The genes doublecortin (DCX) and LIS1 are required for proper cortical neuronal migration and differentiation in humans. Here, we study the expression pattern of the encoded proteins of these genes in developing human brain. LIS1 stained virtually all migrating neurons throughout periods of development. Initially, DCX extensively overlapped with Reelin in early preplate stage in radially oriented columns of cells in the ventricular zone, whereas at later stages, the majority of DCX-positive cells were horizontally oriented. During the cortical plate stage, two opposite patterns of DCX expression were found: in radially oriented apical processes, presumably of pyramidal cells in the cortical plate, and in non-radially oriented mono- or bipolar neurons with migratory morphologies in the deep compartments of the cerebral wall. The extensive co-localization of DCX and Calretinin in non-radially oriented neurons suggested a non-pyramidal phenotype. These cells assumed a more vertical orientation upon entering the subplate. In addition, DCX was expressed by cells in the subpial granular layer and by Cajal–Retzius cells. In a 19 week human fetal cortex with a LIS1 mutation, the number of Reelin-expressing Cajal–Retzius cells was reduced and their morphology was abnormal. DCX was expressed by cells in all regions, but in extremely low numbers, suggesting that LIS1 deficiency adversely affects the migration and differentiation of DCX- and Reelin-positive neurons.
During the process of neuronal migration, neurons are generated in the ventricular zone (VZ) of the cortical wall and ganglionic eminence and reach their destination by both radial and non-radial migration (Rakic, 1972; O’Rourke et al., 1992; Anderson et al., 1997; Lavdas et al., 1999). The first post-mitotic cells settle in the external region of the cortical wall and form the preplate (Rickmann and Wolff, 1981). With the generation and migration of additional neurons, the cortical plate (CP) appears within the preplate (PP), thus splitting the PP into the marginal zone (MZ) and the subplate (SP) (Marin-Padilla, 1978). Formation of the CP then proceeds by progressive addition of newly deposited neurons through an ‘inside-out’ gradient (Angevine and Sidman, 1961; Rakic, 1974). While there is general agreement about these basic events, the molecular characteristics of different populations of migrating neurons, especially in human brain, are not well understood.
Mutations in the human doublecortin (DCX) gene on chromosome Xq22.3 (des Portes et al., 1998; Gleeson et al., 1998) or the LIS1 gene on chromosome 17p13.3 (Reiner et al., 1993) lead to a massive arrest of cortical migration resulting in classical lissencephaly and in double cortex syndrome (subcortical laminar heterotopia). Mutations in the reelin gene on chromosome 7q22 lead to a similar cortical phenotype to lissencephaly, but with the additional feature of cerebellar hypoplasia (Hong et al., 2000). Double cortex syndrome is characterized by an abnormal band of neurons in the white matter underlying an apparently normal cortex, presumably a result of arrested migration of a subset of cortically or subcortically derived neurons. Classical lissencephaly is characterized by a severely thickened cortical gray matter and absence of cortical gyri. Both the DCX and LIS1 genes encode for microtubule-associated proteins (Sapir et al., 1997; Francis et al., 1999; Gleeson et al., 1999a), but their role in migration is unknown. Additionally, while their expression in the agyric rodent cortex has been well characterized, their expression in developing human cortex is less well understood.
Here we examine the expression of DCX together with other cortical markers, LIS1, Reelin (Reln) and Calretinin (CR), by immunohistochemistry in embryonic and fetal human brains covering the in utero period of cortical development. DCX immunoreactivity is present initially in columns of radially oriented cells in the VZ that are co-extensive with similar Reln-immunoreactive (ir) columns and later in predominantly tangentially oriented neurons in the VZ, SVZ and IZ, many of which are also positive for CR, a marker of cortical interneurons. The selective expression of DCX in first radially and then tangentially oriented neurons suggests a possible selective requirement for DCX during specific stages or in specific neuronal subclasses. Furthermore, in 19 gestation week (GW) human cortex with a LIS1 mutation, we found abnormal Cajal–Retzius cell morphology, suggesting that it is required for development of these Reln-expressing cells, which may contribute to the migration defect. There were also fewer DCX-ir cells in this brain, suggesting that LIS1 deficiency may adversely affect the migration or differentiation of DCX- and Reln-expressing neurons.
Materials and Methods
The brains of 41 human fetuses, aged 5.5–40 GW, were fixed in Bouin’s fixative, embedded in paraffin and cut into series of 10 μm thick sections, mostly in a coronal plane. The following ages were studied: Carnegie stages (CS) 16 (5 GW), 16/17 (5.5 GW), 17 (6 GW), 17/18 (6 GW), 18 (6.5 GW), 19 (6.5 GW), 20 (7 GW), 21 (7.5 GW) and 22 (8 GW); fetal stages (GW): 9.5, 10, 11, 11.5, 12, 13, 13.5, 14, 15, 16, 17, 17.5, 18, 18.5, 19, 20, 22, 24, 28, 29, 32 and 40. Part of this material was also used in previous studies (Meyer and Goffinet, 1998; Meyer and Wahle, 1999; Meyer et al., 2000). All brains were derived from spontaneous or medically induced abortions in accordance with Spanish legislation, and supervised by the Ethical Committee of the University Hospital La Laguna. A 19 GW human lissencephaly brain with a documented LIS1 mutation was obtained from Baylor College of Medicine (kindly provided by Dr G. Clark). For immunohistochemistry, sections were incubated in primary antibody overnight in a humid chamber. After rinsing, they were incubated in the corresponding biotinylated secondary antibodies (DAKO, Glostrup, Denmark), diluted 1:200 in Tris-buffered saline (TBS), washed and incubated in ABC complex (DAKO). Bound peroxidase was revealed using 0.05% DAB, 0.06% nickel ammonium sulfate and 0.01% hydrogen peroxide in TBS. The sections were dehydrated, cleared and mounted with Eukitt (O. Kindler GmbH, Freiburg, Germany). The primary antibodies include a polyclonal antibody raised against human Doublecortin (1:500) (Gleeson et al., 1999b), a polyclonal anti-LIS1 antibody (1:250) (sc-7577; Santa Cruz), a rabbit polyclonal antibody against human CR (1:3000) (Swant, Bellinzona, Switzerland) and a mouse monoclonal anti-Reln antibody 142 (1:500) (de Bergeyck et al., 1998). For double staining with CR and DCX we used a mouse monoclonal antibody against CR (1:1000) (Swant). To define the relationship between the proliferative ventricular and subventricular zones and DCX expression, we used a mouse monoclonal antibody against Ki-67 (MIB-1; DAKO), a marker of cell proliferation (Brown and Gatter, 1990). Antibody specificity was controlled by omitting the primary antibodies.
Double immunostaining of CR and DCX was carried out at 14 GW, to explore the spatial overlap of both proteins in deep cortical compartments. Since both antibodies yielded similar cytoplasmic staining, this procedure did not allow us to clearly address co-localization of CR and DCX in individual neurons, but was used to determine whether neurons positive for one antibody were negative for the other. After the first primary antiserum (mouse anti-CR 1:2000 or rabbit anti-DCX 1:500), sections were incubated with the corresponding biotinylated secondary antibodies and with ABC (1:150) and processed with DAB and ammonium nickel sulfate as described above, which yielded a black reaction product. After thorough washes, the sections were incubated overnight in the second primary antibody (anti-DCX or anti-CR), followed by the corresponding secondary antibodies. After washing, the sections were immersed in 0.05% DAB in TBS, pH 7.6; the reaction product appeared yellow. The presence of only yellow immunostaining in a neuron indicated that it was positive for the second antibody but negative for the first (staining black). Double immunostaining with Ki-67 and DCX was done as described for CR and DCX; Ki-67-positive cell nuclei stained black, DCX-positive cytoplasm was yellow. For double immuno-fluorescence, sections were incubated in the primary antibodies (mouse anti-CR 1:1000 and rabbit anti-DCX 1:250) overnight in a humid chamber. After thorough rinsing, they were successively incubated in Cy2-conjugated anti-mouse IgG (Amersham, Arlington Heights, IL), diluted 1:250 in phosphate-buffered saline (PBS), and in Cy3-conjugated anti-rabbit IgG (Amersham) diluted 1:500 in PBS, each for 3 h in the dark and at room temperature. After rinsing in PBS, the sections were coverslipped with a 1:1 solution of phosphate buffer and glycerol.
DCX Expression During Preplate Stage
DCX was expressed in the developing telencephalon from the earliest stages tested. At CS 16 (5.5 GW), shortly after formation of the telencephalic vesicle, DCX was present in horizontally oriented neurons located in a narrow marginal layer between the pia and ventricular zone (VZ) (Figs 1A and 7). The VZ was DCX-negative with the exception of radially oriented columns of DCX-ir cells spanning the whole width of the neuroepithelium. These columns were confined to the rostral aspect of the cortical wall. In single sections through this region, 3–4 cell columns were spaced at remarkably consistent intervals of ~100 μm.
We compared DCX staining with that of Reln, another marker of early preplate neurons. At CS 16, Reln-ir neurons were less numerous and more weakly immunostained than DCX-ir cells; they were mainly located below the pial surface and lay only occasionally deeper in the VZ (Fig. 1B). Superimposition of adjacent DCX- and Reln-immunostained sections revealed a spatial relationship between increased numbers of Reln-ir cells in the marginal layer and DCX-ir columns in the VZ. This suggests that some cells in the DCX-ir radial columns may express Reln when they arrive at the pial surface. Reln-ir and DCX-ir neurons in the preplate were morphologically similar, with both displaying horizontally elongate somas and similar size ranges, between 8 and 12 μm.
The radial columns of DCX-positive cells in the VZ of rostral cortical territories were particularly prominent at CS 19 (6.5 GW). They lay interspersed at intervals of 100–150 μm in the DCX-negative neuroepithelium (Fig. 1C,D) and were continuous within the preplate. The preplate was wider at CS 19 and contained numerous DCX-ir neurons. Reln staining in adjacent sections showed similar columns in VZ (Meyer et al., 2000) in the same locations (Fig. 1E), suggesting extensive overlap of both proteins in cells within the columns and in the PP.
At CS 18 (6.5 GW), the intermediate zone (IZ) appeared in the lateral aspect of the cortex near the cortico-striatal angle (Figs 2A and 7). The IZ contained heavily stained horizontally arranged DCX-ir neurons immersed in a dense fiber plexus. Cell number and fiber density in the IZ increased during development over stages CS 20 (7 GW), 10 GW and 16 GW (Fig. 2B–D) and progressively extended over the entire cortical wall following a lateral to medial gradient. The orientation of cells and fibers parallel to the VZ suggested tangential migration. The origin of the cells appeared to be the ganglionic eminences (Fig. 2A).
DCX in the Cortical Plate (CP)
The condensation of the ‘pioneer plate’ precedes the appearance of the CP (Meyer et al., 2000). Pioneer neurons emit the first efferent fibers (cortico-fugal fibers) that course in the upper IZ at stage 22 (7.5 GW). Pioneer neurons and their axons displayed moderate DCX immunoreactivity (Figs 3A and 7). After the emergence of the cortical plate in the lateral cortex at 9 GW (Figs 3B and 7), the initial pioneer plate staining became replaced by a distinct CP staining pattern that persisted almost unchanged until mid-gestation. Its main characteristic was the DCX-positivity of fasciculated radially oriented processes, particularly intensely stained in the upper part of the CP, and continuous with a more diffuse neuropil staining in the MZ (Figs 3B,C and 7). Morphology and orientation of DCX-ir processes resembled apical dendrites and terminal bouquets of pyramidal cells, while the cell somas were usually unstained.
Around mid-gestation, from 18 to 24 GW, high DCX expression was also observed in cell somas displaying pyramidal shapes (Fig. 3D). Positivity was prominent in large pyramidal cell bodies in prospective layer V of frontal and parietal regions, with some staining in cell somas of smaller non-pyramidal neurons (Fig. 3D). However, after mid-gestation the overall intensity of DCX immunoreactivity decreased, and was almost undetectable in the perinatal cortex (data not shown).
In early cortical plate stages (8–14 GW) DCX was also expressed by Cajal–Retzius cells in the MZ (Figs 3C and 7), the same population of cells that are strongly Reln-ir (Meyer and Goffinet, 1998; Meyer and Wahle, 1999; Meyer et al., 2002). From 14 GW onwards, DCX was present also in subpial granule cells, thus showing a more generalized distribution than Reln in the MZ (Fig. 3C).
DCX in VZ and SVZ
From CS 16 to 19 (5–6.5 GW), the neuroepithelium in the VZ was DCX-negative, with the exception of the radial clusters described above. Only at CS 20 (7GW), shortly before the emergence of the pioneer plate and CP, did the first DCX-positive fibers and cells appear in the VZ; they increased in number at CS 21 (7.5 GW) and CS 22 (8 GW) (Figs 4A and 7 at 8 GW). They were heterogeneous in size and shape, comprising large vertical neurons (Fig. 7), as well as horizontal neurons extending a single leading process and small rounded cells. The incipient subventricular zone (SVZ) was occupied by a horizontal plexus of thick, intensely DCX-ir fibers (Figs 4A and 7). Most ir fibers in the VZ seemed to derive from this plexus, following variable orientations and directions. Whereas a subset of fibers in the SVZ and IZ presumably derived from the DCX-positive cells in VZ and SVZ, the thick fiber diameter and magnitude of the plexus suggested an additional, possibly extra-cortical, origin.
From 12 to 14 GW, number and packing density of DCX-ir cells in VZ and SVZ remained low (Fig. 4B), but increased abruptly after this age (Fig. 4C). Small rounded cells appeared in the VZ and particularly in the SVZ, mostly horizontally oriented with leading processes extending 50–100 μm. The cell body as well as the leading process was DCX-ir.
The prominence of DCX immunoreactivity in VZ and SVZ raised the question of whether the protein was expressed in dividing cells. Double staining experiments using DCX and Ki-67, a marker of proliferative cells, revealed that DCX was not co-expressed with Ki-67 and indicated that it is present in post-mitotic neurons in VZ and SVZ (Fig. 5A,C).
From 15 to 19 GW, the VZ and SVZ were progressively invaded by DCX-ir cells (Fig. 4C). The orientation of DCX-ir cells and processes was extremely variable, although most commonly they were horizontally or obliquely oriented. At 27 GW the SVZ has transformed into the subependymal zone (SE). DCX-ir cells in the SE tended to be aggregated into large clusters and were variably oriented (Fig. 4D). At late gestational ages, 32 GW (Fig. 4E) and 40 GW, the proliferative VZ had disappeared and the SE had transformed into a highly vascularized compartment, occupied by groups of small cells which continued to express DCX into late pre-natal and early post-natal life.
DCX and LIS1 in IZ
The IZ is a transient fiber-rich migration compartment, traversed by migrating neurons en route from the VZ/SVZ to the CP. It expands dramatically during the period of maximum migration, from about 12 to 20 GW, and subsequently transforms into the subcortical white matter. DCX expression in the lower IZ was localized to horizontal fiber bundles and mostly horizontally oriented neurons (Fig. 6B,D,E,G). The most salient observation in the IZ was the DCX-negativity of a large number of migrating neurons. To characterize these neurons, we compared DCX and LIS1 expression during the period of maximum migration, from 13 to 19 GW. LIS1 was more widely expressed than DCX, by migrating neurons displaying both radial and horizontal orientations (Fig. 6A,C). In particular, a divergence of DCX and LIS1 immunostaining was observed at the level of the IZ, where DCX-negative zones were occupied by clusters of LIS1-positive neurons, the radial orientation of which was in contrast with the predominantly horizontal arrangement of DCX-ir elements (Fig. 6A,C versus B,D). The dissociation between DCX and LIS1 in IZ was less evident after 20 GW, a time at which migration is largely complete, because the staining intensity of DCX-positive fibers in the IZ decreased and the LIS1 signal became less conspicuous in IZ. At 16 GW, DCX expression remained high in non-radially oriented neurons and processes (Fig. 6G). DCX expression in the subplate (SP) was limited, possibly because the human SP reaches its maximum extension after mid-gestation when DCX expression begins to decline. DCX-ir neurons in the SP tended to be radially oriented, particularly in the convex regions of the cortical wall (Fig. 6H).
DCX and CR in VZ and SVZ
To further characterize the non-radially oriented DCX-ir neurons in VZ and SVZ during the 14–16 GW period, we compared DCX and CR expression. CR is expressed in interneurons in human cortex (del Rio and DeFelipe, 1996) and appears very early in development (Meyer and Wahle, 1999; Meyer et al., 2000). CR was not present in the cortical VZ before 14 GW, but from 14 to 16 GW, both DCX-ir and CR-ir neurons steeply increased in number in deep cortical compartments, displaying similar morphologies and distributions. Both markers stained rather small non-radially oriented neurons extending prominent leading processes. Comparison of adjacent serial sections demonstrated an extensive spatial overlap of DCX- and CR-ir cells, particularly in the prominent cell clusters in the SVZ at 14 GW (Fig. 6E,F). Nevertheless, double staining experiments revealed the presence of DCX-positive cells in VZ and SVZ that were unstained with CR (Fig. 5B,D–I). Conversely, CR-positive, DCX-negative neurons were not observed (Fig. 5F–I), indicating that CR-ir neurons were a subset of a larger heterogeneous population of DCX-ir cells. A major difference between the DCX and CR distributions was in the fiber staining: the dense DCX-positive plexus in the lower IZ was CR-negative (Fig. 6E,F).
DCX in Miller–Dieker Lissencephaly
We compared staining of DCX and Reln in a 19 GW fetus with Miller–Dieker (MD) lissencephalic cortex and a documented LIS1 deletion, to test whether there is altered expression or abnormal localization of immunoreactive cells. Reln staining revealed a reduced number of Cajal–Retzius cells in the MZ as compared to a normal, age-matched fetus (Fig. 8A,B). At 18–20 GW, Cajal–Retzius cells normally develop vertical cell bodies and descend into the deep MZ (Meyer and Gonzalez-Hernandez, 1993; Meyer and Goffinet, 1998); in the MD fetus, Cajal–Retzius cells remained close to the pia, retaining their early horizontal orientation, and did not exhibit the characteristic morphology associated with this age. No Reln expression was found outside the MZ in the MD fetus cortex.
In the VZ of the MD brain, DCX was expressed by mostly horizontal cells (Fig. 8D), but was not observed in the SVZ. Again, comparison with an age-matched control fetus (Fig. 8C) demonstrated that only a fraction of the normal population was present in the mutant. We conclude that even though DCX- and Reln-expressing cell populations were present in the correct cortical compartments, LIS1 deficiency severely decreased the number of immunoreactive cells, although it was not possible to determine if the defect was in cell migration, differentiation or positioning.
The expression of specific developmental genes is required for successful migration of neurons from the ventricular zones of the telencephalic vesicle to their final destination in the cortical plate. One of these genes is DCX, mutations of which lead to X-linked lissencephaly in hemizygous males or double cortex syndrome in heterozygous females. Our systematic analysis of embryonic and fetal stages shows that DCX displays complex expression patterns during human pre-natal development. Expression starts very early in the preplate, reaches highest levels during the period of cortical migration and declines in the last weeks of gestation, when most neurons have completed migration. Distinct cell populations residing in various compartments of the cortical wall express DCX with precise chronological regulation, suggesting that it is involved in a variety of developmental events. Altogether, this timetable suggests potential roles of DCX not previously suspected from analysis of rodent tissue.
One of the crucial time points in corticogenesis is the emergence of the cortical plate at 8/9 GW (Sidman and Rakic, 1982; O’Rahilly and Müller, 1994; Meyer et al., 2000). During the period of cortical plate migration, DCX displays two apparently opposite expression patterns: it is expressed by radially arranged processes in neurons positioned in the cortical plate and by non-radially oriented migratory neurons in the deep compartments of the cortical wall (VZ, SVZ and IZ). The nonradial component is particularly prominent from approximately 12 to 18 GW, the period of maximum migration.
DCX is Expressed in Preplate Cells
We show that DCX is expressed from the earliest preplate stages onwards, and may even precede the expression of Reln. Reln is a large extracellular matrix glycoprotein, secreted by Cajal–Retzius cells in the MZ of the developing cortex (D’Arcangelo et al., 1995; Ogawa et al., 1995), where it plays a crucial role in the establishment of the inside-out gradient of the cortical plate (Caviness and Sidman, 1973). In human preplate, Reln is present as early as CS 16, preceding the other cellular components of the developing cortex (Meyer et al., 2000). Comparison of the Reln and DCX distributions suggests that the two proteins may be co-expressed in the first cohort of Cajal–Retzius cells in the incipient preplate. However, because two color co-labeling was not performed in this study, we cannot state with certainty that individual Cajal–Retzius cells express both Reln and DCX.
Because the formation of the preplate is a critical stage in cortex development that guides formation of later structures, if DCX and LIS1 play a critical role in formation of this structure then a block in the formation of the preplate may underlie some of the pathology in lissencephaly. While our results suggest that DCX is expressed in early Cajal–Retzius neurons, tissue from developing human brain with a DCX mutation was not available to test for defective preplate or Cajal–Retzius positioning. However, our results provide the first evidence that the Cajal–Retzius neurons are depleted and fail to differentiate normally in developing brain with a LIS1 mutation. In the future, it will be interesting to look for a defect in migration of Cajal–Retzius neurons in human lissencephaly, which might be part of the mechanism for the development of the disorder.
The conspicuous radial columns of DCX-ir cells in the VZ of CS 16–19 suggests a possible role for the protein in migration of these neurons. These radial columns may possibly represent sites of focal production of a subset of Cajal–Retzius cells (Meyer et al., 2002) or regions where radial glia may give rise to neurons (Noctor et al., 2001). It is also possible that the DCX-positive columns in the extremely immature telencephalon at CS 16 precede in time the appearance of radial glia.
In the more advanced preplate (CS 20 and 21), after the disappearance of the radial columns, new cell populations are added that express DCX. These cells appear to be part of the so-called monolayer (Meyer et al., 2000), formed by Relnnegative horizontally oriented neurons arriving in the preplate.
Radial Organization of DCX-positive Elements in the Cortical Plate
Our finding in human brain that DCX expression is highest in the upper layers of the cortical plate, where radial migration stops, is in agreement with studies in rodent (Gleeson et al., 1999a). Several events take place during this final step in radial migration: the leading process disconnects from the radial glia fiber, makes contact with the MZ and then shortens and thickens while the cell body rapidly moves through the upper cortical plate via a mechanism of somal translocation (Nadarajah et al., 2001). Terminal somal translocation and the accompanying changes in the leading process may be independent of radial glia and may require specific regulatory mechanisms of microtubule dynamics. The presence of DCX in radial processes in the upper cortical plate may thus be involved in the control of microtubule rearrangement necessary for the arrest of migration and the transition from a migratory mode to a stationary mode. The arrest of migration may relate to the presence of Reln, which is expressed by Cajal–Retzius cells in the MZ and is thought to play an instructive role in the final positioning and architectonic arrangement of cortical plate neurons (Pearlman and Sheppard, 1996; Walsh and Goffinet, 2000). The recent description of DCX interaction with the μ1 and μ2 subunits of the adapter complexes AP-1 and AP-2 (Friocourt et al., 2001) that function in clathrin-mediated cellular vesicular traffic could provide a potential mechanism of DCX–Reln interaction, as Reln receptors are thought to be internalized via clathrin-mediated mechanisms (Li et al., 2001).
On the other hand, many DCX-ir processes in CP resemble differentiating apical dendrites of post-migratory pyramidal cells, with the strongest expression in the distal branches in upper CP and MZ (Francis et al., 1999). As emphasized by Marin-Padilla (1998), pyramidal neurons are initially anchored to the MZ by their apical dendrites and appear to grow continuously to keep pace with the increasing width of the CP. Thus DCX may mediate microtubule plasticity in these growing apical dendrites or other cellular processes. In keeping with its role of a developmental protein, DCX-positive staining decreases in the last third of gestation. Additionally, there is clear evidence for DCX expression in mature neurons in deep cortical layers, where its function is unknown.
Non-radial Orientation of DCX-positive Cells in VZ, SVZ and IZ
The prevailing non-radial orientation of DCX-ir cells in VZ and SVZ raises the possibility that these neurons originate in the ganglionic eminence (Anderson et al., 1997; Lavdas et al., 1999). The predominantly non-radial orientation of DCX-positive cells in deep cortical compartments has not been observed in the embryonic mouse. However, the enormous width of the IZ and SP is one of the distinctive features of the developing human cortex, which may require specific long-distance migration mechanisms. We examined the cortico-striatal sulcus at different ages, but there was no evidence for a ventral to dorsal stream of DCX-positive cells invading the neocortical VZ through tangential migration. At 16 GW, the number of DCX-positive cells had increased on both sides of the cortico-striatal sulcus, but without a clear indication of a preponderant migratory direction based upon the direction of the leading process.
Tangential migration, independent of the radial glia substrate, occurs at all levels of the cortical wall (O’Rourke et al., 1992, 1995), but is particularly prominent in VZ, SVZ and IZ (O’Rourke et al., 1997). DCX is expressed by distinct cell populations that may travel through a tangential migration mode, which is in keeping with the substantial co-expression of CR in these neurons. In adult human neocortex, CR is expressed by nonpyramidal neurons, most of which are GABAergic (del Rio and DeFelipe, 1996). CR expression in cortical neurons begins very early in development, sometimes even in VZ neurons (Meyer and Wahle, 1999; Meyer et al., 2000). During the migration period, CR and DCX expression patterns show similar time courses and distributions; both proteins are expressed by cells in the VZ/SVZ displaying non-radial orientations. At the upper levels of the IZ and in the subplate, the horizontal orientation is replaced by an oblique or vertical orientation, suggesting that tangential migration is most common at the level of SVZ and lower IZ, but may change towards a radial mode as cells approach the CP. The switch in migration mode proposed here for DCX-ir neurons has also been reported in a recent study of GABAergic interneurons, shown to derive from the VZ/SVZ of the human dorsal telencephalon (Letinic et al., 2002). Altogether, the non-radial orientation of DCX-ir neurons and the substantial overlap of DCX and CR suggest that these cells are interneurons.
The Dissociation of DCX and LIS1 in IZ
One of the most striking findings of our study is the DCX-negativity of a large proportion of migratory neurons, that are nonetheless LIS1-ir. Mutations in the LIS1 or DCX gene produce nearly identical phenotypes, both genes encoding for microtubule-associated proteins (Sapir et al., 1997; Francis et al., 1999; Gleeson et al., 1999a), and the two proteins display possible interactions (Caspi et al., 2000), raising the likely scenario that protein expression could overlap in human brain. LIS1 is widely expressed by migrating neurons in early developing human cortex, showing particularly high levels in Cajal–Retzius cells and in the ventricular neuroepithelium (Clark et al., 1997). In the age group examined in our study, corresponding to the period of maximal migration, LIS1 was expressed by virtually all migratory cells in regions from the ventricular neuroepithelium to the MZ, although staining intensity varied across the different compartments. Among the neurons displaying moderate staining intensities for LIS1, we identified clusters of radially oriented neurons in the IZ, presumably representing pyramidal cells en route from the VZ to the CP, which were clearly negative for DCX. The dissociation of DCX and LIS1 in the IZ suggests a difference in the molecular mechanisms that control radial and tangential migration modes. Whereas LIS1 is widely expressed by both radially and tangentially migrating neurons and may be involved in the general motility of a cell, DCX may regulate microtubule dynamics specifically required by tangential or radial glia-independent migration.
DCX-positive Fiber Systems in IZ and SVZ
In addition to non-radially oriented neurons, DCX labels a horizontal fiber system in the incipient SVZ and IZ from the preplate stages onwards. The periventricular fiber plexus precedes CP formation and remains prominent during the migration period, but the cells of origin could not be determined here. In part, it may arise from the DCX-positive cells that appear at stage 20 in the SVZ, but its continuity with a similar plexus in the ganglionic eminence together with the thickness of the fibers point to an additional origin in basal telencephalic regions. These fibers become very prominent in SVZ and lower IZ over the following weeks, but have so far not been described in the developing human cortex. They are not the axons of pioneer plate and cortical plate neurons, which course more superficially and are only weakly DCX-ir. The plexus also precedes in time the appearance of the first corticofugal and thalamo-cortical fibers and may represent a fiber system analogous to the long leading processes extended by migrating basilar pontine neurons (Yee et al., 1999). The very intense DCX fiber staining in IZ has also been observed in embryonic mouse brain (Francis et al., 1999). Interestingly, DCX has been found to be enriched in the proximal parts of neuritic growth cones (Francis et al., 1999), which along with the prominent fiber staining suggests a possible role of DCX in axonal growth and extension of longrange fiber pathways.
DCX and Reln in LIS1 Deficiency
The pattern of expression of both Reln and DCX was abnormal in a 19 GW brain with MD syndrome and a documented LIS1 deletion, suggesting that proper LIS1 dosage is required for proper differentiation or migration of both Reln- and DCX-positive neurons. In this brain, the basic localizations of the proteins were preserved, but the number of immunoreactive neurons was quite strikingly low. These results are based upon staining of a single brain, which was not processed identically to the other human brains used in this study and thus will need to be replicated in other developing MD syndrome brains.
We thank André Goffinet for the generous gift of the 142 antibody and Gary Clark for the human lissencephaly tissue. G.M. is supported by EU grant QLG3-CT-2000-00158 (Concorde), C.G.P.G. is supported by a grant of the Ministerio de Educación, Cultura y Deportes and J.G.G. is supported by a Searle, Klingenstein, Merck Fellowship and by NINDS NS42749 and NS41537.