Abnormal Expression and Functional Characteristics of Cyclic Adenosine Monophosphate Response Element Binding Protein in Postmortem Brain of Suicide Subjects

SUICIDE IS a major public health concern. Although several studies demonstrate alterations in neurotransmitter receptors (particularly serotonergic and adrenergic receptors) in the postmortem brain of subjects who commit suicide (hereafter referred to as suicide subjects),¹ the precise molecular mechanisms associated with suicidal behavior remain unclear. Physiological responses mediated by protein phosphorylation and subsequent alterations in gene transcription are recently considered some of the most important events in the cellular signaling. In the adenylyl cyclase–cyclic adenosine monophosphate (cAMP) signaling pathway, this role is subserved by enzyme protein kinase A (PKA), which exists as a tetramer holoenzyme that consists of 2 regulatory and 2 catalytic subunits. On occupancy by cAMP, the tetrameric PKA complex dissociates into dimeric regulatory subunits and 2 monomers of catalytic subunits and causes the release of the active catalytic subunits. The catalytic subunits then translocate into the nucleus, where they phosphorylate nuclear substrates, including transcription factors,² and mediate physiological responses.

Given the significance of PKA in phosphorylation events, recent studies have focused on the role of PKA in various psychiatric disorders, including suicidal behavior. For example, Dwivedi et al³ demonstrated a reduced ability of cAMP to bind to the regulatory subunit of PKA in the postmortem brain of suicide subjects. Earlier, Rahman et al⁴ reported lower titrated cAMP binding in various structures of postmortem brains of subjects with bipolar disorder. In addition, Shelton et al⁵ and Manier et al⁶ found lower β-adrenergic–stimulated PKA activity in fibroblasts, and Perez et al⁷ have shown lower expression of PKA regulatory subunits in platelets of depressed patients. Protein kinase A has further been shown to be altered in rat brain after long-term antidepressant treatment.^{8 - 11} These studies thus suggest that PKA plays an important role in affective disorders and suicidal behavior. The functional significance of this alteration in PKA, however, depends on the phosphorylation of specific substrates.

One of the most important substrates of PKA is the transcription factor cAMP response element binding protein (CREB). This 43-kd transcription factor is a member of the leucine zipper family,¹² which binds to the consensus motif 5′-TGACGTCA-3′ found in the promoters of many neuronally expressed genes¹³ and thereby activates or represses the transcription of target genes. Phosphorylation of CREB at serine 133 by PKA is critical for its activation. In this active form, CREB regulates many aspects of neuronal functioning, including excitation of nerve cells,¹⁴ central nervous system development,¹⁵ and long-term synaptic plasticity.¹⁶ In addition, recent studies showing that brain-derived neurotrophic factor–induced cell survival depends on CREB activation and that CREB regulates the expression of brain-derived neurotrophic factor drew attention to the role of this transcription factor in neuroprotection and survival of neurons.¹⁷

Recently, CREB has been implicated in the mechanism of action of antidepressants such that long-term treatment with several classes of antidepressants in rats increases the expression of CREB in the brain.¹⁸ Furthermore, results of a postmortem brain study suggest that CREB expression is increased in the temporal cortex of antidepressant-treated depressed subjects and is decreased in antidepressant-free depressed subjects.¹⁹ This finding suggests that CREB may be involved in depressive behavior. Very recently, Conti et al²⁰ have shown that CREB is critical to target gene regulation after long-term antidepressant treatment. Because a significant population of suicide subjects consists of patients with affective disorders, and because abnormalities in PKA were found in these subjects as well as in the postmortem brains of suicide subjects, it is critical to examine the status of CREB in suicidal behavior. CREB status becomes more important in view of recent findings that suggest that activation of protein kinase C²¹ and extracellular signal-regulated kinase 1/2 (ERK1/2)²² is reduced in the postmortem brains of suicide subjects, and several studies suggest that not only PKA but also serine/threonine kinases such as protein kinase C,²³ ERKs,²⁴ and calciumcalmodulin kinase (CaM) kinase II and IV²⁵ mediate their functional responses through phosphorylation of CREB.

In the present study, therefore, we have extensively examined various aspects of CREB by determining the messenger RNA (mRNA) and protein expressions and functional characteristics by examining cAMP response element (CRE)-DNA binding activity in the prefrontal cortex and hippocampus of suicide subjects. In addition, we have determined catalytic activity of PKA in these subjects, as translocation of PKA is closely associated with activation of CREB. We have also investigated thoroughly whether the alterations in CREB expression are related to depression or are associated with suicide, irrespective of psychiatric diagnosis.

After informed consent, brain tissues were collected by the Brain Collection Program of the Maryland Psychiatric Research Center, Baltimore, in collaboration with the medical examiner's office of the State of Maryland. Brain samples were free of any neuropathological abnormalities or human immunodeficiency virus antibodies. Toxicologic data were obtained by analysis of urine and blood samples from these subjects. Brain pH was measured by homogenizing a piece of cerebellum in 15 volumes of distilled water, and the acidity was measured by means of a pH meter.

The psychiatric diagnosis of suicide or control status was evaluated by 2 senior psychiatrists based on the Diagnostic Evaluation After Death²⁶ and the Schedule for Clinical Interviews for the DSM-IV.²⁷ Family members gave permission for clinical records to be obtained from mental health treatment providers when there was a history of mental health treatment and in all cases of suicide. The present studies were performed in the prefrontal cortex (Brodmann area 9 [BA 9]) and hippocampus obtained from the right hemisphere of brains of 26 suicide subjects and 20 nonpsychiatric controls who died of natural or accidental causes. After the brain had been removed from the cranium, the area of interest was defined from the Brodmann atlas. The hippocampal brain area consisted of 0.5-cm coronal slices of the middle of the hippocampus, including dentate gyrus and area CA1 through CA4. The demographic characteristics of control and suicide subjects are given in Table 1. All patients with major depressive disorder had unipolar disease and no psychotic features. One of the suicide subjects with schizoaffective disorder (subject 15) was depressed at the time of death. All 3 suicide subjects with schizophrenia had paranoid schizophrenia. Their ages of onset were 19(subject 22), 18 (subject 23), and 26 years (subject 24). Their duration of illness was 1, 2, and 14 years, respectively. The protocols for tissue sampling and retrospective assessments were approved by the institutional review boards of the University of Maryland, Baltimore, and the University of Illinois at Chicago.

Nuclear fractions from brain tissues were extracted as described earlier.²² We used commercially available oligonucleotides (Stratagene, La Jolla, Calif) incorporating regulatory elements of the CREB sequence. The probe was end labeled with radioactive phosphorus (γ³²P)–tagged adenosine triphosphate using thyroxine polynucleotide kinase. Binding reactions were performed by incubating 10 µg of nuclear extract with 1 µg of homopolymers of polydeoxyinosinic acid and polydeoxycytidylic acid (poly[dI-dC]) and 6 µg of bovine serum albumin in a reaction mixture (20mM HEPES [pH, 7.9], 1mM dithiothreitol, 0.3mM EDTA, 0.2mM ethyleneglycotetraacetic acid, 80mM sodium chloride, 10% glycerol, and 0.2mM phenylmethylsulfonyl fluoride[PMSF]) for 15 minutes at room temperature. Approximately 40 000 disintegrations per minute of ³²P-labeled CREB oligonucleotide were added and incubated for another 30 minutes. A 4.0% nondenaturing polyacrylamide gel was used to resolve DNA-protein complexes. We quantitatively estimated bands of the DNA-protein complex on the autoradiogram. To show the specificity of CREB-DNA binding activity, nuclear extract from BA 9 of a control subject was incubated in the presence of excess CREB antibody (0.76 µg) for 18 hours at 4°C and then used for the gel mobility shift assay, which resulted in the formation of a supershift band on the gel mobility assay (Figure 1A). The specificity of the CREB-DNA binding activity was also determined in the presence of unlabeled CRE oligo and mutated CREB oligonucleotide (AC→TG substitution in the CRE binding domain) (Santa Cruz Biotechnology Inc, Santa Cruz, Calif). The CRE band was completely blocked by addition of unlabeled CRE oligo, but was not influenced by mutated CRE (Figure 1B).

Basal and cAMP-dependent PKA activity was determined in nuclear fractions.²⁸ This procedure is based on the phosphorylation of a specific substrate, kemptide (Leu-Arg-Arg-Ala-Ser-Leu-Gly), using the transfer of γ-phosphate of γ³²P-labeled adenosine triphosphate by PKA in the presence and absence of 100µM cAMP. The phosphorylated substrate was separated from residual γ-³²P-labeled adenosine triphosphate by spotting 20 µL of reaction mixture onto P81 phosphocellulose paper washing with 75mM orthophosphoric acid and finally with acetone. Data are expressed as picomoles of γ³²P-labeled phosphate transferred to kemptide per minutes per milligram of protein.

The procedure for Western blotting has been described in detail.²² Equal volumes of nuclear fractions containing 15 mg of protein were electrophoresed on 10% (weight to volume ratio) polyacrylamide gel. Total CREB was detected using antiserum C-21 (overnight at 4°C at a dilution of 1:3000) raised against epitopes corresponding to amino acids 295 to 321 of the carboxy terminus of human CREB-1, which detects CREB regardless of its phosphorylation state (Santa Cruz Biotechnology Inc). The specificity of the antiserum samples was determined by using 100-fold excess blocking peptide (relative to the molarity of the antiserum) corresponding to the epitope used to generate CREB antiserum. To examine whether the 43-kd band is indeed CREB, we simultaneously electrophoresed the nuclear fractions obtained from 2 different cell lines (Jurkat and A431) that express CREB and human postmortem brain samples. Both cell lines and the postmortem brain samples showed immunolabeling at 43 kd after probing with CREB antibody (Figure 2B). This finding suggests that the 43-kd band is CREB protein. To reduce the interblot variability, the membranes were subjected to stripping buffer (Chemicon International Inc, Temecula, Calif) and incubated with β-actin antibody (1:3000 dilution; 2 hours at room temperature). The optical density of each sample was corrected by the optical density of the corresponding β-actin band.

Total RNA was isolated by means of cesium choride ultracentrifugation.²² The yield of the total RNA was determined by measuring the absorbency of an aliquot of the precipitated stock at a wavelength of 260/280 nm. Samples with a ratio below 1.8 were rejected. We assessed the degradation of mRNA using denaturing agarose gel electrophoresis and evaluating the sharpness of 28S and 18S rRNA bands.

We used internal standards to determine the quantitation of CREB and neuron-specific enolase (NSE) mRNA. Cloning and synthesis of internal standards have been described in our previous publication.²² Primer pairs 5′GTTCAGTCTTCCTGTAAGGAC (forward) and 5′GTTAGCCAGCTGTATTGCTCC (reverse) were designed to allow amplification of 256 to 585 base pairs (bp) for CREB²⁹ (GenBank accession number X55545), and primer pairs 5′GGGACTGAGAACAAATCCAAG (forward) and 5′CTTCCAAGGCTTCACTGTTCTC (reverse) for amplification of 295 to 675 bp for NSE³⁰ (GenBank accession number M22349). The internal primer sequences for CREB and NSE were 5′GAATGACAGATCTTCTGATGCACC (414-437 bp) and 5′GGCAACAA GCTCGAGATGCAGGAGTTC(478-504 bp), respectively. The underlined bases indicate the Bgl II and XhoI restriction sites for CREB and NSE, respectively, whereas boldface italicized bases indicate the mutation sites. Decreasing concentrations of CREB (100-6.25 pg) or NSE (50-3.125 pg) internal standard complementary RNAs were added to 1 µg of total RNA. The polymerase chain reaction (PCR) mixture was amplified for 32 cycles. After amplification, aliquots were digested with BglII (CREB) or with XhoI (NSE) in triplicate and run by means of 1.5% agarose gel electrophoresis. The results were calculated as the counts incorporated into the amplified complementary RNA standard divided by the counts incorporated into the corresponding mRNA amplification product vs the known amount of CREB or NSE internal standard added to the test sample. The results are expressed as attomoles per microgram of total RNA.

Data analyses were performed using the SPSS 8.0 statistical software (SPSS Inc, Chicago, Ill). All values are reported as mean ± SD. We analyzed differences in mRNA and protein levels of CREB, CRE-DNA binding activity, PKA activity, age, and postmortem interval (PMI) between suicide and control subjects using the independent-sample t test. The relationships among mRNA and protein levels of CREB, CRE-DNA binding activity, PKA activity, and PMI, age, and sex were determined by means of Pearson product moment correlation analysis. P values were 2 tailed. Statistical differences among suicide subjects with a history of major depression, those with a history of other psychiatric disorders, and controls were evaluated by means of 1-way analysis of variance. Multiple comparisons were conducted, and we report unadjusted P values. We have indicated the Bonferroni-adjusted α level and the rationale for its adjustment.

The detailed demographic characteristics of control and suicide subjects are given in Table 1. We found no significant differences in age (t₄₄ =0.59 [P = .56]) or PMI (t₄₂ = 0.072 [P = .94]) between control and suicide subjects. The ratio of male to female subjects in the control group was 16:4, whereas in the suicide group the ratio was 17:9. Brain pH was determined to assess the agonal state. The mean brain pH of control and suicide subjects were 6.1 ± 0.4 and 6.1 ± 0.36, respectively, which was not significantly different (t₄₄ = 0.53 [P = .59]).

Figure 3A shows findings of a representative gel electrophoresis of the competitive reverse transcriptase–PCR (RT-PCR) of CREB in BA 9 from 1 control, whereas Figure 3B shows results of competitive RT-PCR analysis for CREB, where the point of equivalence represents the amount of CREB mRNA present. Messenger RNA levels of CREB were significantly decreased in BA 9 and hippocampus of suicide subjects compared with controls (Table 2). In the BA 9 and hippocampus, the magnitude of decrease in CREB mRNA expression was approximately 50%.

To establish whether neuronal RNA contributes equally to the total RNA pool, we determined the mRNA level of NSE in BA 9 of suicide and control subjects. Results of a representative gel electrophoresis of quantitative RT-PCR of NSE in BA 9 from 1 control is shown in Figure 3C. Results of representative competitive RT-PCR analysis for NSE are presented in Figure 3D, where the point of equivalence represents the amount of NSE mRNA present. Comparison studies show that there were no significant differences in mRNA levels of NSE between control and suicide subjects in BA 9 (Table 2). The ratio of CREB vs NSE mRNA shows that the CREB mRNA level was still significantly decreased in BA 9 of suicide subjects when expressed as a function of the respective NSE mRNA content to correct for nonspecific loss of mRNA owing to a putative neuronal damage (Table 2).

Immunolabeling of CREB was determined in nuclear fraction. The purity of the nuclear fraction was determined by examining the presence of nuclear marker histone H2B and cytosol marker PKA RII subunit in nuclear and cytosol fractions. The immunolabeling of histone H2B was intense in the nuclear fraction, whereas there was no apparent immunolabeling in the cytosol fraction. On the other hand, the immunolabeling of the PKA RII subunit was intense in the cytosol fraction, whereas there was no PKA RII subunit in the nuclear fraction (data not shown). To examine whether CREB was specifically localized in the nuclear fraction, we immunolabeled CREB in cytosol and nuclear fractions and observed that the immunolabeling was present only in the nuclear fraction (Figure 2A).

A representative Western blot showing the immunolabeling of CREB protein in BA 9 from 2 suicide and 2 control subjects is given in Figure 2B, and mean values of ratio of optical density of CREB and β-actin in BA 9 and hippocampus of control and suicide subjects are given in Table 2. We observed that immunolabeling of CREB was significantly decreased in nuclear fractions of the BA 9 and hippocampus obtained from suicide subjects. As with mRNA levels of CREB, the magnitude of decrease in CREB protein was at the level of approximately 50%.

Since we observed decreases in mRNA and protein expression of CREB in BA 9 and hippocampus of suicide subjects, we next examined the functional status of CREB by determining CRE-DNA binding activity using a gel mobility shift assay. A representative autoradiogram showing CRE-DNA binding activity in BA 9 of 4 control and 4 suicide subjects is depicted in Figure 1C. As can be seen in Figure 1C and Table 2, CRE-DNA activity was decreased in the nuclear fraction of BA 9 and hippocampus of suicide subjects compared with controls. The extent of this decrease was approximately 37% to 40%.

To examine the interrelationships between decreased mRNA and protein expression of CREB and decreased CRE-DNA binding activity, we correlated mRNA levels of CREB with CREB protein levels and CREB protein levels with CRE-DNA binding activity. We observed a significant correlation between CREB mRNA and protein levels in the cortex (r = 0.73; P<.001) and hippocampus (r =0.62; P<.001). We also observed that CREB protein levels and CRE-DNA binding activity were significantly correlated in the cortex(r = 0.42; P = .006) and hippocampus (r = 0.42; P =.004).

Since PKA is one of the physiological kinases that phosphorylates CREB, and because translocation of catalytic subunits of PKA from cytosol to nucleus has been shown to be closely related to CREB phosphorylation, we determined basal and cAMP-stimulated PKA activity in the same nuclear fraction in which we determined the protein levels of CREB and CRE-DNA binding activity. Initial observation shows that PKA activity was higher in the hippocampus than in BA 9. This is consistent with our previous results in membrane and cytosol fractions of rat brain.²⁸ The results given in Table 2 show that basal and cAMP-stimulated PKA activity were significantly decreased in BA 9 and hippocampus of suicide subjects compared with controls.

We evaluated the effects of confounding variables such as age, PMI, brain pH, and sex with respect to mRNA and protein levels of CREB, CRE-DNA binding activity, and basal and cAMP-stimulated PKA activity. We did not observe any significant effects of age in the cortex (CREB mRNA level, r = 0.05 [P = .71]; CREB protein level, r = 0.13 [P = .38]; CRE-DNA binding activity, r = 0.15 [P =.29]; basal PKA activity, r = 0.069 [P = .65]; and cAMP-stimulated PKA activity, r =0.001 [P = .99]) or the hippocampus (CREB mRNA level, r = 0.001 [P = .99]; CREB protein level, r = 0.22 [P = .88]; CRE-DNA binding activity, r = 0.12 [P = .41]; basal PKA activity, r = 0.12 [P = .41]; and cAMP-stimulated PKA activity, r = 0.11 [P = .39]). We also did not observe any significant effects of PMI in the cortex (CREB mRNA level, r = 0.11 [P = .47]; CREB protein level, r = 0.01 [P = .92]; CRE-DNA binding activity, r = 0.9 [P = .52]; basal PKA activity, r = 0.01 [P = .95]; and cAMP-stimulated PKA activity, r =0.03 [P = .85]) or the hippocampus (CREB mRNA level, r = 0.85 [P = .58]; CREB protein level, r = 0.14 [P = .37]; CRE-DNA binding activity, r = 0.15 [P = .34]; basal PKA activity, r = 0.15 [P = .30]; and cAMP-stimulated PKA activity, r = 0.03 [P = .81]). Controls included 17 men and 9 women; suicide subjects, 16 men and 4 women. The correlation analysis showed no significant relationship between sex and various measures in the cortex (CREB mRNA level, r = 0.20 [P = .17]; CREB protein level, r = 0.93 [P = .54]; CRE-DNA binding activity, r = 0.03 [P = .81]; basal PKA activity, r = 0.15 [P = .31]; and cAMP-stimulated PKA activity, r = 0.12 [P =.42]) or in the hippocampus (CREB mRNA level, r =0.17 [P = .25]; CREB protein level, r = 0.04 [P = .80]; CRE-DNA binding activity, r = 0.18 [P = .22]; basal PKA activity, r = 0.14 [P =.34]; and cAMP-stimulated PKA activity, r = 0.14[P = .35]). Similarly, we found no significant correlation between brain pH and CREB mRNA level (cortex, r =0.12 [P = .41]; hippocampus, r = 0.11 [P = .46]), basal PKA activity (cortex, r = 0.09 [P = .56]; hippocampus, r = 0.21 [P = .45]), cAMP-stimulated PKA activity (cortex, r = 0.26 [P = .08]; hippocampus, r = 0.06 [P = .66]), CREB protein level (cortex, r =0.17 [P = .24]; hippocampus, r = 0.07 [P = .62]), or CRE-DNA binding activity (cortex, r = 0.07 [P = .63]; hippocampus, r = 0.15 [P =.34]).

To examine whether the differences in mRNA and protein levels of CREB, CRE-DNA binding activity, and PKA activity between control and suicide subjects were related to depression or were present in all suicide subjects irrespective of diagnosis, we examined the effect of major depression on these measures(Table 3). For this purpose, we divided suicide subjects into those with a diagnosis of major depression that was active at the time of death, those with diagnoses of other psychiatric disorders, and those with no psychiatric illness. Of the 26 suicide subjects, 11 had a diagnosis of major depression; 5, adjustment disorders; 3, schizophrenia; 2, schizoaffective disorder; and 3, no psychiatric illness. In the other 2 suicide subjects, the diagnosis was not available. As given in Table 3, mRNA and protein levels of CREB and CRE-DNA binding activity were not different between suicide subjects with major depression and those with other psychiatric disorders. However, both groups showed significant differences compared with controls in both BA 9 and hippocampus (Table 3). On the other hand, PKA activity showed a different pattern. Basal and cAMP-stimulated PKA activity were significantly decreased only in those suicide subjects who had a history of major depression compared with controls. No significant differences in PKA activity were observed between suicide subjects who had a history of other psychiatric disorders compared with controls.

We next determined whether these decreases were related to antidepressant treatment. Of the 11 suicide subjects with major depression, 4 showed antidepressant toxicologic findings at the time of death (subjects 7-10). We observed that the mean mRNA and protein levels of CREB, CRE-DNA binding activity, and PKA activity were virtually identical among depressed suicide subjects with positive findings for antidepressant. Give the small number of suicide subjects with antidepressant toxicologic findings, it is premature to state that antidepressants do not influence CREB and PKA.

This investigation led to interesting findings such that the absolute amount of CREB mRNA was significantly decreased in the prefrontal cortex and hippocampus of suicide subjects. Because basic regions in CREB direct its translocation from the cytoplasm to the nucleus (irrespective of its state of dimerization),³¹ where it binds to the promoter region of the CRE-containing gene and causes transcriptional activation, we determined protein levels of CREB and the ability of CREB to bind DNA-containing consensus CRE. We observed that protein levels of CREB and CRE-DNA binding activity were significantly decreased in BA 9 and hippocampus of suicide subjects.

In the nuclear fraction similar to that in which we determined CREB protein levels, we examined PKA activity, because PKA catalytic subunits translocate to the nucleus after dissociation from regulatory subunits on cAMP stimulation and phosphorylate and activate CREB. In the nuclear fraction of the prefrontal cortex and hippocampus, we observed that basal and cAMP-stimulated PKA activity were significantly decreased in suicide subjects. These results thus suggest that PKA activity, CREB expression, and functional characteristics of CREB are abnormal in the postmortem brains of suicide subjects. Previous studies have shown that CREB is expressed in nuclei of almost all of the cell types in the brain with no apparent regional specificity.^{32 - 33} In contrast, catalytic subunits of PKA are also expressed throughout the brain, with the highest expression in neuronal cell layers,³⁴ although glial localization has also been reported.³⁵

Recent studies suggest that CREB may be involved in depression,³⁶ as CREB expression is decreased in the temporal cortex of depressed subjects.¹⁹ Long-term treatment of rats with several classes of antidepressants increases the expression of CREB and CRE-DNA binding activity in the hippocampus.¹⁸ Chen et al³⁷ have shown recently that long-term antidepressant treatment induces CRE-mediated gene expression in a neuroanatomically differentiated pattern in the rat brain and that the overexpression of CREB in the hippocampus results in an antidepressant effect in learned-helpless rats. These findings raise the possibility that CREB expression and CREB-mediated functions may be decreased in depression and that antidepressant treatment may alleviate depression by reversing this decrease. To investigate whether the observed decrease in CREB expression and CRE-DNA binding activity in the postmortem brain of suicide subjects was related to depression, we divided suicide subjects into those who had a history of major depression and those who had other psychiatric disorders. Messenger RNA and protein expression of CREB and CRE-DNA binding activity were decreased in all suicide subjects regardless of diagnosis, including those suicide subjects who had no psychiatric illness. Our results thus suggest that the decreased expression of CREB may not be related specifically to major depressive disorder but seems to be a suicide-related phenomenon. This conclusion is further supported by our observation that CREB expression was decreased, even in the postmortem brains of suicide subjects with schizophrenia, which is quite contrary to the findings of Dowlatshahi et al.¹⁹ Their report found that CREB expression was not altered in the temporal cortex of these patients. Changes in CREB may be related to stress, as there is a strong relationship between stress and suicidal behavior,^{38 - 39} and a dysregulated stress system has been shown in suicide subjects in studies showing increased corticotropin-releasing hormone levels in cerebrospinal fluid,⁴⁰ altered ratio of corticotropin-releasing hormone I/II receptor mRNA in the pituitary,⁴¹ down-regulation of corticotropin-releasing factor receptors in the prefrontal cortex,⁴² and increased levels of the precursor molecule of corticotropin mRNA in the pituitary⁴³ of suicide victims. Glucocorticoids suppress CREB phosphorylation and activation in neurons containing corticotropic-releasing hormone in rat brain,⁴⁴ and decreased CREB response in suicide subjects may be related to elevated cortisol levels; however, further studies are needed to confirm this speculation.

Protein kinase A activity was decreased only in those suicide subjects who had a history of major depression. This finding suggests that PKA-mediated functions may be abnormal in depression and that this abnormality may not be related to suicide. Using commercially available phospho-CREB antibodies, we were unable to detect the phosphorylated form of CREB. Nonetheless, decreased PKA activity suggests a strong possibility that PKA-mediated phosphorylation, including CREB, may be decreased, at least in those suicide subjects with a history of major depression. However, this does not exclude the possibility that CREB phosphorylation may not be altered in the postmortem brain of other suicide subjects, as other serine/threonine kinases such as protein kinase C,²³ ERK1/2,²⁴ and CaM kinase II and IV²⁵ have been shown to be involved in phosphorylating and/or activating CREB, and we have previously reported that activation of protein kinase C²¹ and ERK1/2²² is abnormal in the postmortem brain of suicide subjects.

The precise mechanism behind the decreased CRE-DNA binding activity and expression of CREB in the postmortem brain of suicide subjects is unclear; however, the most likely possibility for decreased CRE-DNA binding activity seems to be related to a low availability of CREB protein. Another factor that may be responsible for decreased CRE-DNA binding activity could be decreased levels of CREB-binding proteins, as CREB-binding protein has been shown to regulate CRE-DNA binding activity by interacting with phosphorylated CREB.⁴⁵ On the other hand, the regulation of the mRNA expression of CREB is quite complex. There are several mechanisms through which CREB transcription can be regulated. One such mechanism is through PKA, which phosphorylates CREB and other CREB/activating transcription factor–like proteins, some of which are involved in regulation of the transcription of CREB. The decreased catalytic activity of PKA, as observed in the present investigation, may cause the decreased phosphorylation of CREB/ATF-like proteins, and thereby cause the decreased expression of CREB. Also, CREB is constitutively bound to gene promoters, and CREB phosphorylation causes conformational changes and, consequently, recruits CREB-binding proteins and regulates gene transcription.⁴⁶

Regardless of the mechanism, given the critical role played by CREB in modulating gene transcription and thereby the regulation of neuronal functioning, our findings of decreased expression and functional characteristics of CREB in the brain of suicide victims may be important, as the hippocampus has been implicated in cognitive behavior, control of emotion, learning, and memory, and stress-related psychiatric disorders and other vegetative processes, whereas the dorsolateral prefrontal cortex (BA 9) plays a role in mood regulation⁴⁷ and has been associated with several neurobiological abnormalities implicated in the suicidal behavior and affective disorders. In addition, recent studies suggest that CREB is a possible regulator of a general survival program in neurons⁴⁸ that occurs through brain-derived neurotrophic factor, whose transcription is regulated by CREB.^{49 - 50} It has been shown that stress decreases the expression of brain-derived neurotrophic factor in rat hippocampus.⁵¹ Furthermore, atrophy of the hippocampal neurons during stress in the rat brain and reduced hippocampal and prefrontal cortical volume in depressed patients^{52 - 55} have been reported. Because stress plays a role in suicidal behavior, and a significant population of suicide subjects shows a history of affective disorders, reduced CREB may contribute to the atrophy of the neurons in the brains of suicide subjects.

In conclusion, given the significance of CREB in mediating various physiological functions through gene transcription, our findings of decreased expression and functional characteristics of CREB in the postmortem brain of suicide raise a strong possibility that CREB serves as an important vulnerability factors in suicidal behavior.

Corresponding author and reprints: Yogesh Dwivedi, PhD, Psychiatric Institute, Department of Psychiatry, University of Illinois at Chicago, 1601 W Taylor St, Chicago, IL 60612 (e-mail: ydwivedi@psych.uic.edu).

Submitted for publication June 1, 2002; final revision received July 18, 2002; accepted July 24, 2002.

This work was supported by career development award KO1 MH 01836 (Dr Dwivedi) and grant RO1MH48153 (Dr Pandey) from the National Institute of Mental Health, Rockville, Md; and a Young Investigator Award from the American Foundation For Suicide Prevention (Dr Dwivedi), New York, NY.

We thank John Smialek, MD, chief medical examiner, and Dennis Chute, MD, assistant medical examiner, for their cooperation in the collection of brain samples; Terri U'Prichard, MA, for performing the psychological autopsies; Boris Lapidus, MD, for the dissection; and Barbara Brown, BS, and Miljana Petkovic, MS, for their help in organizing the brain tissues.