1989); however, physiological evidence for the afferent drive of transplanted neurons after electrical stimulation of host rat visual cortex was obtained for the first time (see also below, Harvey et al., 1982). In an alternate way of measuring the influence of host afferents on graft function, we examined metabolic activity in E15 tectal grafts 6–22 weeks after transplantation and combined this with the tracing of host retinal input into the grafts after horseradish peroxidase (HRP) or WGA-HRP eye injections (Dyson et al., 1988a). Using radioactive 2-deoxyglucose and cytochrome oxidase staining, we identified focal areas of high metabolic activity in grafts, but only in localized superficial regions that received a high density of retinal input. Other, patch-like regions deep within grafts or in non-connected grafts did not show enhanced activity. In some cases, the levels of cytochrome oxidase activity and 2-DG uptake in the superficial patches were qualitatively similar to those seen in the host superficial SC. Thus, we were able to show, for the first time, that the activity of cells in grafted fetal tissue can be modified long-term by host innervation; in the case of tectal grafts, this input was via a sensory system capable of increasing metabolic activity in areas of a transplant that received appropriate host innervation.

Evidence suggesting homology between the localized, retino-recipient fiber-free regions in fetal tectal grafts and the SGS of normal SC has been presented. This homology is further supported by additional studies that revealed a remarkable number of shared features between normal SGS and the patches in grafts. For example, typical of the normal SGS, the localized regions in grafts stained intensely for acetylcholinesterase (AChE) (Harvey and MacDonald, 1985) and contained a high density of α-bungarotoxin binding sites (Tan and Harvey, 1987). The selective expression of these markers in tectal grafts was apparent even in localized regions that were not innervated by host retinal axons, indicating a degree of autonomous differentiation within the maturing graft neuropil, not dependent on extrinsic non-collicular influences. Furthermore, immunohistochemical analysis of the glial architecture in mature E15 tectal grafts revealed a prolonged increase in astrocyte reactivity and large numbers of differentiated oligodendroglia, associated with a dense network of myelinated axons (Harvey et al., 1993). Most importantly, characteristic of the SGS of normal SC, localized areas that stained intensely for AChE contained few oligodendroglia and relatively little myelin, indicating that the factors that influence the distribution of myelinating cells in the SC may also be operative in developing fetal tectal grafts.

It is important to note here that the pattern of glial organization and degree of reactivity in tectal grafts is very different from that seen in fetal cerebral cortex heterotopically transplanted onto the neonatal rat midbrain (Björklund et al., 1983; Syková et al., 1999). Compared to homotopic cortex-to-cortex grafts, the embryonic cortex grafted onto the midbrain developed bands of white matter and intense astrocytic reactivity that persist for many months. Increased microglial activity, enhanced deposition of proteoglycans, and altered expression of Alzheimer-related proteins were also seen, affecting a number of physiological parameters (Syková et al., 1999). Thus, although neuronal development in cortex-to-midbrain grafts appears to be relatively normal (Jaeger and Lund 1980, 1981), other aspects of graft development are highly abnormal, and functional sequelae resulting from such changes need to be taken into account in any heterotopic graft trials. Based on these fetal graft data, and given the importance of astrocytes and microglia in regulating synaptic function, plasticity, and neural circuit activity (Cornell et al., 2022; Lyon and Allen, 2022; Oliveira and Araque, 2022), the interaction between resident (or donor) glia and transplanted neural precursors or neurons derived from precursors should be taken into account when planning homotopic or heterotopic graft trials (Barker et al., 1996; Fjodorova et al., 2017; Fainstein and Ben-Hur, 2018; Hastings et al., 2022; Limone et al., 2022; Voronkov et al., 2022).

In Situ Hybridization Studies

A comprehensive in situ hybridization study using oligonucleotide probes labeled at the 3 prime ends with [³⁵S]dATP revealed the distribution of cells in the adult rat SC expressing mRNAs for seven neurotransmitters/neuromodulators (Harvey et al., 2001). A novel finding was the existence of a sub-laminar distribution of neuropeptides within the superficial SC layers. While cells expressing glutamic acid decarboxylase mRNA (GAD, marker for gamma aminobutyric acid) were the most abundant and distributed throughout the depth of the rat SGS, in the superficial SC the expression of mRNA for preprotachykinin (PPT - substance P), cholecystokinin (CCK) and somatostatin (SOM) was largely restricted to sub-laminae: PPT cells most superficial, CCK cells in an intermediate zone, and SOM cells located deeper in SGS, adjacent to stratum opticum (Harvey et al., 2001). Figure 3 (right-hand panel) presents data obtained using these radioactive probes showing the characteristic distribution of GAD, PPT, CCK, and SOM-expressing cells in rat SGS. In these animals, under anesthesia, one eye had been removed at birth (Harvey et al., 1995); the panel on the left (Figure 3) is contralateral to the enucleated side and shows a smaller SC and noticeably reduced levels of expression of GAD, PPT and CCK mRNAs in the SGS, but with only minor if any impact on SOM expression. Quantitative analysis of total cell counts in tissue sections and density of silver grains in autoradiogram films confirmed these qualitative findings. Overall these data show that retinal input enhances the expression of GAD and some neuropeptide genes in the superficial SC, but that there is ongoing basal expression of these genes in the absence of visual drive from the contralateral eye. This is an important observation of relevance to the graft work discussed below.

Figure 3.

In situ hybridization autoradiographic images of coronal sections of the adult rat superior colliculus (SC).

Images show mRNA expression for (A) the precursor for substance P (preprotachykinin, PPT), (B) cholecystokinin (CCK), (C) glutamic acid decarboxylase (GAD), the synthesizing enzyme for gamma aminobutyric acid and (D) somatostatin (SOM). This animal had been unilaterally enucleated at birth; note the relatively decreased label in the superficial SC layers on the left of the images, contralateral to the enucleated eye. All oligonucleotide probes were labeled at the 3 prime ends with [³⁵S]dATP (for details see Harvey et al., 2001). As described elsewhere (Harvey et al., 2001), there was sub-lamination within the stratum griseum superficiale of neurons expressing PPT, CCK, or SOM. Scale bar: 500 µm. SGS: Stratum griseum superficiale; SO: stratum opticum. Unpublished data.

Gene expression was analyzed in E15 tectal tissue transplanted onto the midbrain of neonate rat hosts (Harvey et al., 1994). An example is shown in Figure 4. Two AChE dense patches are labeled (arrows, Figure 4A), one at the graft surface and one deeper within the mature graft neuropil. Adjacent sections were probed for expression of GAD, PPT, or SOM mRNA. In strong support of the homology between AChE patches in grafts and normal rat SGS (Figure 3), these localized regions contained many GAD-expressing cells (arrow in Figure 4B). It is noteworthy that there was a differential distribution of PPT- (Figure 4C) and SOM- (Figure 4D) expressing cells in the superficial AChE patch. The CCK cells are located towards the graft surface and the SOM-expressing cells are situated more deeply, typical of their sub-laminar distribution in normal SGS.

Figure 4.

In situ hybridization autoradiographic images of coronal sections of transplanted fetal tectal tissue.

(A) Coronal section of a mature E15 tectal graft showing two patches stained intensely for acetylcholinesterase (AChE, arrows). (B–D) Adjacent sections processed for autoradiography showing the distribution of transplanted neurons expressing (B) glutamic acid decarboxylase (GAD), (C) the precursor for substance P (preprotachykinin, PPT), and (D) somatostatin. Oligonucleotide probes were labeled at the 3 prime ends with [³⁵S]dATP (Harvey et al., 2001). Note increased GAD expression in the AChE dense regions, and in the superficial patch, the PPT-expressing neurons are dorsal to those expressing SOM. Scale bars: 500 µm. Unpublished data.

One further example of mRNA expression in a tectal graft is shown in Figure 5; here the AChE patch is deep within the graft (Figure 5A). Again, note the concentration of GAD mRNA-expressing cells in this patch (Figure 5B). The patterns of expression of GAD, SOM, PPT and CCK mRNA expression in and around the AChE dense region are shown at higher power in Figure 5C–F respectively, and are digitally mapped in Figure 5G. Note that cells expressing low levels of PPT mRNA are located towards the center of this deep patch, CCK-expressing cells have a more intermediate location, and cells expressing relatively high levels of SOM mRNA form a peripheral ring around the patch border (Figure 5D and G). These features again reflect the differential distribution of these peptide-expressing cells in tectal grafts, in a manner markedly similar to that seen in normal superficial SC. The difference here is that the circular distribution of SOM-expressing cells around the edge of this deep AChE dense patch suggests that the normal horizontal lamination pattern had been lost and the presumptive SGS had curled into a discrete ball. As described earlier, deeply located fiber-free/AChE dense patches in grafts do not receive host retinal input; the sustained expression of these various mRNAs in fetal tectal transplants is consistent with the basal level of expression seen in normal rat superficial SC deprived of retinal input from birth (Figure 3).

Figure 5.

Expression of different neuropeptide mRNAs in a fetal tectal transplant.

(A) Coronal section of a mature E15 tectal graft showing a central region stained intensely for acetylcholinesterase (AChE, arrow). (B–F) Adjacent sections processed for autoradiography showing the distribution of transplanted neurons expressing either (B, C) glutamic acid decarboxylase (GAD), (D) somatostatin (SOM), (E) the precursor for substance P (preprotachykinin, PPT), or (F) cholecystokinin (CCK) mRNA. (G) These various neurons mapped throughout the graft using an MD1 digitizer (Harvey et al., 2001). The AChE dense patch contains a high density of GAD-expressing neurons. Note that within this patch PPT expressing neurons tend to be found in the center, SOM-expressing neurons around the periphery, and CCK-expressing neurons are mostly located between the two types. Although this is a circular patch not a layer, this differential distribution is remarkably similar to that seen in the stratum griseum supericiale of the normal superior colliculus. Scale bars: 500 µm in A, B, 250 µm in C–F, 500 µm in G. C: Cerebellum; IC: inferior colliculus. Unpublished data.

In Vitro Manipulation of Tectal Grafts

The data described in the preceding paragraphs clearly demonstrate: (i) that neurons destined for the SGS of normal SC aggregate together in localized regions in fetal tectal grafts, (ii) that there is a level of intrinsic organization within these patches that resembles the sub-lamination seen in normal SGS, and (iii) that when grafted onto the SC in neonate rat hosts, ingrowing retinal axons specifically innervate localized, presumptive SGS regions, but only if these regions are located adjacent to the graft surface.

These characteristic features of fetal tectal transplants are remarkably robust, even when the tissue is obtained from older donors or is manipulated in some way prior to transplantation. Thus, AChE dense/fiber-free patches were also seen in fetal tectal grafts obtained from E18, E20, or newborn donors (Majda and Harvey, 1987). Viability and graft size were inversely related to donor age, indicating increased loss of cells in grafts derived from older donors. Indeed, only one small graft was identified when using tectal tissue from newborns. Although grafts from older fetal donors were smaller in volume, the selective host retinal innervation of circumscribed, superficial AChE dense patches remained. Similarly, E15 tectal tissue cultured as explants for 3–14 days in vitro (DIV) prior to grafting developed localized regions in Nissl, neurofibrillar and AChE stained material when examined many weeks post-transplantation. All 3 DIV and 7 DIV grafts survived and grew substantially in volume, but only 50% of 14 DIV grafts survived and all were small in size. Importantly, in all cases, the specificity of host retinal innervation to AChE dense patches was retained (Harvey et al., 1988). Note that, whereas all E15 tectal tissue cultured for 7 DIV survived and matured after transplantation, only one small surviving graft was identified using newborn donors (an equivalent post-conception age to E15 plus 7 days in vitro); thus the developmental age at which tissue is initially obtained is a key factor in transplant viability and growth. The results also demonstrate that many types of developing tectal neurons survive for a period of time without extrinsic input and that those cells destined for the superficial layers continue to express a specific identity that allows them to be selectively recognized by ingrowing host retinal axons.

In further in vitro/in vivo experiments, E15/E16 tectal tissue was dissociated into a cell suspension and then re-pelleted via centrifugation prior to transplantation via a pipette onto the SC of newborn hosts (Vukovic et al., 2007). The mature cell architecture in these reaggregated grafts was markedly similar to that seen in direct, undissociated grafts, most notable was the consistent development of the previously described localized AChE dense patches containing mostly small close-packed neurons (Lund and Harvey, 1981; Harvey and MacDonald, 1985). These localized regions were seen both deep and superficially within the graft neuropil; moreover, host retinal innervation to these grafts was, as before, confined to the superficially located patches. These observations suggest that cells destined for the SGS have a specific affinity for each other in the reaggregated transplants and that the dissociation-reaggregation procedure does not affect retinotectal target recognition. Note here that this approach allows the incorporation of other cell types during the reaggregation phase in order to analyze how these added cells influence tectal graft differentiation and connectivity (Vukovic et al., 2007). Adding olfactory ensheathing glia to fetal tectal tissue had little impact on the selective pattern of host retinal innervation of grafts; however, the addition of Schwann cells resulted in more scattered retinal input not confined to AChE patches, with axons sometimes growing among clusters of co-grafted Schwann cells. To date, this is the only example of loss of retina-target specificity in tectal grafts, suggesting that Schwann cells added to the CNS may promote axon growth but may also interfere with target recognition.

Other Afferent Inputs and Efferent Connections

The high level of selectivity shown by host retinal axons for the circumscribed fiber-free, AChE dense regions in tectal grafts were less obvious in other types of afferent input from the host brain. After lesions of the host occipital cortex, degenerating profiles were scattered throughout the graft neuropil but were sparse within the fiber-free patches, even in the deeper patches that did not receive retinal input (Lund and Harvey, 1981). Electron microscopy revealed the presence of host cortical asymmetric synapses with dendritic spines and small dendrites of cells within grafts (Harvey et al., 1982). In these pioneering neural graft experiments, microelectrodes were used to record from single neurons in transplants and the host occipital cortex was excited via an array of stimulating electrodes. Approximately 12% of grafted cells were activated after host cortical excitation, some with latencies suggestive of monosynaptic drive (Harvey et al., 1982). Taken together, with the 2-deoxyglucose and cytochrome oxidase data linking increased metabolic activity in grafts with host retinal innervation (Dyson et al., 1988a), the cortical physiological data indicate a complex capability of host brains to influence the functionality and behavior of grafted fetal neurons. Tectal grafts also send some axons out into the host brain, with efferents identified in the host midbrain, pretectum, and pontine tegmentum (Lund et al., 1982). While no long-distance projections were evident to the thalamus, pons, or spinal cord, the host functional drive of grafted neurons might indicate that novel host-graft-host connections are possible in this system.

In the early study on retinal and cortical input into tectal transplants, degeneration in grafted tissue was heaviest when lesions affected the extrastriate visual cortex (Lund and Harvey, 1981). To further examine cortex-graft projections, HRP or fluorescent tracers were injected into mature tectal grafts to reveal the anatomical distribution of retrogradely labeled host cortical neurons (Harvey and Lund, 1981; Worthington and Harvey, 1990). Confirming the anterograde degeneration data, the heaviest projections into grafts arose from lamina V neurons in extrastriate visual regions areas 18a and 18b, with less label in area 17 and even more sparse projections from the auditory, cingulate, and somatosensory cortex. In the best case, approximately 6000 graft-projecting neurons were counted in one host cortex (Harvey and Lund, 1981). To discern the presence of any topography in the cortical projection into grafts, three different tracers were injected into different parts of mature E15 tectal grafts and the distribution of neurons labeled with each tracer was mapped (Worthington and Harvey, 1990). Statistical analysis revealed that the pattern of cortical innervation of grafts from both areas 17 and 18 was non-random, but no coherent topographic maps typical of the normal corticotectal projection were found. In addition to host cortical innervation, careful mapping revealed that about 50 host regions could potentially contain retrogradely labeled neurons after graft HRP injections. Areas that commonly contained at least some neurons that sent axons into grafts included the pretectum, parabigeminal nucleus, and brachial region of the inferior colliculus (Harvey and Lund, 1981).

Areas 18a and 18b normally project to the intermediate layers of rat SC while area 17 axons distribute mostly within the SGS and dorsal stratum opticum. This may explain why few degenerating cortical axons were seen in the fiber-free patches in tectal grafts. The fact that the more deeply projecting extrastriate axons had a greater capacity for growing into tectal graft neuropil suggests that mere proximity to the host-graft interface is not sufficient to explain the density or selectivity of host innervation. Indeed, only a relatively small number of cells in the host SC closest to the point of connection was retrogradely labeled after HRP tectal graft injections (Harvey and Lund, 1981). Comparison of tectal graft connectivity with the afferent and efferent connections of fetal retina and cortex transplanted onto the neonate rat midbrain reinforces the conclusion that such grafts are not merely “passive receivers” of nearby host axons but display region-specific selectivity (Lund et al., 1982). Tissue-specific differences are also apparent in the distribution within the host brain of efferents from fetal retinal, cortical or tectal tissue transplanted onto the rat midbrain (Jaeger and Lund, 1980; Lund et al., 1982; McLoon and Lund, 1983; Steedman et al., 1983). Other studies involving the use of, for example, fetal cortex derived from different parts of the cerebral hemisphere have also revealed subtle differences in the pattern of projections into the host that relate to the areal origin of the grafts (Gaillard and Roger, 2000; Pinaudeau et al., 2000).

The intermediate layers of the rat SC contain a dense network of fibers immunopositive for choline acetyltransferase (ChAT) (Tan and Harvey 1989). Far fewer ChAT-positive profiles are seen in SGS, most of these likely originating in the dorsal and ventral sub-nuclei of the ipsilateral parabigeminal nucleus (Tan and Harvey, 1989). Cholinergic axons were scattered throughout tectal grafts, the pattern of staining qualitatively resembling that seen in the intermediate layers of normal SC. In contrast, serotoninergic axons are normally found in high density in rat SGS. However, within tectal grafts, these host-derived fibers were broadly scattered throughout the neuropil with no evidence of a higher density of innervation within the fiber-free, AChE-dense patches homologous to SGS (Harvey and MacDonald, 1987). This clearly differs from the highly selective innervation of these regions by host retinal afferents, suggesting that the two fiber systems do not recognize the same cues within the developing graft neuropil. This in turn implies differences in the way that these afferent systems establish their pattern of innervation in normal SC.

Transplantation into Older Host Brains – Regeneration

When transplanted onto the intact SC of rats aged between 3 and 21 days (Figure 6), ingrowth of ³H proline labeled retinal axons was consistently seen in fiber-free patches in fetal tectum grafted in day 3 hosts and also occasionally in tissue grafted into 7-day-old hosts (Figure 6A and B). A retinal projection into fetal tectal tissue transplanted onto the unlesioned SC in 14-day hosts was rarely seen (Figure 6E), and even when there was clear host-graft continuity (Figure 6D) input was sparse and restricted to the host-graft interface. This decrease in innervation plasticity parallels the maturation of retinal input into normal SC, which in the rat continues until about postnatal day 5 or 6 and originates primarily from late-born retinal ganglion cells (Dallimore et al., 2002). In contrast, after lesions of the host occipital cortex, large numbers of degenerating profiles (Fink-Heimer stain) were evident in tectal tissue grafted into 7 (Figure 6C) and 14-day hosts (Figure 6F), although the density was reduced and ingrowth more limited in older hosts. In tissue transplanted to 14-day-old host rats, occasional retrogradely labeled pyramidal neurons in the host visual cortex were seen after injections of HRP into grafts. The extended period of host-graft plasticity shown by the cortex compared to the retina presumably reflects the relative immaturity of corticotectal projections in the postnatal rat (de Carlos and O’Leary, 1992).

Figure 6.

E15 tectum grafted onto the superior colliculus (SC) of older hosts.

(A–C) 7-day-old hosts. (A) Neurofibrillary stained coronal section showing graft (T) at the junction between inferior colliculus (IC) and SC. Note the small fiber-free region (arrowed) (B), adjacent autoradiographic section to “A” showing a patch of ³H proline labeled retinal axons (arrow) localized to the fiber-free region arrowed in A. (C) Widespread distribution of degenerating profiles in a tectal graft after lesion of the host visual cortex. (D–F) 14-day-old host. (D) Sagittal neurofibrillary stained section showing a small interface (arrow) between a graft (T) and underlying host stratum griseum superficiale (SGS). (E) Autoradiographic section adjacent to D showing lack of ingrowth of ³H proline labeled retinal axons into the graft neuropil (arrow). (F) Higher power view of this interface showing degenerating profiles in the graft after a host visual cortex lesion. Scale bars: 500 µm in A, 200 µm in B, E, 50 µm in C, F, 150 µm in D. Unpublished data.

When afferents that normally project to the SC are deprived of this target, enhanced axon ingrowth into transplanted fetal tissue can be the result. Here again, the emphasis is on host retinal axon ingrowth into tectal grafts, in this case, the extent and pattern of host retinal innervation after central lesions in juvenile or adult rats. After unilateral lesions of the brachium of the SC and pretectal region in older (10–18 days old) rats, there was spontaneous regrowth of retinal axons for several millimeters in glia-connective tissue membranes that from over the lesion site (Dyson et al., 1988b). Ultrastructural and immunohistochemical analysis revealed that regenerating retinal axons were often closely associated with astrocytic processes and that a proportion of the axons was remyelinated by oligodendroglia (Dyson et al., 1988b). Retinal axons that reached the superficial layers of post-lesion SC either failed to grow into the SC neuropil or penetrated for only a short distance. However, when fetal (E15) tectal tissue was transplanted into these brachial lesion cavities, many regenerating retinal axons grew into graft tissue and – as with transplantation into newborn hosts – they specifically innervated localized AChE, fiber-free patches within the grafts (Harvey et al., 1987). Similar to the lack of ingrowth into the embryonic cerebral cortex transplanted onto the neonatal midbrain (Lund et al., 1982), no retinal innervation was evident when the fetal cortex was transplanted into the brachial lesion sites (Harvey et al., 1987).

The spontaneous regeneration of identified retinal axons within newly formed glia-connective tissue membranes that form over lesion sites was also seen after unilateral brachial lesions in adult rats (Harvey and Tan, 1992). Remarkably, this axon regrowth extended for up to 6 mm from the rostral edge of the injury site. When embryonic tectal tissue was grafted into these brachial lesion cavities, regenerating adult retinal axons that contacted the developing grafts were again found to form dense projections that were confined to localized AChE dense patches in the transplant neuropil. There is evidence that at least some guidance cues are still present in the deafferented adult SC (Bähr and Wizenmann 1996; Rodger et al., 2005). Consistent with this, adult retinal ganglion cell axons stimulated to regenerate through autologous peripheral nerve grafted onto the transected optic nerve and distally inserted into the SC only form dense arbors and presumed excitatory synaptic contacts with neurons in the SGS (Aguayo et al. 1990; Yin et al., 2019). Overall, these observations show that (i) at least some adult retinal axons possess an intrinsic ability to regrow for some distance after injury, and (ii) that adult retinal axons retain the capacity to preferentially innervate presumptive target regions, both within the immature and mature SC.

Conclusion

There is now a considerable body of research investigating the normal developmental mechanisms that influence neurogenesis, neuronal polarity, process expression, axon guidance, laminar-specific innervation, and the subsequent recognition and formation of synaptic contacts with appropriate target neurons (e.g., Williams et al., 2010; Banerjee et al., 2016; Villalba et al., 2021; Zang et al., 2021; Qi et al., 2022). Historically, the experimental fetal graft approach has proved to be a useful tool in studying the limits of neural plasticity and capacity for repair in developing and adult brains. In this context, one aim of this review has been to revisit studies that examined the differentiation and connectivity of fetal tectal tissue transplanted into the developing or mature rat brain. In summary, these visual system experiments have provided information on (i) intrinsic programs of cytoarchitectural and neurochemical tissue differentiation, (ii) retinal ganglion cell axon-axon interactions, (iii) selective affinities between some axon populations and their targets during both development and adult regeneration, (iv) the existence of host functional drive of grafted tectal neurons, and (v) the relative impact that introduced non-neural cells (Schwann cells) can have on the specificity of retinal axon-target recognition.

In a broader context, using either solid or dissociated embryonic neural grafts it has proved possible to obtain: (i) a better understanding of the spatial and temporal factors that affect the differentiation of grafted cells into neurons and/or glia, and (ii) how to influence the effective formation of new, functionally effective, graft-host interconnections. Based on a wide range of experimental fetal graft studies there have been several clinical transplant trials using donor human embryonic tissue. However, because of ethical concerns most therapeutic strategies now focus on using neural precursors, neurons derived from human pluripotent stem cells, or neurons transdifferentiated from other cell types, in order to replace lost or damaged neurons and hopefully to restore behaviorally appropriate functional circuits (e.g., Ferrari et al., 2018; Han and Xu, 2020; Aversano et al., 2022; Limone et al., 2022; Van den Bos et al., 2022; Xu et al., 2022).

Because these studies cannot necessarily rely on the intrinsic differentiation of specific neuronal types, known to occur within embryonic tissue transplants, the precursor/stem cell approach usually requires external agents and/or genetic manipulation to drive appropriate neuronal differentiation or trans-differentiation prior to transplantation. Using modern molecular techniques, cells can be genetically marked to enable careful monitoring of long-term viability and integration. Often however, a specific cell type is generated, unlike fetal neural tissue grafts that can contain the entire gamut of developing neuronal and glial classes, potentially important given the role of the local microenvironment in the differentiation of neural precursors (Snyder et al., 1997; Wang et al., 1998). In addition, based on earlier embryonic graft studies, and given the impact of glia on integration, synaptogenesis, and neural processing (Cornell et al., 2022; Lyon and Allen, 2022; Oliveira and Araque, 2022), the presence as well as the reactive state of the donor or host astrocytes and microglia (Barker et al., 1996; Harvey et al., 1997; Limone et al., 2022; Voronkov et al., 2022) may be an important consideration when designing stem cell/precursor cell graft studies. Indeed, while dissociated cell suspensions are often used, increasingly the cells are integrated into three-dimensional polymer networks and involve the creation of organoids and more complex architectures (Wang et al., 2016; Roth et al., 2021; Eichmüller and Knoblich, 2022; Kelly and Paşca, 2022; Penna et al., 2022).

In summary, while the tectal transplant research described here provides pertinent information about development and plasticity in visual pathways, a more general aim of this retrospective is to highlight how the review of the historical fetal graft literature may aid in an appreciation of the factors that influence survival, differentiation, connectivity, and functionality of engineered cells and organoids when they are transplanted into the CNS.

Additional file: Open peer review reports 1 ^{(94.1KB, pdf)} and 2 ^{(95.1KB, pdf)} .