DNA Fragmentation Decreased in Schizophrenia but Not Bipolar Disorder

DURING THE past 20 years, remarkable progress has been made in identifying changes in the brain related to the pathophysiology of schizophrenia and, more recently, bipolar disorder. Both structural brain imaging¹ and postmortem² studies have demonstrated volume loss in subjects with schizophrenia. Similarly, a reduction in the numerical density of neurons has also been demonstrated in several regions, including the anterior cingulate cortex^{3 - 4} and hippocampal formation⁵ of subjects with schizophrenia and bipolar disorder. Because no study to date has presented quantitative evidence for an increase in astroglial cells, it has been concluded that the major psychotic disorders are probably not neurodegenerative in nature.⁶ It is important to emphasize, however, that shrinkage and neuronal loss can result not only from a necrotic process like that seen in Alzheimer disease, but also from apoptosis, a process that is associated with the normal turnover of cells during development.^{7 - 10}

A dysregulation of apoptotic mechanisms is believed to play a role in a variety of neuropsychiatric disorders,^{11 - 12} even those like Alzheimer disease that show a preponderance of necrotic changes.¹³ Apoptosis and necrosis can be distinguished histopathologically from one another on the basis of their energy requirements, histological profiles, plasma membrane changes, phagocytic mechanisms, presence or absence of an inflammatory reaction, and patterns of DNA breakdown.¹⁴ The damage to DNA noted in apoptosis is believed to involve a DNA fragmentation factor that is turned on by activated caspase enzyme.¹⁵ The resulting DNA nicks that are induced in the nucleotide strands can appear as either single-stranded or double-stranded breaks.¹⁶ If a cell is unable to repair such damage with a DNA polymerase enzyme, these structural DNA changes may persist indefinitely in surviving neurons. In the current study, we have assessed the presence of DNA fragmentation to ascertain whether apoptosis might play a significant role in the pathophysiology of schizophrenia and bipolar disorder. To this end, we have applied a method for in situ end-labeling (ISEL) of single-stranded DNA breaks using the Klenow fragment of DNA polymerase I¹⁷ in anterior cingulate cortex.

The cohort was obtained from the Harvard Brain Tissue Resource Center at McLean Hospital and consisted of tissues from 18 healthy controls, 18 schizophrenic subjects, and 10 subjects with bipolar disorder matched for age, postmortem interval, sex, and freezer storage time (Table 1). The tissue from 1 case was lost during processing, reducing the number of schizophrenic samples to 17. Exposure to neuroleptic drugs was assessed, using the chlorpromazine-equivalent dose (in milligrams) during the year prior to death as an index. Five subjects in the bipolar disorder group were either neuroleptic-naive or neuroleptic-free at the time of death, and only one of the schizophrenic subjects was neuroleptic-free. As presented in Table 2, the average age, postmortem interval, and sex ratios were remarkably similar for the 3 groups. The freezer storage times were also matched across the 3 groups, and this is reflected in the respective B-numbers that were assigned to each brain at the time it was donated (Table 1). Neuroleptic exposure was expressed as the average±SD chlorpromazine-equivalent dose for the subjects with schizophrenia (420 ± 839 mg) and the subjects with bipolar disorder (339 ± 416 mg) during the year prior to death. All tissues included in this study were subjected to a neuropathological evaluation at both the gross and microscopic levels. Specifically, the presence of senile plaques and neurofibrillary tangles was assessed in cases obtained prior to 1999 using the CERAD¹⁸ criteria, and thereafter, using the Braak criteria.¹⁹

Single-stranded DNA breaks were visualized using the so-called Klenow method for ISEL. The Klenow method was selected over the terminal deoxynucleotidyl transferase–mediated biotin-deoxyuridine triphosphate nick end-labeling (TUNEL) method for localizing double-stranded DNA breaks because previous work in other laboratories has suggested that it may be more sensitive for assessing apoptotic changes²⁰ in the cortex,¹⁷ although the 2 techniques show a high degree of spatial²¹ and temporal¹⁷ overlap in the distribution of staining in some regions such as the hippocampus. The Klenow technique has also been found to show equivalent sensitivity to the localization of the DNA-binding domain of poly (adenosine diphosphate–ribose) polymerase as a probe for detecting DNA breaks in intact cells undergoing apoptosis.²² Briefly, fresh blocks of anterior cingulate cortex (Brodmann area 24) that were obtained at a point rostral to the genu of the corpus callosum were prepared as previously described.²³ The tissue blocks were sectioned on a cryostat at a thickness of 10 µm, mounted on glass slides, postfixed with 4% paraformaldehyde in 0.1M phosphate buffer, treated with Proteinase K/Cytopore, preincubated in hydrogen peroxide, and incubated in a mixture containing a compound of the Klenow fragment of DNA polymerase I and biotinylated–uridine triphosphate. A standard avidin-biotin peroxidase reaction was used to visualize the ISEL. All slides were run in parallel and were codified to conceal the identity of the subjects.

As shown in Figure 1, ISEL-positive cells in the tissue sections were identified by the presence of a characteristic brown staining in the nuclei. The characteristic brown diaminobenzidine (DAB; Ventana Medical Systems Inc, Tucson, Ariz) reaction product appeared either as either (1) diffuse nuclear staining, (2) chromatin clumps, or (3) nuclear blebs. In most cases, the vast majority of nuclei showed no ISEL-positive staining. The slides were codified and analyzed under strictly blinded conditions using a Leitz Laborlux (LEICA Microsystems, Wetzlar, Germany) bright-field microscope equipped with a solid-state video camera interfaced to a Bioquant Image Analysis System. Initially, a column of cortex (width, 300 µm) was delineated using a 4× objective lens, and the sampling field extended across the 6 layers from the pial surface above layer I to the interface of layer VI with the underlying white matter. Using a 40× objective lens, the position of each ISEL-positive nucleus was marked using a X,Y,Z-encoder (Boeckler Instruments, Petershagen, Germany) interfaced with the BIOQUANT System (BIOQUANT Image Analysis Corp, Nashville, Tenn) (Figure 2). The number of nuclei showing staining that was diffuse or that appeared as chromatin clumps or nuclear blebbing was determined for each layer of each case (Figure 2). Although there was no specific way of distinguishing between neurons and glia in the ISEL preparation, most of the labeled nuclei were larger than those typically seen in glial cells, and this suggested that the data reported herein may have been derived from neuronal cells, particularly since few if any labeled cells were found in the subcortical white matter. The data were expressed as a numerical density (ie, the number of nuclei per square millimeter of sampling field) for each type of ISEL-positive nucleus. The latter data for the respective cases were averaged across the 3 groups and expressed as a mean (SE). Using a frequency histogram analysis, the data were found to be distributed in a nonnormal manner. Accordingly, the nonparametric Kruskal-Wallis test was used to assess the significance of differences in the means across the 3 groups. A 2 × 2 contingency table analysis was also used to evaluate differences in the distribution patterns.

As shown in Figure 3, there was a significant reduction (71%) in the numerical density of ISEL-positive chromatin clumps in the schizophrenia group when compared with either the healthy control or bipolar groups (P = .001). The numerical density of nuclei showing either diffuse staining or staining of nuclear blebs showed no differences among the 3 groups. When the data were broken down according to layer, there was an 80% reduction in the numerical density of nuclei showing ISEL-positive chromatin clumps in layers III (P = .03) and V or VI (P was not significant). In layer II, nuclei showing diffuse ISEL-positive staining and positive chromatin clumps were reduced in the subjects with schizophrenia and in those with bipolar disorder, but these differences were not significant. For ISEL-positive nuclear blebs in layer II, there was also a much lower density in the subjects with schizophrenia; however, the difference was not significant.

When the data were expressed in the form of frequency distributions, the subjects with schizophrenia had many more cases with no ISEL-positive nuclei. In the control and bipolar groups, approximately 30% to 40% of the subjects showed no labeling of chromatin clumps, while 76.5% of the subjects with schizophrenia showed an absence of this staining. Using a 2 × 2 contingency table analysis, the distribution of data for the schizophrenia group was significantly shifted to the left when compared with the healthy controls (χ² = 4.37; P = .04); however, the subjects with bipolar disorder did not show a significant difference when compared with the controls (χ² = 0.34; P = .56).

As noted previously, the 3 groups were well matched for age, postmortem interval, sex ratio, and freezer storage time, and it seems quite unlikely that the differences in the subjects with schizophrenia could be explained by any confounding effects related to these variables. The potential relationship between the numerical density of nuclei showing ISEL-positive staining was also evaluated with respect to neuroleptic exposure. Using simple correlations, the chlorpromazine-equivalent dose showed no relationship between ISEL staining and the data for both the schizophrenia and bipolar groups. Additionally, the bipolar group was broken down according to subjects who were neuroleptic-free or neuroleptic-naïve, and then compared with those who were treated with these agents. The averages for these subgroups showed no difference with respect to exposure to antipsychotic drugs. Another potential confounding effect that was considered was the presence of Alzheimer disease in the healthy controls and in subjects with bipolar disorder. There were cases in both groups (3 and 2 subjects, respectively) showing a minimal number of plaques and/or tangles. The schizophrenia group (n = 5) also showed such changes, though these were of moderate severity. None of the subjects showing plaques and tangles received a neuropathological diagnosis of senile dementia of the Alzheimer type. Furthermore, the data for subjects with and without plaques and tangles did not show differences that could account for the findings reported here (eg, for the controls, the average densities of nuclei with ISEL-positive clumps in layer III were 3.5 and 2.2, respectively, and for the schizophrenic subjects, these average values were 0.26 and 0.53, respectively).

In summary, we report here for the first time (to our knowledge) that there is a selective decrease of single-stranded DNA breaks in neurons of anterior cingulate cortex in schizophrenic subjects, but not in subjects with bipolar disorder. Since both patient groups were well matched for age, postmortem interval, sex ratio, and freezer storage time, it seems unlikely that these variables could account for the differences observed in the subjects with schizophrenia. Moreover, a comparison of neuroleptic-treated and neuroleptic-free subjects with bipolar disorder demonstrated no differences in ISEL staining, suggesting that the decreased occurrence of single-stranded DNA breaks in the subjects with schizophrenia may not be due to a medication effect. Finally, the possibility that the current results may be attributable to the presence of Alzheimer disease in the healthy controls and bipolar subjects has been ruled out by detailed neuropathological evaluation and by the observation that Alzheimerlike changes were not associated with any difference in the distribution of ISEL labeling.

The 3 nuclear patterns of ISEL labeling observed in this study seemed to represent different stages of apoptosis. First, because the diffusely stained nuclei were generally larger and showed no clumping, this was interpreted as being an early stage of DNA fragmentation — one in which significant chromatin condensation had not yet occurred. Second, the nuclei showing labeled clumps seemed to represent a stage in which discrete regions of DNA fragmentation and/or condensation of chromatin had occurred. This latter stage is not associated with DNA laddering, a hallmark feature of end-stage apoptotic cell death.²⁰ Finally, those nuclei showing ISEL-positive blebbing of the nuclear envelope appeared to be in a late stage of DNA fragmentation—one in which the degree of positive staining was much greater than in the 2 previous categories. The aggregation of labeled chromatin around the periphery of the nucleus and the obvious protrusions into the surrounding cytoplasm suggested that the nuclei were in a state of dissolution. Taking these observations together, the ISEL-positive chromatin clumps might represent an intermediate phase of DNA fragmentation, one in which either cell survival or death may be possible sequelae. Like the labeled clumps, the density of nuclei showing ISEL-positive blebs was somewhat lower in the schizophrenia group, suggesting the possibility that a DNA repair mechanism might have arrested the apoptotic cascade.

There are a variety of mechanisms through which a decreased fragmentation of DNA could occur in a disease group as compared with healthy subjects. The first and most obvious possibility is that neurons exposed to oxidative stress could have died and dropped out of the population under study. Support for this possibility comes from the results of a study in which a reduced density of neurons in the anterior cingulate cortexes of subjects with schizophrenia was reported⁴ ; however, the fact that a similar, but even more robust reduction was also noted in subjects with bipolar disorder argues against this possibility. In view of this latter finding, it is particularly striking that the tissues of subjects with bipolar disorder included in the current study did not show any change in DNA fragmentation. If cell death were indeed responsible for the decrease of DNA fragmentation observed in the tissues from subjects with schizophrenia in this study, then a similar and perhaps more striking decrease of ISEL staining should have been observed in the bipolar group as well.

An alternative possibility is that there may be altered regulation of intracellular signaling and mitochondrial pathways that are associated with apoptotic cell death in response to oxidative stress.⁹ A variety of proteins are known to be proapoptotic and include Bax,²⁴ a cytochrome c/Apaf-1 complex,^{25 - 26} and caspase-dependent DNA fragmentation factor.^{27 - 29} Other factors, however, promote cell survival, and include Bcl-2,³⁰ brain-derived neurotrophic factor,^{31 - 32} and glial-derived neurotrophic factor.³³ Both proapoptotic and prosurvival proteins are subject to a variety of regulatory changes in response to oxidative stress.^{34 - 35} In experimental models^{36 - 38} and human degenerative disorders,¹³ changes in the expression of proapoptotic and prosurvival factors are complex and can involve changes in several different factors at any given time. For example, in programmed cell death, Bax is increased and Bcl-2 is decreased,⁹ while in Alzheimer disease, an increase of Bcl-2 expression parallels a similar change for caspase-dependent DNA fragmentation factor.¹³

A recent model of schizophrenia has postulated that intrinsic circuits within the anterior cingulate cortex may be exposed to excessive amounts of excitatory activity, a change that would result in variable degrees of oxidative stress.⁶ Interestingly, recent postmortem schizophrenia studies have demonstrated paradoxical changes in 2 prosurvival factors: a decrease of Bcl-2 expression (proapoptotic) in temporal cortex⁸ and an increase of brain-derived neurotrophic factor expression (prosurvival) in cingulate cortex.³² Several other studies have reported that oxidative enzymes associated with mitochondria, such as the malate shuttle system³⁹ and complex IV or cytochrome c oxidase, are also decreased in the frontal cortexes of subjects with schizophrenia.⁴⁰ The latter change would promote survival if it were accompanied by a decrease of complex formation with APAF-1 and activation of caspase 9.²⁵ In the basal ganglia, a similar reduction of cytochrome c oxidase expression has also been found, but it was negatively correlated with emotional and cognitive impairments detected prior to death.⁴¹ Finally, evidence from both a primate model^{39 ,41} and from postmortem studies³⁹ have indicated that neuroleptic exposure is associated with increases, rather than decreases, of mitochondrial markers for oxidative metabolism. Taking these findings together, decreases in mitochondrial function in subjects with schizophrenia is probably neither epiphenomenal in nature nor related to antipsychotic drug treatment. Indeed, a case can be made for such findings being related to corticolimbic dysfunction in this disorder.

We report a marked reduction of single-stranded DNA breaks in the anterior cingulate cortexes of subjects with schizophrenia, which is a change that may help to explain why subjects with schizophrenia have shown less striking reductions in neuronal density when compared with those with bipolar disorder.⁴ Overall, the findings reported here are consistent with a down-regulation of intracellular signaling and oxidative pathways occurring in subjects with schizophrenia. Although it is tempting to speculate that this dramatic reduction of DNA fragmentation might represent an adaptive cellular response to promote cell survival, it could also represent a failure of cingulate neurons to mount an appropriate response to an oxidative challenge. Unfortunately, available evidence does not allow for a distinction to be made between these 2 mutually exclusive possibilities. Future studies will be directed toward understanding how an altered expression of proteins that promote cell death and survival may be related to neuronal pathology in schizophrenia.

Corresponding author: Francine M. Benes, MD, PhD, McLean Hospital, 115 Mill St, Belmont, MA 02478 (e-mail: benesf@mclean.harvard.edu).

Submitted for publication June 14, 2002; final revision received October 1, 2002; accepted October 4, 2002.

This work was supported by grants MH00423 and MH42261 from the National Institutes of Mental Health, Bethesda, Md.