Declaration of interest: None. The authors declare that they have no competing financial interests.
Ahmad Khundakar, PhD, Edwardson Building, Institute for Ageing and Health, Campus for Ageing and Vitality, Newcastle University, Newcastle upon Tyne, NE4 6BE, UK. Email: email@example.com
Aims: This study immunohistochemically examined the orbitofrontal cortex for three possible candidates in hypoxic/ischemic signaling: the cytokine transforming growth factor-β, the glucose transporter-1 and the neuron-specific oxygen-binding protein neuroglobin.
Methods: Post-mortem tissue from 20 depressed and 20 non-depressed individuals was obtained and the expression of the three proteins was analyzed using image analysis software.
Results: No significant changes were found in transforming growth factor-β or neuroglobin in the orbitofrontal cortex between depressed and non-depressed individuals. There was, however, a trend towards a reduction in glucose transporter-1 in the depressed group.
Conclusions: This study does not clearly support the hypothesis that hypoxic/ischemic processes are behind the pathological deficits in the frontal-subcortical circuitry associated with depression and therefore does not provide evidence to support the ‘vascular depression’ hypothesis.
IT IS NOW over a decade since the ‘vascular depression’ hypothesis was first published.1 The theory proposed that vascular disease could have a significant effect on the pathology and severity of depression, especially in the elderly. Though the mechanisms behind changes in vasculature and the onset of depression remain unclear, accumulating evidence has indicated the possible involvement of cytokines in the cause of depression.2–5 The hypothesis has been substantiated by neuroimaging and neuropathology studies5,6 that have shown increased white matter lesions in the frontal-subcortical circuitry and that these are ischemic. Furthermore, neuroimaging studies have indicated reduced volume7,8 and reduced activity9 in the orbitofrontal cortex circuitry in major depression and such changes are associated with impaired function.10
Based around the ‘vascular depression’ hypothesis and possible cytokine involvement, this immunohistochemical study examined the expression of transforming growth factor (TGF)-β1, neuroglobin and the glucose transporter (GLUT)-1 in the orbitofrontal cortex of post-mortem tissue from elderly depressed and non-depressed individuals. TGF-β1 has been shown to orchestrate numerous critical cellular functions (including the modulation of immune response and angiogenesis), and dysfunction in its signaling processes has been linked to several pathological disorders,11 including depression.12,13In vitro and in vivo studies have shown that the small hemeprotein member of the globin family, neuroglobin, plays an important role in the protection of brain neurons from ischemic and hypoxic damage.14–16 GLUT-1 is highly expressed in the endothelium of cerebral microvessels and facilitates transportation of glucose across the blood–brain barrier.17 It has therefore been utilized as a marker of endothelial cell expression, vessel integrity and angiogenesis.18,19
The overall aim of this study therefore was to study the expression of TGF-β1, neuroglobin and GLUT-1 in depressed and non-depressed patients in the orbitofrontal cortex. We hypothesized that the expression of these molecules would be increased in depression compared with a matched control group.
Post-mortem tissue was taken from 20 elderly subjects, who had been diagnosed by full clinical assessment during life and reviewed by the Newcastle Clinicopathological committee after death as having suffered at least one well-documented episode of major depression but not dementia or any other major psychiatric disorder. In addition, 20 non-depressed comparison subjects who met the same criteria as depressed subjects but had not experienced any episode of major depression were assessed. The causes of death among patients in this study have been published previously,20,21 and the two groups were evenly matched for vascular risk factors. Permission for post-mortem research was granted by the subjects and ethical approval was granted for the use of the tissue in this study.
The right hemisphere of each subjects' brain was fixed in 10% formalin post-mortem and dissected using the standard procedure. The orbitofrontal cortex was excised by carefully selecting the coronal slice from each subject to include Brodmann area 11 and 12, according to a standard map,22 and embedded into paraffin wax blocks. Three 10-µm sections were randomly sampled from 15 serial sections taken from each block, which had been cut using a rotary microtome (ThermoFisher Ltd, Waltham, MA, USA) onto large 3 × 2′ 3-aminopropyl-triethoxy-silane-coated slides. Tissue pH and brain weight, the delay from death to formalin fixation and the duration of fixation in paraffin blocks before cutting were also recorded. The quality of the sections was carefully checked for consistency and all slides were coded by a third party (A.T.) so that the analysis of the sections was carried out blind to patient diagnosis.
Immunohistochemistry was performed on the sections according to a standardized procedure. Paraffin-embedded sections were briefly dewaxed in xylene, rehydrated through alcohol washes and microwaved in 0.01% citrate buffer (pH 6.0) to optimize antigen retrieval, then quenched in hydrogen peroxide and covered in a blocking solution to minimize non-specific labeling. The sections were then covered in primary antibody for 1 h. The antibodies comprised of the polyclonal GLUT-1 antibody at 1:100 concentration (Affinity Bioreagents Ltd, Golden, CO, USA), the monoclonal TGF-β1 antibody at 1:40 concentration (Novocastra Ltd, Newcastle upon Tyne, UK) and the monoclonal neuroglobin antibody at 1:250 concentration (Abcam Ltd, Cambridge, UK).
Based on previous studies from this group,20,21 five images from each section (three per subject) were taken within the region of interest (orbitofrontal cortex) using an objective lens with a magnification of ×10 on a Zeiss Axioplan 2 light microscope, connected to a three-chip charge coupled device true-color video camera (JVC KY F55B). Two analytical methods were employed to measure changes in immunoreactivity based on the expression of the respective protein. A pilot study showed that TGF-β1 and neuroglobin protein were expressed in the soma of neurons (Fig. 1), whereas GLUT-1 was predominantly expressed in the endothelial cells of vessel walls (Fig. 1). Changes in TGF-β1 and neuroglobin expression were therefore assessed by obtaining an estimation of the area fraction of tissue positively labeled for the specific antibody using a standard analytical software program, as described previously20,21 (Image Pro. Plus, version 4.0; Media Cybernetics, Silver Spring, MD, USA). A previously used analytical procedure, Buffon's needle method,23 was used for assessing changes in GLUT-1 expression. This method estimates the length of an object (in this case, the vessel wall) from the number of intersections through a set of lines of known length superimposed in random orientation across the region of interest. A pilot study was first undertaken to establish the ideal length of the geometric probe in relation to the number of vessel intersections in order to achieve a large but reasonable sampling number. The length density (length of a profile per unit area) of the object of interest (vessel) was analyzed.
Statistical comparisons of the two groups were carried out using the unpaired Student's t-test, anova and Pearson correlations. In addition, two-way t-tests were also used to compare the depressed and control groups on their demographic variables. All comparisons were carried out using spss software (Version 15.0; spss, Somers, NY, USA).
The magnitude of experimental error of the sampling design was assessed by coefficient of error (CE) calculation according to a Gundersen–Jensen formula.24 The mean CE calculations showed that they were made with a high degree of precision and were of a similar magnitude for TGF-β (1.1%), neuroglobin (1.5%) and GLUT-1 (0.16%).
Demographic information on the study subjects is summarized in Table 1. There were no significant differences between the groups in age (P = 0.78), duration of tissue fixation (P = 0.65) or post-mortem interval (P = 0.30) (Table 1). The causes of death were similar in the two groups, with only two depressed subjects dying by suicide. The clinical features of depressed subjects showed that in most cases they had experienced severe depressive illnesses as all except one had required inpatient care for their depression and most had received at least one course of electroconvulsive therapy. They had all received standard antidepressant treatment regimes, with selective-serotonin reuptake inhibitor or tricyclic antidepressants singly or often in combination with other agents.
Table 1. Patient characteristics for depressed and control groups
Depression group (n = 20)
Control group (n = 20)
Data are expressed as means ± SD.
CI, confidence interval.
−5.87 to 4.47
Age of depression onset, years
Fixation period, months
−19.80 to 6.70
Post-mortem delay, hours
−19.18 to 6.08
TGF-β1 was abundantly expressed in the soma of neurons of the orbitofrontal cortex (Fig. 1). The total area fraction of immunoreactivity for TGF-β1 was than examined quantitatively. There was no significant change in the area fraction for TGF-β1 between depressed patients and non-depressed controls (P = 0.656) in the orbitofrontal cortex (Fig. 2).
Neuroglobin was also expressed clearly in the soma of neurons in the orbitofrontal cortex (Fig. 1). There was no significant difference between neuroglobin immunoreactivity between the two groups (P = 0.249; Fig. 2).
GLUT-1 was extensively expressed in the vasculature of the orbitofrontal cortex (Fig. 1). Figure 3 shows the difference between GLUT-1 immunoreactivity in vessels in the orbitofrontal cortex in depressed individuals and non-depressed controls. Although there was no significant difference between the two groups, there was a trend towards a decrease in GLUT-1 immunoreactivity in the depressed group (P = 0.057).
There were no significant correlations between any of these measures and age or post-mortem interval (all P > 0.3). TGF-β1 was negatively correlated with pH (r = −0.49, P = 0.022) and there was a trend for GLUT-1 to be negatively correlated with duration of fixation (r = −0.29, P = 0.068); there were no other significant correlations with the expression of these molecules. Although these may represent chance findings we repeated our analyses co-varying for these and our findings remained non-significant.
The main finding in this study was that no significant changes were found in TGF-β1, neuroglobin and GLUT-1 immunoreactivity in the orbitofrontal cortex between depressed and non-depressed patients. The findings therefore do not provide clear support for the vascular depression hypothesis, which predicts the presence of hypoxia-induced changes that result in the prefrontal circuitry in late-life depression.
Previous studies have reported an upregulation in GLUT-1 immunoreactivity in brain microvasculature response to hypoxic25,26 and ischemic conditions.27 Glucose metabolism is altered due to the increased energy demand as a result of hypoxic conditions.27 Glucose is therefore transported to the microvascular endothelium by the family of facilitative glucose transporters, including GLUT-1.28 We had hypothesized an increase in endothelial cell GLUT-1 expression in elderly depressed patients as a compensatory measure to ongoing hypoxic conditions. A possible explanation for the slight reduction in GLUT-1 immunoreactivity in this study could be that damage to the endothelial cell membrane or blood–brain barrier dysfunction may have limited the GLUT-1 response in the microvessels.
The cytokine, TGF-β1, is a versatile molecule that has been shown to be important in many critical physiological processes, including neuron proliferation, migration, survival, and astrocyte differentiation.11 It has been termed an inherently ‘bipolar’ molecule, as it is associated with both exaggerated immune excitability, as well as regulation of innate and adaptive immune cells.29 TGF-β1 has been clinically associated with post-stroke activity since increased immunoreactivity was found in infarcted and penumbral regions in post-mortem tissue.30 Non-clinical studies have also found upregulation of the TGF-β1 isoform in neurons in acute experimental cerebral ischemia.31,32 Such post-stroke alterations have widespread changes in members of the TGF-β superfamily, such as glial-cell-line-derived neurotrophic factor (GDNF), which mediate numerous pleiotropic effects on growth tissue, including cellular proliferation and differentiation, as well as inflammation.33 Recent studies have also shown a higher level of TGF-β1 in the blood of patients with major depression, which was reduced after 6 weeks of antidepressant treatment,12,13 suggesting an increase in TGF-β1 may be state-related. As most of our subjects were taking antidepressants at the time of death, it could the case that TGF-β1 in the orbitofrontal cortex has been downregulated by such treatment. A previous study conducted in the same patients as this study found increases in intercellular cell adhesion molecule (ICAM)-1 in the dorsolateral prefrontal cortex.20 In accordance with TGF-β1, ICAM-1 expression has been shown to be increased by ischemia in vitro.34 Whilst this study apparently confounds earlier findings, TGF-β1 is a complex molecule and its expression and response to treatment may differ from that of ICAM-1 that acts in many ways and in conjunction with many signaling molecules and processes.
Neuroglobin was chosen for this study because of its apparent neuroprotective effects after hypoxia,14,15 and for the suggestion that it plays an important role in the transport of oxygen to metabolically active neuronal tissue.35 However, previous studies have revealed that the period of hypoxic exposure may play an important part in neuroglobin recruitment. Sun et al.14 showed acute hypoxic challenge resulted in an increase in the neuroglobin transcription and immunoreactivity in cultured primary cortical neurons. Nevertheless, a subsequent in vivo study observed no change in global neuroglobin expression after chronic periods of hypoxic exposure.36 This may therefore indicate that neuroglobin is recruited in acute hypoxic challenges but may not be altered by changes in oxygen tension over a chronic period, such as might be present in subjects with late-life depression and white matter disease.
The orbitofrontal cortex is thought to play a major role in mood regulation, aspects of emotion and ‘social intelligence’. Functional imaging studies have shown both altered activity9 and decreased volume in orbitofrontal cortex of major depressive disorder patients.7,8 However, stereological analysis of the cellular components within the orbitofrontal cortex appears to suggest anatomical variability in the morphological changes in late-life depression patients. A reduction in pyramidal neuron size has been found in the orbitofrontal cortex, as well as layers 3 and 5, of the rostral portion of the orbitofrontal cortex;37 however, this group found no morphological changes in any layers of the caudal portion of the orbitofrontal cortex in a similar cohort of elderly major depressive disorder patients conducted in this study.38 Likewise, this group found reductions in pyramidal neuron volume in the dorsolateral prefrontal cortex overall, as well as layers 3 and 5, of the dorsolateral prefrontal cortex in a similar cohort of patients.39 Thus, the upregulation in the inflammatory markers, ICAM-1 and interleukin-1β, found in previous studies in the dorsolateral prefrontal cortex21 indicated evidence of possible vascular damage as suggested in the ‘vascular depression’ hypothesis in the dorsolateral prefrontal cortex; whether this applies to the orbitofrontal cortex is a matter of conjecture. The orbitofrontal cortex and the frontal cortex as a whole are cytoarchitectonically and functionally heterogeneous and variability in hypoxic damage may occur in different subregions in response to pathological events. Functional magnetic resonance imaging studies have shown varying patterns of activity within the orbitofrontal cortex, with abnormal glucose activity metabolism40 and bilateral reductions in gray matter volume noted in the medial portion.41 Due to inherent issues in the use of post-mortem human tissue, the whole of the orbitofrontal cortex was not available for sampling. It is therefore difficult to establish whether that lack of change in hypoxia-related markers in this study applies to the whole of the orbitofrontal cortex or the caudal subregion only. It is also apparent that the dorsolateral prefrontal cortex, and other cortical (e.g. anterior cingulate cortex) and subcortical (e.g. caudate nucleus) areas associated with affective function should be examined; this will form the basis of subsequent studies from this group.
Our study benefited from the thorough clinical assessments of the cases, detailed post-mortem examinations to exclude degenerative pathologies and the application of valid morphometric methods with a high degree of precision. There are, however, several possible reasons why the three markers measured were unchanged in this series of experiments, which was not what we had hypothesized. As a post-mortem study our investigation was necessarily limited in being cross-sectional and thus a study at only one stage of a chronic illness. The fluctuating course of depression and changes over time in cerebral perfusion and oxygenation make it possible that at other points the hypothesized changes would have been present. Alternatively, the markers used in the experiments may also only detect acute changes in vascular activity, so the possible gradual, chronic hypoxic changes suggested in the ‘vascular depression’ hypothesis may have been too subtle to detect in this study. This seems to be the case for neuroglobin and may be the case for our other markers too. It is also possible, though we think unlikely, that antidepressant treatment may have confounded our study. Finally, our findings here may be genuine and our previous findings due to random variation and not to real hypoxia-ischemia in our depressed subjects. We think this is unlikely, however, because not only were the differences robust20,21 but were also supported by our study using a different methodology showing white matter hyperintensities were ischemic in these subjects.42
In conclusion, this study found no significant alterations in the immunoreactivity of TGF-β1, neuroglobin and GLUT-1 in the orbitofrontal cortex between depressed and non-depressed patients.
We would like to express our sincere thanks to the NIHR Biomedical Research Centre for Ageing and Age-related Disease for their support for this study. We also thank the staff at the Newcastle Brain Tissue Resource for their considerable assistance during the study.