Abstract

We have recently begun to gain a clearer understanding of the phasing and patterns of colonization of the developing human brain by microglia. In this study we investigated the distribution, morphology and phenotype of microglia specifically within the wall of the human telencephalon from 12 to 24 gestational weeks (gw), a period that corresponds to the development of thalamocortical fibres passing through the transient subplate region of the developing cerebral wall. Sections from a total of 45 human fetal brains were immunoreacted to detect CD68 and MHC class II antigens and histochemically reacted with RCA-1 and tomato lectins. These markers were differentially expressed by anatomically discrete populations of microglia in the cerebral wall: two cell populations were noted during the initial phase of colonization (12–14 gw): (i) CD68++ RCA-1+ MHC II− amoeboid cells aligned within the subplate, and (ii) RCA-1++ CD68− MHC II− progenitors in the marginal layer and lower cortical plate that progressively ramified within the subplate, without seemingly passing through an ‘amoeboid’ state. At this stage microglia were largely absent from the germinal layers and the intermediate zone. From 14 to 15 gw, however, MHC class II positive cells were also detected within germinal layers and in the corpus callosum, and these cells, which coexpressed CD68 antigen (a marker associated with phagocytosis), further populated the lower half of the telencephalon from 18 to 24 gw. These findings are discussed in relation to developmental events that take place during the second trimester within the wall of the telencephalon.

Introduction

Microglia are resident mononuclear phagocytes of the nervous system. These cells begin to colonize the human brain and spinal cord before the start of the second trimester (12th week of gestation) (Rezaie and Male, 1999, 2002a,b; Rezaie, 2003). They disseminate throughout all parts of the nervous system, where they occupy defined spatial territories, and progressively differentiate to immunologically quiescent, ramified cells with the passage of time. From birth onwards, the infant brain will harbour microglia essentially in their mature ‘adult’ forms and distribution patterns (Rezaie, 2003).

On the basis of morphology, microglia within the normal central nervous system (CNS) have been further subdivided into two distinct populations, namely amoeboid and ramified cells. Studies in rodents have shown that amoeboid microglia are widely distributed throughout the brain during embryonic and neonatal development. These cells predominantly occur close to blood vessels, particularly within the white matter, and become increasingly ramified as they differentiate into microglia (Perry et al., 1985; Kaur and Ling, 1991). Morphological accounts of microglia within the human CNS indicate that a population of amoeboid microglia are already present in limited numbers within the brain during mid-to-late first trimester (Choi, 1981; Fujimoto et al., 1989; Andjelkovic et al., 1998; Zecevic et al., 1998). Amoeboid cells have been considered to represent the precursors of ramified microglia (Imamoto and Leblond, 1978), mainly because a rise in the number of ramified cells parallels a decline in the number of amoeboid cells during development (Ling and Tan, 1974). However, it is not yet clear how many of the progenitors actually differentiate into amoeboid microglia in situ. Furthermore, not all amoeboid microglia in the fetus and neonate will survive into adulthood and many undergo apoptosis (Perry, 1996), although this is another topic that has recently been the subject of debate (Wang et al., 2002). Unconfirmed estimates have placed the figure for the percentage of microglial progenitors that persist and differentiate to form amoeboid and then ramified microglia somewhere in the region of 30% in rodents. Studies that address this point have been mainly hampered by difficulties in continually tracking the phenotypically heterogeneous population of microglia and their progenitors during development.

In addition to the morphological heterogeneity of microglia, heterogeneity in their expression of cell surface markers has also been emphasized by several previous studies (Sasaki et al., 1988; Hutchins et al., 1990; Lawson et al., 1990; deGroot et al., 1992; Ogawa et al., 1993; Becher and Antel, 1996; Rezaie and Male, 1999). Such heterogeneity hinders attempts to distinguish the intrinsic population of microglia from recruited monocytes or macrophages not only during development, but also in the adult. One criteria that has been proposed for discriminating these cell types relates to the levels of expression of multiple surface markers on microglia. For example, rodent microglia constitutively express low levels of CD45 and negligible CD11b, as detected by flow cytometric analysis (Sedgwick et al., 1991). These are designated CD11b−CD45low, whereas monocytes and macrophages are typically CD11b+CD45high. Becher and Antel (1996) separately demonstrated that immediately ex vivo human cultured microglia displayed a CD11b+CD45low phenotype and undetectable levels of CD14, a finding also noted by Peterson et al. (1995). A more recent attempt at multiple phenotypic characterization of human microglia has been reported by Dick et al. (1997), whereby freshly isolated resident human microglia could be defined by the following flow cytometric phenotype: CD11b+CD45lowCD4−CD11chighMHC class II+CD26−CD14−. With respect to the developing nervous system, Hutchins et al. (1990) showed that human fetal microglia downregulate expression of CD68 and CD64 antigens (phagocyte-associated markers) and become more ramified with developmental progression. Penfold et al. (1991) further noted distinct subpopulations of CD68- and CD45-positive microglia in the human fetal retina. Fetal microglia have further been shown to express MHC class II antigens and identified using histochemistry with the lectins Lycopersicon esculentum (tomato lectin) and Ricinus communis agglutinin-1 (RCA-1) (see Rezaie, 2003, for references). Recently, we reported that microglia in the fetal and adult human brain can be distinguished from other mononuclear phagocytes since they lack expression of CD163 antigen (equivalent of ED-2 antigen in rodents) (Rezaie and Male, 2003). Nevertheless, the lack of a reliable and specific marker for microglia continues to confound attempts to distinguish between the transient populations of these cells during development and resident microglia of the adult brain.

What determines this phenotypic heterogeneity of microglia? Put simply, the phenotype of microglia depends on their location, state of differentiation and potential physiological functions. Factors released by astrocytes and neurons and contact-mediated interactions with these cells are known to affect microglial morphology, function and phenotype (Rezaie and Male, 2002a). The substrate to which microglia adhere also has an important influence on their phenotype (Giulian et al., 1995). The blood–brain barrier (BBB) is further known to influence microglial phenotype (Becher and Antel, 1996). In the adult CNS, microglia express higher levels of phenotypic markers at sites lacking a BBB, and this also occurs in aged CNS where presumably limited barrier breakdown occurs, allowing leakage of plasma proteins. A full barrier forms relatively late in development, long after the majority of neurons have formed their connections (Davis et al., 1994). However, cerebral endothelial cells have an intrinsic ability to form ‘tight junctions’ when they first invade the brain during development (Saunders, 1992), and naturally occurring plasma proteins have not been found to leak across even the most immature cerebral vessels, although the ‘BBB’ of the fetus and newborn is much more permeable to smaller molecules. Hence microglia in different locations will be in contact with different levels of serum molecules, which is partly related to their distance from the blood vessels and the ependymal barrier. Moreover, different areas of the brain may develop their barrier properties at different times — there appears to be an ependymal-to-cortical gradient in barrier formation, at least in rodents, since injected proteins are still visible in subependymal layers at embryonic day 16 (Engelhardt and Risau, 1995). The morphology and immunophenotype of fetal microglia may therefore relate partly to the timing and extent of barrier development in each region of the brain.

Phenotypic heterogeneity is a phenomenon that is seen in other populations of mononuclear phagocytes within fetal tissues, that may also reflect the localization and associated specialized function of these cells (Naito et al., 1990; Bardadin et al., 1991; Higashi et al., 1992; Takeya and Takahashi, 1992; Wijffels et al., 1994). However, within these tissues, there is similar uncertainty whether heterogeneity in phenotype reflects variation within a single cell population, or identifies the coexistence of several distinct populations of resident macrophages.

We and others have previously characterized the extent of microglial colonization in the developing human brain and spinal cord (see Rezaie, 2003, for references). In this paper, we extend our previous observations by defining the distribution of microglia specifically within the wall of the telencephalon from gestational weeks (gw) 12 to 24, using immunohistochemistry with antisera against CD68 and MHC class II antigens (markers used routinely to detect microglia in diagnostic neuropathology), and histochemistry with tomato lectin (Lycopersicon esculentum) and RCA-1 (Ricinus communis agglutinin-1), which are known to detect both developing as well as resting adult microglia (Rezaie, 2003). We have differentiated at least two distinct populations of microglia based on immunophenotypic and morphological characteristics, and discuss the potential roles of these cells in relation to the development and structural organization of the telencephalon.

Materials and Methods

Forty-five human fetal brains were selected for examination (Table 1). Gestational ages were consistent with standard clinical criteria estimated according to crown–rump, crown–heel and foot lengths where available. All fetal materials were derived from cases of elective or spontaneous termination of pregnancy with prior informed parental consent, and approval of the local ethical committee. These cases showed no signs of hypoxic-ischaemic injury or other histopathological abnormalities in the CNS. Fetal brains were either immersion fixed in 4% paraformaldehyde solution for ∼48 h at room temperature, cryoprotected in 30% sucrose Tris-buffered solution and stored frozen at −70 to −85°C prior to use, or fixed in neutral buffered formalin, processed and embedded in paraffin wax as indicated in Table 1. Serial sections of the brain were taken at thicknesses of 20 μm (paraffin sections) to 40 μm (cryostat sections) in coronal and/or sagittal planes, and the wall of the telencephalon (corresponding to that shown on Fig. 1) was assayed for immuno- and lectin histochemistry.

Figure 1.

Coronal (Nissl-stained) sections of the human fetal forebrain from representative cases, to show the progressive expansion of the wall of the telencephalon (T) between 12 and 24 gestational weeks (gw). One hemisphere is indicated. Scale bar represents ∼9 mm.

Figure 1.

Coronal (Nissl-stained) sections of the human fetal forebrain from representative cases, to show the progressive expansion of the wall of the telencephalon (T) between 12 and 24 gestational weeks (gw). One hemisphere is indicated. Scale bar represents ∼9 mm.

Table 1

Details of human fetal cases investigated in this study

Case
 
Gender
 
Age/weeks
 
Cause of death
 
Tissue block
 
12 induced abortion coronal frozen 
12 spontaneous miscarriage coronal frozen 
12–13 induced abortion coronal paraffin 
– 13 spontaneous miscarriage coronal frozen 
13 spontaneous miscarriage sagittal frozen 
13–14 spontaneous miscarriage coronal paraffin 
14 uterine rupture coronal paraffin 
– 14 spontaneous miscarriage coronal frozen 
14 spontaneous miscarriage coronal frozen 
10 – 15 – coronal frozen 
11 15 induced abortion sagittal frozen 
12 16 spontaneous miscarriage coronal frozen 
13 – 16 – coronal frozen 
14 16 spontaneous miscarriage coronal frozen 
15 16 spontaneous miscarriage coronal frozen 
16 17 spontaneous miscarriage coronal frozen 
17a 17 – coronal frozen 
18a 17 – coronal frozen 
19 – 17 spontaneous miscarriage coronal frozen 
20 18 spontaneous miscarriage coronal frozen 
21 18 spontaneous miscarriage coronal frozen 
22 18 spontaneous miscarriage coronal frozen 
23 18 spontaneous miscarriage coronal frozen 
24 18 spontaneous miscarriage coronal frozen 
25 19 induced abortion horizontal frozen 
26 19 – coronal frozen 
27 19 spontaneous miscarriage coronal frozen 
28a 19 spontaneous miscarriage coronal frozen 
29a 19 spontaneous miscarriage coronal frozen 
30 19–20 induced abortion sagittal frozen 
31 20 spontaneous miscarriage coronal frozen 
32 20 induced abortion coronal frozen 
33 20 spontaneous miscarriage horizontal frozen 
34 – 20 – coronal frozen 
35 20 spontaneous miscarriage coronal frozen 
36 20 induced abortion coronal frozen 
37 22 spontaneous miscarriage horizontal frozen 
38 22 induced abortion coronal frozen 
39 – 22 induced abortion coronal frozen 
40 23 spontaneous miscarriage coronal frozen 
41 23 premature delivery coronal frozen 
42 23 induced abortion coronal frozen 
43 – 24 induced abortion coronal frozen 
44 24 spontaneous miscarriage coronal frozen 
45
 
M
 
24
 
induced abortion
 
coronal frozen
 
Case
 
Gender
 
Age/weeks
 
Cause of death
 
Tissue block
 
12 induced abortion coronal frozen 
12 spontaneous miscarriage coronal frozen 
12–13 induced abortion coronal paraffin 
– 13 spontaneous miscarriage coronal frozen 
13 spontaneous miscarriage sagittal frozen 
13–14 spontaneous miscarriage coronal paraffin 
14 uterine rupture coronal paraffin 
– 14 spontaneous miscarriage coronal frozen 
14 spontaneous miscarriage coronal frozen 
10 – 15 – coronal frozen 
11 15 induced abortion sagittal frozen 
12 16 spontaneous miscarriage coronal frozen 
13 – 16 – coronal frozen 
14 16 spontaneous miscarriage coronal frozen 
15 16 spontaneous miscarriage coronal frozen 
16 17 spontaneous miscarriage coronal frozen 
17a 17 – coronal frozen 
18a 17 – coronal frozen 
19 – 17 spontaneous miscarriage coronal frozen 
20 18 spontaneous miscarriage coronal frozen 
21 18 spontaneous miscarriage coronal frozen 
22 18 spontaneous miscarriage coronal frozen 
23 18 spontaneous miscarriage coronal frozen 
24 18 spontaneous miscarriage coronal frozen 
25 19 induced abortion horizontal frozen 
26 19 – coronal frozen 
27 19 spontaneous miscarriage coronal frozen 
28a 19 spontaneous miscarriage coronal frozen 
29a 19 spontaneous miscarriage coronal frozen 
30 19–20 induced abortion sagittal frozen 
31 20 spontaneous miscarriage coronal frozen 
32 20 induced abortion coronal frozen 
33 20 spontaneous miscarriage horizontal frozen 
34 – 20 – coronal frozen 
35 20 spontaneous miscarriage coronal frozen 
36 20 induced abortion coronal frozen 
37 22 spontaneous miscarriage horizontal frozen 
38 22 induced abortion coronal frozen 
39 – 22 induced abortion coronal frozen 
40 23 spontaneous miscarriage coronal frozen 
41 23 premature delivery coronal frozen 
42 23 induced abortion coronal frozen 
43 – 24 induced abortion coronal frozen 
44 24 spontaneous miscarriage coronal frozen 
45
 
M
 
24
 
induced abortion
 
coronal frozen
 

Where cases of spontaneous miscarriage are recorded, no fetal congenital abnormalities were present. There was no evidence of hypoxic-ischaemic injury or haemorrhage, either within the germinal matrix or associated with the meninges in any of these cases. Gestational age is given in weeks and was determined according to a number of parameters including crown–rump measurements, anatomically by foot length or by history of the last menstrual period.

a

These cases were twin births at 17 gw and 19 gw respectively.

–, Clinical record not available.

For immunohistochemistry, a three-step ABC-horseradish peroxidase (ABC-HRP) procedure was followed according to standard protocols. Briefly, consecutive cryostat sections were air-dried at room temperature for 2 h and paraffin-embedded sections dewaxed for 15 min in xylene, before immersion in methanol solution containing 2.5% of a 30% hydrogen peroxide solution for 2 h at room temperature. Non-specific binding was blocked for 2 h by incubating sections with normal rabbit serum (1:10 dilution, Dako), and these were next immunoreacted with monoclonal murine antisera directed against human CD68 (clone PG-M1, 1:50 dilution, Dako) and MHC class II antigen (recognizing DP-DQ-DR, clone CR3/43, 1:200, Dako) for 24 h at room temperature, to detect macrophages and microglia. Biotinylated rabbit anti-mouse antibody (1:200 dilution, 2 h at room temperature, Dako) was applied to sections following thorough washes with Tris-buffered saline solution (TBS, pH 7.6, 3 × 5 min). After a further three washes with TBS, sections were incubated with ABC-HRP (Vectastain Elite Kit prepared 1 h prior to use, Vector Laboratories) for a further 2 h at room temperature, prior to washing with TBS. Immunoreactivity was visualized with 3,3′-diaminobenzidine (DAB, with hydrogen peroxide) as chromogen, using cobalt chloride for enhancement. Cobalt chloride also gave additional definition to cellular profiles within cryostat sections of fetal brains, making nuclear counterstaining unnecessary. Where required, however, nuclei were counterstained with haematoxylin or methyl green solutions. Sections were dehydrated in graded alcohols, cleared in xylene and coverslipped in the usual manner. All sections were assayed in the same batch in order to ensure consistent labelling. Negative control sections were included where the primary and/or secondary antibody was replaced by normal serum alone.

Lectin histochemistry was carried out according to previously established protocols (Rezaie et al., 1997, 1999), incubating sections with biotinylated tomato lectin or RCA-1 (Vector Laboratories, UK) at 1:100–1:500 dilution, overnight at room temperature, followed by three washes in PBS and incubation with ABC-HRP (Vectastain Elite Kit, Vector Laboratories) for 2 h at room temperature. The reaction was visualized with DAB and cobalt chloride as outlined above. Negative control sections were incubated either with tomato lectin solution containing 400 mM N-acetylglucosamine, or with RCA-1 solution containing 400 mM lactose and galactose (inhibitory substrates).

Serial 120 μm thick free-floating sections from cases at mid-trimester (18–19 gw) were used for dual labelling with RCA-1 lectin and CD68 antisera. These were first incubated with the lectin according to the protocol described above, and the reaction visualized with DAB (brown). Subsequently the sections were rinsed in Tris-buffered saline solution containing 0.025% of levamisole (Sigma), incubated with CD68 antisera overnight and the reaction visualized using an alkaline phaosphatase system and 5-bromo-4-chloro-indolyl-phosphate/nitro-blue-tetrazolium (BCIP/NBT), 1:50 (Vector Laboratories), according to established protocols (Ulfig et al., 1998).

Results

The terminology used throughout this paper in reference to the histological layers of the cerebral wall of the telencephalon at second trimester is in accordance with previous established definitions (Boulder Committee, 1970; Chan et al., 2002). The subplate (SP) is particularly noteworthy, as a transient structure located immediately below the cortical plate (CP), which harbours thalamocortical fibres, as these develop and make their connections within the laminae of the CP. Figure 1 shows representative sections of the human fetal forebrain, to illustrate the rapid expansion of the telencephalon during the period of the second trimester in the cases under investigation.

CD68 and MHC Class II Immunoreactivity

An overview of the developmental stages of human fetal microglia and their associated expression of surface markers has been presented elsewhere (Rezaie and Male, 1999; Rezaie, 2003). An overall finding from the present study was that microglia were distributed throughout dorsal and lateral aspects of the telencephalon, but much fewer cells could be detected in ventral and medial aspects throughout the period under investigation.

Between 12 and 14 gw, immunohistochemistry with CD68 labelled an early population of mononuclear phagocytes residing within the marginal layer, interspersed throughout the lower CP and at the boundary between the CP and SP (Fig. 2). Immunoreactivity with CD68 was particularly intense on round cells within the upper marginal layer, underlying the meninges (Fig. 2A,B). The majority of these latter cell types possessed an uneven surface and closely resembled macrophages in morphology (identified by their relatively larger size and irregular ‘ruffled’ surface membrane). Many cells were located around (but not usually within) delicate, small-diameter and thin-walled meningeal vessels (not shown). Deeper in the marginal layer, and within the upper limits of the CP, CD68-positive cells were smaller and amoeboid (Fig. 2C). CD68-positive cells within the SP (Fig. 2DF) were clearly also amoeboid and demonstrated surface and cytoplasmic labelling with this marker. Figure 2G,H shows CD68-positive cells aligned at the boundary of the CP more clearly. Some of these CD68-positive cells were located around blood vessels that penetrated the CP (Fig. 2E), and an occasional isolated cell could be seen closely associated with such vessels (Fig. 2I). Between 14 and 17 gw, a different laminar distribution of CD68-positive cells was evident: these were specifically confined to the lower aspect of the SP and exhibited unipolar, bipolar, multipolar morphology (Fig. 2J), or were occasionally dividing (not shown). The majority of these ramifying microglia possessed between two and six processes of variable length, extending up to several micrometres (Fig. 2F and 2J, inset). The population of ramifying microglia came to occupy progressively more of the full width of the SP between 18 and 22 gw (Fig. 3C). Overlapping these developments, between 14 and 24 gw, CD68-positive cells occurred in two other locations, namely the germinal layers (VZ and SVZ) and the corpus callosum (Figs 2K,L and 3A,D,E), but were conspicuously absent from the bulk of the CP.

Figure 2.

CD68-positive cells in the human fetal telencephalon, 12–17 gw. (A) CD68-immunoreactive cells are located within the marginal layer (ML) below the meninges, interspersed throughout the cortical plate (CP) with particular preference for lower cortical layers, and within the subplate (SP) immediately subjacent to the CP between 12 and 13 gw. (B, C) Higher power photos of CD68-immunoreactive cells located within the ML (indicated by closed circles in A). The majority of these rounded mononuclear phagocytes resemble monocytes/macrophages in their morphology (B, and inset). They are frequently detected around small meningeal vessels, particularly overlying the marginal layer. Occasionally, more differentiated, CD68-immunoreactive cells of smaller, amoeboid morphology can be seen within the marginal layer itself (C, arrows) and within the upper layers of the cortical plate. (DF) Higher power photos of CD68-immunoreactive cells confined to the subplate (indicated by arrowheads in A). These cells are clearly amoeboid in morphology. Some associate with radial blood vessels (BV) passing through the cortical plate and subplate (E), and occasionally appear more ramified, extending processes several micrometers away from the cell body (F). Expression of CD68 is both at the cell surface and of a vesicular pattern within the cytoplasm, in keeping with previous descriptions of macrophages. (G, H) Clear examples of amoeboid CD68-positive cells located at the boundary between the cortical plate and the subplate, at 13 and 14 gw. Small CD68-positive cells are also associated with radiating blood vessels passing through the cortical plate (I) and can be found lying freely within the lower aspect of the subplate between 13 and 15 gw, where some are dividing (not shown) and others beginning to sprout processes (J, inset). By 16–17 gw, CD68-positive cells are particularly dense within the corpus callosum and in the ventricular and subventricular zones (K, L), but are not detectable in the cortical plate (not shown).Nuclei counterstained with haematoxylin or methyl green. Scale bar represents ∼80 μm in (A), 12 μm in (B), 20 μm in (CF), 140 μm in (G, L), 30 μm in (H, I), 70 μm in (J), 310 μm in (K).

Figure 2.

CD68-positive cells in the human fetal telencephalon, 12–17 gw. (A) CD68-immunoreactive cells are located within the marginal layer (ML) below the meninges, interspersed throughout the cortical plate (CP) with particular preference for lower cortical layers, and within the subplate (SP) immediately subjacent to the CP between 12 and 13 gw. (B, C) Higher power photos of CD68-immunoreactive cells located within the ML (indicated by closed circles in A). The majority of these rounded mononuclear phagocytes resemble monocytes/macrophages in their morphology (B, and inset). They are frequently detected around small meningeal vessels, particularly overlying the marginal layer. Occasionally, more differentiated, CD68-immunoreactive cells of smaller, amoeboid morphology can be seen within the marginal layer itself (C, arrows) and within the upper layers of the cortical plate. (DF) Higher power photos of CD68-immunoreactive cells confined to the subplate (indicated by arrowheads in A). These cells are clearly amoeboid in morphology. Some associate with radial blood vessels (BV) passing through the cortical plate and subplate (E), and occasionally appear more ramified, extending processes several micrometers away from the cell body (F). Expression of CD68 is both at the cell surface and of a vesicular pattern within the cytoplasm, in keeping with previous descriptions of macrophages. (G, H) Clear examples of amoeboid CD68-positive cells located at the boundary between the cortical plate and the subplate, at 13 and 14 gw. Small CD68-positive cells are also associated with radiating blood vessels passing through the cortical plate (I) and can be found lying freely within the lower aspect of the subplate between 13 and 15 gw, where some are dividing (not shown) and others beginning to sprout processes (J, inset). By 16–17 gw, CD68-positive cells are particularly dense within the corpus callosum and in the ventricular and subventricular zones (K, L), but are not detectable in the cortical plate (not shown).Nuclei counterstained with haematoxylin or methyl green. Scale bar represents ∼80 μm in (A), 12 μm in (B), 20 μm in (CF), 140 μm in (G, L), 30 μm in (H, I), 70 μm in (J), 310 μm in (K).

Figure 3.

Distribution of microglia in the wall of the telencephalon, 17–19 gw. (A) Composite low-power photographs spanning the cerebral wall at 17 gw to show the dense focal distribution of CD68-positive cells (arrowheads) in the corpus callosum and lower aspect of the subplate. Microglia within the subplate (B, C), subventricular and ventricular zones (D, E) are clearly identifiable by 19 gw. In the ventricular and subventricular zones (D, E) fetal microglia occur as dense aggregates (corresponding to Del Rio-Hortega's ‘fountains’), whereas within the subplate, microglia have already begun to differentiate into ramified cells (B). Scale bar represents ∼520 μm in (A), 12 μm in (B), 310 μm in (CE).

Figure 3.

Distribution of microglia in the wall of the telencephalon, 17–19 gw. (A) Composite low-power photographs spanning the cerebral wall at 17 gw to show the dense focal distribution of CD68-positive cells (arrowheads) in the corpus callosum and lower aspect of the subplate. Microglia within the subplate (B, C), subventricular and ventricular zones (D, E) are clearly identifiable by 19 gw. In the ventricular and subventricular zones (D, E) fetal microglia occur as dense aggregates (corresponding to Del Rio-Hortega's ‘fountains’), whereas within the subplate, microglia have already begun to differentiate into ramified cells (B). Scale bar represents ∼520 μm in (A), 12 μm in (B), 310 μm in (CE).

In summary, three spatially separate bands of CD68-positive cells could be identified in the wall of the telencephalon, underlying the CP, during the mid-second trimester period: those located in the (i) lower SP, (ii) the subventricular zone/corpus callosum and (iii) the ventricular zone. With the passage of time, CD68-positive cells were also detected within the upper SP. Likewise, in addition to the lower germinal layers, the intermediate zone was further and progressively populated by CD68-positive cells until the entire expanse of the wall of the telencephalon (with the exception of the CP) was colonized by 24 gw. The cells residing within the SP (particularly the upper aspect) expressed CD68 antigen only faintly and had begun to ramify more extensively (Fig. 3B).

Microglia in the SP did not express MHC class II antigens (Fig. 4A). Instead, MHC class II antigen was specifically expressed by microglia in the corpus callosum, subventricular and ventricular zones of the telencephalon from 14 to 17 gw onwards (Fig. 4A, EH). MHC class II separately labelled blood vessels weakly in the SP from 20 gw (Fig. 4B) and perivascular cells associated with radiating cortical blood vessels initially from 20 gw and more specifically around 24 gw (Fig. 4C,D).

Figure 4.

MHC class II positive cells in the human fetal telencephalon at 20 gw. Whereas blood vessels are weakly labelled in the subplate (A, B), microglia within this region do not express MHC class II molecules, and expression of this marker is also absent from the cortical plate (A). From 20 gw onwards, however, small MHC class II positive perivascular cells were detected within small-to-medium diameter blood vessels that pass through the subplate (C), and some begin to associate with and become closely apposed to the surface of such blood vessels in the subplate (D). Parenchymal microglia located specifically in the corpus callosum (E, G) and within periventricular areas (F, H) are MHC class II-positive. Scale bar represents ∼1150 μm in (A), 370 μm in (B), 140 μm in (E, F), 12 μm in (C, D, G, H).

Figure 4.

MHC class II positive cells in the human fetal telencephalon at 20 gw. Whereas blood vessels are weakly labelled in the subplate (A, B), microglia within this region do not express MHC class II molecules, and expression of this marker is also absent from the cortical plate (A). From 20 gw onwards, however, small MHC class II positive perivascular cells were detected within small-to-medium diameter blood vessels that pass through the subplate (C), and some begin to associate with and become closely apposed to the surface of such blood vessels in the subplate (D). Parenchymal microglia located specifically in the corpus callosum (E, G) and within periventricular areas (F, H) are MHC class II-positive. Scale bar represents ∼1150 μm in (A), 370 μm in (B), 140 μm in (E, F), 12 μm in (C, D, G, H).

Lectin Histochemistry

RCA-1 labelled blood vessels, microglial progenitors and differentiating microglia between 12 and 24 gw. Microglia were clearly distinguishable from blood vessels that were also stained. Specifically, RCA-1 (and tomato lectin albeit more weakly, data not shown) highlighted a population of small and rounded progenitor cells within the marginal layer, CP and SP of the wall of the telencephalon between 12 and 14 gw, which were morphologically distinct from the CD68-immunoreactive cells discussed above (Fig. 5). These cells were small in diameter, presented with round or ovoid morphologies, and were found interspersed in the upper aspect of the marginal layer, where they occasionally occurred as paired cells (Fig. 5B, and inset to the right). RCA-1-positive cells could also be seen associated with blood vessels penetrating the CP which were also labelled, although they could only very infrequently be detected within blood vessels (see Fig. 5, colour panel I). However, these cells were detected at various stages of ‘detachment’ from radial vessels within the lower aspects of the CP (Fig. 5C,D). There were clear morphological differences between progenitors located within the CP and those located to the SP, as shown in Figure 5ER. Cells within the CP (typically in the lower part) were dividing (Fig. 5JM), and those within the SP now bore processes. This progressive transformation took place largely between 12 and 15 gw, and the different ‘stages’ are shown more clearly in colour panels I–III to the right of Figure 5. By 15 gw, progenitors in the lower CP now exhibited between one and three delicate processes (Fig. 5, colour panel II), whereas cells within the upper and lower aspects of the SP were in a more advanced state of differentiation, and beginning to resemble adult microglia in morphology (Fig. 5, colour panel III).

Figure 5.

RCA-1 positive progenitors within the cortical plate and subplate between 12 and 15 gw. Lectin histochemistry with RCA-1 clearly identified blood vessels and microglial progenitors within the telencephalon between 12 and 14 gw. Microglial progenitors and differentiating cells were found predominantly in the lower cortical plate and subplate (A, arrowheads), whereas the intermediate zone, subventricular and ventricular areas were largely devoid of these cells at this stage of development. RCA-1-labelled cells located within the marginal layer were small in diameter and more rounded or ovoid in appearance (BD). Occasionally they were found in pairs (B, and inset on the right), and some were clearly associated with blood vessels that penetrated the cortical plate (C, D). RCA-1-positive cells appeared to attach to, and progressively detach from radial vessels. (ER) A clear transition in the morphology of RCA-1-positive progenitors was seen when progressing from the lower cortical plate to the upper aspect of the subplate, with scattered cells (EI) detected particularly within the lower cortical plate, some of which were dividing (JM) and progressively more differentiated, process-bearing cells found within the upper aspect of the subplate region (NR) in advancing towards the intermediate zone. The differentiation of RCA-1-positive progenitors was more advanced between 14 and 15 gw. The transition from small rounded progenitors in the cortical plate, some of which were associated with blood vessels, or were dividing (colour panel I), and their replacement by unipolar and bipolar forms within the lower cortical plate and upper aspect of the subplate (colour panel II), and further differentiation to multipolar ramified cell types within the subplate (colour panel III) is clearly shown to the right of this figure. Methyl green counterstain. Abbreviations: ML, marginal layer; CP, cortical plate; SP, subplate; IZ, intermediate zone; SVZ, subventricular zone; VZ, ventricular zone. Scale bar represents ∼150 μm in (A), 25μm in (BY, colour panel I), 30 μm in colour panels II and III.

Figure 5.

RCA-1 positive progenitors within the cortical plate and subplate between 12 and 15 gw. Lectin histochemistry with RCA-1 clearly identified blood vessels and microglial progenitors within the telencephalon between 12 and 14 gw. Microglial progenitors and differentiating cells were found predominantly in the lower cortical plate and subplate (A, arrowheads), whereas the intermediate zone, subventricular and ventricular areas were largely devoid of these cells at this stage of development. RCA-1-labelled cells located within the marginal layer were small in diameter and more rounded or ovoid in appearance (BD). Occasionally they were found in pairs (B, and inset on the right), and some were clearly associated with blood vessels that penetrated the cortical plate (C, D). RCA-1-positive cells appeared to attach to, and progressively detach from radial vessels. (ER) A clear transition in the morphology of RCA-1-positive progenitors was seen when progressing from the lower cortical plate to the upper aspect of the subplate, with scattered cells (EI) detected particularly within the lower cortical plate, some of which were dividing (JM) and progressively more differentiated, process-bearing cells found within the upper aspect of the subplate region (NR) in advancing towards the intermediate zone. The differentiation of RCA-1-positive progenitors was more advanced between 14 and 15 gw. The transition from small rounded progenitors in the cortical plate, some of which were associated with blood vessels, or were dividing (colour panel I), and their replacement by unipolar and bipolar forms within the lower cortical plate and upper aspect of the subplate (colour panel II), and further differentiation to multipolar ramified cell types within the subplate (colour panel III) is clearly shown to the right of this figure. Methyl green counterstain. Abbreviations: ML, marginal layer; CP, cortical plate; SP, subplate; IZ, intermediate zone; SVZ, subventricular zone; VZ, ventricular zone. Scale bar represents ∼150 μm in (A), 25μm in (BY, colour panel I), 30 μm in colour panels II and III.

Dual Labelling of Microglia with RCA-1 and CD68

The distribution of heterogeneous populations of microglia were more clearly defined at mid-second trimester through dual labelling of tissue sections using RCA-1 lectin (brown) and immunohistochemistry with CD68 (blue), shown in Figure 6. In particular, cells could be identified that were predominantly expressing CD68 antigen (blue in Fig. 6A,C,D) in the ventricular and subventricular zones and the corpus callosum, or predominantly detected with RCA-1 lectin (brown in Fig. 6B,DF), particularly ramified cells within the subventricular and intermediate zones (Fig. 6DF, closed arrowheads). Ramified cells expressing both of these markers were located within the lower aspects of the SP and in the intermediate zone during this period (Fig. 6E,F, open arrowheads). These double-labelling studies serve to strengthen the argument for the existence of heterogeneous populations of microglia during the second trimester. Nevertheless, further investigations are necessary to clarify the precise origin of microglial progenitors.

Figure 6.

Dual labelling of microglial populations with RCA-1 lectin and CD68, 18–19 gw. Lectin histochemistry with RCA-1 (brown) combined with immunohistochemistry against CD68 (blue), identified populations of cells that were predominantly blue (expressing CD68 antigen, A, C, D) particularly located within the ventricular and subventricular zones and the corpus callosum, or predominantly brown (RCA-1 positive cells, B, D, E, F): ramified cells within the subventricular and intermediate zones (D, E, F closed arrowheads). Ramified cells expressing both of these markers were located within the intermediate zone (E, open arrowheads) and lower aspects of the subplate during this period (F, open arrowheads). For orientation, the cortical plate is to the top of panel B, and to the top right of panels E and F. Scale bar represents ∼110 μm in (A), 150 μm in (B), 60 μm in (C) and 120 μm in (DF).

Figure 6.

Dual labelling of microglial populations with RCA-1 lectin and CD68, 18–19 gw. Lectin histochemistry with RCA-1 (brown) combined with immunohistochemistry against CD68 (blue), identified populations of cells that were predominantly blue (expressing CD68 antigen, A, C, D) particularly located within the ventricular and subventricular zones and the corpus callosum, or predominantly brown (RCA-1 positive cells, B, D, E, F): ramified cells within the subventricular and intermediate zones (D, E, F closed arrowheads). Ramified cells expressing both of these markers were located within the intermediate zone (E, open arrowheads) and lower aspects of the subplate during this period (F, open arrowheads). For orientation, the cortical plate is to the top of panel B, and to the top right of panels E and F. Scale bar represents ∼110 μm in (A), 150 μm in (B), 60 μm in (C) and 120 μm in (DF).

Figures 7 and 8 schematically summarize the progressive colonization of the cerebral wall of the human telencephalon by phenotypic and morphologically different populations of microglia during the second trimester. The phenotypic expression of these markers on microglia is indicated qualitatively on these figures as follows: (−) absence of phenotypic marker, (+/−) weak or absent expression, (+) moderate expression, (++) intense expression. Figure 7 highlights the differential expression of CD68 antigen and RCA-1 binding sites on microglia, with emphasis predominantly on the distribution of RCA-1-positive cells. Figure 8 illustrates the differential expression of CD68 and MHC class II antigen expression by microglia, with emphasis predominantly on the distribution of CD68-positive cells.

Figure 7.

Schematic illustrations to show the progressive colonization of the cerebral wall of the human telencephalon by microglia between 12 and 24 gw. Part I: differential expression of CD68 antigen and RCA-1 binding sites. Initially between 12 and 14 gw, two morphologically and phenotypically distinct populations of cells (population A and population B) can be detected within the marginal layer, cortical plate (CP) and subplate (SP). Cell population A are amoeboid, CD68++ RCA-1+ and reside within the marginal layer and immediately subjacent to the cortical plate. These cells resemble macrophages morphologically. Population B are small RCA-1++ CD68- progenitor cells within the marginal layer and cortical plate, which are progressively replaced by more ramified cells within the subplate. These cells do not resemble macrophages, but do resemble early forms of microglia that are found in the mature nervous system. Between 16 and 24 gw, amoeboid CD68++ cells progressively accumulate additionally within the ventricular and subventricular zones (VZ, SVZ) and corpus callosum (CC), and more ramified cells can be found dispersed ubiquitously throughout the subplate (SP) and intermediate zones (IZ) of the rapidly expanding telencephalon. Abbreviations: CP, cortical plate; SP, subplate; IZ, intermediate zone; SVZ, subventricular zone; CC, corpus callosum; VZ, ventricular zone.

Figure 7.

Schematic illustrations to show the progressive colonization of the cerebral wall of the human telencephalon by microglia between 12 and 24 gw. Part I: differential expression of CD68 antigen and RCA-1 binding sites. Initially between 12 and 14 gw, two morphologically and phenotypically distinct populations of cells (population A and population B) can be detected within the marginal layer, cortical plate (CP) and subplate (SP). Cell population A are amoeboid, CD68++ RCA-1+ and reside within the marginal layer and immediately subjacent to the cortical plate. These cells resemble macrophages morphologically. Population B are small RCA-1++ CD68- progenitor cells within the marginal layer and cortical plate, which are progressively replaced by more ramified cells within the subplate. These cells do not resemble macrophages, but do resemble early forms of microglia that are found in the mature nervous system. Between 16 and 24 gw, amoeboid CD68++ cells progressively accumulate additionally within the ventricular and subventricular zones (VZ, SVZ) and corpus callosum (CC), and more ramified cells can be found dispersed ubiquitously throughout the subplate (SP) and intermediate zones (IZ) of the rapidly expanding telencephalon. Abbreviations: CP, cortical plate; SP, subplate; IZ, intermediate zone; SVZ, subventricular zone; CC, corpus callosum; VZ, ventricular zone.

Figure 8.

Schematic illustrations to show the progressive colonization of the cerebral wall of the human telencephalon by microglia between 12 and 24 gw. Part II: differential expression of CD68 and MHC class II antigens. CD68 and MHC class II clearly highlighted different populations of microglia within the telencephalon between 12 and 24 gw. CD68 clearly detected amoeboid microglia confined immediately below the cortical plate from 12 to 13 gw as shown previously (Fig. 7). However from 14 to 22 gw, CD68-positive cells were mainly restricted to the lower aspect of the subplate, where they had begun to differentiate into process-bearing forms, and by 24 gw, these cells had distributed towards the cortical plate, occupying the upper aspect of the subplate. Significantly, these CD68++ cells within the subplate did not express MHC class II antigen. Instead, MHC class II positive (and CD68++) cells were restricted to the ventricular and subventricular zones of the developing telencephalon from 14 to 24 gw. Abbreviations: CP, cortical plate; SP, subplate; uSP, upper aspect of subplate; lSP, lower aspect of subplate; IZ, intermediate zone; SVZ, subventricular zone; CC, corpus callosum; VZ, ventricular zone.

Figure 8.

Schematic illustrations to show the progressive colonization of the cerebral wall of the human telencephalon by microglia between 12 and 24 gw. Part II: differential expression of CD68 and MHC class II antigens. CD68 and MHC class II clearly highlighted different populations of microglia within the telencephalon between 12 and 24 gw. CD68 clearly detected amoeboid microglia confined immediately below the cortical plate from 12 to 13 gw as shown previously (Fig. 7). However from 14 to 22 gw, CD68-positive cells were mainly restricted to the lower aspect of the subplate, where they had begun to differentiate into process-bearing forms, and by 24 gw, these cells had distributed towards the cortical plate, occupying the upper aspect of the subplate. Significantly, these CD68++ cells within the subplate did not express MHC class II antigen. Instead, MHC class II positive (and CD68++) cells were restricted to the ventricular and subventricular zones of the developing telencephalon from 14 to 24 gw. Abbreviations: CP, cortical plate; SP, subplate; uSP, upper aspect of subplate; lSP, lower aspect of subplate; IZ, intermediate zone; SVZ, subventricular zone; CC, corpus callosum; VZ, ventricular zone.

Discussion

The phagocytic and immunological capacities of microglia are now well recognized and numerous studies have emphasized their potential involvement in diseases affecting the nervous system (Rezaie and Male, 2002b; Rezaie, 2003). Considerably less is known, however, regarding the roles that microglia play during development, and the origin of these cells has sparked countless debates over the past two decades (Rezaie and Male, 2002a). In recent years, we have begun to gain a clearer understanding of the phasing and patterns of colonization of the developing human brain by microglia (Rezaie, 2003). In the present work we have demonstrated that the cerebral wall of the telencephalon is colonized by at least two populations of cells destined to give rise to microglia during the second trimester. To be more specific, we have shown that the cerebral wall is colonized initially during the start of the second trimester by a population of RCA-1-positive microglial progenitors that are localized within the marginal layer underlying the mesenchyme/meninges surrounding the brain, and take up residence specifically within the SP where they are replaced progressively by process-bearing forms. A second population of cells expressing CD68 are more typical of the ‘amoeboid microglia’ widely discussed in the literature with characteristic morphology more closely resembling that of macrophages. These cells preside in the deep marginal layer and at the boundary between the CP and SP at the start of the second trimester. From 14–17 gw they can be found within the lower aspect of the SP, where they also appear to differentiate and progressively populate the upper aspect of this region by 24 gw.

Thus, two putative types of progenitors have been identified based on morphological and phenotypical presentation during the initial phase of colonization. The first type appear as small rounded cells with the phenotype: RCA-1++ CD68− MHC II− (they are also negative for the markers CD11b, CD45 and CD64, and only weakly positive for tomato lectin, data not shown). These cells clearly differentiate early within the SP of the human fetal telencephalon. The second variety are CD68++ RCA-1+ MHC II− (they are also only weakly positive or negative for CD11b, CD45, CD64 and tomato lectin, data not shown) and typically adopt amoeboid macrophage-like forms. The second type of cell can be found from 12 to 14 gw in the marginal layer, immediately subjacent to the CP and at later stages (14–24 gw) in the SP, and additionally in the ventricular and subventricular zones, in the corpus callosum and in the developing white matter of the intermediate zone. These progenitors divide within the germinal layers, disperse throughout the lower layers and the intermediate zone, and differentiate within this region to more ramified cells and downregulate expression of markers at the cell surface. By contrast, MHC class II antigen was specifically detected on microglia in the corpus callosum and germinal layers (VZ, SVZ) of the telencephalon from 14 to 17 gw onwards. Colonization was well underway by 16 gw, and by 22–24 gw microglia had distributed widely throughout the developing human fetal telencephalon and began to take on ramified forms characteristic of adult microglial cells, probably constituting a merger of the two populations of progenitor cells. However, amoeboid microglia still prevailed within the ventricular zone, subventricular zone and corpus callosum even up to the beginning of the third trimester. Microglia located within the SP maintained their territorial fields, and were more ramified, with a downregulated phenotype (expression of surface/cytoplasmic markers).

The stages of colonization within the wall of the telencephalon are summarized in Figures 7 and 8, where the initial phase of migration of lectin-positive progenitors inwards from the marginal layer is met during the latter half of the second trimester by an advancing population of CD68 and MHC class II positive microglia migrating outwards from the germinal layers (ventricular and subventricular zones). Placing these findings in the context of microglial development according to the literature, the CD68++ RCA-1+ MHC II− and CD68++ RCA-1+ MHC II++ cells with morphologies more typical of macrophages (large, rounded vesicle-laden cells with an uneven or ‘ruffled’ surface membrane) can be referred to as ‘amoeboid’ microglia by virtue of their phenotype, morphology and localization (particularly within the germinal layers and corpus callosum). Likewise, the population of RCA-1++ CD68+/− MHC II− and RCA+ CD68+ MHC II− cells with small ovoid or round cell body and bearing 2–5 delicate and long, branched processes can be referred to as ‘differentiating’, ‘transforming’ or ‘early ramified’ microglia, again in keeping with their phenotype, morphology and localization to the SP and intermediate zones. Finally, the population of RCA-1++ CD68− MHC II− cells of small, rounded (undifferentiated) appearance that mainly resided in the marginal layer, the CP and SP at the outset of the second trimester are here referred to as microglial progenitors. These definitions are summarized in Figure 9. The term ‘mononuclear phagocyte’ encompasses all these different characteristic forms of developing microglia. However, the precise relationship between amoeboid cells and early ramified microglia is difficult to determine from these studies in situ. As already discussed, the amoeboid ‘state’ appears to be bypassed at least during the first phase of colonization of the developing telencephalon that takes place before 16 gw, where by far the majority of microglial progenitor cells begin to differentiate morphologically and ramify without apparently passing through an amoeboid stage.

Figure 9.

Criteria used to define populations of microglia in the human fetal telencephalon during the second trimester, based on phenotypic and morphological characteristics, and in the context of widespread reference in the literature.

Figure 9.

Criteria used to define populations of microglia in the human fetal telencephalon during the second trimester, based on phenotypic and morphological characteristics, and in the context of widespread reference in the literature.

The marked heterogeneity in phenotype of these microglia during the second trimester may reflect a fundamental difference in ontogeny between the cell types, or relate to their state of differentiation and specialized functions, regionally within the human fetal telencephalon. The migration and maturation of microglia, known to be influenced by factors derived from astrocytes (Rezaie and Male, 2002a), coincides both with the development of radial glia and the maturation of astrocytes within respective brain areas (Rezaie et al., 2003). In particular, the differentiation of astrocytes also takes place predominantly in the SP and intermediate zone of the telencephalon towards the end of the second trimester, and the differentiation of these cells follows a similar spatially interspersed pattern in these regions as that reported for microglia (Rezaie et al., 2003). Nevertheless, we have noted that astrocytes appear to differentiate at a slightly later stage than described for microglia (Rezaie et al., 2003), a finding that will need additional investigation, particularly with respect to whether microglia may themselves exert a direct influence over the regional differentiation of astrocytes with which they co-localize. This particular hypothesis is intriguing since an overwhelming number of studies in vitro have focused on the contrary relationship, i.e. the influence of astrocytes on microglial differentiation (Rezaie and Male, 2002a). Whether microglia can exert important roles in the maturation not only of astrocytes, but also of oligodendrocyte progenitors, remains to be established.

Our earlier findings also showed that GFAP-positive astrocytes associate closely with blood vessels within the SP and intermediate zone of the telencephalon from 19 gw onwards, and that this could indicate the commencement of regional barrier formation between the blood and brain (Rezaie et al., 2003). It has been reported that perivascular cells which invest small to medium-sized blood vessels in the adult human brain can be readily identified by their intrinsically high levels of expression of MHC class II antigen (Graeber et al., 1992). In this study, we observed that MHC class II positive perivascular cells first began to invest radiating cerebral vessels in the SP and intermediate zone of the telencephalon from around 20 gw onwards and more prevalently by the end of the second trimester. These observations correspond well with the concept that a fetal BBB has begun to establish by the start of the third trimester, but warrant further investigation.

Separately, microglial colonization has been considered to be a response towards programmed neuronal death or axonal degeneration during development (Perry et al., 1985; Rakic and Zecevic, 2000). However, the distribution of microglia does not conform solely to that of programmed cell death during fetal development (Rezaie, 2003), and this idea is not consistent with the overall patterns of distribution and differentiation of microglial progenitors shown in this study. Therefore, it appears likely that the process underlying colonization of the cerebral wall of the telencephalon by microglia is primarily a developmentally regulated step rather than exclusively a response to cell death in a simplistic sense. What, therefore, preferentially attracts microglial progenitors to the SP during the initial period of the second trimester?

Between 13 and 36 gw, this transient structure serves as a ‘waiting area’ for unsettled thalamocortical, corticostriatal and associative and commissural pathways that are involved in area specification of the cortex, participating in the induction of a new area (Kostovic and Judas, 2002). Furthermore, the SP acts as a reservoir for maturing neurons and transient synapses that may also exert influence on the laminar arrangement of the CP (Ulfig et al., 2000; Kostovic and Judas, 2002). Microglial colonization of the SP therefore coincides with the development of SP neurons, synaptogenesis and thalamocortical projection fibres that transiently reside within this region (Kostovic and Judas, 2002). The SP zone is also characterised by a large extracellular space that is rich in extracellular matrix components, particularly of the chondroitin sulphate family, laminin, fibronectin, tenascin, as well as chemoattractant and chemorepellant factors (Kostovic and Judas, 2002). It is known that macrophages and microglia are capable of synthesizing and responding to several proteoglycans including keratan, chondroitin and dermatan sulphate proteoglycans under physiological and pathological conditions (Edwards et al., 1990; Jander et al., 2000; Jones and Tuszynski, 2002; Jones et al., 2002; Moon et al., 2002). Keratan, heparan and chondroitin sulphate proteoglycans have been shown to regulate neuronal outgrowth and survival (Kostovic and Judas, 2002). Thus, the repellant/anti-adhesive properties of the CP and the attractant/neurite-promoting properties of the SP and intermediate zone, are reliant on specific and differential expression of bound proteoglycans (Kostovic and Judas, 2002). Likewise, cytokine and growth factors that are bound to specific sulphated proteoglycans can promote the migration of mononuclear phagocytes (Hayashi et al., 2001). Proteolytic activity may also establish chemoattractant gradients for microglia. In fact, in our own previous studies, we have found that chemokines that are known to selectively recruit monocytes and macrophages (including monocyte chemoattractant protein-1/MCP-1 and RANTES) are specifically expressed in the SP and CP of the developing telencephalon during the second trimester, and we have proposed that these may act as developmental signals for recruiting microglial progenitors to the developing cortex (Rezaie and Male, 1999). Therefore, it would appear that the SP is not only a site which promotes the recruitment of microglial progenitors, but also contributes towards their early differentiation.

It is not known whether microglia perform more specialized functions in the developing telencephalon. However, speculations on such functions include synaptic remodelling or sorting (Zecevic et al., 1989), maintenance of the CP border through phagocytosis or elimination of debris and of apoptotic cells (Ferrer et al., 1990; Rakic and Zecevic, 2000), or the sorting, guidance or elimination of axonal fibers or neurons (Innocenti et al., 1983; Ume et al., 1983; Perry et al., 1985). Towards, the latter half of the second trimester, it may also be possible that microglia are involved in the processes of gliogenesis, modelling of white matter tracts and perhaps myelinization (Zecevic et al., 1998), or even the development of a BBB (Mato et al., 1996). These proposals await further investigations.

The work presented here and in several recent published articles offers a clearer picture of the morphology, phenotype and distribution of human fetal microglia in the nervous system during development (reviewed in Rezaie, 2003). Previous accounts have generally held the view that microglial progenitors enter the CNS from within and around the germinal layers: ventricular and subventricular zones. A striking similarity in the majority of these prior investigations was the characteristic ‘pattern of differentiation’ associated with the progressive outward migration and ramification of microglia from such original portals of entry. However, it has emerged from this study that microglial progenitors are intimately associated with the parenchymal wall of penetrating radial blood vessels, and appear to invade the parenchyma of the telencephalic wall inwards from the marginal layer, and align within the SP region of the developing telencephalon during the initial period of the second trimester (the appearance of microglia within the SP contrasted sharply with the lack of these cells within the germinal layers and intermediate zone). At later stages of development (14–16 gw onwards), there is probably additional supplementary infiltration of progenitors via blood vessels and proliferation within germinal layers. These findings are in keeping with studies carried out in lower vertebrates and mammals (Sorokin et al., 1992; Alliot et al., 1999; Herbomel et al., 2001; Kaur et al., 2001), which indicate that early ‘fetal macrophages’ first colonize the cephalic mesenchyme from whence they invade the brain, prior to and concomitant with the commencement of vascularization of the nervous system. Subsequently, these cells penetrate the CNS parenchyma in association with the invading vessels. The existence of a dual lineage for microglia (cells derived from fetal macrophages versus bone marrow progenitors) is a topic that is currently under debate (Provis et al., 1996, 1997; Rezaie et al., 1997; Andjelkovic et al., 1998; Alliot et al., 1999; Rezaie and Male, 1999, 2002a,b; Kaur et al., 2001).

In conclusion, we have now begun to understand how glial populations colonize the developing human telencephalon during the second trimester. This is a period of development that coincides with heightened structural reorganization of the CNS (Chan et al., 2002), and may be particularly prone to pathological injury. Our findings have direct bearing as a basis for future clinical diagnostic assessments of glial responses that could be directed towards pathological insults received by the fetus during mid-intrauterine life. The finding of morphologically and phenotypically distinct microglia within the telencephalon, and the coordinated manner in which microglia infiltrate the telencephalon during this period, necessitates further studies into the functional roles of these cells during development.

The authors would like to thank Dr U. Saretzki (Department of Pathology, Klinikum Schwerin, Germany), and Dr N. Cairns and M.N. Khan (Brain Bank, Institute of Psychiatry, London) for human materials. Sabine Cleven, Anke Sund, Jana Müller and Frank Neudörfer are gratefully acknowledged for expert technical support. This study was approved by the ethical committees of the University of Rostock (Ethikkommission der Ärtztekammer Mecklenburg-Vorpommern ref: IIHV 02/3002) and the Open University (Human Participants and Materials Ethical Committee ref: HPMEC/03/#41/1). This work was presented in preliminary form as a poster at the 20th Workshop of the German Anatomical Society, October 2003.

References

Alliot F, Godin I, Pessac B (
1999
) Microglia derive from progenitors originating from the yolk sac, and which proliferate in the brain.
Dev Brain Res
 
117
:
145
–152.
Andjelkovic AV, Nikolic B, Pachter JS, Zecevic N (
1998
) Macrophages/microglial cells in human central nervous system during development: an immunohistochemical study.
Brain Res
 
814
:
13
–25.
Bardadin KA, Scheuer Pj, Peczek A, Wejman J (
1991
) Immunocytochemical observations on macrophage populations in normal fetal and adult human liver.
J Pathol
 
164
:
253
–259.
Becher B, Antel JP (
1996
) Comparison of phenotypic and functional properties of immediately ex vivo and cultured human adult microglia.
Glia
 
18
:
1
–10.
Boulder Committee (
1970
) Embryonic vertebrate central nervous system: revised terminology.
Anat Rec
 
166
:
257
–261.
Chan WY, Kostovic I, Takashima S, Feldhaus C, Stoltenburg-Didinger G, Verney C, Yew D, Ulfig N (
2002
) Normal and abnormal development of the human cerebral cortex.
Neuroembryology
 
1
:
78
–90.
Choi BH (
1981
) Haematogenous cells in the central nervous system of developing human embryos and fetuses.
J Comp Neurol
 
196
:
683
–694.
Davis EJ, Foster D, Thomas WE (
1994
) Cellular forms and functions of brain microglia.
Brain Res Bull
 
34
:
73
–78.
deGroot CJA, Huppes W, Sminia T, Kraal G, Dijkstra CD (
1992
) Determination of the origin and nature of brain macrophages and microglial cells in mouse central nervous system using non-radioactive in situ hybridization and immunoperoxidase techniques.
Glia
 
6
:
301
–309.
Dick AD, Pell M, Brew BJ, Foulcher E, Sedgwick JD (
1997
) Direct ex vivo flow cytometric analysis of human microglial cell CD14 expression: examination of central nervous system biopsy specimens from HIV-seropositive patients with other neurological diseases.
AIDS
 
11
:
1699
–1708.
Edwards IJ, Wagner WD, Owens RT (
1990
) Macrophage secretory products selectively stimulate dermatan sulphate proteoglycan production in cultured arterial smooth muscle cells.
Am J Pathol
 
136
:
609
–621.
Engelhardt B, Risau W (
1995
) Development of the blood-brain barrier. In: New concepts of a blood–brain barrier (Greenwood J, Begley DJ, Segal MB, eds), pp. 11–31. New York: Plenum Press.
Ferrer I, Bennet E, Soriano ET, delRio T, Fonseca M (
1990
) Naturally occurring cell death in the cerebral cortex of the rat and removal of dead cells by transitory phagocytes.
Neuroscience
 
39
:
451
–458.
Fujimoto E, Miki A, Mizoguti H (
1989
) Histochemical study of the differentiation of microglial cells in the developing human cerebral hemispheres.
J Anat
 
166
:
253
–264.
Giulian D, Bartel S, Broker J, Li X, Kirckpatrick JB (
1995
) Cell surface morphology identifies microglia as a distinct class of mononuclear phagocytes.
J Neurosci
 
15
:
7712
–7726.
Graeber MB, Streit WJ, Buringer D, Sparks DL, Kreutzberg GW (
1992
) Ultrastructural localization of major histocompatibility complex (MHC) class II positive perivascular cells in histologically normal human brain.
J Neuropathol Exp Neurol
 
51
:
303
–311.
Hayashi K, Kadomatsu K, Muramatsu T (
2001
) Requirement of chondroitin sulphate/dermatan sulphate recognition in midkine-dependent migration of macrophages.
Glucoconj J
 
18
:
401
–406.
Herbomel P, Thisse B, Thisse C (
2001
) Zebrafish early macrophages colonize cephalic mesenchyme and developing brain, retina, and epidermis through an MCSF receptor-dependent invasive process.
Dev Biol
 
238
:
274
–288.
Higashi K, Naito M, Takeya M, Ando M, Araki S, Takahashi K (
1992
) Ontogenic development, differentiation and phenotypic expression of macrophages in fetal rat lungs.
J Leukoc Biol
 
51
:
444
–454.
Hutchins KD, Dickson DW, Rashbaum WK, Lyman WD (
1990
) Localization of morphologically distinct microglial populations in the developing human fetal brain: implications for ontogeny.
Dev Brain Res
 
55
:
95
–102.
Imamoto K, Leblond CP (
1978
) Radioautographic investigation of gliogenesis in the corpus callosum of young rats. II. Origin of microglial cells.
J Comp Neurol
 
180
:
139
–164.
Innocenti GM, Clarke S, Koppel H (
1983
) Transitory macrophages in the white matter of the developing visual cortex II. Development and relations with axonal pathways.
Brain Res
 
313
:
55
–66.
Jander S, Schroeter M, Fischer J, Stoll G (
2000
) Differential regulation of microglial keratan sulphate immunoreactivity by proinflammatory cytokines and colony stimulating factors.
Glia
 
30
:
401
–410
Jones LL, Tuszynski MH (
2002
) Spinal cord injury elicits expression of keratan sulphate proteoglycans by macrophages, reactive microglia and oligodendrocyte progenitors.
J Neurosci
 
22
:
4611
–4624.
Jones LL, Yamaguchi Y, Stallcup WB, Tuszynski MH (
2002
) NG2 is a major chondroitin sulphate proteoglycan produced after spinal cord injury, and is expressed by macrophages and oligodendrocyte progenitors.
J Neurosci
 
22
:
2792
–2803.
Kaur C, Ling EA (
1991
) A study of the transformation of amoeboid microglial cells into microglia labelled with the isolectin Griffonea simplicifolia in postnatal rats.
Acta Anat
 
142
:
118
–125.
Kaur C, Hau AJ, Wu CH, Ling EA (
2001
) Origin of microglia.
Microsc Res Tech
 
54
:
2
–9.
Kostovic I, Judas M (
2002
) The role of the subplate zone in the structural plasticity of the developing human cerebral cortex.
Neuroembryology
 
1
:
145
–153.
Lawson LJ, Perry VH, Dri P, Gordon S (
1990
) Heterogeneity in the distribution and morphology of microglia in the normal adult mouse brain.
Neuroscience
 
39
:
151
–170.
Ling EA, Tan CK (
1974
) Amoeboid microglial cells in the corpus callosum of neonatal rats.
Arch Histol Jpn
 
36
:
265
–280.
Mato M, Ookawara S, Sakamoto A, Aikawa E, Ogawa T, Mitsuhashi M, Masuzawa T, Suzuki H, Honda M, Yazaki Y, Watanabe L, Luoma J, Yla-Herttuala S, Fraser I, Gordon S, Kodama T (
1996
) Involvement of specific macrophage-lineage cells surrounding arterioles in barrier and scavenger function in brain cortex.
Proc Natl Acad Sci USA
 
93
:
3269
–3274.
Moon LD, Asher RA, Rhodes KE, Fawcett JW (
2002
) Relationship between sprouting axons, proteoglycan and glial cells following unilateral nigrostriatal axotomy in the adult rat.
Neuroscience
 
109
:
101
–117.
Naito M, Takahashi K, Nishikawa SI (
1990
) Development, differentiation and maturation of macrophages in fetal mouse liver.
J Leukoc Biol
 
48
:
27
–37.
Ogawa M, Araki M, Naito M, Takeya M, Takahashi K, Yoshida M (
1993
) Early changes of macrophage-like immunoreactivity in the rat inferior olive after intraperitoneal 3-acetylpyridine injection.
Brain Res
 
610
:
135
–140.
Penfold PL, Madigan MC, Provis JM (
1991
) Antibodies to human leukocyte antigens indicate subpopulations of microglia in human retina.
Vis Neurosci
 
7
:
383
–388.
Perry VH (
1996
) Microglia in the developing and mature central nervous system. In: Glial cell development: basic principles and clinical relevance (Jessen KR, Richardson WD, eds), pp. 123–140. London: Bios Scientific.
Perry VH, Hume DA, Gordon S (
1985
) Immunohistochemical localization of macrophages and microglia in the adult and developing mouse brain.
Neuroscience
 
15
:
313
–326.
Peterson PK, Gekker G, Hu S, Sheng WS, Anderson WR, Ulevitch RJ, Tobias PS, Gustafson KV, Molitor TW, Chao CC (
1995
) CD14 receptor-mediated uptake of monopsonised mycobacterium tuberculosis by human microglia.
Infect Immun
 
63
:
1598
–1602.
Provis JM, Diaz CM, Penfold PL (
1996
) Microglia in human retina: a heterogeneous population with distinct ontogenesis.
Perspect Dev Neurobiol
 
3
:
213
–222.
Provis JM, Leech J, Diaz CM, Penfold PL, Stone J, Keshet E (
1997
) Development of the human retinal vasculature: cellular relations and VEGF expression.
Exp Eye Res
 
65
:
555
–568.
Rakic S, Zecevic N (
2000
) Programmed cell death in the developing human telencephalon.
Eur J Neurosci
 
12
:
2721
–2734.
Rezaie P (
2003
) Microglia in the human nervous system during development.
Neuroembryology
 
2
:
18
–31.
Rezaie P, Male D (
1999
) Colonization of the developing human brain and spinal cord by microglia- a review.
Microsc Res Tech
 
45
:
359
–382.
Rezaie P, Male D (
2002
a) Differentiation, ramification and distribution of microglia within the central nervous system examined.
Neuroembryology
 
1
:
29
–43.
Rezaie P, Male D (
2002
b) Mesoglia and microglia: a historical review of the concept of mononuclear phagocytes within the central nervous system.
J Hist Neurosci
 
11
:
325
–374.
Rezaie P, Male D (
2003
) Microglia in fetal and adult human brain can be distinguished from other mononuclear phagocytes through their lack of CD163 expression.
Neuroembryology
 
2
:
130
–133.
Rezaie P, Cairns NJ, Male DK (
1997
) Expression of adhesion molecules on human fetal cerebral vessels: relationship to microglial colonization during development.
Dev Brain Res
 
104
:
175
–189.
Rezaie P, Patel K, Male DK (
1999
) Microglia in the human fetal spinal cord — patterns of distribution, morphology and phenotype.
Dev Brain Res
 
115
:
71
–81.
Rezaie P, Ulfig N, Male D (
2003
) Distribution and morphology of GFAP-positive astrocytes in the human fetal brain at second trimester.
Neuroembryology
 
2
:
50
–63.
Sasaki A, Hirato J, Nakazato Y, Ishida Y (
1988
) Immunohistochemical study of the early human fetal brain.
Acta Neuropathol
 
76
:
128
–134.
Saunders NR (
1992
) Ontogenic development of blood-brain barrier mechanisms. In: Physiology and pharmacology of the blood-brain barrier (Bradbury MW, ed.), pp. 327–371. New York: Springer-Verlag.
Sedgwick JD, Schwender S, Imrich H, Dorries R, Butcher GW, terMeulen V (
1991
) Isolation and direct characterization of resident microglial cells from the normal and inflamed nervous system.
Proc Natl Acad Sci USA
 
88
:
7438
–7442.
Sorokin SP, Hoyt RF Jr, Blunt DG, McNelly NAF (
1992
) Macrophage development II. Early ontogeny of macrophage populations in brain, liver, and lungs of rat embryos as revealed by a lectin marker.
Anat Rec
 
232
:
527
–550.
Takeya M, Takahashi K (
1992
) Ontogenic development of macrophage subpopulations and Ia-positive dendritic cells in fetal and neonatal rat spleen.
J Leukoc Biol
 
52
:
516
–523.
Ulfig N, Nickel J, Bohl J (
1998
) Transient features of the thalamic reticular nucleus in the human fetal brain.
Eur J Neurosci
 
10
:
3773
–3784.
Ulfig N, Neudorfer F, Bohl J (
2000
) Transient structures of the human fetal brain: subplate, thalamic reticular complex, ganglionic eminence.
Histol Histopathol
 
15
:
771
–790.
Ume DA, Perry VH, Gordon S (
1983
) Immunohistochemical localization of a macrophage-specific antigen in mouse retina: phagocytosis of dying neurons and differentiation of microglial cells to form a regular array in the plexiform layers.
J Cell Biol
 
97
:
253
–257.
Wang CC, Wu CH, Shieh JH, Wen CY (
2002
) Microglial distribution and apoptosis in fetal rat brain.
Dev Brain Res
 
139
:
337
–342.
Wijffels JF, De Rover S, Beelen RH, Kraal G, van Rooijen N (
1994
) Macrophage subpopulations in the mouse spleen renewed by local proliferation.
Immunobiol
 
191
:
52
–64.
Zecevic N, Bourgeois JP, Rakic P (
1989
) Changes in synaptic density in motor cortex of rhesus monkey during fetal and postnatal life.
Dev Brain Res
 
50
:
11
–32.
Zecevic N, Andjelkovic A, Matthieu JM, Tosic M (
1998
) Myelin basic protein immunoreactivity in the human embryonic CNS.
Dev Brain Res
 
105
:
97
–98.

Author notes

1Department of Biological Sciences, Faculty of Science, The Open University, Walton Hall, Milton Keynes, MK7 6AA, UK, 2Department of Neuropathology, Institute of Psychiatry, King's College London, DeCrespigny Park, London SE5 8AF, UK, 3Department of Histopathology, Addenbrooke's Hospital, Hills Road, University of Cambridge, Cambridge CB2 2QQ, UK and 4Neuroembryonic Research Laboratory, Institute of Anatomy, University of Rostock, Gertrudenstrasse 9, D-18055 Rostock, Germany