The medulla oblongata is involved in many different and complex functions as it contains nuclei involved in cardiac and respiratory control, nervous centers providing sensory and motor innervation of peripheral structures, groups of neurons receiving inputs related to the special senses of hearing, vestibular function and taste, or providing/receiving efferent/afferent cerebellar projections. There have been a few studies in the literature detailing morphometric parameters, such as neuronal density, nuclear volume, and total neuron number, in the medullary nuclei considered in this study (e.g., Nara et al.,1989,1991; Konrat et al.,1992; O'Kusky and Norman,1992,1995; Huang et al.,1993; Lamont et al.,1995; Diaz et al.,1996; Pamphlett and Treloar,1996; Lopez et al.,1997; Suarez et al.,1997; Alvarez et al.,1998,2000; Tang et al.,2001/2002; Kinney et al.,2002; Ma et al.,2005,2006; Lasn et al.,2006). Most studies showed intrinsic bias because of counting in a two-dimensional plane and applying of correction factors to estimate neuron numbers in three-dimensional space (Abercrombie,1946). On the contrary, unbiased stereological methods using a three-dimensional probe and not relying on any assumptions about size and shape of the objects have been developed. Moreover, all previous studies limited the morphometric analysis to one or two nuclei, without providing the possibility to compare morphometric data of different medullary nuclei, and few data are available about comparisons between infants and adults. The aim of this work was to provide a morphometric analysis, through the method of the optical disector, of a wide series of medullary nuclei in both adult and infant casistics.
In the literature, comprehensive and comparative morphometric studies of infant and adult medullary nuclei performed with unbiased methods are still lacking. In this study, the unbiased quantitative method of the optical disector was applied to analyze neuronal densities, nuclear volumes, and total neuron numbers of the hypoglossal nucleus (XII), dorsal motor nucleus of the vagus (DMNV), nucleus tractus solitarii (NTS), medial vestibular nucleus (MedVe), cuneate nucleus (Cu), nucleus of the spinal trigeminal tract, principal inferior olivary nucleus (PION), medial inferior olivary nucleus (MION), and dorsal inferior olivary nucleus (DION) in adults (16 male, six female; mean age: 37 years) and infants (five male, five female; mean age: 5 months). In both infants and adults, higher neuronal densities were found in the more ventrally located nuclei of the spinal trigeminal tract (mean values (coefficient of variation): 20,947 (0.29) and 8,990 (0.18) neurons/mm3, respectively) and inferior olivary complex (PION: 20,010 (0.15) and 9,076 (0.10); MION: 18,667 (0.20) and 9,989 (0.13); DION: 22,424 (0.17) and 10,986 (0.20), respectively) than in the nuclei of the medullary tegmentum, that is, XII (2,747 (0.39) and 1,026 (0.31)), DMNV (2,876 (0.19) and 1,553 (0.26)), NTS (7,993 (0.17) and 2,877 (0.13)), MedVe (7,010 (0.17) and 2,918 (0.12)), and Cu (2,563 (0.23) and 1,038 (0.16)). All the medullary nuclei showed higher volumes and lower neuronal densities in adults than in infants, without statistically significant differences in total neuron numbers, probably because of postnatal development of the neuropil and microvascularization. Anat Rec, 2009. © 2009 Wiley-Liss, Inc.
MATERIALS AND METHODS
This study was performed on 32 brainstems sampled during autopsy from 22 adults and 10 infants. Table 1 shows demographic data and causes of death of all the subjects. The study was approved by the local Ethical Committee and was performed according to the Italian laws on autopsied human tissues. Autopsies were performed within 36 hr of death. In all cases, macroscopic and microscopic examination revealed the absence of acute, chronic, localized, or diffuse brain pathology. Brainstems were fixed in 10% formalin for 7 days. They were cut into 5-mm slices perpendicular to the brainstem axis. Each tissue block was dehydrated in a series of alcohol solutions of 50%, 70%, 80%, and 95% (12 hr in each) and in 99% alcohol, xylene, and regular paraffin solutions (24 hr in each). The paraffin-embedded blocks were cut serially and exhaustively into 10-μm-thick transverse sections at a calibrated microtome. Final section thicknesses were measured with microcator by focusing from the top to the bottom surface. The final mean section thickness corresponded to 10 μm. Systematic and uniformly random sampling of the medullary nuclei in their full rostral-caudal extent was performed. In each case, every 80th or 160th section was stained with haematoxylin-eosin and examined for morphometric analysis, with a random start (selected by lottery) within the first most caudal 80 or 160 sections of the medulla. In particular, for the nucleus of the spinal trigeminal tract (NSTT) (the only nucleus sampled by every 160th section), the first or second sections in the series used for other nuclei (sampled by every 80th section) were randomly chosen as starting sections. Adjacent sections were also stained for Nissl substance using 0.1% cresyl violet solution and with Klüver-Barrera and azan-Mallory.
|Case||Gender||Age (adults: years) (infants: months)||Death-autopsy interval (hr)||Cause of death|
|Mean (CV)||37 (0.3)||29 (0.07)|
|Mean (CV)||5 (0.6)||30 (0.07)|
Examination of the sections was performed under a Leica DM4500B microscope (Leica Microsystems, Wetzlar, Germany) equipped with a motorized object table and a microcator with digital readout for measuring movements in the Z-direction with 0.5 μm precision. The microscope was connected with a Leica DFC320 high-resolution digital camera (Leica Microsystems) and a computer equipped with softwares for image acquisition and analysis (QWin, Leica Microsystems) and CAST-Grid stereology software (Olympus, Denmark).
The nuclei examined were the hypoglossal nucleus (XII), dorsal motor nucleus of the vagus (DMNV), nucleus tractus solitarii (NTS), medial vestibular nucleus (MedVe), cuneate nucleus (Cu), NSTT, principal inferior olivary nucleus (PION), medial inferior olivary nucleus (MION), and dorsal inferior olivary nucleus (DION) (Figs. 1 and 2). The boundaries of these nuclei were mainly defined according to Paxinos (1990) and Paxinos and Huang (1995). In transverse sections, XII appears as a round nucleus characterized by large neurons and located just beneath the floor of the fourth ventricle in ventromedial position with respect to DMNV and NTS. DMNV is located between XII and NTS. It shows a principal part and dorsal and medial fringes, which are scarcely populated and less reactive to common stainings (Paxinos,1990; Huang et al.,1993). In this study, morphometric analysis of the DMNV was limited to the principal part, which is easily recognizable because of quite large dimensions of its neurons (Fig. 2A). Boundaries of the NTS were defined according to detailed descriptions, drawings, and micrographs contained in Paxinos (1990) and Paxinos and Huang (1995). The tractus solitarius was always clearly identifiable and obviously was not considered as part of the NTS. Boundaries of the MedVe were defined according to criteria of Diaz et al. (1996), Suarez et al. (1997), and Alvarez et al. (1998). MedVe is the largest and longest of the human vestibular nuclei, extending beneath the floor of the fourth ventricle (Olszewski and Baxter,1954). It was identified with respect to the other vestibular nuclei on the basis of its morphological characteristics and topographical relationships, although in some regions (for instance, at its rostral pole) boundaries are not clearly distinguishable because of intermingling of neurons of different vestibular nuclei (Diaz et al.,1996). In transverse sections, MedVe has a triangular shape, with longest side parallel to the floor of the fourth ventricle, and it shows higher cellular density and less myelinated fibers than the other vestibular nuclei (Brodal,1974; Diaz et al.,1996). It is medially and caudally located with respect to lateral and superior vestibular nuclei, respectively. The inferior vestibular nucleus is also lateral to the MedVe at a more caudal level. Moreover, rostrally, the MedVe is located in a ventrolateral position with respect to the nucleus of the abducens nerve. Caudally, it is mainly ventromedially limited by the NTS (Fig. 2A). Its caudal pole is found at the level of the rostral pole of the gracile nucleus. Dorsolateral boundary of the Cu was defined by the presence of the cuneate fasciculus; it was possible to identify ventromedial limit of the Cu on the basis of its cytoarchitectonic characteristics. The lateral border of the NSTT is clearly given by the spinal trigeminal tract; its medial boundary is recognizable on the basis of cytoarchitectonic features of the nucleus (Fig. 2B). As regards inferior olivary complex, it is not difficult to distinguish its different subnuclei due to clear anatomical borders (Fig. 2C) (Lasn et al.,2006). Haematoxylin-eosin staining usually permitted precise delineation of the medullary nuclei. Further confirmation derived from comparison with the adjacent serial sections, stained with Nissl, Klüver-Barrera, and azan-Mallory. Sections were first analyzed at primary magnifications of 10–40× for delineation of the nuclei. Morphometric findings were derived from right or left nuclei chosen systematically at random.
Counting of neurons was performed using the optical disector method. The disector represents a probe which samples isolated particles with a uniform probability in three-dimensional space irrespective of size, shape, or orientation (Sterio,1984; Gundersen et al.,1988; Pakkenberg and Gundersen,1988,1997; West,1993). Table 2 lists the stereological parameters followed in this work for calculating neuronal densities with optical disector method. Systematic and uniformly random sampling of the fields of vision was performed with XY-step = 200 × 200, 400 × 400, or 700 × 700 μm2 depending on medullary nucleus (Table 2). Counting was performed with a high-numerical aperture (NA = 1.4) 100× oil-immersion objective at a final magnification of 1,700. An unbiased counting frame of known area (15,162 μm2) was superimposed on an optical section. The upper right corners of the counting frames were taken as associated points. The height of the optical disector was 9 μm. The focal plane (or optical section) was moved through the thickness of the section, producing a continuous series of overlapping sections within which counting could be carried out with the following disector counting rules. The specific unit chosen for counting was the nucleolus. Profiles in contact with the upper right edges of the frame were considered to be inside the frame, and those in contact with the lower left edges to be outside it. All neuron nucleoli seen in focus in the first, most superficial, look-up plane were disregarded, the thickness of the guard area being 1 μm. Then, all neuron nucleoli which came into focus through the height of the disector were counted. The total number of neuronal profiles (Q−) in each disector was counted. The data obtained in each disector and section were then summed to provide the total number of profiles observed (∑Q−). According also to preceding studies on the same medullary nuclei (Machaalani and Waters,2003; Kiryu-Seo et al.,2005), neurons were distinguished from glial cells on the basis of their larger size, clearly defined neuronal cytoplasm, with Nissl substance, and membrane-bound nucleus with a clear nucleolus (Toft et al.,2005). For each medullary nucleus, the neuronal density was calculated as follows:
where Vol(dis) was the volume of each disector, that is, the product of the area of the counting frame and the height of the disector (in this case, 15,162 μm2 × 9 μm = 136,458 μm3); ∑P was the total number of points (upper right corners of the counting frames) hitting the region of interest.
|Frame area (μm2)||XY steps (μm2)||Mean number of disectors sampled (∑P)||Mean number of neurons sampled (∑Q−)|
An estimate of the total number of neurons (N) was then performed in a two-step process, which involved both the numerical densities of the neurons, NV, and the volumes of the medullary nuclei, V. For each nucleus, the total neuron number (N) was estimated as follows:
where a(p) is the area associated with each sampling point (100,000 or 150,000 μm2, depending on medullary nucleus), is the mean distance between two consecutive studied sections (0.8 or 1.6 mm), n is the number of sections studied for each nucleus, and is the sum of points hitting a given target. Table 3 shows the stereological parameters followed for volume estimations using Cavalieri's principle. All morphometric analyses were performed by one investigator (P. A.), which was blind with regard to subjects.
|Mean number of sections hitting the nucleus (n)||Intersectional distance (μm)||Area per point (μm2)||Mean number of points counted (∑P)|
For each nucleus, mean values and coefficients of variance (CV = standard deviation/mean) were calculated for the various sample populations. Results were expressed as means with CV shown in brackets. Systematic, uniform, random sampling allows unbiased, efficient, and precise estimates. In this work, the precision of the estimates of neuronal densities, nuclear volumes, and total neuron numbers from each sample was estimated as the coefficient of error (CE). Various error estimators of volume, neuronal densities, and total neuron numbers by Cavalieri's, fractionator, or disector methods have been reported in the literature (Gundersen and Jensen,1987; Braendgaard et al.,1990; Cruz-Orive,1990,1999,2004,2006; Guntinas-Lichius and Neiss,1996; Gundersen et al.,1999; Kieu et al.,1999; Gual-Arnau and Cruz-Orive,2006). In this work, the CE for the neuronal density, that is, CE(NV), was calculated according to the following formula (Gundersen and Jensen,1987; Braendgaard et al.,1990; Tang and Nyengaard,1997; Zhang et al.,2008):
where n is the number of sections sampled.
The CE for the estimate of the volume, that is, CE(V), is function of the noise effect and systematic uniform random sampling (SURS) variance for the sums of areas: the noise effect represents the uncertainty deriving from point counting; SURS variance for the sum of areas is the uncertainty of sampling between sections, as estimates on different sections may vary. The CE(V) was calculated according to the following formulas (Gundersen et al.,1999; Christensen et al.,2007; Eriksen and Pakkenberg,2007; Fabricius et al.,2008):
where and where b and a represent the mean section boundary length and mean section area, respectively. The shape coefficient may be estimated from few sections and is usually quite stable for a given type of object (Cruz-Orive,1999).
The CE for the estimate of the total neuron number, that is, CE(N), was derived from CE(V) and CE(NV) as follows (Braendgaard et al.,1990):
Mean CEs, that is, and were calculated as follows (Gundersen and Jensen,1987):
where m is the number of cases. are listed in Table 4.
|NV (n/mm3) (CV) [ ]||V (mm3) (CV) [ ]||N (n) (CV) [ ]||NV (n/mm3) (CV) [ ]||V (mm3) (CV) [ ]||N (n) (CV) [ ]|
|XII||1,026 (0.31)[0.08]||12.3 (0.24)[0.03]||11,971 (0.19)[0.09]||2,747 (0.39)[0.08]||5.4 (0.30)[0.04]||13,396 (0.12)[0.09]|
|DMNV||1,553 (0.26)[0.06]||11.5 (0.25)[0.03]||17,218 (0.23)[0.07]||2,876 (0.19)[0.10]||6.1 (0.23)[0.04]||16,887 (0.12)[0.11]|
|NTS||2,877 (0.13)[0.07]||38.2 (0.14)[0.02]||109,553 (0.17)[0.07]||7,993 (0.17)[0.11]||15.0 (0.17)[0.03]||117,741 (0.13)[0.11]|
|Cu||1,038 (0.16)[0.15]||21.7 (0.14)[0.02]||22,603 (0.23)[0.15]||2,563 (0.23)[0.11]||8.1 (0.26)[0.04]||19,974 (0.22)[0.12]|
|MedVe||2,918 (0.12)[0.06]||30.6 (0.12)[0.02]||88,783 (0.14)[0.06]||7,010 (0.17)[0.05]||13.7 (0.16)[0.03]||93,846 (0.11)[0.06]|
|NSTT||8,990 (0.18)[0.07]||25.0 (0.14)[0.04]||220,666 (0.13)[0.08]||20,947 (0.29)[0.09]||11.8 (0.21)[0.06]||235,020 (0.17)[0.10]|
|PION||9,076 (0.10)[0.03]||82.9 (0.10)[0.01]||748,163 (0.09)[0.03]||20,010 (0.15)[0.06]||36.9 (0.11)[0.02]||736,002 (0.17)[0.06]|
|MION||9,989 (0.13)[0.08]||6.0 (0.23)[0.04]||58,510 (0.16)[0.09]||18,667 (0.20)[0.11]||3.0 (0.13)[0.06]||53,995 (0.14)[0.12]|
|DION||10,986 (0.20)[0.05]||3.3 (0.24)[0.08]||35,539 (0.23)[0.09]||22,424 (0.17)[0.10]||1.8 (0.22)[0.08]||38,054 (0.13)[0.12]|
The Mann-Whitney U-test was performed to verify differences in morphometric parameters (neuronal density, nuclear volume, and total neuron number) between groups. To reveal differences between the morphometric parameters of the different medullary nuclei, statistical analysis was performed in both adult and infant cases by the Kruskal–Wallis test and Dunn's multiple comparison test. P < 0.05 was considered to be statistically significant. Statistical calculations were carried out by Prism 3.0.3 (GraphPad Software, San Diego, CA).
Figure 3 and Table 4 show the estimates of neuronal densities, nuclear volumes, and total neuron numbers with reference to adult and infant series. Statistically significant differences were not found in the above estimates between males and females and between SIDS cases and other infant cases. The Mann-Whitney U test also did not reveal any statistically significant differences in total neuron numbers between adults and infants for the medullary nuclei considered. All the medullary nuclei showed lower nuclear volume and higher neuronal density in infants than in adults (P < 0.05).
In both adult and infant cases, the Kruskal–Wallis test revealed that the differences in morphometric parameters between the nuclei reached statistical significance (P < 0.0001). In infants, the inferior olivary complex and NSTT showed higher neuronal densities than DMNV, XII, and Cu. Statistically significant difference was also found between DION and MedVe. In adults, the neuronal densities of the inferior olivary complex (PION, MION, and DION) were higher than neuronal densities of DMNV, XII, Cu, and NTS. Neuronal densities of DION and MION were also higher than that of MedVe. Neuronal density of NSTT was higher than neuronal densities of DMNV, XII, and Cu. NTS and MedVe showed higher neuronal densities than XII and Cu.
In infants, PION showed higher volume than Cu, DMNV, XII, MION, and DION; NTS presented higher volume than XII, MION, and DION; volumes of MedVe and NSTT were higher than volumes of MION and DION; and Cu had higher volume than DION. In adults, PION showed higher volume than NSTT, Cu, XII, DMNV, MION, and DION; NTS showed higher volume than Cu, XII, DMNV, MION, and DION; MedVe presented higher volume than XII, DMNV, MION, and DION; volumes of NSTT and Cu were higher than volumes of MION and DION; and XII had higher volume than DION.
In infants, PION showed higher total neuron number than MION, DION, Cu, DMNV, and XII; NSTT showed higher total neuron number than DION, Cu, DMNV, and XII; NTS had higher total neuron number than Cu, DMNV, and XII; MedVe presented higher total neuron number than DMNV and XII; total neuron number of MION was higher than that of XII. In adults, PION showed higher total neuron number than MedVe, MION, DION, Cu, DMNV, and XII; NSTT showed higher total neuron number than MION, DION, Cu, DMNV, and XII; NTS had higher total neuron number than DION, Cu, DMNV, and XII; MedVe presented higher total neuron number than Cu, DMNV, and XII; total neuron number of MION was higher than those of DMNV and XII; and total neuron number of DION was higher than that of XII.
Estimates of cell total numbers in various tissues have for much time been carried out applying indirect and biased methods based on counts in a two-dimensional plane. In fact, the probability for objects of being hit by a single section is proportional not only to their number but also to their size and form. Thus, direct and unbiased methods for cell counting have been developed, the so-called disector (Sterio,1984) and fractionator (Gundersen,1986). The basic principle of these methods is that the probability that an object is hit by a section but is not hit by a parallel section is independent from object size and form. The physical disector method relies on parallel physical sections separated by a distance smaller than the minimum particle height; objects counted in a section but not in the other one are considered. In the optical disector method (Gundersen,1986), relatively thick sections are used, through which two or more parallel optical sections are performed by moving the plane of focus (reviewed in Gundersen et al.,1988; Pakkenberg and Gundersen,1988; West,1993). The fractionator method (Gundersen,1986) relies on the same basic counting principles as the disector; the estimate of the total object number derives from multiplication of counted particles with fixed fractions of sections and fields counted. In this work, the optical disector method was applied, because of the advantage of eliminating the difficult and time-consuming task of identifying corresponding portions in different physical sections. Most morphometric studies on medullary nuclei, with few exceptions (Tang et al.,2001/2002; Lasn et al.,2006), did not apply the disector or fractionator methods, indicating a not yet complete acquaintance of the importance of the unbiased principles at the basis of such methods. A research group compared analyses of total neuron numbers in MedVe with two- and three-dimensional probes (optical fractionator), finding 18% higher total neuron number with the unbiased method (Lopez et al.,1997; Tang et al.,2001/2002). Moreover, it must be considered that preceding studies concern one or few nuclei, lacking a comprehensive and contemporary study of many medullary structures. The use of different methods in the earlier studies limits the reliability of the comparison of morphometric data from different nuclei. Moreover, morphometric data regarding NTS, NSTT, and accessory inferior olivary nuclei are completely lacking. To the best of our knowledge, this is the first study applying the unbiased optical disector method to detail reliable morphometric data of a relatively wide series of medullary nuclei. As it concerns possible methodological limitations of this study, we must consider that tissue shrinkage and low guard zones may have introduced some bias (Dorph-Petersen et al.,2001). On the other hand, higher guard zones in the available sections would have reduced disector heights probably introducing another bias and, in this work, nucleoli were quantified, which provide better resolution being split less likely than nuclei (Coggeshall and Chung,1984; Keuker et al.,2004). Moreover, in this kind of morphometric analysis, it would be better to count more than 100 points or neurons of any structure but this general rule was not followed for some nuclei (Tables 2 and 3). However, CEs and CVs listed in Table 4 show that also for the nuclei more sparsely sampled the precision of the estimates was sufficient relative to the rather large observed coefficients of variation.
Previous works reported noteworthy differences between the various medullary nuclei in terms of vascularization (Porzionato et al.,2004a,2005) and responses to hypoxic-ischemic injuries (Gilles,1969; Revesz and Geddes,1988; De Caro et al.,2000,2003; Parenti et al.,2005; Porzionato et al.,2004b,2008a). Our study also revealed statistically significant differences between morphometric parameters of the different medullary nuclei. In particular, the nuclei of the medullary tegmentum (XII, DMNV, NTS, MedVe, and Cu) showed lower neuronal densities than the NSTT and the nuclei of the inferior olivary complex, which are located in a more ventral position (Fig. 1). Further analysis could consider the relationships between morphometric data of neuronal populations and microvascularization.
In the literature, sex-related differences have mainly been reported for telencephalic structures, such as neocortical neurons in Pakkenberg and Gundersen (1997). In this work, analogous differences were not found in the medullary nuclei, although a possible confounding role of age cannot be excluded. Subtle abnormalities have also been reported in central and peripheral centers involved in cardiorespiratory regulation (Carey and Foster,1984; Filiano and Kinney,1992; O'Kusky and Norman,1992; Kinney et al.,2002; Macchi et al.,2002; Porzionato et al.,2008b). In particular, O'Kusky and Norman (1992) found lower neuronal density in XII in SIDS cases with respect to controls without significant changes in total neuron numbers. Conversely, Macchi et al. (2002) found increased neuronal density in XII with corresponding reduction in DMNV in one SIDS case. Lower neuronal densities in SIDS cases have also been reported for the inferior olivary complex (Kinney et al.,2002). In this work, statistically significant differences between SIDS and other infant cases were not found. Nevertheless, the presence of SIDS cases in the infant casistics must be considered.
As regards comparison between adults and infants, we have not found statistically significant differences in the total neuron numbers in all the nuclei examined. In the literature, higher total neuron number was only found in the medial cuneatus nucleus of an adult than in a 2-month infant and was ascribed to continuation of the development of this nucleus also after birth (Ma et al.,2005). However, the small number of specimens examined in that study strongly limited the reliability of the finding. The absence of statistically significant differences in total neuron number between adults and infants indicates the complement of the development of the medullary nuclei considered, at least for total neuron number, in the prenatal period. On the contrary, both literature data and our findings showed an increased volume and corresponding decreased neuronal density in adult medullary nuclei with respect to infant ones. Such changes may be ascribed to development of the neuropil and microvascularization (Bourrat and Sotelo,1984; Porzionato et al.,2004a,2005). To fully evaluate the development of the medullary nuclei regarding synaptogenesis and maturation of the neuropil, also with reference to the different patterns of dendritic arborization (isodendritic, allodendritic, and idiodendritic), further studies considering morphological and morphometric parameters could be of interest.