Morphogenesis of Lateral Choroid Plexus During Human Embryonic Period

Authors


Correspondence to: Dr. Tetsuya Takakuwa, Human Health Science, Graduate School of Medicine, Kyoto University, 606-8507 Sakyo-ku Shogoin Kawahara-cyo 53, Kyoto, Japan. E-mail: tez@hs.med.kyoto-u.ac.jp

Abstract

The morphological and histological changes of the choroid plexus (CP) during Carnegie stage (CS) 18 and CS23 were presented, based on magnetic resonance imaging data and histological serial section of human embryos from the Kyoto Collection of Human Embryos. The primordium of the CP was initially detected as a small lump at CS19 that grew caudally, so that the CP became crescent shaped. It developed in all directions after CS21, as the dorsal and frontal growth also became prominent. The CP formed a number of undulating surfaces at CS20, irregular bulges at CS21, and then three large clusters with two deep fissures on the caudal surface at CS23. The mean volume of the CP was 0.282±0.141 mm3 at CS19; it reached 16.8±8.77 mm3 at CS23. Additionally, the histology was different depending on the regions of the CP at all stages after CS20. The epithelium and angioblasts in the center of the stroma were proliferated in the proximal region, whereas the epithelium was differentiated and lobulated in the distal region where the blood vascular system was organized. The histological differentiation was mapped on the CP reconstructed from histological serial sections. The data suggested the correlation between morphological information obtained from magnetic resonance data sets and distribution of the differentiation. With the help of morphological analysis and histological findings, we have been able to categorize each CP into specific stages. These findings will be useful in clinical evaluation of development during the embryonic period. Anat Rec, 296:692–700, 2013. © 2013 Wiley Periodicals, Inc.

Abbreviations used
CP

choroid plexus

CS

Carnegie stage

DV

diencephalic ventricle

JB

junctional base

LV

lateral ventricle

MR

magnetic resonance

3D

three-dimensional.

The human brain is arguably one of the most complicated organs in living systems. It develops after the dynamic and gross changes during the embryonic period (Moore et al., 2008). Originating from a simple neural tube, this elaborate structure is formed by a series of differentiation processes (Bayer and Altman, 2002; Huang et al., 2009). Neural tube closure in the human embryo initiates at multiple sites in Carnegie stage (CS)10 (22 days after fertilization) and completes at CS12 (26 days after fertilization) (Nakatsu et al., 2000). The cerebral vesicle, formed at the cranial end of the neural tube, differentiates and forms the prosencephalic ventricle, the mesencephalic ventricle, and the rhombencephalic ventricle at CS13. The prosencephalic ventricle is divided into diencephalic ventricle (DV) and telencephalic ventricle at CS14, which become the lateral ventricle (LV).

The choroid plexus (CP) is a special organ that produces cerebrospinal fluid. It has a unique structure, differentiated from ependymal cells, and is responsible for an important biological barrier system, forming an interface between the blood and the cerebrospinal fluid. The CP is distributed in four parts, one in each of the LVs, one in the DV, and one in the rhombencephalic ventricle. Precise studies about the development of the CP have been published for a number of species: rabbit (particularly vascularization of the CP) (Strong, 1956, 1964a, 1964b), sheep (Jacobsen et al., 1983), mouse (Knudsen, 1964; Sturrock, 1979), and rat (Wislocki and Ladman, 1958; Cancilla et al., 1966). In humans, primordium of the CP forms on the roof of the telencephalon medium (choroid plate) as a fold at CS17 and CS18 (Müller and O'Rahilly, 1990; O'Rahilly and Müller, 1990). The primordium develops to CP of the bilateral LVs and DV accompanies with the differentiation of those ventricles. Ariens Kappers (1958) has categorized the development of the lateral ventricular CP between 6 weeks and 5.5 months of gestation into three histogenetic phases, according to the histological findings such as epithelium type and the presence of cytoplasmic glycogen. Netsky and Shuangshoti, (1975) also observed the CP of the fetus and the neonate at 1 month after birth, then extended this to four stages (Table 1). Though their study was excellent, the information about embryonic period was insufficient. According to their categorization, only the first and a part of the second stage may include the period during CS19 and CS23 although the cerebral vesicles develop dramatically in those periods (O'Rahilly and Müller, 2006).

Table 1. Stage of differentiation of human lateral ventricular choroid plexus
 Stage IStage II
Time of developmentSeventh week of gestationNineth week of gestation
EpitheliumPseudostratified tall; Predominantly central nucleiLow columnar; apical nuclei
GlycogenAbsent?Abundant
VilliAbsent; lobules presentSparse primary villi
TubulesMay be presentSeveral
StromaLoose mesenchymeExtremely loose mesenchyme, small amount of connective tissue fibers
Stromal blood vesselsIslets of nucleated blood cells surrounded by ill-defined vascular walls, blood islets located subepitheliallyDefinite vascular walls, capillaries located subepithelially, large blood vessels in central interstitium
Size of plexus in relation to ventricleMinuteExtremely large

In this study, the morphogenesis of the CP of the LV region between CS19 and CS23 were precisely observed applying two approaches—three-dimensional (3D) reconstruction using magnetic resonance (MR) imaging data and histological observation. The data provided demonstrate the dramatic growth of CP during the embryonic period in each CS.

MATERIALS AND METHODS

Human Embryo Specimens

Approximately, 44,000 human embryos comprising the Kyoto Collection are collected and stored at the Congenital Anomaly Research Center of Kyoto University (Nishimura et al., 1968; Shiota, 1991; Yamada et al., 2004). In most cases, pregnancy was terminated during the first trimester for socioeconomic reasons, under the Maternity Protection Law of Japan. Some of the specimens (∼20%) were undamaged, well-preserved embryos. When the aborted embryos were brought to the laboratory, they were measured, examined, and staged using the criteria provided by O'Rahilly and Müller (1987). Approximately, 500 well-preserved human embryos, diagnosed as externally normal from CS13 to CS23, were subjected to histological serial sections. Another set of 1,200 well-preserved human embryos were selected for MR microscopic imaging. The conditions used to acquire the serial sections and the MR images of the embryos are described elsewhere (Nishimura, 1975; Matsuda et al., 2003; Yamada et al., 2006; Shiota et al., 2007).

For this study, 49 samples distributed between CS19 and CS23, consisting of 10 samples for each stage (except CS22 for which there were nine samples) were selected from the 1,200 MR image data sets, based on the criteria described previously (Nakashima et al., 2011) for 3D reconstruction and morphometric analysis. Forty-one samples distributed between CS18 and CS23, consisting of seven samples for CS18 and CS22, six samples for CS19, five samples for CS20 and CS23, and 11 samples for CS21, were selected for histological analysis, based on the following criteria: no obvious damage or significant anomaly is present in the cerebral vesicle, ventricle, abdominal organs, and external appearance.

3D Reconstruction and Morphometric Analysis

3D MR image data sets for each embryo were initially obtained as 256 × 256 × 512 voxel data (Matsuda et al., 2007). Sequential 2D images were resectioned digitally by using ImageJ64™ (ver. 1.44, National Institutes of Health, Bethesda, Maryland) and saved as Analyze file formats (.hdr, .img). The LV and the CP were segmented for 3D reconstruction using FSL View of FMRIB Software Library™ (ver. 4.1.9, Analysis Group, FMRIB, Oxford, United Kingdom). Histological serial sections for 3D reconstruction were digitized using a film scanner (DiMAGE Scan Multi PRO AF-5000, Konica Minolta, Tokyo, Japan). 3D objects of CP were computationally reconstructed and morphometrically observed in software Amira™ (ver. 5.4.0, Visage Imaging, Berlin, Germany). Volume of the CP was calculated using the software OsiriX™ (ver. 4.0, Pixmeo SARL, Geneva, Switzerland).

Histological Observations

Histological serial sections stained with Hematoxylin and Eosin were observed microscopically by two of the authors (Naoki Shiraishi and Tetsuya Takakuwa). The histological findings of the CP were evaluated according to the previous study (Shuangshoti and Netsky, 1966) (Table 1), that is 1) type of epithelium and the presence of glycogen in the cytoplasm, 2) position and mitosis of the nucleus in the epithelium, 3) connectivity and lobulation of the stroma, 4) the presence of angioblasts, and 5) the presence of the capillaries. The degree of differentiation was categorized into three stages using the histological findings, that is Dif: histologically differentiated (simple columnar epithelium, apical nuclei, and vacuoles in the cytoplasm), Undif: histologically undifferentiated (pseudostratified epithelium, frequent mitosis, and many angioblasts in the stroma), and Mix (the histological findings of both Dif and Undif mixed in various combinations).

RESULTS

Morphogenesis of the Reconstructed CP

The CP was arising from areas close to the interventricular foramen and close to the medial eminence depression by the amygdala in CS18. The primordium of CP was detected by histology, as described later, but not by 3D reconstruction (Fig. 1). The primordium formed a small lump at CS19, and then developed in the LV and DV. The left and the right CP developed as mirror images, implying no noticeable morphological differences between the two. The growth of the CP followed that of the LV. The LV formed a C-shape between CS18 and CS23. The inferior horn grew like the “horn of a ram” (O'Rahilly et al., 1987; Drews, 1995), constructing an arch directed backward. A similar development was also observed for the growth toward the caudal direction between CS18 and CS21 (Figs. 1, 2). The CP developed laterally and caudally at CS20. It was thin and slightly curved (crescent shaped). It grew caudally toward the inferior horn of the LV and developed in all directions after CS21. Growth in the dorsal direction was prominent after CS22. Further, the CP rapidly grew frontally and filled one-fourth of the LV at CS23 (Supporting Information video S1).

Figure 1.

Representative 3D image of the CP (red) within the LV (yellow). (A) Frontal, lateral, and dorsal view between CS18 and CS23 are shown. Asterisk (*) in the dorsal view indicates the JB. (B) Orientation used in this article: The red, green, and blue arrows indicate lateral, dorsal, and caudal views, respectively.

Figure 2.

Direction of growth of the CP between CS19 and CS23. For comparison, CP between CS19 and CS23 was merged at the JB, with the same magnification. Dorsal and lateral views are shown. Asterisk (*) in the dorsal view indicates the JB. The viewpoint was indicated in the illustration of the ventricles.

The surface of the CP changed according to the development and differentiation. The CP formed a number of undulating surfaces at CS20, and then formed irregular bulges on the caudal–dorsal surface at CS21. The bulges became more complicated and deep in CS22 (than in CS21); they were more on the caudal surface, whereas a few were on the frontal surface. Several large clusters on the caudal surface were observed in place of the bulges of the surface at CS23 (Fig. 3A). Three large clusters with two deep fissures were observed in 7 out of 10 cases (Fig. 3B). These clusters were distributed in a medial to lateral order.

Figure 3.

Representative gross 3D images of the CP. (A) The feature of the dorsal surface between CS20 and CS23. The viewpoint was indicated in the illustration of the ventricles. (B) Dorsocaudal view of right lateral CP in CS23. Three large clusters (I, II, and III) with two deep fissures (red dot) are indicated in schema. The similarity between these clusters and the microvascular architecture in the fetal period has been discussed in Discussion section. Asterisk (*) in the dorsal view indicates the JB.

The CP facing in DV was too small to be recognized in the MR imaging data at all stages analyzed in this study.

Volume of CP and LV

The mean volume of the CP was 0.282±0.141 mm3 (mean±SD) at CS19. It exponentially increased until CS23 when it reached 16.8±8.77 mm3 (Fig. 4). The CP increased in volume by approximately 60-fold from CS19 to CS23. The mean volume of the LV was 6.10±1.43 mm3 at CS19. It exponentially increased until CS23 when it reached 74.3±28.0 mm3. The volume ratio of CP to LV increased from 4.84±2.52% to 21.5±4.13% between CS19 and CS23. That is, the CP was increasing in volume relatively faster than the LV. The intrastage variation of the CP to LV ratio was limited. There was no significant difference in volume between the right and the left CP (data not shown). The present data were consistent with the morphological observations described above.

Figure 4.

Calculated volumes of LV and CP between CS18 and CS23. The volume was calculated as described in Materials and Methods section. Volume at each CS is shown as mean±standard deviation (mm3).

Histology of the CP

In all cases, the primordium of the CP of the LV region appeared at CS18 (Fig. 5A-a and Table 2). The border, as evident by histology, was defined as “junctional base (JB)” and used as the anatomical reference in this study. The CP was covered with pseudostratified tall epithelium. The epithelium was thinner than the neuroepithelium (Fig. 5A-b,c). The nuclei of the epithelium were large and centrally located in almost all areas. The frequency of nuclear mitosis in the epithelial cells was comparable with that of the neuroepithelium. The stroma was loose and many nucleated red-stained cells were found at the subepithelial regions; these cells were defined as angioblasts (Ellen, 1922).

Figure 5.

Histology of the CP in LV stained with Hematoxylin and Eosin. Each magnification is indicated in parentheses. (A) Coronal section of the CP at CS18 (No. 1633). (a) The primordium of CP form on the roof of the telencephalon medium (choroid plate) (squared) (×40) (b) (×200). (c) The junction between CP and ventricular wall. Thickness of the pseudostratified epithelium differs between them. The junction was defined as JB in this study for the reference point (×400). (B) Coronal section of the CP at CS19 (No. 4424). (a) The CP forming a “Club” shape. (×100) Arrowhead (▴) indicates the paraphysis. (b) The epithelium showing nonuniform thickening (×400). (c) Angioblasts with endothelial cells (arrows) (×400). (C) Coronal section of the CP at CS20 (No. 1770). (a) The CP forming “fist-shape” (arrow). Note that the roof of the telencephalon grows and divides between LVs (L) and DVs (D). The epithelium facing the DV is tall, indicating ventricular wall (neuroepithelium) but not CP (×40). Arrowhead (▴) indicates the paraphysis. (b) The epithelium in the CP of LV, showing interlobular clefts (arrows) (×200). (c) Angioblasts with vascular wall in the stroma (arrows) (×400). (D) Coronal section of the CP at CS21 (No. 2314). The specimen belongs to the late half of CS21. (a) More complex lobulation of the CP and appearance of villous structure (primary villi) (×20). (b) The region close to the junction with ventricle wall (square in (D-a)) (×200). (c) The region distant from the junction (square of doublet in (D-a)). Villous structure was observed (×200). (E) Sagittal section of the CP at CS23 (No. 4381). (a) Gross view (×20) (b) The region close to the JB (square in (E-a)). Tubular and villous structure was observed (×200). (c) The region distant from JB (square of doublet in (E-a)) (×200).

Table 2. Acquision of histological feature of the CP from CS 18–23
StructureStateChangeCS
18 19 20 21 22 23 
  1. a

    Percentage is given in parentheses.

EpitheliumTypeOnly pseudostratified7(100)a6(100)4(80.0)5(45.5)0 0 
Pseudostratified+low columnar0 0 1(20.0)6(54.5)6(85.7)1(20.0)
Completely low columnar but stalk0 0 0 0 1(14.3)4(80.0)
GlycogenNo presence7(100)6(100)5(100)10(90.9)3(42.9)0 
Presence0 0 0 1(9.09)4(57.1)5(100)
NucreiPositionPredominantly central7(100)6(100)4(80.0)3(27.3)0 0 
Central and apical mixed0 0 1(20.0)8(72.7)5(71.4)1(20.0)
Predominantly apical0 0 0 0 2(28.6)4(80.0)
MitosisPresence7(100)6(100)5(100)9(81.8)6(85.7)0 
No presence0 0 0 2(18.2)1(14.3)5(100)
StromaConnectivityLoose7(100)6(100)5(100)11(100)1(14.3)0 
Extremely loose0 0 0 0 6(85.7)5(100)
LobulationNo lobulation7(100)6(100)2(40.0)1(9.09)0 0 
Lobulation0 0 3(60.0)2(18.2)0 0 
Lobulation+tubules0 0 0 6(54.5)1(14.3)0 
Lobulation+tubules+villi0 0 0 2(18.2)6(85.7)5(100)
Blood vesselsAngioblastNumerous7(100)6(100)5(100)6(54.5)1(14.3)0 
A few0 0 0 5(45.5)6(85.7)5(100)
Capillaries located subepitheliallyNo presence7(100)6(100)5(100)1(9.09)1(14.3)0 
Presence0 0 0 10(90.9)6(85.7)5(100)
   7 6 5 11 7 5 

The primordium became club shaped in four out of the six cases (66.7%) at CS19. The epithelium was not uniform in thickness, that is the basal membrane (boundary between epithelium and stroma) was irregular (Fig. 5B-a). The pseudostratified epithelium intruded into the stroma in spots. In those spots, the epithelium became thinner as the rows of pseudostratified cells reduced in number (Fig. 5B-b). A mitotic figure was observed at the luminal side of the epithelium, at the same degree as in the neuroepithelium. Numerous angioblasts were observed in the central region of the stroma, and endothelial cells were occasionally observed around them (Fig. 5B-c).

The fist-shaped CP protruded toward LV in all cases at CS20 (Fig. 5C-a). The irregularity of epithelium was conspicuous in four out of the five cases (80.0%). Interlobular clefts were observed in three out of the five cases (60.0%) (Fig. 5C-b). The CP was mostly covered with pseudostratified tall cells, but a few tall columnar epitheliums with apically elevated nuclei were present at the distant region from the JB. Mitotic nuclei were present at the apical side of the epithelial cells as much as at the neuroepithelium. Numerous angioblasts were located subepithelially, indicating the formation of the capillary and vein (Fig. 5C-c).

The lobulation of the CP proceeded in CS21 (Fig. 5D-a); some tubules were observed in 8 out of the 11 cases (72.7%) and primary villi appeared in 2 out of the 11 cases (18.2%). The tall columnar epithelium covered an apparently much wider area at CS21 than at CS20. In the epithelium, the nuclei were distributed at the apical border in 8 out of the 11 cases (72.7%). The pseudostratified epithelium had a few lobulations and frequent mitosis, and the stroma had many angioblasts in the region close to the JB (Fig. 5D-b), whereas the tall columnar epithelium in distal region had almost apically elevated nuclei, few mitosis, many lobulations, and the stroma had a few angioblasts and frequent capillary walls (Fig. 5D-c). These findings suggest that the proximal region of CP was undifferentiated and proliferated, whereas the distal region was differentiated at CS21.

Pseudostratified epithelium and numerous angioblasts were still observed in the proximal region at CS22, whereas the other epithelium formed a single layer of cylindrical, ciliated cells with apically lined oval nuclei. Vacuoles were observed in the basal side of the cytoplasm within the epithelial cells in four out of the seven cases (57.1%). These epithelial cells may have cytoplasmic glycogen (Shuangshoti and Netsky, 1966). The CP lobulated in a complex manner and many tubules and primary villi were observed in six out of the seven cases (85.7%). In the distal region, the stroma was extremely loose, angioblasts were few, capillaries were formed subepithelially, and veins were scattered in the central interstitium (data not shown).

The CP expanded with almost no lobulation in the distal region in CS23 (Fig. 5E-a). Almost all the CPs were covered with a single layer of low columnar epithelium, except the region in the stalk or the tips of the interlobular clefts (Fig. 5E-b,c). Nuclei were located apically and mitosis was rarely observed in the single layer of low columnar epithelium. The stroma was extremely loose and angioblasts were rarely seen.

Histological Differentiation Map on Reconstructed CP

Histological observation suggested that the histological feature in the proximal region was different from that in the distal regions of the CP after CS20. In this context, the histological differentiation was then mapped on the CP reconstructed from histological serial sections.

The 3D reconstructed image from histological serial sections at CS21 revealed features similar to that obtained from MR data sets (Fig. 6, Supporting Information videos S2,S3); it revealed a crescent shape with thickness in the dorsoventral direction, and an irregular bulge formed at the caudal–dorsal region.

Figure 6.

3D histological reconstitution map of left CP at CS21. 3D image of CP at CS21 was reconstructed using histological serial sections (No. 2314), and mapped the histological differentiation; red: undifferentiated, blue: differentiated, green: mixed. The definition of the differentiation was described in Materials and Methods section. Yellow: the junction between CP and ventricle wall. The reconstructed CP from MR sets were shown for comparison. Asterisk (*) indicates the JB.

The histological differentiation map revealed the Undif located at the frontal–ventral region, close to the JB (Fig. 6); the Dif located at the frontal–dorsal region, which is distant from the JB; and the Mix located at the caudal region. The ratio of the Undif and Dif to the total CP in volume was 27.0% and 28.7%, respectively. The junction between the CP and the medial wall of LV become broad, and stretched from the JB (asterisk, Fig. 6) to the caudal regions.

A 3D image at CS23 was also reconstructed from histological serial sections (Supporting Information video S4). Undulation and large clusters were observed in the caudal surface of the histological reconstructed CP at CS23; these were also detected on the 3D image from MR imaging data (Fig. 3, Supporting Information video S1). Dif, located at the distal region of CP, occupied approximately 81.2% of the CP by volume. Mix occupied the remaining proximal region. Undif appeared rudimentary at the proximal region of CP. The junction between the CP and the medial wall of the LV continued from the JB (asterisk, Fig. 6) to the caudal regions, which is similar to that at CS21.

DISCUSSION

The morphogenesis of CP in the human embryonic period have not been fully described so far. Shuangshoti and Netsky (1966) described histological findings in 18 cases of fetus including several cases of embryo with crown-rump length of 13–30 mm. O'Rahilly and Müller (1990) described the morphogenesis of cerebral ventricles between CS11 and CS23. The histological feature of CP was described in that study as a specific region of the cerebral ventricle between CS18 and CS23. The emergence of the primordium of CP was precisely described. However, morphological and histological data corresponding to each CS remain questionable. In this study, the morphogenesis of the lateral ventricular CP between CS18 and CS23 using 3D reconstruction and precise histological observation are described (Fig. 7). The 3D reconstruction model of the CP revealed unique and discernible morphology at each CS.

Figure 7.

Summary of histological findings of CP between CS18 and CS23. CP/LV ratio; ratio of volume of CP to volume of LV. Standard deviation (SD) of them is in parentheses. Note that intrastage variation of the ratio was limited except at CS23.

As previously described, the primordium of CP was observed only histologically at CS18. The CP mainly grew caudally along the inferior horn of the LV between CS18 and CS21; it grew in all directions after that stage. The direction of growth does not seem to be restricted to the shape of the ventricle because of leaving enough room within it during the embryonic period. The CP filled approximately one-fourth of the LV at CS23. The speed of growth is maintained until the CP filled approximately three-fourths of the LV at 50 mm crown-rump length (around 11 weeks of gestation) (Shuangshoti and Netsky, 1966).

The surface of the CP was also unique at each stage. The CP had a number of undulating surfaces at CS20, irregular bulges at the caudal–dorsal surface at CS21, and numerous deep bulges on the caudal surface at CS22. They formed a large cluster at CS23. The distribution of the clusters observed in 3D reconstruction at CS23 was similar to that of microvascular architecture found in human fetuses at 20 weeks of gestation as previously reported (Zagorska-Swiezy et al., 2008). In that study, the CP was divided into five clusters based on the microvascular pattern and density—the anterior part, the glomus, the posterior part, the villous fringe, and the free margin. Three large clusters observed at the caudal surface in reconstructed CP at CS23 may correspond to the villous fringe, glomus, and posterior part because of their location and morphological features. Analysis of the development of microvascular architecture during embryonic period will be necessary to decide whether the clusters observed in this study are identical to those structures or not.

Each histological finding of CP was similar to those described in the previous studies (Netsky and Shuangshoti, 1975; Müller and O'Rahilly, 1990). However, the histology may be different depending on the regions of the CP. The epithelium and angioblasts in the center of the stroma were proliferated in the proximal region, whereas the epithelium was differentiated and lobulated, and had abundant cytoplasmic glycogen in the distal region where the blood vascular system was organized, that is capillaries were formed subepithelially and small veins were scattered in the central interstitium. A mitotic figure was rarely found. A similar tendency was observed at all stages after CS20. These data suggest that the CP proliferated mainly at the proximal regions and the differentiating components may have been pushed out to the peripheral region one after another.

3D reconstruction from histological serial images was used in this study because that approach has a definite advantage. In this approach, histological information can be mapped on the 3D image. 3D reconstruction from MR data sets is accurate in morphology (Nishimura, 1975; Matsuda et al., 2003; Yamada et al., 2006; Shiota et al., 2007). Comparison of the construction with the images obtained from MR reconstruction suggests that the histologically undifferentiated region may correspond to the area with smooth surface (mainly located at the frontal–medial part of CP) in MR reconstruction, whereas the histologically differentiated and mixed regions may correspond to the area with a lot of bulges, undulation, and/or clusters (mainly located caudal–lateral part of CP) in MR reconstruction. It is notable that the reconstructed CP from the two different approaches with different specimens was similar to each other. As the comparison in this study was limited in number and stages, further analysis may be necessary.

The JB defined in this study was valuable for anatomical reference. The JB arose as the border between the telencephalic ventricle wall and the primordium of the CP at LV region at CS18. The junction was very limited until CS20. With the development of the CP of LV regions, the junction between the CP and the wall of the LV become broad, elongating from the JB in the caudal direction after CS21. That junction may be necessary for abundant blood supply to the CP (Millen and Woollam, 1953).

The CP of the DV and paraphysis may arise around the JB because the CP of LV and DV are continuous in the mammalian brain (Netsky and Shuangshoti, 1975; Dziegielewska et al., 2001). O'Rahilly and Müller (1990) observed it at CS18 for the first time. In this study, the primordium faces both LV and DV at CS 20. Paraphysis was detected at least after CS19 at the roof of the DV and the epithelium became thin and irregular around that, whereas the stroma remained undifferentiated. This study did not determine whether structures around the paraphysis were CP of the DV or part of the paraphysis.

The CP is a special organ with a unique morphology. It produces cerebrospinal fluid. Neoplasia (papilloma) of the CP causes serious diseases like hydrocephalus (Eisenberg et al., 1974). This study reveals that the CP has a characteristic differentiation and morphogenesis phase between CS18 and CS23 (Fig. 7). With the help of morphological analysis and histological findings, we have been able to categorize each CP into specific stages. These findings will be useful in clinical evaluation of development during the embryonic period.

ACKNOWLEDGEMENTS

The authors are deeply indebted to Executive Vice President of Kyoto University, Kohei Shiota, for providing the invaluable MR data and to Ms. Saki Ueno for technical assistance.

Ancillary