The present work represents the first study of the glial cytoarchitecture during the embryonic development of a cartilaginous fish using markers of radial glia and cell proliferation. We analysed the expression of GFAP, BLBP, GS, and PCNA by means of immunohistochemistry in the developing telencephalon of
S. canicula in S25–S34 embryos, as well as in posthatching (early) juveniles. On the basis of major morphological characteristics, Ballard et al. (
1993) established 3 periods during the development of
S. canicula: (1) early development, from the day of fertilization until S26 of development; (2) intermediate development, from S27 to S31; (3) and late development, from S32 to hatching. More recently, equivalences among brain developmental stages of
S. canicula and mouse, birds and zebrafish have also been established (Rodríguez-Moldes et al.
2017). From the day of fertilization until hatching, the catshark embryogenesis takes around 6 months (depending on water temperature), which is a lengthy period compared to the 72-h of zebrafish embryogenesis. Therefore, sharks provide a good time window for studying certain developmental processes that might be ignored by a rapid development (Quintana-Urzainqui et al.
2015; Rodríguez-Moldes et al.
2017). In addition, sharks exhibit an evaginated telencephalon, instead of an everted telencephalon as in teleost fishes, which ease comparisons between developmental periods and developmental events in sharks and other tetrapods.
We show that the expression pattern of the three glial markers studied follows a sequential expression pattern along development: early (GFAP), intermediate (GFAP and BLBP) and late (GFAP, BLBP and GS). We found that these markers mainly label RGCs in the ventricular zone of the developing telencephalon and that their expression remains in juveniles. In addition, some non-ventricular positive cells were found during development (BLBP- and GS-immunoreactive cells) and in juveniles (GS-inmunoreactive cells) which do not have morphology of RGCs. Ventricular zone RGCs usually co-express all glial markers studied in late development, but exhibit some degree of heterogeneity in the dorsal telencephalon. In addition, double immunofluorescence experiments have shown a high degree of colocalization between the proliferation marker (PCNA) and glial markers such as GFAP or BLBP. The comparative aspects across vertebrates and the developmental implications regarding the appearance of these cells will be discussed below.
GFAP, BLBP and GS are expressed in cells with morphology of RGCs during embryonic development and in postnatal individuals
In the telencephalic hemispheres of the lesser spotted dogfish, GFAP immunoreactivity is detected in embryos from early developmental stages (S25) until the end of embryonic period (S32–S34), as well as in posthatching juveniles. In mouse embryos, GFAP mRNA was first detected in the germinal zones of the forebrain on E15 and it increases on E17 (Fox et al.
2004), which correspond to late embryonic stages (Rodríguez-Moldes et al.
2017). Mamber et al. (
2012) showed that GFAPδ transcripts are expressed from E12 in radial glia, but GFAP-immunoreactive structures are only detected from E18 onwards. RGCs expressing GFAP were also detected in the forebrain of late embryonic stages of
Xenopus laevis (S36) (Messenger and Warner
1989). Studies in zebrafish embryos have shown that the expression of GFAP in the central nervous system (CNS) begin at the time of formation of the first somite (Thisse et al.
2001) but, contrary to that observed in
S. canicula, mouse and
Xenopus its expression decreases during the embryonic period (Marcus and Easter
1995). The early developmental expression of GFAP in chondrichthyes and teleosts contrasts with its later expression in mammals. In mammals, the establishment of the RGCs scaffold during forebrain development necessary for neuronal migration occurs previously to the neurogenic period (Martynoga et al.
2012). Quintana-Urzainqui et al. (
2015) established that neurogenesis and neuronal migration in the catshark telencephalon mainly occurs during the intermediate developmental period (S27–S31). Like in mammals, GFAP expression in the telencephalon of the catshark occurs before neurogenesis; therefore, the presence and distribution pattern of GFAP-positive radial processes in the telencephalic hemispheres at these stages indicate that they could act as guide-posts during neuronal migration, as do RGCs in the mammalian cortex (Rakic
1972). Regarding the distribution of GFAP within glial cells, in early embryos GFAP is mainly located in subpial endfeet. As embryonic development progresses, GFAP is observed in cell bodies lining the ventricle and its radial processes and subpial endfeet. These GFAP-immunoreactive cells morphologically resemble glial cells described in the developing cerebral cortex of mammals (for review see: Alvarez-Buylla and Kriegstein
2013). The distribution pattern of GFAP in catshark embryos also resembles that shown in studies of amphibians (Messenger and Warner
1989) and teleosts (Marcus and Easter
1995; Arochena et al.
2004). A similar distribution pattern is found in early juveniles, though GFAP expression decreases with respect to that observed in embryos: the thickness of the ependyma decreases with respect to the ventricular zone of embryonic stages, and less GFAP immunoreactive ependymocytes or tanycytes are visualized in contact with the ventricle. In any case, radial glial processes and subpial endfeet still contain GFAP.
The distribution pattern of GFAP observed in the telencephalon of posthatching juveniles is in agreement with the GFAP distribution pattern reported in the adult brain of reptiles (Lazzari and Franceschini
2006), amphibians (Kirkham et al.
2014) and diverse bony fishes (carp: Kálmán
1998; trout: Díaz-Regueira and Anadón
1998; grey mullet: Arochena et al.
2004; zebrafish: Grupp et al.
2010; Than-Trong and Bally-Cuif
2015; african lungfish: Lazzari and Franceschini
2004) and chondrichthyes (sharks, rays and skates: Wasowicz et al.
1999; Kálmán and Gould
2001; Ari and Kálmán
2008b; and chimaeras: Ari and Kálmán
2008a), in which it has been shown that ependymal and radial glial cells are the predominant astroglial cell type. However, GFAP positive cells are found in the ventral telencephalon of catshark posthatching juveniles, which is not the case in teleost fish, such as zebrafish (Ganz et al.
2010; März et al.
2010), where GFAP expression is essentially absent. Zebrafish and catshark show many differences in their embryonic development; zebrafish is fast developmental species, while catshark is slow developmental species, and the telencephalon morphogenesis process is also different between both species. In addition, galeomorph sharks have relatively large brains, similar in size to those of the mammal and bird brains with similar body weights (Northcutt
1981) and its size brain increases from posthaching period until adulthood. During mammalian embryonic development the length of radial glia processes has been related with the radial growing of the neural tube, in addition, the number of radial glial cells determine the size of a brain region, and at the same time radial glia cells act as guides in the neuronal migration process (for review see Götz
2013). The presence of GFAP in the ventral telencephalon of catshark may be related with the continuous growth of the brain.
In the adult brain of skates, GFAP-positive cells with stellate morphology typical of astrocytes were also found (Kálmán and Gould
2001). Although we have not observed astrocyte-like cells in the telencephalic hemispheres of juveniles, GFAP-immunoreactive cells with astrocytic morphology can be visualized in other brain areas.
BLBP-immunoreactivity in RGCs of the catshark telencephalic hemispheres is detected after GFAP-immunoreactivity, around S28 (present results). At this developmental stage BLBP-immunoreactivity is weak and restricted to ventricular cells, but as development proceeds (S31–S32) its expression increases considerably and can also be observed in cells away of the ventricular zone (present results). In mammals, BLBP is a marker of radial glia and its expression is first visualized in RGCs present in the caudolateral regions of the forebrain on E12.5 and, by E14.5–E16.5, BLBP expression expands to medial regions of the forebrain (Anthony and Heintz
2008). Rodríguez-Moldes et al. (
2017) established equivalences between catshark S27–S31 and mouse E10.5–14.5 stages, which suggests striking similarities in the temporal expression of BLBP between catshark and mouse. In
Xenopus BLBP-positive cells line the telencephalic ventricles at stages 37/38 (Moreno and González
2017). These
Xenopus stages belong to a late embryonic period; however, as they show active neurogenesis it can be comparable to the catshark intermediate period. In the zebrafish forebrain, BLBP positive cells were described lining the ventricles in adults and 21 dpf juveniles (Grupp et al.
2010). We are not aware of studies of BLBP expression during zebrafish embryogenesis, precluding further comparisons.
A subpial layer of BLBP immunoreactive cells has been detected from S32–33 onwards. GFAP-immunoreactive subpial astrocytes have been described in the telencephalon of adult rays (Kálmán and Gould
2001), however, the morphology of the catshark BLBP-positive subpial cells resembles more to RGCs than to ordinary astrocytes, suggesting interspecific differences. In contrast with other vertebrates, the blood–brain barrier in cartilaginous fishes is not formed by endothelial cells but is formed by glial cell processes (Bundgaard and Cserr
1981). Whether the submeningeal glial cells observed in cartilaginous fishes contribute to form the blood–brain barrier need to be further investigated.
In catshark embryos, numerous BLBP-positive cells line the telencephalic ventricles and their immunoreactive radial processes span the telencephalic walls from the ventricular zone to the pia. Scattered BLBP-immunoreactive cells outside the ventricular zone were also observed. However, in juveniles BLBP expression was restricted to ventricular glial cells and their subpial endfeet (present results). In mammals, BLBP is transiently expressed in the RGCs in the embryonic ventricular zone. It has been proposed that BLBP is involved in the establishment of the radial processes of RGCs in the developing brain (Feng et al.
1994). Studies performed in zebrafish (März et al.
2010; Diotel et al.
2016) and
Xenopus (D’Amico et al.
2011) have reported the presence of BLBP in RGCs of juvenile and adult brains, with a distribution pattern similar to that observed in the telencephalon of juvenile catshark. Likewise, BLBP expression in the catshark telencephalon might be related with the formation of the radial glia scaffold that persists in juveniles.
GS expression in the telencephalon of catshark appears in the later embryonic period previously to hatching, and the distribution pattern evidenced during embryonic development is similar to that of BLBP. In mammals GS is present in astrocytes and RGCs; in rat embryos, GS immunoreactivity was first observed in ventral ependymocytes of caudal neural tube at embryonic day 14 (E14) and radial glia shows GS immunoreactivity at E16 (Akimoto et al.
1993). As in the rodent brain, in the catshark telencephalon GS is expressed in the later period of embryonic development. However, in teleosts and reptiles GS occurs earlier in development (Romero-Alemán et al.
2003; Grupp et al.
2010). The enzyme GS has a critical role in the metabolism of neurotransmitters released by neurons such as glutamate, which is uptaken by glia and converted by GS to glutamine to prevent neurotoxicity (Norenberg
1979). GS has been reported as a marker of mature glial cells in the retina of this catshark species (Bejarano-Escobar et al.
2012; Sánchez-Farías and Candal
2016). In juveniles GS was detected in ependymal cells (or tanycytes), which is in agreement with results observed in the adult human and mouse brain GS (Norenberg
1979; Norenberg and Martinez-Hernandez
1979; Bernstein et al.
2014); moreover, GS positive cells with morphology of RGCs were also found in the adult brain of amphibians (Kirkham et al.
2014) and fishes (teleosts: Grupp et al.
2010; Than-Trong and Bally-Cuif
2015; chondrichthyes: Ari and Kálmán
2008a,
b).
We can conclude that GFAP, BLBP and GS label cells with radial glia morphology during different developmental periods in the telencephalon of catshark. The onset expression timing of the three glial markers during catshark embryonic development shows many similarities with mammals. Present results reinforce the hypothesis that the basic developmental processes in this species (main molecular events, main morphogenetic processes) are quite similar to those of mammals (see also Rodríguez-Moldes et al.
2017 and references therein). However, in the catshark telencephalon RGCs are maintained after hatching, as in other species of fishes, which has been related with the high neurogenic potential of the glia in adult fishes (reviewed by Than-Trong and Bally-Cuif
2015). Present results also indicate that the lesser spotted dogfish is an excellent model to study the development of the glial system in an evo-devo context.
BLBP and GS adventricular positive cells may belong to the oligodendroglial lineage
In embryonic and juvenile stages of catshark, scattered GS-positive cells are observed in the mantle zone of the medial and the caudal dorsal pallium (pars centromedialis). These cells also express BLBP in embryos, but not in juveniles (present results). They are characterized by their large size, the eccentric nucleus and a thick process, features that resemble to those of oligodendrocytes. Del Río Hortega (
1928) made the first description of oligodendrocytes and classified them into four types attending to their size, shape, processes or even their interaction with axons (reviewed by Pérez-Cerdá et al.
2015). We found that GS-positive cells of the caudal dorsal pallium in juvenile fit well with type-3 oligodendrocytes, which are described by Del Rio Hortega as cells with a “bulky cell body” with one to four processes. In reptiles, it was shown that most of the cells positive for S-100 (also used as a marker of oligodendrocytes) also express GS during the embryonic period (Romero-Alemán et al.
2003). Interestingly, in adult humans and rodents, GS is expressed by subpopulations of oligodendrocytes (Bernstein et al.
2014). Ultrastructural studies focused on the mesencephalon of a lizard indicated that GS-positive cells may be oligodendrocytes (Monzón-Mayor et al.
1998). Oligodendrocytes with different size and shape were found in the brain of rainbow trout (Pérez et al.
1996) using NADPH diaphorase enzymohistochemistry, and these are morphologically similar to the GS-positive mantle cells of the catshark telencephalon. Differences in the size of oligodendrocytes has been related with the size of axons they ensheath (Díaz-Regueira and Anadón
1998). Fibres in the catshark telencephalon are mostly unmyelinated, except in the compact bundles that run through the medial pallium and the centromedial dorsal pallium (Smeets et al.
1983). Curiously, we have also observed similar cells positive for GS in the rhombencephalon of this specie associated with myelinated tracts of large diameter such as the medial longitudinal fascicle (results not shown). GS immunohistochemistry also revealed adventricularly disperse cells in the subpallium with a different morphology. These bipolar cells are orientated radially and horizontally in the subpallium and exhibit thin processes. They might correspond to either radial glial cells that have lost contact with the ventricular zone or to type-4 oligodendrocytes (or Schwannoid cells) of the classification of Del Rio Hortega, which use to show bipolar processes adhered to thick axons (for review see: Pérez-Cerdá et al.
2015). Numerous thick axons, some of them myelinated, run in the basal forebrain bundles through the catshark subpallium, including the area superficialis basalis (Smeets et al.
1983).
Since we have detected GS-positive cell type that morphologically resemble oligodendrocytes in regions that contain myelinated fibres we have carried out double immunofluorescence techniques using antibodies against GS and chondroitin sulfate proteoglycan NG2 (a marker of oligodendrocyte precursor cells, OPCs). Though GS-positive cell populations do not express NG2 (data not shown), we cannot exclude the possibility that these cells belong to the oligodendroglial lineage. Further investigations using markers of late progenitor pro-oligodendrocytes and mature oligodendrocytes such as O4-antigen and myelin proteins (Ono and Ikenaka
2013) might shed light about the nature of these glial cells.
Subsets of proliferating RGC are present in the catshark embryonic telencephalic ventricular zone
In mammals, the development of the forebrain involves multiple types of progenitor cells, including NECs, RGCs, different pools of intermediate progenitor cells and basal radial glia cells. Apical progenitor cells (or RGCs) are primary progenitor cells generated from NECs; they undergo self-renewing division in the ventricular surface, producing neurons and generating different pools of intermediate progenitors that themselves divide and give rise to differentiated neurons or different types of glial cells (astrocytes, oligodendrocytes and ependymal cells) at sequential stages in development. Although by the end of development most of the RGCs transform into astrocytes, in some brain areas RGCs persist after birth and function as primary progenitor cell (Götz et al.
1998; Marshall et al.
2003; Spassky et al.
2005; Kriegstein and Alvarez-Buylla
2009; Taverna et al.
2014; Beattie and Hippenmeyer
2017).
In mammals, the transition from NECs to apical RGCs involves the expression of several glial markers, including BLBP; several studies have shown that activation of the BLBP promoter in NECs promotes radial glia differentiation (for review see: Pinto and Götz
2007). In rodents most of the precursor cells in the embryonic ventricular zone are RGCs (Noctor et al.
2002). We have detected that most of the ventricular cells express BLBP in S29–S30 catshark embryos, suggesting that these are RGCs progenitors and that the transition from NECs (BLBP-negative) cells to RGCs occurs before these stages. Other feature of the apical RGCs of the mammalian dorsal telencephalon is the expression of the paired-box transcription factor 6 (Pax6) (Englund et al.
2005). Previous studies of our group have found expression of Pax6 in the dorsal telencephalon of catshark at S29 (results not shown). This data suggests that most of the BLBP-positive cells in the dorsal telencephalon also co-express Pax6.
On the other hand, our results indicate that the surface of the catshark telencephalic ventricles in later embryonic stages consists of subsets of RGCs; double immunofluorescence against GFAP, BLBP and GS reveals the presence of subpopulations of RGCs in the ventricular surface (see Fig.
7 and Suppl. Figure 1). Due to the incompatibility among the antibodies used, double immunofluorescence between GFAP and BLBP was not carried out. However, according to the distribution pattern we observed, a high degree of colocalization between GFAP and BLBP is expected. A molecular heterogeneity of RGCs is present in the developing telencephalon of mammals, with subsets of radial glial cells differing in the expression of the GLutamate ASpartate Transporter (GLAST), BLBP and the Radial Glial cell marker 2 (RC2), and whose proportions change throughout the neurogenic period (Hartfuss et al.
2001). In catshark embryos, we have not appreciated regional differences in the distribution of the subsets of radial glial although variations in thickness of the ventricular zone were observed.
Most of the RGCs expressing GFAP and BLBP are PCNA-positive, which indicates that they are proliferative (present results). Previous studies in this catshark have reported a high proliferative potential in S32–S34 embryos (Quintana-Urzainqui et al.
2015). Using fluorescently-tagged endogenous proliferating cell nuclear antigen (PCNA), Zerjatke et al. (
2017) showed that differences in the intensity of the PCNA immunolabelling reflect different phases of the cell cycle. Based on this fact, GFAP- and BLBP-positive cells of the catshark telencephalon with intense PCNA are probably in S phase, while nuclei with a weak PCNA immunolabelling probably indicate cells that are in other phases of the cell cycle. In the mammalian telencephalon, nuclei undergoing the S phase form several layers at the basal surface of the ventricular zone. On the other hand, nuclei in M phase are close to the ventricle, and nuclei in G1 and G2 phases are in the mid region. This process is known as “interkinetic nuclear migration” (for review see: Kriegstein and Alvarez-Buylla
2009). Such zonation of the ventricular zone can be observed on the basis of the intensity of PCNA immunoreactivity in the telencephalon of catshark.
On the other hand, the number of proliferating ventricular cells seems higher in the pallium than in the subpallium. Differences in the proliferative potential between the dorsal and ventral telencephalon were also described in rodent embryos (Malatesta et al.
2003). Our results are in agreement with studies in other vertebrates showing proliferative RGCs. In mouse embryos, most of the proliferating ventricular cells have morphological and molecular characteristics of radial glia and express BLBP (Noctor et al.
2002; Anthony et al.
2004). Proliferating BLBP-positive cells are also present in the ventricular zone of
Xenopus embryos (Moreno and González
2017) and teleosts (Grupp et al.
2010). Furthermore, in the developing dorsal telencephalon of reptiles ventricular radial glial cells undergo cell divisions (Clinton et al.
2014). Like in mammals (Hartfuss et al.
2001; Noctor et al.
2002), RGCs represent the majority of precursor cells in the catshark telencephalon. Our results also indicate that different cycling cells are present in the ventricular zone. Double GFAP/ GS and BLBP/GS immunolabelling clearly showed the presence of different RGCs subpopulations, and double immunofluorescence for PCNA and GFAP or BLBP revealed some proliferating cells that do not exhibit glial cell markers. Curiously, nuclei of most of these proliferating cells used to be strongly PCNA-immunoreactive, which means that cells are in S phase, and in consequence committed to undergo mitosis. Despite the actual nature of these cells is not known, they might correspond with some kind of intermediate progenitor cells.
Fate mapping experiments using Cre recombinase driven by RGCs promoters BLBP or GFAP in mammals indicate that most neurons throughout telencephalon originate from RGCs that express GFAP and BLBP and, interestingly, GFAP-positive cells also produce oligodendrocytes and astrocytes (Malatesta et al.
2003; Casper and McCarthy
2006; Anthony and Heintz
2008; Farkas and Huttner
2008; Martynoga et al.
2012). Besides, in rodents RGCs give rise to most of the ependymal cells between E14 and E16, and these cells complete their maturation after birth (Spassky et al.
2005). If these results were extensive to basal vertebrates such as sharks, it would suggest that BLBP/GFAP-positive proliferating cells may generate both neurons and different types of glial cell at the end of the embryonic period. At this point, further investigations are needed to elucidate the identity of the progeny originated by the proliferating RGCs present in the catshark telencephalon.