Introduction

Idiopathic Parkinson’s disease (PD) is a common age-associated neurodegenerative disorder clinically characterized by bradykinesia, rigidity, resting tremor, and abnormal postural reflexes. Many PD patients eventually develop cognitive decline [1, 25] and, in some, the earlier clinical manifestations may include sleep disorders [10, 42]. The neuropathology of PD is characterized by degeneration of the substantia nigra (SN) with loss of neurons and the accumulation of α-synuclein aggregates within neurons and neurites, known as Lewy bodies and Lewy neurites, respectively [17, 23, 39]. PD is now considered to be part of the spectrum of Lewy body diseases [14]. Although the pathology of PD is not limited to the SN, as it also involves other brain stem nuclei, hippocampus, amygdala, neocortex, and olfactory bulb [3, 5, 12, 24], the degeneration of the SN pars compacta and its dopaminergic projection to the striatum is the most salient morphological feature of PD and plays a preeminent role in the motor manifestations that characterize the disease.

Since PD is an age-associated disorder, it is important to assess how the loss of neurons associated with normal ageing relates to the disease. Previous morphometric studies of the human SN, both non-stereological [15, 31, 32, 43, 44] and stereological [7, 29, 38, 41] agree that there is a loss of neuromelanin-containing (pigmented) SN neurons in normal ageing, but disagree on the rate of this loss. Some studies have estimated the loss as 4.3% per decade [7], whereas other studies have reported almost 10% [28]. There is no consensus either as to what happens with the size of SN neurons in normal ageing. Previous non-stereological studies have reported a decrease in the nuclear volume of SN neurons in ageing [16, 30] and enlargement of the cell bodies as demonstrated on Golgi stains [11]. Stereological studies have yielded divergent results. Although Ma et al. [28] reported smaller pigmented neurons in the SN of older subjects, Cabello et al. [7], using the rotator method, found an increase in the volume of these neurons. The variations in the outcomes of earlier studies may be attributed to the use of different morphometric techniques, variability in the demarcation of the SN, and to the use of samples of convenience with limited neuropathological characterization, because the majority of these studies were performed before α-synuclein immunostains became available for assessing the presence of PD or Lewy body disease. Because of the significant discrepancies of previous studies both on the rate of age-associated loss and the changes in size of SN neurons, we believe additional studies are warranted. In the present study, we used stereological approaches to examine the number and volume of neuromelanin-containing (pigmented) and tyrosine hydroxylase (TH) positive (+) neurons in the SN, including pars compacta and reticulata, in a large number of autopsy brains that included those from patients with PD as well as well-characterized clinically normal older controls from the Baltimore Longitudinal Study of Aging (BLSA). All specimens underwent a thorough neuropathological examination including α-synuclein and tau immunostains to identify and exclude cases of Lewy body disease or tauopathies, respectively. Our main finding is that the pigmented and TH (+) neurons of the SN decrease in number in the course of normal ageing, while at the same time their cell bodies grow larger. In contrast, SN neurons in PD are markedly reduced in number and show decreased neuronal volumes. In concert, these observations suggest that the increased neuronal volumes in the SN of older controls may represent a mechanism to compensate for the loss of neurons.

Materials and methods

Subjects

Demographic and neuropathological information is summarized in Table 1.

Table 1 Demographic and stereological data
Full size table

Controls. Young controls (n = 7) were 18–21 years and had no histories of neurological disease. Autopsies showed no abnormalities of the nervous system and no α-synuclein or tau lesions. Middle-aged controls (n = 9) were 43–59 years with no history of neurological disease. Autopsies showed no abnormalities of the nervous system and no α-synuclein or tau lesions. Older controls (n = 7) were 76–96 years and were from the Baltimore Longitudinal Study of Aging (BLSA). They underwent annual evaluations that included a series of neuropsychological tests [26], neurological exam, interval medical history, medication review, and a structured subject and informant interview for clinical dementia rating (CDR) [21, 37]. In BLSA controls, the mean interval between the last clinical evaluation and death was 8.6 months. We excluded subjects with histories of PD, other movement disorders, cognitive decline, dementia, cerebrovascular disease or any neurological abnormalities. Autopsies revealed no α-synuclein lesions. CERAD neuritic plaque scores were 0 to A [36], and neurofibrillary Braak scores were II to III [4].

Parkinson’s disease patients (n = 8) were 68–75 years. The majority were followed longitudinally at the Johns Hopkins University Morris Udall Parkinson’s Disease Center, and the diagnosis of PD was confirmed at autopsy [6, 17, 20]. These individuals had histories of PD for 7–20 years, with or without dementia, and their brains had α-synuclein lesions in the SN and other brain regions [34]. There were no significant lesions indicative of other neurodegenerative disorders, i.e., tauopathies [13] or multiple system atrophy [18, 40].

Neuropathology

All autopsies were conducted by the Neuropathology Division of the Johns Hopkins University. Brains were fixed in 10% buffered formaldehyde for 2 weeks and examined grossly on coronal sections. For neuropathologic diagnosis, standard tissue blocks from cerebrum, brain stem, cerebellum, and spinal cord were dissected and processed for paraffin sections and hematoxylin-eosin staining. Selected sections were stained with silver (Hirano method) [50] and immunostained for α-synuclein (Synuclein-1, dilution 1:500 from Transduction Laboratories) and phosphorylated tau (PHF-1 antibody, dilution 1:100; a kind gift of Dr. P. Davies) as previously described [35]. The α-synuclein immunostains were used to rule out Lewy body disease and phosphorylated tau (PHF1) immunostains were used to exclude possible cases of Progressive Nuclear Palsy (PSP) or other tauopathies [13].

Stereology

In all of the brains, we dissected 5 or 6 coronal tissue blocks of brain stem and diencepahalon that contained the entire SN. The span of the dissected tissues exceeded the rostral and caudal extents of the SN. The tissues were processed in an automated processor, dehydrated through alcohols, followed by sequential immersions in cedarwood oil, methyl salicylate, and xylene, embedded in paraffin and cut serially at 50 μm. This thickness was chosen because the sections undergo significant shrinkage during the staining processes. Then, we selected a set of sections that included every 20th section with a random start within the first 20 sections. This yielded ∼12–14 sections, which were stained with cresyl violet (Nissl) for all four groups (Fig. 1). In addition, tissue sections sets from young and older control groups were immunostained with tyrosine hydroxylase (TH) polyclonal antibody (1:50 concentration, Pel-Freez) (Fig. 1). The borders of the SN were outlined using a 5× objective following the landmarks proposed by [45]. Neurons from one side of the SN, including pars compacta and pars reticulata, were counted using the optical fractionator probe [46, 47]. Stereological measurements were performed with a Zeiss light microscope equipped with a 100×, NA 1.30, oil Plan neofluor ∞/0.17 objective and interfaced with a StereoInvestigator system (MBF Bioscience, Williston, VT, USA). The images were captured with a microFire video camera (Optronics, Goleta, CA, USA). The thickness of the sections was measured by focusing on the top of the section, setting the Z-axis to 0, and then refocusing to the bottom of the section and recording the actual thickness. The actual thickness of the histological sections was in the 30–44 μm range. The total number of SN neurons (pigmented) was estimated using the Stereo Investigator Optical Fractionator probe; the disector height was 25 μm. This height was chosen based on the actual thickness of the sections plus an allowance for top and bottom guards. The counting frame was 50 μm × 35 μm, and the sampling grid was 400 μm × 400 μm. Using the nucleator probe [19], the volume of pigmented neurons was estimated by placing six rays centered on the nucleolus and intersecting the cell membrane.

Fig. 1
figure 1

This panel of photomicrographs compares the sizes of representative pigmented neurons of the SN in young controls (a), older controls (b), middle-aged controls (c), and PD cases (d). The lower figures display examples of TH (+) neurons in young controls (e) and older controls (f). All micrographs are at the same magnification (see bars) and show the hypertrophy of SN neurons in older controls (b, f) and the atrophy in PD (d). Paraffin sections were cut at 50 μm and stained with Cresyl Violet (Nissl method) (a–d) or immunostained for TH (e, f)

Full size image

Statistical analyses

We compared the number of neurons in the four study groups by one-way ANOVA followed by Tukey’s post-test using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA, USA).

Since analysis of data distribution for neuronal volumes revealed significant departures from normal distribution with significant right skewness, we used parametric and non-parametric (Kruskal–Wallis) ANOVA to test the null hypothesis. With significant main effects, ANOVA was followed by post hoc tests (Kolmogorov–Smirnov) to analyze differences between particular groups. The Kolmogorov–Smirnov test was chosen because it is particularly sensitive to changes in the shapes of distribution. The results are represented as arbitrary categories of small (0–9,000 μm3), medium (9,001–18,000 μm3), large (18,001–27,000 μm3), and very large (27,001–36,000 μm3) neurons to simplify the description of changes in the shape of distributions. This range of neuronal sizes from small to very large neurons included more than 97% of all neurons and was divided into four equal groups to represent small, medium, large, and very large size neurons. Neurons with cell body volumes greater than 36,000 μm3 were omitted from the figures representing the Kolmogorov–Smirnov test because of negligible percentages observed (Figs. 6, 7). Statistical analyses were also performed for the whole range of raw (not-clustered) data. These analyses were performed using Statistica 6.0, StatSoft, Inc. Tulsa, OK, USA.

Results

The results of the estimation of the total number of pigmented and TH (+) neurons in one side of the SN and their volumes are displayed in Table 1. Among controls, there is a significant age-associated loss of pigmented SN neurons (Figs. 2, 3a). This loss of neurons is 28.3% in older controls compared with younger controls. In cases of PD the loss of pigmented neurons is massive, reaching 80.4% compared with younger controls and 73% compared with older controls. We also found a 36.2% loss of TH (+) neurons in the older controls compared to younger controls (P < 0.001) (Fig. 3b). The regression analysis demonstrates a significant (P < 0.05) increase in the size of pigmented SN neurons in the course of normal ageing (Fig. 4). The mean volume of SN neurons appears 39% larger in older controls compared to younger controls (Fig. 5). Furthermore, the distribution analysis of cell body volume revealed significant (Kolmogorov–Smirnov test, P < 0.01) hypertrophy of pigmented neurons in older controls and atrophy in PD cases (Figs. 6, 7a). We also observed hypertrophy of TH (+) neurons in older controls compared to younger controls. The mean volume of TH (+) neurons was 23% larger in older controls. Although the difference of the means was not significant, the analysis of cell body volume distribution revealed a significant difference between the groups (Kolmogorov–Smirnov test, P < 0.01) (Fig. 7b), with the older controls having a larger proportion of large neurons. The hypertrophy of SN neurons in normal ageing and their atrophy in PD is illustrated in photomicrographs in Fig. 1.

Fig. 2
figure 2

Linear regression of the number of pigmented neurons (closed circles) versus age is significant (P = 0.0060) with R 2 = 0.337; CV = 0.20. The open triangles correspond to the mean number of pigmented neurons in the SN of Parkinson’s disease cases

Full size image
Fig. 3
figure 3

These scatter plots show the total number of estimated pigmented neurons in one side of the substantia nigra for each study group (a) and for tyrosine hydroxylase (TH) (+) neurons in young and older controls (b). The horizontal bars indicate the mean values. One-way ANOVA revealed significant differences present between the number of pigmented neurons in younger and older controls (P < 0.01) and between each control group and the PD group (P < 0.001) (a). There is a 36.2% loss of TH (+) neurons in the older control group, also a significant difference (P < 0.001) (b)

Full size image
Fig. 4
figure 4

Linear regression of the volumes of neurons of the three control groups (PD excluded) versus age is significant (P = 0.046) with R 2 = 0.18; CV = 0.31

Full size image
Fig. 5
figure 5

Scatter plot comparing the mean volumes of the substantia nigra neurons among the four study groups. Horizontal bars indicate the means. Although the older controls show an increase in the mean value (+39%), this change is not statistically significant when the mean value is compared with any of the other groups

Full size image
Fig. 6
figure 6

Histogram of the frequency of small, medium, large, and very large pigmented SN neurons in young controls (YC), middle-aged controls (MC), Older Controls (OC), and Parkinson’s cases (PD). Note the higher percentage of very large neurons in older controls and of small neurons in PD. The Y-axis corresponds to the frequency (%) of measured neurons in each size category. The X-axis represents the arbitrary neuronal volume bins in μm3: small (0 to 9,000 μm3), medium (9,001 to 18,000 μm3), large (18,001 to 27,000 μm3), and very large (27,001 to 36,000 μm3)

Full size image
Fig. 7
figure 7

These graphs represent the distribution of cell bodies of SN neurons according to their volumes. a Shows the distribution of pigmented SN neurons in young controls (YC), middle-aged controls (MC), Older Controls (Old), and Parkinson’s cases (PD). Comparison of the curves shows neuronal hypertrophy in older controls and atrophy in PD. Statistical analysis (Kolmogorov–Smirnov Test) demonstrated significant differences (P < 0.01) when comparing older controls to either middle-aged or young controls, and when comparing PD to each control group. Note the displacement of the older controls curve to the right indicating a higher proportion of larger neurons compared to all other groups. In contrast, the curve of PD cases is shifted to the left due to the larger proportion of smaller neurons. b Displays the distribution of TH (+) neurons in young controls (YC) versus older controls (Old). Statistical analysis (Kolmogorov–Smirnov Test) demonstrated significant increase in the proportion of larger neurons in older controls compared to young controls (P < 0.01), as revealed by the displacement of the older controls curve to the right

Full size image

Discussion

Our observations indicate that the human SN undergoes a significant loss of pigmented and TH (+) neurons from the pars compacta and reticulata during normal ageing (Figs. 2, 3). At the same time, there is a significant enlargement of those neurons (Figs. 4, 5, 6, 7). In PD, the loss of neurons from the SN is massive (Figs. 2, 3), and their cell body volumes are reduced (Figs. 6, 7).

Our measurements indicate that the human SN of young controls has a mean of 424,000 pigmented neurons (unilateral). As shown in Table 1 and Fig. 3a, however, we find a wide range of values from ∼350,000 to ∼500,000 neurons. This wide range in the total number of neurons is not unique to the SN as there is a similar variability in the number of neurons in hippocampus [46, 48] and entorhinal cortex [49] of normal subjects. Our estimation of total number of pigmented neurons in younger subjects is higher than that of previous studies (adjusted for one side of the SN): Ma et al. [29] ∼150,000, and Cabello et al. [7] 189,000. These differences can be attributed to several factors including the use of different embedding media, histological stains, stereological methods, but more importantly the delineation of the boundaries of the SN region [41]. In the older controls, we estimated a mean of 305,000 pigmented neurons (unilateral), a number slightly higher than the 275,000 neurons (adjusted for one side of the SN) estimated by Pakkenberg et al. [38] in older controls (mean age: 81).

The mean number of pigmented neurons in older controls (mean age: 87.6) was 28.3% (P < 0.01) less than that of younger controls (mean age: 19.9). Distributed over the intervening 6.5 decades, this loss amounts to an average 4.35% per decade. Compared to non-stereological studies, this loss is less than the 48% loss by age 60 (∼7% per decade) reported by McGeer et al. [32] and the 35% loss of neurons by age 90 observed by Mann et al. [31]. In a previous stereological study of the SN in ageing, Ma et al. [29] examined the brains of 26 normal subjects with an age range from 17 to 90 and demonstrated a significant loss of the total number of pigmented neurons in older subjects. Using paraffin sections and the physical disector method, that study estimated there to be ∼300,000 neurons in the youngest subjects and ∼100,000 neurons in the oldest subjects. The average decline in the number of pigmented neurons per decade was 9.8%. It is important to note that although the older subjects were clinically normal, the brains in that study were not examined for ubiquitin or α-synuclein lesions. In a subsequent stereological study, using plastic sections and the optical disector, Cabello et al. [7] examined the brains of 28 males aged 19–92. They observed a significant decrease in the total number of pigmented neurons as a function of age of approximately 4.3% per decade, an estimate identical to our study. Nevertheless, the difference between the 7 youngest (mean age: 27; 378,000 neurons) and the 7 oldest (mean age: 82; 256,000) subjects was not significant. Although the absolute numbers of pigmented neurons in the SN in young subjects in our series are different from those reported in two previous stereological studies [7, 29], our results are consistent with an age-associated loss of pigmented neurons in the SN [7, 29, 41]. Although it should be noted that other stereological studies have found no age-related losses of pigmented SN neurons, those same studies have reported marked decreases in the number of neurons expressing specific phenotypic markers [8,9].

There are few morphometric studies of TH (+) neurons of the SN in normal ageing. Kubis et al. [27] did not find a significant loss of TH (+) in the SN, nor in other midbrain regions, in a series of 21 control subjects who died at ages 44–110. In contrast, we have shown a 36% (P < 0.001) loss of TH (+) SN neurons in older controls compared to younger subjects. This loss is slightly higher than the 28.3% observed for pigmented neurons. Our observations are consistent with the observation that in normal ageing there is a marked loss of dopamine transporter-immunoreactive neurons in the human SN [28], and that all dopaminergic neurons that express DAT mRNA are also neuromelanin positive [2].

As discussed in the Introduction, there is no consensus with regard to the changes in the volume of SN neurons in normal ageing. Non-stereological studies have reported a decrease in the nuclear volume of SN neurons in ageing [16, 30] and enlargement of the cell bodies as demonstrated on Golgi stains [11]. Stereological studies have yielded divergent results. Although Ma et al. [28, 29] reported smaller pigmented neurons in the SN of older subjects, Cabello et al. [7], using the rotator method, found an increase in the volume of these neurons. Our measurements, using the nucleator [19], a different stereological probe than that used by Cabello et al. [7], are consistent with the latter assessment. Indeed, our observations indicate that there is an increase in the volume of SN in older subjects and that although the comparison by ANOVA of the mean neuronal volumes in younger versus older controls does not reach statistical significance, the analysis of frequency distributions of neuronal volumes does show a significant difference between younger and older subjects (Figs. 6, 7). Importantly, examination of the frequency distribution histograms reveals that the increase in the mean volume of SN in older subjects is not the result of selective loss of small size neurons, but rather a genuine hypertrophy of the cell body of pigmented nerve cells.

A previous morphological study of the TH (+) neurons in the human SN found no correlation of cell body size with age [27]. Although stereological studies of the SN have not examined the volume of TH (+) neurons in normal ageing, there is evidence for a close correspondence between pigmented and TH (+) neurons [2, 7]. Based on this notion, one would expect a hypertrophy of TH (+) neurons in normal ageing. Indeed, we found that TH (+) neurons become larger in normal ageing, a change similar to that observed in pigmented neurons. The mechanism responsible for this neuronal hypertrophy is not directly addressed in the present study; nevertheless, some theoretical possibilities can be entertained. Progressive accumulation of neuromelanin and other pigments could account for the neuronal enlargement. However, this does not explain why the neuronal hypertrophy is present in older subjects, whereas no difference is detected when we compare young and middle-aged subjects. A second possibility is that the increased neuronal size is a response to injury [33]. A third possibility, and the one we favor, is that as SN neurons are lost with age, the remaining neurons take over and reinnervate deafferented targets, predominantly in the striatum. This compensatory hypothesis, also proposed by Cabello et al. [7], is supported by the observation that the total volume of SN pigmented neurons, i.e., volume × number of neurons, stays stable across the age groups. A similar mechanism has been proposed for the hypertrophy of cortical neurons in asymptomatic Alzheimer’s disease (D. Iacono, personal communication). Age-associated hypertrophy has been reported in the pigmented neurons of the human locus coeruleus by Iwanaga et al. [22], who attributed the expansion of the somatic cytoplasm to the enveloping of synaptic terminals.

As expected, the loss of SN neurons in PD cases is massive (Figs. 2, 3a) and the remaining neurons are atrophic (Figs. 6, 7). Most of the cases that we examined were in the end stage of the disease. The numbers of remaining pigmented neurons in the SN were in the 20–30% range of a healthy SN, a number consistent with Pakkenberg et al. [38]. Notably, the two PD cases with the higher number of neurons in the SN were those with the shortest duration of disease. Inspection of scatter plots of the mean volume of SN neurons in PD and controls (Fig. 5) shows that the volume of the few remaining SN-pigmented neurons in PD was not significantly different from that of middle-aged subjects. However, further analysis of the volume frequency distribution (Figs. 6, 7a) reveals that the SN of PD cases has a larger proportion of small neurons compared with all of the control groups. This atrophy is opposite to the neuronal hypertrophy of normal ageing and supports the notion that the changes of the SN in PD are not a manifestation of accelerated ageing, but a specific pathologic disorder.

As a result of the neuronal atrophy and the massive loss of neurons, the total volume, (i.e., the product of number and volume) of SN-pigmented neurons in PD was drastically reduced to ∼20% compared with controls. This observation suggests that the normal functioning of the nigrostriatal system requires a certain total volume, not number, of SN-pigmented neurons below which motor impairment begins to manifest.

In conclusion, our observations indicate that in the course of normal ageing, neuromelanin-containing and TH (+) neurons of the SN decrease in number, but at the same time their cell bodies enlarge. This finding suggests the presence of a compensatory mechanism within individual SN neurons that allows for normal motor function despite the loss of neurons in normal ageing. Presumably, this compensatory mechanism breaks down or is overwhelmed by the pathological events of PD leading to the onset of the characteristic motor disturbances.