Heat Shock Protein 12A Shows Reduced Expression in
the Prefrontal Cortex of Subjects with Schizophrenia
Julie L. Pongrac, Frank A. Middleton, Lansha Peng, David A. Lewis, Pat Levitt, and Károly Mirnics
Background: Deoxyribonucleic acid microarray analyses of dorsolateral prefrontal cortex (DLPFC) area 9 from 10 matched pairs of
schizophrenic and control subjects revealed a consistent and significant decrease (p ? .001; mean log2 signal difference ? ?.58) in
transcript expression for a gene clone KIAA0417. This database entry has been recently annotated as two highly homologous members
of a heat-shock protein family (HSPA12A and HSPA12B).
Methods: We followed up our initial results by in situ hybridization in subjects with schizophrenia, major depression, and a chronic
haloperidol-treated nonhuman primate model. Furthermore, we investigated the distribution of HSPA12A and HSPA12B transcripts
across the human and nonhuman primate brain.
Results: We found that HSPA12A (but not HSPA12B) is highly expressed in the human brain and shows a neuron- and region-specific
transcript distribution, with strongest expression in the frontal and occipital cortical regions. HSPA12A messenger ribonucleic acid was
significantly reduced (p ? .01; mean log2 optical density difference ? ?.84) across subjects with schizophrenia but not in the DLPFC
of subjects with major depression or in monkeys chronically treated with haloperidol.
Conclusions: The data are consistent with metabolic alterations in schizophrenia, reflected in selective changes in the expression of
certain genes encoding proteins involved in cellular metabolism or metabolic responsiveness.
haloperidol, in situ hybridization
includes an intricate interplay between genetic and environmen-
tal factors that alters neural function. The dorsolateral prefrontal
cortex (DLPFC) is a brain region that shows consistently altered
microstructure and function in schizophrenia (Lewis and Lieber-
Deoxyribonucleic acid (DNA) microarrays, with their ability
to assess the expression of tens of thousands of genes simulta-
neously (Lockhart et al 1996; Schena et al 1995), are becoming a
method of choice for uncovering complex expression patterns
associated with human brain disorders (Marcotte et al 2001;
Mirnics et al 2001). Transcriptome studies of schizophrenia have
primarily focused on the DLPFC (Hakak et al 2001; Mimmack et
al 2002; Mirnics et al 2000a; Tkachev et al 2003; Vawter et al 2002)
and have revealed complex expression changes in known genes
involved in both neural (Mirnics et al 2000a, 2000b; Pongrac et al
2002; Vawter et al 2002) and glial (Hakak et al 2001; Mimmack et
al 2002; Pongrac et al 2002; Tkachev et al 2003) function;
however, microarrays also contain probes against many partially
annotated transcripts and transcript fragments (e.g., hypothetical
genes, expressed sequence tags), and although virtually all
microarray studies to date reported expression differences in
multiple uncharacterized messenger ribonucleic acids (mRNAs),
none has been systematically examined to date. These expres-
chizophrenia is a complex brain disorder that affects 1% of
the population (Carpenter and Buchanan 1994; Gottesman
1991). The etiology of schizophrenia is complex and likely
sion changes are intriguing, but their interpretation is challenging
because typically no information is available about the function
or anatomical distribution of the gene products. In our initial
microarray study (Mirnics et al 2000a), the hypothetical gene
KIAA0417 (UniGene Hs.12385) showed altered expression in the
DLPFC of subjects with schizophrenia. A recent update to the
UniGene (Wheeler et al 2004) separated and re-annotated the
Hs.12385 cluster into two novel groups of genes: HSPA12A
(Hs.372597) and HSPA12B (Hs.368340). These two novel genes,
based on sequence and domain homology to HSP70, belonged
to the family of 70-kd heat shock proteins (HSP70).
HSP70 family members consist of a highly conserved N-
terminal adenosine triphosphatase domain and a variable C-
terminal substrate-binding domain, which is responsible for the
substrate specificity of different family members (Bukau and
Horwich 1998). Mammalian HSP70 family members are generally
cytoplasmic, although nuclear, mitochondrial, and endoplasmic
reticulum localization of isoforms have been described (for
reviews, see Hohfeld et al 2001; Takayama et al 2003; Voos and
Rottgers 2002). When cellular proteins are denatured under
stress, synthesis, folding, or assembly, HSP70 proteins bind short
stretches of exposed hydrophobic amino acids, thus preventing
them from irreversible aggregation. Once detached from HSP70,
if the misfolded polypeptides do not regain their correct confor-
mation, they will be degraded by the ubiquitin–proteasome
HSPA12A shows a strong domain homology to HSP70
(pfam00012.5 NCBI [National Center for Biotechnology Informa-
tion]); however, this domain is atypical and split into two parts
(Han et al 2003), separated by spacer amino acids 245–311. The
functional significance of this split is currently unknown, but it
raises the possibility of a functional specialization of HSPA12A
within the group of HSP70 proteins. This notion is further
strengthened by the uniqueness of the HSPA12A C-terminal
substrate binding domain (Han et al 2003).
Alterations in HSP70 levels have been associated with a
diverse array of unrelated disorders, ranging from digestive
cancers to neuropsychiatric disorders. Because altered HSP70
expression levels have been implicated in both pathophysiology
(Kilidireas et al 1992; Kim et al 2001; Schwarz et al 1999; Wang et
al 2003) and treatment (Gabriele et al 2003; Garcia-Osta et al
Pennsylvania; Departments of Psychiatry and Neuroscience and Physi-
ology (FAM), State University of New York Upstate Medical University,
Syracuse, New York; and Vanderbilt Kennedy Center for Research on
Human Development (PL), Vanderbilt University, Nashville, Tennessee.
Address reprint requests to Karoly Mirnics, M.D., University of Pittsburgh
15261; E-mail: firstname.lastname@example.org.
Received May 26, 2004; revised August 11, 2004; accepted September 13,
BIOL PSYCHIATRY 2004;56:943–950
© 2004 Society of Biological Psychiatry
2003; Hashimoto et al 1996; Sharp et al 1992, 1994) of schizo-
phrenia, further characterization of changes in heat shock pro-
teins are necessary to understand the precise role they play. To
address how HSPA12A might contribute to the pathophysiology
in schizophrenia we 1) examined the regional distribution of the
two novel HSPA12 isoforms in the human brain; 2) verified the
microarray data using in situ hybridization; and 3) tested these
findings for their diagnostic specificity by examining the expres-
sion of HSPA12A mRNA in subjects with major depressive
disorder (MDD). Finally, we examined the potential of chronic
neuroleptic exposure to modulate HSPA12A transcript levels in a
nonhuman primate model.
Methods and Materials
Characteristics of Human Subjects
All human tissue was obtained under the auspices of the
University of Pittsburgh Conte Center for the Neuroscience of
Mental Disorders Brain Bank. Permission to review medical and
psychiatric records of the subjects and to conduct structured
interviews with relatives was obtained by written informed
consent. These procedures have been approved by the Univer-
sity of Pittsburgh’s Institutional Review Board for Biomedical
Research as part of the Human Brain Bank Core of the Center for
Neuroscience of Mental Disorders (MH45156).
Ten pairs of schizophrenic and control subjects were matched
for gender, except subject pair 557c and 537s. The entire group
of schizophrenic and control subjects did not differ in mean
(?SD) age at time of death (45.2 ? 10.4 years and 44.4 ? 11.9
years, respectively), postmortem interval (PMI; 18.5 ? 6.9 hours
and 17.4 ? 5.2 hours, respectively), brain pH (6.8 ? .3 and 6.8 ?
.2, respectively), or tissue storage time at ?80°C (47.6 ? 10.4
months and 40.4 ? 10.6 months, respectively) at the time the
studies were initiated. Of the subjects with schizophrenia, 8 of 10
were receiving antipsychotic medication at the time of death, and
5 had a history of alcohol abuse or dependence. For a description
of individual demographic data, see Middleton et al 2002. Ten
subjects with MDD without psychosis were matched to control
subjects for gender. These groups did not differ in mean (?SD)
age at time of death (52.7 ? 13.1 years and 52.1 ? 13.1 years,
respectively), PMI (14.9 ? 5.3 hours and 15.7 ? 5.5 hours,
respectively), brain pH (6.81 ? .17 and 6.72 ? .30, respectively),
or tissue storage time at ?80°C (39.0 ? 17.4 months and 39.9 ?
13.2 months, respectively). Of the depressed subjects, 2 had a
history of alcohol dependence, and 6 died by suicide. Consensus
DSM-IIIR diagnosis was determined from clinical records, toxi-
cology studies, and structured interviews with surviving relatives
(Volk et al 2000a).
Microarray Experiments and Data Analysis
Tissue preparation, nucleic acid isolation, sample labeling,
microarray hybridization, and data analysis have been described
previously (Middleton et al 2002; Mirnics et al 2000a). Balanced
differential expression (BDE) values were calculated for each
array by dividing the log2 transformed balanced fluorescent
signal intensity of a control subject with signal intensity observed
for the matching schizophrenic subject. Z scores were calculated
as Z ? (individual gene BDE – mean BDE of each array)/SD of
each array. Significances of the mean KIAA0417 expression ratios
were calculated against the mean BDE or Z score distribution of
all genes on each array with a t test in Microsoft Excel (Microsoft,
Multiple tissue Northern membranes (Clontech, Palo Alto,
California) that represent eight different brain regions were
hybridized with DNA probes designed to detect HSPA12A
(XM_048898), HSPA12B (NM_052970) and ?-actin gene se-
quences.33P-labeled DNA probes were prepared with Random
Primed DNA labeling kit (Roche Applied Science, Indianapolis,
Indiana) according to the manufacturer’s instructions. Mem-
branes were successively hybridized with either of the DNA
probes, exposed to X-MAR (Kodak, New Haven, Connecticut)
film, stripped, and re-probed with a DNA probe for ?-actin as a
loading control. The same probes were used in the Northern blot
and in situ hybridization experiments. On the basis of NCBI
BLAST searches, the designed probes showed ?30% nucleic acid
sequence similarity to any other known human transcript.
In Situ Hybridization
In situ hybridization was performed with 10 matched pairs of
subjects with schizophrenia and control subjects. Six pairs were
identical to those reported in microarray experiments. Due to
tissue constraints, 4 new, comparably matched pairs were sub-
stituted. Postmortem human DLPFC was identified according to
surface landmarks and sulci and verified by histology. Sections
(20 ?m) were cut from tissues containing the right DLPFC (area
9) with a cryostat at ?20°C, mounted on to gelatin-coated glass
slides, and stored at ?80°C until use. HSPA12A complementary
RNA [35S]-labeled riboprobes were generated by polymerase
chain reaction amplification of complementary (c)DNA obtained
from normal human brain (forward primer sequence: GCAAG-
GAGGTTCAGGGAAGA; reverse primer sequence: AGGGCAAG-
GTATGGGGTCA) and synthesis of approximately 3 ng of probe
(?2 ? 106disintegrations per minute) per each section. Methods
used for hybridization have been described previously (Camp-
bell et al 1999; Mirnics et al 2000b).
Slides were exposed to BioMax MR film (Kodak) for 8–22
hours and then dipped and exposed to autoradiographic emul-
sion (NTB-2, Kodak) for 3–5 days at 4°C. Scion Image (version
4.0b; Scion Corporation, Frederich, Maryland) was used to obtain
high-resolution scans of each film image for quantification.
Darkfield images were captured from the developed slides.
Slides were coded as to render the investigator performing the
analysis blind to the diagnosis of the subjects. Subject pairs were
processed in parallel. Three sections per subject per area were
analyzed. Control hybridization with sense probe did not result
in detectable signal.
Owing to the absence of labeling in the white matter,
quantification was performed by subtracting the average white
matter optical density (OD) from the average signal OD mea-
sured in five nonoverlapping rectangular regions from each of
three sections per tissue block. These rectangular regions
spanned layers I–VI. Relative expression differences were calcu-
lated by subtracting the ODs of schizophrenic from control
subjects that were matched. To assess expression differences, we
performed analyses of covariance (ANCOVAs) with diagnosis as
the main effect and brain pH and storage time as covariates.
Subject pair was used as a blocking factor. Statistical significance
was assessed using groupwise Student’s t tests (18 degrees of
Four pairs of male cynomolgus (Macaca fascicularis) mon-
keys, matched for age and weight, were used in this study (Pierri
et al 1999). One animal in each pair was treated for 9–12 months
944 BIOL PSYCHIATRY 2004;56:943–950
J.L. Pongrac et al
with the antipsychotic medication haloperidol decanoate. The
animals treated with haloperidol decanoate had serum levels in
the therapeutic range observed for the treatment of schizophre-
nia. Animals receiving haloperidol also received benztropine
mesylate at a level effective for managing extrapyramidal symp-
toms. Dorsolateral prefrontal cortex sections from each pair were
analyzed by in situ hybridization in the fashion described above.
In both humans and monkeys, Brodmann’s area 9 was identified
according to cytoarchitectonic criteria on Nissl-stained sections
from each subject.
In situ hybridization conducted with monkey brain tissue
followed the protocols for the human samples described
above. Because we used a human probe for the monkey in
situ hybridizations, we also sought evidence in the NCBI
databases for sequence divergence between the HSPA12A and
HSPA12B inmonkey (http://genome.wustl.edu/projects/
chimp/). On the basis of the identified sequences and blast
alignments, we mapped HSPA12A to a chimpanzee contig
8.836 (97% identity to human) and HSPA12B to contig 706.19
(98% identity). On the basis of these data, we concluded that
1) the divergence between these two genes is comparable
between the nonhuman primates and humans; and 2) the
employed human in situ hybridization probes are specific for
the monkey tissue. A groupwise Student’s t test (6 degrees of
freedom) was performed.
Microarrays Reveal Decreased Expression of KIAA0417 in the
DLPFC of Subjects with Schizophrenia
The data from our previous cDNA microarray study reported
that KIAA0417 was abundantly expressed in the DLPFC of all
control and schizophrenic subjects. This data set also revealed a
significant decrease in the expression of KIAA0417 in the subjects
with schizophrenia (Figure 1A) (mean average log2 ratio of
control/schizophrenic pairs ? ?.58, p ? .001). This finding was
also significant by Z score analysis (mean pairwise Z score ?
?1.38, p ? .001).
HSPA12A and HSPA12B Transcripts Show Regional
The microarray probe sequence for KIAA0417 was initially
mapped to UniGene cluster Hs.12385, which owing to heter-
ogeneity within the cluster has recently been split into two
Figure 1. (A) KIAA0417 transcript (later annotated
as HSPA12 isoform A) expression was significantly
panel, average Z ratio ? ?1.38; p ? .001). In both
panels, the x axis represents subject pairs. The left
panel y axis denotes log2 fluorescent signal inten-
sity ratios for each pairwise comparison, and the
regardless of the analysis performed, KIAA0417
showed a marked reduction across the dorsolateral
prefrontal cortex of all schizophrenic subjects (a
value of 1.0 corresponds to a twofold change on a
log2 scale). (B) KIAA017 has been recently anno-
which share a high deoxyribonucleic acid and pro-
tein sequence homology. This homology is promi-
(C), whereas the protein sequence seems to be
highly conserved throughout the whole coding re-
croarray clone showed that the microarray probe
sequence was derived from the HSPA12A gene.
gene showed differential expression across the hu-
man brain regions, with the highest expression in
the frontal and occipital cortices. In contrast,
sion across all the studied brain regions. Beta-actin
hybridization was used as a loading reference. PFC,
nology Information; Hml, homology; Prot, protein;
mRNA, messenger ribonucleic acid.
J.L. Pongrac et al
BIOL PSYCHIATRY 2004;56:943–950 945
new clusters, Hs.372597 and Hs.368340, annotating two novel
genes HSPA12A and HSPA12B, respectively (Figure 1B). Pair-
wise BLAST alignment (Altschul et al 1997) of representative
mRNA and protein sequences revealed a high degree of
sequence homology between these two HSPA12 family mem-
HSPA12B in five mRNA regions exceeded 70% (Figure 1C),
whereas the overall protein sequence similarity was 77% for
the human HSPA12 isoforms (Figure 1D). This raised a
concern that the microarray signal (although the probe was
designed against HSPA12A) might be partially derived from
HSPA12B cross-hybridization. Because there are no data
available on the distribution and abundance of the two
isoforms in the central nervous system, and because cross-
hybridization greatly depends on relative abundance, we
decided to investigate the expression levels of HSPA12A and
HSPA12B across eight different brain regions. Northern hy-
bridizations were performed with multiple tissue Northern
membranes (MTN; Clontech) against minimally conserved
mRNA sequences of HSPA12A and HSPA12B. Each probe
labeled only a single band (HSPA12A ?5.9 kb and HSPA12A
?3.1 kb), arguing for a strong specificity of the probes used.
HSPA12A was highly expressed in the human brain. This
isoform showed difference in the mRNA regional distribution,
with highest expression observed in the occipital and frontal
cortices and low expression in the spinal cord and cerebellum
(Figure 1E). In contrast, HSPA12B transcript was only sparsely
present in the brain, and the expression levels were uniform
across the investigated brain regions.
In Situ Hybridization Confirms Decreased Expression of
HSPA12A in the DLPFC of Subjects with Schizophrenia
Because microarray data sets can be associated with type I
errors, verification of findings is essential (Mirnics 2002). The
microarray data were verified on the same set of subjects with in
situ hybridization with35S-labeled riboprobes. In both control
and schizophrenic subjects, the HSPA12A transcript was strongly
expressed in gray matter, particularly in layers III and V–VI, with
no labeling in layer I of the DLPFC or the underlying white matter
(Figure 2A). The morphology of the stained cells suggests that
this transcript is expressed in both projection neurons and
interneurons but not in glial cells.
situ hybridization against HSPA12A. Note that the
density of the labeling is most prominent in deep
layer III, whereas the white matter and layer I label-
ing were absent. (B) Dorsolateral prefrontal cortex
in situ hybridization autoradiograms for HSPA12A
gene originating from two matched pairs of schizo-
graph of deep layer III in situ hybridization with
HSPA12A riboprobes. Darkfield-illuminated image
(red) is superimposed over the Nissl-stained image
clustering over large cells that show a classic pyra-
midal shape (apical toward top). (D) The quantifica-
tion of the in situ hybridization autoradiograms re-
vealed a significant decrease in the HSPA12A signal
across10 subjects pairs
? ?.84, p ? .01). The x axis denotes subject pairs,
the y axis represents optical density (OD) average
log ratios for the subject pairs (three runs com-
bined). (E) Ten subject pairs (patients with major
depression and matched control subjects) did not
report significant differences in HSPA12A transcript
levels (overall average log2 ratio ? ?.09, p ? .95).
946 BIOL PSYCHIATRY 2004;56:943–950
J.L. Pongrac et al
When compared with matched control subjects, 9 of 10
subjects with schizophrenia showed a decrease of HSPA12A
expression (mean OD log ratio range .27–1.8) (Figure 2B–D).
Across the 10 pairs of subjects analyzed, the autoradiographic
images of brain sections of subjects with schizophrenia con-
tained significantly less signal (overall mean log ratio ? ?.84,
p ? .01). In summary, the in situ hybridization findings
reported a mean decrease comparable to that seen in the
microarray data set (–.84 and ?.58, respectively). On the same
six pairs of subjects used in both experimental approaches,
the microarray-reported expression changes were significantly
correlated to the expression decreases found by in situ
hybridization (r ? .56, p ? .05).
HSPA12A Expression Is Unchanged in the DLPFC of Subjects
To test the diagnostic specificity of the change observed in the
subjects with schizophrenia, we performed in situ hybridization
on the DLPFC of 10 pairs of subjects with MDD and matched
control subjects. Subjects with MDD (Figure 2E) showed no
change in HSPA12A transcript levels (mean log2 OD ratio ?
?.09, p ? .95), suggesting that the expression change observed
in schizophrenia is not characteristic of all brain disorders;
however, this finding does not exclude the possibility that
HSPA12A transcript levels might be specifically altered in more
than one brain disorder (Tkachev et al 2003).
HSPA12A Expression Levels Do Not Change in Response to
Chronic Neuroleptic Exposure in a Nonhuman Primate Model
Gene expression change of HSPA12A in subjects with
schizophrenia might be related to the disease process or might
reflect a consequence of the pharmacologic treatment of the
disorder. In our study, 8 of 10 subjects with schizophrenia
were receiving antipsychotic medications at the time of death.
Importantly, the two subjects not receiving antipsychotic
medication showed a less prominent decrease in HSPA12A
transcript from the rest of the subjects with schizophrenia in
the microarray study (H537s and H622s; Figure 1A) but not in
the in situ hybridization study (Figure 2D). To further test the
effect of medication on the HSPA12A gene expression, we
decided to use a nonhuman primate model of chronic anti-
psychotic treatment (Pierri et al 1999). First, using in situ
hybridization, we established the expression pattern of
HSPA12A mRNA in the monkey brain. Within the frontal lobe,
HSPA12A transcript showed pronounced labeling in dorsome-
dial regions, with diminished labeling in the lateral regions.
More posteriorly, there was moderate expression in lateral and
medial temporal lobe regions, including the entorhinal cortex
and hippocampus. Subcortically, HSPA12A was expressed at
relatively low levels in the dorsal thalamus, with moderate
levels in multiple basal ganglia structures (caudate, putamen,
substantia nigra pars compacta), but it displayed prominent
and patterned expression in the lateral geniculate nucleus
(Figure 3A and 3B). The cellular distribution of HSPA12A
radioactive label in the monkey DLPFC closely mimicked the
distribution pattern that was observed in the human DLPFC
tissue. When compared with matched control subjects, four
monkeys chronically treated with haloperidol showed no
difference in HSPA12A transcript expression in the DLPFC
(overall mean log2 OD ratio ? ?.10, p ? .53) (Figure 3C). A
power analysis, assuming the same effect size as in humans,
suggests that at p ? .05 we would have detected a haloperidol-
induced expression change more than 90% of the time.
The present study of a novel protein HSPA12A established
that 1) HSPA12A, but not HSPA12B, is highly expressed in the
human and nonhuman primate brain in a region- and cell
type-specific pattern; 2) HSPA12A transcript levels are signifi-
cantly reduced in the DLPFC of subjects with schizophrenia, both
by DNA microarray and in situ hybridization assessment; 3)
subjects with MDD do not show altered transcript levels of
HSPA12A; and 4) HSPA12A expression is unchanged in monkeys
that received chronic neuroleptic medication.
On the basis of the overall data, we also believe that the
obtained findings are related to the disease processes of schizo-
phrenia rather than technical confounds, such as brain pH and
agonal state of the studied subjects (Harrison et al 1995; Tomita
et al 2004). Importantly, the pH of the samples used is in the
range that does not lead to skewed gene expression profile.
Furthermore, the cause of death was sudden in the majority of
the studied cases. Finally, the impact of this variable was
accounted for in our statistical analyses (ANCOVA). Similarly, the
ANCOVA analysis did not find significant association between
HSPA12A expression changes and age, PMI, gender, race, tissue
storage time, or medication history.
There are no systematic studies to date that have assessed the
expression changes of HSP transcripts or proteins in schizophre-
HSPA12A levels were unchanged as a result of the neuroleptic treatment. PFC, prefrontal cortex.
J.L. Pongrac et al
BIOL PSYCHIATRY 2004;56:943–950 947
nia; however, several independent studies report that a signifi-
cant percentage of subjects with schizophrenia produce high
levels of autoantibodies against HSP60 (Kilidireas et al 1992;
Leykin et al 1999; Schwarz et al 1999) and HSP70 (Kim et al
2001). Furthermore, the percentage of schizophrenic patients
with high levels of antibody to HSP70 was decreased significantly
after 6 weeks of antipsychotic treatment (Kim et al 2001).
Although it is not clear how and whether these findings are
related to expression changes of HSP transcripts, the combined
data suggest that the disturbance of HSP function might contrib-
ute to the pathogenesis of schizophrenia.
HSPA12A and Heat Shock Proteins as Putative Schizophrenia
Altered HSPA12A expression levels in subjects with schizo-
phrenia might reflect either genetic or nongenetic causality. The
HSPA12A gene maps to 10q26.12, a cytogenetic region that might
represent a common susceptibility locus for both schizophrenia
and bipolar affective disorder (Ewald et al 2002; Kelsoe et al
2001; Mowry et al 2000). Interestingly, several other genes
harboring HSP domains are also located on putative schizophre-
nia susceptibility loci (Brzustowicz et al 2002; Lewis CM et al
2003) (HSPA6 and HSPA7 on 1q21–22, HSPA4 and HSPA9B on
5q31, HSPA1A and HSPA1B on 6p21). Consistent with the
multigenic susceptibility model of schizophrenia, if HSPA12A
genetic susceptibility plays role in the pathogenesis of schizo-
phrenia, it would likely represent a common and necessary, but
not sufficient, predisposing factor to the disease (Mirnics and
HSPA12A Expression Changes as an Epigenetic Response
Related to Homeostatic Changes in Schizophrenia
The observed schizophrenia-associated HSPA12A expression
alteration might not be DNA sequence-based. How can this
nongenetic expression change occur? In the same subjects, we
previously reported a transcript downregulation of multiple
metabolic genes (Middleton et al 2002), and some of these genes
were related to energy metabolism and mitochondrial function.
Isoforms of the 70-kd heat shock–related protein (and potentially
HSPA12A) effectively control normal mitochondrial function
(Voos and Rottgers 2002) by facilitating protein import into
mitochondria (Murakami et al 1988; Parcellier et al 2003). This is
achieved through adenosine triphosphate hydrolysis by matrix
HSP70, which provides the necessary energy for the complete
translocation of the bulk polypeptide chains through the mito-
chondrial membrane (Voos and Rottgers 2002). In the context of
altered mitochondrial transcript changes in the same subjects,
one can envision that changes in HSPA12A and nuclear-encoded
mitochondrial genes are part of the same transcriptome re-
sponse: altered metabolism in the DLPFC will lead to decreased
mitochondrial transcript synthesis and potentially decreased
mitochondrial function. This will lead to a lesser demand for
HSP70-dependent protein translocation through the mitochon-
drial system, which in turn might drive down HSPA12A transcript
Proteins undergoing structural changes during assembly, fold-
ing, and membrane insertion are protected from ubiquitin/
proteasome degradation by the chaperone function of heat
shock proteins (Bukau and Horwich 1998). Thus, mitochondrial
and heat shock protein expression are likely to be co-regulated
by the expression of ubiquitin/proteasome pathway genes. The
simultaneous downregulation of HSPA12A, mitochondrial genes,
and the ubiquitin/proteasome transcripts we observed in the
same subjects (Middleton et al 2002) are likely to be part of the
same transcriptome response: decreased mitochondrial function
might result in decreased energy production, which will in turn
decrease de novo protein synthesis. This might lead to decreased
metabolic stress, decreasing HSPA12A levels, which in turn might
result in a compensatory downregulation of gene products of the
Finally, there is increasing evidence that heat shock proteins
act as “evolutionary capacitors” buffering mutations from being
expressed because they share key roles in adapting to environ-
mental stress and regulating signal transducers during develop-
ment (Bergman and Siegal 2003; Rutherford and Lindquist 1998).
In a polygenic disease with a neurodevelopmental component,
such as schizophrenia, heat shock protein expression might play
a critical role in counteracting mutations or polymorphisms that
might be directly associated with the disease. In this model,
changes in HSPA12A that we observed might have resulted as an
adaptational response to an environmental stress, subsequently
leading to expression of genetic mutations detrimental to the
development or functioning of brain tissue.
HSPA12A Expression Changes as a Result of a Putative
Interestingly, heat shock proteins are one of the most abun-
dant proteins during development (Fountoulakis et al 2002).
They are involved in many aspects of brain formation. Neural
plate induction, prenatal neuron generation, cellular migration,
postnatal neural differentiation and maturation, cellular signal-
ing, and protein transport are all heat shock protein–dependent
processes (for a review, see Calabrese et al 2002). Importantly,
the NCBI databases suggest that HSPA12A is robustly expressed
in the developing human and rodent brain. In the context of the
developmental role of heat shock proteins and the neurodevel-
opmental hypothesis of schizophrenia (Weinberger 1995), sub-
jects with schizophrenia might not be able to maintain physio-
logic levels of HSPA12A expression, and this would impair the
normal maturation of the HSPA12A-expressing neuronal popu-
lation. Recent studies suggest that Hsp70, cystein string proteins,
GAD(65), and GAD(67) are all associated with synaptic vesicles
and that heat shock protein chaperones modulate glutamic acid
decarboxylase (GAD) activity (Hsu et al 2000). Consequently,
HSPA12A underexpression might have a preferential impact on
the development of GAD-expressing cells and thus might con-
tribute to the ?-aminobutyric acidergic system disturbances seen
in schizophrenia (Lewis et al 1999; Lewis DA et al 2004; Pierri et
al 1999; Volk and Lewis 2002; Volk et al 2000b).
HSPA12A Expression Is Not Modulated by Antipsychotic
Animal studies suggest that HSP70 transcripts and proteins
increase in response to drugs that mimic symptoms of schizo-
phrenia (e.g., phencyclidine) (Hashimoto et al 1996; Sharp et al
1994), and this induction can be prevented by administration of
atypical antipsychotic agents (Gabriele et al 2003; Garcia-Osta et
al 2003; Okamura et al 2003) but not haloperidol (Garcia-Osta et
al 2003; Nakahara et al 1999). In a nonhuman primate model
(Pierri et al 1999) of chronic haloperidol treatment, we did not
see induction or repression of HSPA12A transcript. Although we
recognize that none of the existing animal models can perfectly
model the pharmacotherapy of schizophrenia, we believe that
our data suggest that the altered gene expression of HSPA12A in
subjects with schizophrenia are more likely to be related to the
948 BIOL PSYCHIATRY 2004;56:943–950
J.L. Pongrac et al
disease process rather than to be a consequence of treatment
with antipsychotic medication.
HSPA12A expression changes in subjects with schizophre-
nia might represent convergent pathology with expression
changes in the mitochondrial and ubiquitination systems, and
these combined changes might contribute to the developmen-
tal aspect of the disease; however, the unique structural
organization and the specific regional anatomical distribution
suggest that HSPA12A also might have a highly specialized
function that might go beyond the HSP chaperone function.
To establish the specific function of the HSPA12A gene
product in the brain, specific biological follow-up experi-
ments will be essential. As with all gene expression alteration
studies related to human brain disorders, replicate studies
across different cohorts will be pivotal in establishing the full
extent of HSPA12A transcript changes in schizophrenia. Fi-
nally, HSPA12A might also represent a promising candidate
for genetic association studies of schizophrenia.
This research was primarily supported by a National Alliance
for Research on Schizophrenia and Depression Young Investiga-
tor Award (KM). Additional support was provided by Projects 1
(DAL), 2 (KM), and 4 (PL) of National Institute of Mental Health
(NIMH) Center Grant MH45156 (DAL), and NIMH training
grant T32 MH18273 (FAM).
We thank the colleagues who read and commented on earlier
versions of this manuscript, as well as Dr. J. Pierri for his
involvement in the chronic haloperidol treatment of monkeys.
JLP is currently at the Department of Psychiatry, University of
British Columbia, Vancouver, British Columbia, Canada.
Gapped BLAST and PSI-BLAST: A new generation of protein database
mapping of the schizophrenia susceptibility locus on chromosome
1q22. Hum Hered 54:199–209.
Bergman A, Siegal ML (2003): Evolutionary capacitance as a general feature
of complex gene networks. Nature 424:549–552.
Calabrese V, Scapagnini G, Ravagna A, Giuffrida Stella AM, Butterfield DA
(2002): Molecular chaperones and their roles in neural cell differentia-
tion. Dev Neurosci 24:1–13.
abolished on the Purkinje cell degeneration (pcd) mutant background.
Ewald H, Flint TJ, Jorgensen TH, Wang AG, Jensen P, Vang M, et al (2002):
Search for a shared segment on chromosome 10q26 in patients with
bipolar affective disorder or schizophrenia from the Faroe Islands. Am J
of the fetal brain. Proteomics 2:1547–1576.
Gabriele J, Culver K, Sharma S, Zhang B, Szechtman H, Mishra R (2003):
brain, following quinpirole administration. Synapse 49:261–269.
Garcia-Osta A, Frechilla D, Del Rio J (2003): Reduced basal and phencycli-
dine-induced expression of heat shock protein-70 in rat prefrontal cor-
tex by the atypical antipsychotic abaperidone. Prog Neuropsychophar-
Gottesman II (1991): Schizophrenia Genesis: The Origins of Madness. New
Hakak Y, Walker JR, Li C, Wong WH, Davis KL, Buxbaum JD, et al (2001):
Han Z, Truong QA, Park S, Breslow JL (2003): Two Hsp70 family members
expressed in atherosclerotic lesions. Proc Natl Acad Sci U S A 100:1256–
Harrison PJ, Heath PR, Eastwood SL, Burnet PW, McDonald B, Pearson RC
interval for human brain gene expression studies: selective mRNA vul-
nerability and comparison with their encoded proteins.. Neurosci Lett
of heat shock protein (HSP)-70 in posterior cingulate and retrosplenial
cortex of rat brain by dizocilpine and phencyclidine: Lack of protective
effects of sigma receptor ligands. Addict Biol 1:61–70.
Hohfeld J, Cyr DM, Patterson C (2001): From the cradle to the grave: Molec-
ular chaperones that may choose between folding and degradation.
Hsu CC, Davis KM, Jin H, Foos T, Floor E, Chen W, et al (2000): Association of
L-glutamic acid decarboxylase to the 70-kDa heat shock protein as a
potential anchoring mechanism to synaptic vesicles. J Biol Chem 275:
(2001): A genome survey indicates a possible susceptibility locus for
Kilidireas K, Latov N, Strauss DH, Gorig AD, Hashim GA, Gorman JM, et al
(1992): Antibodies to the human 60 kDa heat-shock protein in patients
with schizophrenia. Lancet 340:569–572.
nia. Schizophr Res 52:127–135.
Lewis CM, Levinson DF, Wise LH, DeLisi LE, Straub RE, Hovatta I, et al (2003):
Genome scan meta-analysis of schizophrenia and bipolar disorder, part
Lewis DA, Lieberman JA (2000): Catching up on schizophrenia: Natural his-
tory and neurobiology. Neuron 28:325–334.
Lewis DA, Pierri JN, Volk DW, Melchitzky DS, Woo TU (1999): Altered GABA
Lewis DA, Volk DW, Hashimoto T (2004): Selective alterations in prefrontal
treatment of working memory dysfunction. Psychopharmacology (Berl)
immune response to human heat-shock protein-60 in schizophrenic
arrays. Nat Biotechnol 14:1675–1680.
Marcotte ER, Srivastava LK, Quirion R (2001): DNA microarrays in neuropsy-
profiling reveals alterations of specific metabolic pathways in schizo-
phrenia. J Neurosci 22:2718–2729.
Gene expression analysis in schizophrenia: Reproducible up-regulation
of several members of the apolipoprotein L family located in a high-
Mirnics K (2002): Microarrays in brain research: Data quality and limitations.
Mirnics K, Middleton F, Marquez A, Lewis D, Levitt P (2000a): Molecular
expression in prefrontal cortex. Neuron 28:53–67.
Mirnics K, Middleton F, Stanwood G, Lewis D, Levitt P (2000b): Disease-
in schizophrenia. Mol Psychiatry 6:293–301.
disorders with microarrays: Schizophrenia as a disease of the synapse.
J.L. Pongrac et al
BIOL PSYCHIATRY 2004;56:943–950 949
Mowry BJ, Ewen KR, Nancarrow DJ, Lennon DP, Nertney DA, Jones HL, et
al (2000): Second stage of a genome scan of schizophrenia: Study of
of at least two distinct cytosolic factors stimulating protein import into
Nakahara T, Kuroki T, Hondo H, Tsutsumi T, Fukuda K, Yao H, et al (1999):
Effects of atypical antipsychotic drugs vs haloperidol on expression of
al (2003): Protective effect of the antipsychotic drug zotepine on dizo-
cilpine-induced neuropathological changes in rat retrosplenial cortex.
Parcellier A, Gurbuxani S, Schmitt E, Solary E, Garrido C (2003): Heat shock
proteins, cellular chaperones that modulate mitochondrial cell death
Pierri JN, Chaudry AS, Woo TU, Lewis DA (1999): Alterations in chandelier
neuron axon terminals in the prefrontal cortex of schizophrenic sub-
Pongrac J, Middleton FA, Lewis DA, Levitt P, Mirnics K (2002): Gene expres-
sion profiling with DNA microarrays: Advancing our understanding of
psychiatric disorders. Neurochem Res 27:1049–1063.
Rutherford SL, Lindquist S (1998): Hsp90 as a capacitor for morphological
evolution. Nature 396: 336–342.
gene expression patterns with a complementary DNA microarray. Sci-
heat shock proteins in schizophrenic patients: Implications for the
mechanism of the disease. Am J Psychiatry 156:1103–1104.
Phencyclidine induction of the hsp 70 stress gene in injured pyramidal
neurons is mediated via multiple receptors and voltage gated calcium
channels. Neuroscience 62:1079–1092.
injured by phencyclidine (PCP), MK801 and ketamine. J Neurosci Res
apoptosis. Oncogene 22:9041–9047.
Tomita H, Vawter MP, Walsh DM, Evans SJ, Choudary PV, Li J, et al (2004):
Effect of agonal and postmortem factors on gene expression profile:
quality control in microarray analyses of postmortem human brain. Biol
Tkachev D, Mimmack ML, Ryan MM, Wayland M, Freeman T, Jones PB, et al
der. Lancet 362:798–805.
schizophrenia: A preliminary study. Schizophr Res 58:11–20.
Volk DV, Austin MC, Pierri JN, Sampson RA, Lewis DA (2000a): Decreased
prefrontal cortical gamma-aminobutyric acid neurons in subjects with
Volk DW, Austin MC, Pierri JN, Sampson AR, Lewis DA (2000b): Decreased
prefrontal cortical gamma-aminobutyric acid neurons in subjects with
Volk DW, Lewis DA (2002): Impaired prefrontal inhibition in schizophrenia:
Relevance for cognitive dysfunction. Physiol Behav 77:501–505.
Voos W, Rottgers K (2002): Molecular chaperones as essential mediators of
Wang XF, Wang D, Zhu W, Delrahim KK, Dolnak D, Rapaport MH (2003):
Studies characterizing 60 kda autoantibodies in subjects with schizo-
phrenia. Biol Psychiatry 53:361–375.
(2004): Database resources of the National Center for Biotechnology
Information: Update. Nucleic Acids Res 32 Database issueD35–D40.
950 BIOL PSYCHIATRY 2004;56:943–950
J.L. Pongrac et al