Introduction

Hypoxic-ischemic encephalopathy (HIE) is one of the major complications in emergency medicine and obstetrics [4, 67, 70, 71]. It can be caused by prolonged mechanical resuscitation, respiratory insufficiency as well as arterial hypotension or shock. It is the leading cause of morbidity and mortality occurring after successful resuscitation of cardiac arrest [29, 32]. The harmful effects of HIE are most severe in the hippocampus, neocortex, and cerebellum [29]. Whether these areas are most affected because of a border zone problem of cerebral perfusion is still a matter of debate. Nearly all patients with HIE remain in permanent need of care, 40% enter a persistent vegetative state [30] and after 1 year approximately 80% have died [29]. Current treatment includes intensive care medicine and rehabilitative therapy [61]. Since 2002, mild hypothermia for at least 24 h is recommended after cardiac arrest [60, 72].

Until 1960, adult neurogenesis was regarded as nonexistent [16]. Since then, different new methods have verified the proliferation of neural progenitor cells and their differentiation into mature neurons within the granule cell layer of the hippocampal dentate gyrus in both adult mammalian [1, 9, 17, 19, 22, 25, 38, 48, 53, 64] and human brains [2, 3, 6, 12, 21, 31, 37, 52, 56].

Neural stem cells (NSC) are located in the subgranular layer of the dentate gyrus [24, 38] where a specific microenvironment exists that is considered to be crucial for neural proliferation [51]. Neuronal precursor cells are probably generated in clusters [22, 25, 53] and migrate a short distance to their designated destination: the granule cell layer of the dentate gyrus [54]. However, whether their integration and differentiation into mature neurons is of functional significance is still a matter of debate [17, 22, 48, 64]. In rodents, a decrease of the magnitude of hippocampal neurogenesis with increasing age is well known [9, 19, 25, 42]. However, recent data suggest that hippocampal neurogenesis in rats appears to remain relatively unaltered in adulthood [18]. In humans, adult neurogenesis was increased in various neurological diseases of different pathophysiology such as Alzheimer’s disease [2], multiple sclerosis [6], subarachnoidal hemorrhage [52], primary intracerebral hemorrhage [56] and epilepsy [15] while hippocampal neurogenesis was shown to be reduced in patients who underwent radiation therapy for central nervous system malignancies [37]. These observations suggest that enhancement of adult neurogenesis in humans may not be a uniform phenomenon, but a differently regulated event depending on the underlying condition [28, 44].

Whether adult neurogenesis is affected in response to HIE is not yet known. This is of interest, since the administration of neuroprotective drugs which interfere with adult neurogenesis and apoptosis and stimulate proliferation and migration of neuronal progenitor cells may be a potential therapeutic option. Recently, two infants with HIE caused by prolonged cardiopulmonary arrest received an intraventricular infusion of nerve growth factor. Improvement of their neurological condition and of cerebral perfusion, as well as an increase of doublecortin expression in the cerebrospinal fluid, were noted [8]. Other potential neuroprotective drugs may be fibroblast growth factor [38], epidermal growth factor [43], heparin-binding EGF-like growth factor [20], granulocyte colony-stimulating factor [35, 46], erythropoietin [57], inhibitors of vascular endothelial growth factor receptor [23], inducible nitric oxide synthase [74], statins [7], antidepressants [11, 45] and sildenafil [73].

The present study is designed to determine whether any change in neuronal proliferation, differentiation or apoptosis occurs in response to HIE. Furthermore, the study investigates whether these effects depend on the patients’ age.

Autopsy cases and methods

Autopsy cases

In this retrospective study, autopsy cases from the Department of Neuropathology, Georg-August-University, Göttingen, Germany, were selected from 1996 to 2004 according to the following criteria: patients that had suffered from a clinically observed and neuropathologically confirmed HIE which had taken place at least 24 h before death and age-matched control cases. Exclusion criteria were stroke, death subsequent to a neurological disease, sepsis, cancer that had been treated with cytostatic agents or radiation before death, and forensic cases.

The HIE group consisted of 22 subjects (9 women, 13 men, median age 64 years, age span 35–85 years) who had later died because of heart failure in 11 cases, because of respiratory insufficiency in 4 cases, because of failure of central regulation in 4 cases and because of shock in 3 cases (Table 1). HIE took place a median 10 days before death (range 1–53 days) because of prolonged mechanical resuscitation (13 cases), acute pulmonary decompensation (6 cases) or acute coronary syndrome (3 cases) (Table 1).

Table 1 Information about HIE cases
Full size table

The control group consisted of 19 subjects (7 women, 12 men, median age 59 years, age span 35–81 years) whose causes of death were either sudden heart failure (16 cases), asphyxia (2 cases) or shock (one case) (Table 2). Resuscitation was attempted in 11 cases but, in contrast to the HIE group, death had occurred within 1.5 h after start of resuscitation. The cerebral hypoxia before death could have resulted in neuronal damage but all control cases lacked neuropathological abnormalities, particularly no evidence of HIE was detected (Table 2). Routine neuropathological examinations were performed by the Department of Neuropathology, Georg-August-University, Göttingen, Germany. The study was approved by the Ethics Committee of the Medical Faculty of the Georg-August-University in Göttingen, Germany.

Table 2 Information about control cases
Full size table

Immunohistochemistry

Sections of the hippocampal formation were stained immunohistochemically for the cell proliferation maker proliferating cell nuclear antigen (PCNA) [26, 33], for TUC-4 (TOAD-64/Ulip/CRMP-4) as a marker for immature granule cells during axon guidance and outgrowth [36, 41, 50, 66] and for calretinin as a marker for young postmitotic neurons [5, 49]. One section was stained per each marker and case. The formalin-fixed, paraffin-embedded brain tissue was cut in 2-μm-thick sections, dried, deparaffinizied and pretreated with 15 min of microwaving in citric acid buffer (10 mmol/L, pH 6.0).

PCNA and calretinin stainings were performed in phosphate-buffered saline; after treating with 3% H2O2 for 10 min and blocking with 10% fetal calf serum for 30 min, monoclonal mouse anti-PCNA antibody (1:200, Chemicon, Temecula, CA, USA) or polyclonal rabbit anti-calretinin antibody (1:1,000, Swant, Bellinzona, Switzerland) were allowed to bind for 120 min at room temperature. The primary antibodies were detected by biotinylated sheep anti-mouse antibody (1:200, Amersham, Buckinghamshire, UK) or biotinylated donkey anti-rabbit antibody (1:200, GE Healthcare, Buckinghamshire, UK). Avidin peroxidase (Sigma-Aldrich, St. Louis, MO, USA) was used as conjugated enzyme and DAB (Roche, Mannheim, Germany) as chromogenic substrate.

TUC-4 staining was performed in Tris-buffered saline; after blocking with 10% fetal calf serum for 30 min, polyclonal rabbit anti-TUC-4 antibody (1:200, Chemicon, Temecula, CA, USA) was allowed to bind overnight at 4°C. Detection was carried out by means of a monoclonal mouse anti-rabbit antibody (1:50, Dako, Kopenhagen, Denmark), a rabbit anti-mouse antibody (1:50, Dako, Kopenhagen, Denmark) and alkaline phosphatase anti-alkaline phosphatase complex (1:50, Dako, Kopenhagen, Denmark). Newfuchsin (Dako, Kopenhagen, Denmark) was used as chromogenic substrate.

All sections were counterstained with hemalum (Merck, Darmstadt, Germany). Control sections were treated similarly, with or without primary antibodies.

Apoptotic granule cells

Apoptotic granule cells in the dentate gyrus were detected by in situ tailing (IST) of fragmented deoxyribonucleic acid combined with the use of morphological criteria (condensed, shrunken nucleus and narrow cytoplasm) [14, 39]. For IST, brain tissue was cut, dried, and deparaffinizied as described above. The sections were first incubated with 100 μg/ml proteinase K (Sigma-Aldrich, St. Louis, MO, USA) at 37°C for 15 min, followed by incubation with a tailing mixture of 10 μl tailing buffer, 1 μl digoxigenin deoxyribonucleic acid labeling mix, 2 μl cobalt chloride, 0.5 μl terminal transferase (all Roche, Mannheim, Germany) ad 50 μl distilled water for 60 min and additionally with 10% FCS for 20 min at 37°C. Finally, sheep anti-digoxigenin antibody (FAB fragments, 1:250, Roche, Mannheim, Germany) in Tris-buffered saline was allowed to bind for 120 min at room temperature. 4-Nitroblue-tetrazoliumchloride/5-bromine-4-chloride-3-indolylphosphate was used as chromogenic substrate. The sections were counterstained with nuclear fast red-aluminium hydroxide (Roche, Mannheim, Germany). Negative control sections were treated similarly but without tailing mixture or the anti-digoxigenin antibody. As positive controls, sections of the hippocampal formation of a newborn marmoset monkey with a high rate of apoptotic neurons were used.

Light microscopy

All sections were inspected by light microscopy (BX 51 TF, Olympus, Hamburg, Germany) equipped with digital camera technology (ColorView II Camera Set, Olympus Hamburg, Germany, Workstation Standard, Soft Imaging System, Münster, Germany). The area of the dentate gyrus (granule cell layer and border of the subgranular layer) was determined by planimetry with the software AnalySIS 3.2 (Soft Imaging System, Münster, Germany). The numbers of immunoreactive cells within the determined areas were counted by a blinded investigator at a magnification of 60×. The cell densities were expressed as number of immunoreactive cells per mm2 of measured area.

Statistical analysis

Data are presented as box-whisker plots with maximum, 75th percentile, median, 25th percentile and minimum and were compared by Mann–Whitney’s two-tailed nonparametric U test. The relation between the density of immunoreactive cells and the age of the patient was assessed with Spearman’s nonparametric two-tailed rank correlation. All statistical analyses were carried out using GraphPad Prism Software version 4 (GraphPad Software, San Diego, CA, USA). A P value of <0.05 was considered statistically significant.

Results

In this study, firstly an HIE group was compared to a control group with normal brains regarding precursor cell proliferation, expression of TUC-4 and calretinin and density of apoptotic granule cells. Secondly, these parameters were related to the age of the autopsy cases in both groups.

Neither in HIE nor in control cases, was the density of the immunoreactive cells of any marker used in this study correlated with the intervals from death to fixation. The age of the paraffin blocks also did not correlate with the densities of immunoreactive cells for any stainings employed in this study (Tables 1, 2). No statistically significant relations were noted between recent corticosteroid treatment, depression, or the presence or absence of previously untreated malignant disease (found accidentally at autopsy) and the densities of immunoreactive cells. In those cases of the control group where resuscitation was attempted, death did occur within 1 h after start of resuscitation and no differences in proliferation, neurogenesis or apoptosis in the dentate gyrus were detected compared to control cases without a history of resuscitation.

Precursor cell proliferation in the dentate gyrus after HIE

The proliferation of putative neuronal precursor cells in the dentate gyrus was determined by staining with the proliferation marker PCNA, an intranuclear cofactor of deoxyribonucleic acid polymerase delta in rats [33] that is expressed from the G1 phase to the G2/M transition of cell division in humans [26]. The median density (/mm2) of proliferating cells in the granule cell layer of the dentate gyrus was higher in the HIE group (factor of 3.66) in comparison to controls (Fig. 1). However, the difference failed to reach statistical significance (P = 0.086). Stained cells were approximately evenly distributed within the dentate gyrus (Fig. 2a).

Fig. 1
figure 1

Comparison of densities (/mm2) of immunoreactive cells for the proliferation marker PCNA, of immature granule cells during axon guidance and outgrowth (TUC-4) and of young postmitotic neurons (calretinin) as well as apoptotic granule cells in the dentate gyrus in cases with HIE and controls with normal brains. Both the density of TUC-4- and of calretinin-labeled young neuronal cells as well as apoptotic granule cells were significantly increased in subjects with HIE in comparison to controls (*P = 0.0002, **P = 0.0001, ***P = 0.014). Data are presented as box-whisker plots (maximum, 75th percentiles, medians, 25th percentiles, minimum)

Full size image
Fig. 2
figure 2

Light photomicrographs of the dentate gyrus of humans with HIE. Expression of the proliferation marker PCNA (a) in a 73-year-old woman after death caused by shock and from acute coronary syndrome the day before. Staining of immature TUC-4 immunoreactive granule cells (b) in a 73-year-old man after death due to heart failure after prolonged mechanical resuscitation 15 days before. Young calretinin-labeled postmitotic neurons (c) in a 61-year-old man after death because of respiratory arrest after prolonged mechanical resuscitation 59 days before. Apoptotic granule cell with typical morphology labeled by in situ tailing (d) in an 81-year-old man after death because of respiratory arrest after prolonged mechanical resuscitation 3 days before. Scale bar 50 μm

Full size image

Increased expression of TUC-4 and calretinin in the dentate gyrus after HIE

The densities of both immature granule cells during axon outgrowth (assessed by TUC-4 immunohistochemistry) and young calretinin-expressing postmitotic neurons were significantly increased in the subgranular and granule cell layer of the dentate gyrus in subjects with HIE.

TUC-4 is expressed in immature granule cells and involved in axon guidance and outgrowth in humans [41], and in rats most probably from the first to the fourth week after mitosis [36, 50]. In the present study, TUC-4-labeled cells were mainly located in the subgranular layer of the dentate gyrus while only a few were located in the granule cell layer of the dentate gyrus (Fig. 2b). Their median density was significantly increased by 2.33 times in the HIE group in comparison to controls (P = 0.0002) (Fig. 1).

Calretinin is a calcium-binding protein that is expressed in young postmitotic neurons of mice for up to 6 weeks [5, 49]. Immunoreactive cells were most frequently located in the subgranular layer of the dentate gyrus (Fig. 2c). The median density of calretinin-labeled cells was significantly increased by 1.95 times in subjects with HIE in comparison to controls (P = 0.0001) (Fig. 1). In a sample of five cases with HIE, analysis of serial sections stained by calretinin and TUC-4 revealed that 11.7% of 162 calretinin-labeled cells expressed TUC-4 in the adjacent section, suggesting the coexpression of both markers.

Increased density of apoptotic granule cells after HIE

IST and morphological criteria (condensed, shrunken nucleus and narrow cytoplasm) [14] were used to detect apoptotic granule cells [39] (Fig. 2d). Apoptotic granule cells were mainly located at the border of the subgranular and granule cell layer. In HIE cases, in median 19.5% of all IST-positive cells showed an apoptotic morphology (range, 0–45.5%). The density of apoptotic granule cells in autopsy cases with HIE was significantly higher than in subjects without neuropathological abnormalities (P = 0.014) (Fig. 1). Regarding apoptotic granule cells, no factor of augmentation could be calculated, because a median of 0 apoptotic granule cells/mm2 was detected in control subjects (Fig. 1).

Densities of immunoreactive cells as related to the age of the patients

No correlation of proliferating cells with age

Within the HIE group the density of PCNA-labeled cells remained approximately constant and did not change significantly with increasing age (Table 3a). Similarly, there was no correlation between the density of proliferating cells within the dentate gyrus and age in control subjects (Table 3b).

Table 3 Relation between age at death and the density of immunoreactive cells or apoptotic granule cells, respectively, and as calculated by Spearman’s rank correlation in: (a) cases with HIE, (b) controls
Full size table

Inverse correlation of TUC-4 immunoreactive cells with age after HIE, but not in controls

The density of immature TUC-4 immunoreactive granule cells was higher in younger subjects in comparison to older cases and correlated inversely with age of death of the subjects with HIE (P = 0.027) (Fig. 3; Table 3a). Although the density of calretinin-labeled granule cells also decreased with age, this correlation failed to reach statistical significance (P = 0.057) (Table 3a). The expression pattern of TUC-4 and calretinin within the dentate gyrus in control cases without neuropathological abnormalities was not significantly influenced by age (Table 3b).

Fig. 3
figure 3

Density of immature TUC-4 immunoreactive granule cells in the dentate gyrus related to the age at death in subjects with HIE, revealing an inverse correlation between the density of TUC-4-labeled cells and age (P = 0.027, Spearman’s rank correlation coefficient = −0.47). Although the density of calretinin-labeled granule cells also decreased with age, this correlation failed to reach statistical significance (P = 0.057, Spearman’s rank correlation coefficient = −0.41)

Full size image

No correlation of apoptotic granule cells with age

The densities of apoptotic granule cells in both subjects with HIE and controls did not change significantly with age (Table 3a, b).

Discussion

The present study investigates the impact of HIE on adult neurogenesis and on neuronal apoptosis in the human hippocampal dentate gyrus. We found that the density of cells differentiating into the neuronal cell lineage, assessed by the expression of the early neuronal markers TUC-4 and calretinin, was higher in autopsy cases with cerebral hypoxia compared to control cases with normal brains. Furthermore, in contrast to controls, an inverse correlation between the density of TUC-4-labeled cells and age was detected in cases with HIE. In addition, the amount of apoptotic granule cells was also significantly increased in subjects with HIE compared to controls but no correlation with age was found.

Similar to our results, HIE induced by four vessel occlusion (4VO) resulted in a 2–3 times increase of adult neurogenesis in the hippocampal dentate gyrus of macaque monkeys [62], a 5–10 times increase in adult gerbils [55] and an increase of 14–27% in newborn rats [65], as detected by 5-bromo-2-deoxyuridine immunohistochemistry combined with different markers for neuronal differentiation. Furthermore, increased mitotic activity of both astrocytes and oligodendrocytes was found in neonatal rats after HIE [58]. In rats and macaque monkeys, the steps of neurogenesis after hypoxia and ischemia from stem cells to new granule cells were similar to the conditions in neurogenesis without exogenous stimuli [34, 37, 62, 63].

In addition to the dentate gyrus, human adult neurogenesis was also observed in the hippocampal CA1 region and in the olfactory bulb, even without previous stimulation [2, 21]. While the human neocortex was shown not to be an area of adult neurogenesis [3], the proliferation, neuronal differentiation and migration of newborn cells from the human subventricular zone is still a matter of debate [12, 15, 52]. An increase of adult neurogenesis was reported in the subventricular zone in macaque monkeys after 4VO [63]. Whether adult neurogenesis after HIE also occurs in the hippocampal CA1 region, the striatum and the neocortex is still controversial [47, 62, 63, 68, 69].

It has to be considered that variables such as different diseases or drug therapies may distinctly influence adult neurogenesis and that delays in the postmortal examinations may influence the quality of immunohistochemistry. In the present study, analysis of the patients’ records revealed that nine cases underwent corticosteroid therapy before death, and two other subjects suffered from depression. In seven cases, the autopsy revealed cancer, but the patients underwent neither cytostatic nor radiation therapy before death. In principle, any of these conditions could have interfered with adult neurogenesis [11, 37, 45, 59]. However, in the present study these cases were equally spread within the groups and comparisons regarding these parameters (absence or presence of corticosteroid therapy, depression, cancer revealed accidentally by autopsy) in HIE and control cases revealed no statistically significant differences. The interval from death to fixation of the autopsy material (0–5 days) and the age of the paraffin-embedded brain sections (3–11 years) (Tables 1, 2) also did not substantially influence the density of immunoreactive cells.

In our study, the median density of PCNA-labeled cells was elevated after HIE, similar to what was noted for TUC-4- and calretinin-immunoreactive cells. However, the increase of PCNA-labeled cells failed to reach statistical significance, whereas the density of TUC-4- and calretinin-labeled cells was significantly increased in HIE cases. This discrepancy most probably was the result of a high interindividual variation of the densities of PCNA-labeled cells and, furthermore, may result from the different periods of the expression of these markers. Whereas PCNA is only expressed from the G1 phase to the G2/M transition of cell division in humans [26], TUC-4 is expressed most probably between the first and the fourth week in rats [36, 50] and calretinin from the first to the sixth week after mitosis in mice [5, 49]. We also tried to carry out immunohistochemistry for Ki-67, nestin and calbindin, but these markers turned out not to be compatible with formalin-fixed, paraffin-embedded human brain tissue.

Regarding the dentate gyrus, there is debate about three presumptive functions of basal adult neurogenesis and enhanced adult neurogenesis: declarative memory [27], mood [11, 45] and recovery after impairment, but in both mammals and humans, areas where adult neurogenesis occurs do not fully recover function after damage [10, 13, 21, 29, 47].

In addition to an increased number of young neurons, an increase of apoptotic granule cells in the dentate gyrus was also observed in cases with HIE. This observation accords with previous data on autopsy cases with HIE [39]. In newborn rats, a similar effect was shown to be only transient after 4VO [65], suggesting that enhanced neuronal apoptosis after HIE probably occurs for a short time only. In this study 19.5% of all IST-labeled cells in HIE cases had the typical apoptotic morphology. The other IST-labeled cells were most probably dead or dying by necrosis or IST staining was a result of post-mortem DNA degradation. The increased dentate granule cell apoptosis after hypoxia may be one cause of memory impairment frequently observed in survivors after cerebral hypoxia [29, 39, 61].

With respect to the magnitude of adult neurogenesis in different age groups, in our study older-aged individuals without brain abnormalities showed neither a decrease in adult neurogenesis nor an increase of apoptotic neurons in the dentate gyrus compared to younger adult individuals. This supports the recent finding that hippocampal neurogenesis appears to remain relatively unaltered during healthy adulthood in rats [18], although a strong decrease of neurogenesis has been observed in older individuals of several other rodent species [9, 19, 25, 42].

In our study, older patients with a history of HIE showed approximately equal levels of hippocampal apoptosis compared to younger individuals with HIE. In contrast, after focal ischemia in aged rats, a larger number of apoptotic cells and an accelerated infarct development was observed in comparison to younger rats [40].

In conclusion, we show here that neurogenesis is increased in HIE. Although proliferation within the subgranular cell layer of the dentate gyrus appeared to remain relatively unaltered with age, older patients with a history of HIE appear to display a lower differentiation rate into neuronal cells. This may be of clinical importance since stimulation of adult neurogenesis by drugs or exercise may be less efficient in older human brains and could possibly impede functional recovery in older individuals.