Cerebral Oxidative Stress and Microvasculature Defects in TNF-α Expressing Transgenic and Porphyromonas gingivalis -Infected ApoE –/– Mice

INTRODUCTION

Oxidative stress is implicated in the onset, progression, and pathogenesis of numerous diseases and conditions, including periodontal disease, neurodegeneration, and aging [1 –8]. Prolonged episodes of oxidative stress produce excessive accumulation of reactive oxygen species (ROS), capable of inducing deleterious cellular and biomolecular damage to the host [1 –5]. Neurons are particularly sensitive to ROS-mediated, oxidative damage, as they contain higher numbers of mitochondria along their cell bodies and axonal/dendritic tree lengths, compared to other cell types. The ROS-mediated oxidative damage within the brain can potentially lead to neuronal dysfunction, with a likely outcome of Alzheimer’s disease (AD) pathology [3 , 9]. In the aging brain and during AD, the APOE4 genetic susceptibility is further capable of increasing the possibility of recurrent infections, leakage of plasma proteins, ROS and metallic ion production (iron from erythrocyte death), adding to a growing pool of oxidative stress mediators within the brain. Furthermore, oral and extra-oral bacteria are documented to be present within AD brains [10], where these are likely to promote additional ROS generation and host tissue damage. During chronic periodontal infection, oral pathobionts, such as Porphyromonas gingivalis (P. gingivalis), drive the host’s immuno-inflammatory responses towards elevated levels of pro-inflammatory cytokines, such as tumor necrosis factor-α (TNF-α), interleukin (IL) IL-1β and IL-6, leading to elevated oxidative stress [11]. This is particularly achieved by the interplay between the phagocytic oxidative burst of the host’s neutrophils and macrophages, coupled with the oxidative stress response initiated by bacteria [12]. Furthermore, live P. gingivalis infection is likely to exert its effect on the host through a diverse range of virulence factors. Apart from its lipopolysaccharide (LPS), which can cause the host to respond through oxidative burst and other mechanisms, proteolytic activity mediated by P. ginivalis gingipains are capable of targeting cell-cell adhesion molecules [13 –15]. It is, therefore, likely that gingipains also contribute to the degradation of endothelial cell tight junction proteins and the subsequent loss of blood-brain barrier (BBB) functional integrity. Oxidative stress from host response also appears to affect a multitude of other activities relevant to AD, including activation of positive feedback between the γ- and β-secretase cleavage of the amyloid-β protein precursor (AβPP) metabolic product, amyloid-β (Aβ) [16], and the formation of the inflammasome assembly which has subsequent effects on the liberation of mature forms of IL-1β/IL-18 family members [17, 18]. These cytokines modulate innate immune responses against microbes by generating a subsequent inflammatory response serving important biological and physiological functions. Inappropriate secretion of cytokines result in pro-oxidant mechanisms especially in neuroinflammatory pathologies. For example, TNF-α has been shown to permeabilize brain microvascular endothelial cells and degrade cell-cell adhesion proteins, such as occludin [19], and it can induce ROS production in cultured cells [20].

Since neurodegenerative diseases are of complex etiologies, often exhibiting mixed pathologies, the contribution of specific genes is best investigated using animal models. The genetically modified apolipoprotein E knockout (ApoE^–/–) mouse is a good example, in order to determine the effects of peripheral infection on a host undergoing high intrinsic stress, due to the constitutive expression of the pro-inflammatory cytokine, TNF-α [21, 22], lipid peroxidation [23], impaired immuno-modulatory function in macrophages and innate/adaptive immune responses [24 –26], and impaired cerebellar microvessels [27 –30]. The TNF-α transgenic mouse represents an additional model in which to explore the direct roles of the proinflammatory cytokine, TNF-α, in the process of local oxidative stress within the brain. In response to chronic bacterial infection, TNF-α release represents a proinflammatory cytokine at the apex of the inflammatory cascade that mediates local inflammatory damage including the production of localized oxidative stress and has been shown to be upregulated in response to P. gingivalis LPS in vivo [13 –15]. In the TNF-α transgenic mouse, TNF-α mRNA is stabilized, and its expression appears to be increased within all tissues, including the brain. This makes the TNF-α transgenic mouse an excellent control model to explore mechanistic insights into how bacterial models of oxidative stress may be mediating their actions [31].

While the cerebral microvessels in TNF-α transgenic and ApoE^–/– mice appear to be resilient to tissue damage, the cerebellar BBB in both mouse models was leaky when tested for mouse IgG at the beginning of the investigation. In addition, the cerebral BBB in ApoE^–/– mice have been shown to become leaky either due to experimentally induced physical injury [28], or as recently reported, inflammation-mediated physical injury following infection with P. gingivalis [32]. Indeed, this study, demonstrated that both TNF-α expressing transgenic mice and ApoE^–/– mice share such leaky cerebellar BBB characteristics, thereby providing a unique opportunity to draw systematic comparisons between common biological and physiological factors involved in stress initiation.

We have previously reported associations between P. gingivalis LPS and AD brains [33] and subsequently established that P. gingivalis is capable of accessing ApoE^–/– mouse brains from its primary gingival location [32, 34]. Once in the brain, P. gingivalis modulates innate immune responses via activation of the complement cascade, resulting in bystander injury of functional pyramidal neurons in the hippocampus, following 24 weeks of chronic infection [34]. A recent investigation using the same ApoE^–/– P. gingivalis-infected mice brains (24 weeks) provided evidence of BBB defects, with numerous age-related granules and cerebral tissue damage mediated via gingipains [32]. This led us to the rationale that acute phase inflammation, in the form of oxidative stress, may precede loss of the BBB integrity in these mice at an earlier timepoint of infection which was absent in ApoE^–/– P. gingivalis-infected mice brains (24 weeks). Therefore, this study sought to critically examine howP. gingivalis infection (at 12 weeks) alters acute innate immune responses by increasing oxidative exposure on cerebral parenchyma, and whether this alteration can potentially lead to tissue injury and subsequent hippocampal BBB permeability which was clearly evident at 24-week infection [32].

MATERIALS AND METHODS

RESULTS

Western blot analysis

Western blot analysis from tissue lysates from a section of the frontal lobe of all the mice tested, confirmed the presence of carbonyl content and oxidative protein damage, at variable molecular weights in the range of 25–70 kDa. Wild-type mouse brain tissue blots (Fig. 1a) consistently exhibited a prominent band around 43 kDa (Fig. 1a, broken arrow), with additional lesser bands of variable intensities at 25–40 kDa and 45–70 kDa positions. The TNF-α transgenic mouse brain blots (Fig. 1b), exhibited prominent bands at around 25 kDa (Fig. 1b, short arrow), 43 kDa (Fig 1b, broken arrow), and 68 kDa (Fig. 1b, arrow head) and lesser intensity bands at 50–65 kDa. Sham-infected mouse brain blots (Fig. 1c), demonstrated a prominent band at around 25 kDa (Fig. 1c, short arrow), with additional bands of highly variable degrees of intensities between 29–40 kDa and 45–70 kDa. P. gingivalis-infected mouse brain blots (Fig. 1d), exhibited numerous prominent bands at around 25 kDa (Fig 1d, short arrow), and 43 kDa (Fig. 1d, broken arrow) and one exclusive to this group between 30–43 kDa* (Fig. 1d, astrix), with lesser bands at 45–70 kDa. These observations were further confirmed by the analysis of intensities for the prominent bands for each experimental group, using ImageJ Software (Fig. 1e). Data is from means of triplicates±SD using Mann-Whitney U test for two independent samples of each 25 kDa, 43 kDa, and 30–43 kDa* and 68 kDa bands. Significant differences between the wild-type and TNF-α transgenic mouse groups; and between the sham-infected and P. gingivalis-infected groups were identified, respectively (Fig. 1e).

OPEN IN VIEWER

Fig.1

Representative OxyBlot profiles demonstrating protein carbonyl band intensities in wild-type (a), TNF-α transgenic (b), sham-infected (c), and P. gingivalis-infected (d) mouse brain lysates. All samples demonstrated protein carbonyl content and the presence of multiple oxidized proteins. Differences in the mean band intensities of the prominent bands at around 25 kDa, 43 kDa, 30–43 kDa*, and 68 kDa between each group using Image J Software (3 replicate blots analyzed). Non-parametric Mann-Whitney U (e) shows normalized data from mean of means of triplicates±SD. Significance indicated (see p values in e) in protein carbonyl and oxidative stress levels between wild-type and TNF-α transgenic mice and the sham-infected and P. gingivalis-infected mice.

Immunohistochemical analysis

The three different types of negative controls employed remained negative for protein carbonyl immuno-stain detection. These were (1) where 2,4-DNPH derivatization was omitted (Fig. 2a), (2) where free carbonyl groups were reduced with 25 mM sodium borohydride in 80% methanol, at room temperature for 30 min, before 2,4-DNPH derivatization (Fig. 2b), and (3) where the primary antibody was excluded (Fig. 2c). Immunostaining of protein carbonyl content within the hippocampus and microvessels of wild-type and TNF-α transgenic mouse brains was similar to that seen in negative controls (Fig. 2d-e), as was the case with the sham-infected mice (Fig. 2f). However, in P. gingivalis-infected mouse brains, protein carbonyl immunostaining was distinctly observed in the vasculature, especially localized within the hippocampal capillaries (Fig. 2g). At higher magnification, the protein carbonyl immunostaining was more distinctly observed in the microvasculature (boxed areas in Fig. 2g are shown at higher magnification in Fig. 2h-i). The microvessels adjacent to the pyramidal neurons of the CA sectors in the hippocampus (Fig. 2j, box) also showed immunostaining localized to the capillaries and endothelial cells (Fig. 2k, l, endothelial cell staining is indicated with arrows).

OPEN IN VIEWER

Fig.2

Compared with the controls (a-c), immunohistochemical detection of protein carbonyl groups within wild-type brains and TNF-α transgenic mouse hippocampi generally demonstrated lower protein carbonyl immuno-staining overall or within the microvessels (d, e). This was also the case in sham-infected mouse hippocampi, although staining was observed within the pyramidal neurons (f). P. gingivalis-infected mouse brains exhibited protein carbonyl immuno-staining in the vasculature, especially localized within the hippocampal capillaries (g). The boxed areas in (g) are shown at higher magnification (h, i), to further demonstrate the microvasculature. The capillaries adjacent to the CA neurons of the hippocampus (j, box) also showed immuno-staining localized to the capillaries and endothelial cells (k and l, arrows denote endothelial cell staining).

The frontal brains of wild-type mice demonstrated protein carbonyl immunostaining, but was much weaker at the optimal antibody dilution used (Fig. 3a, b), whereas the TNF-α transgenic mouse tissues exhibited weaker staining of the myelin bundles (Fig. 3c) and occasional cellular staining (Fig. 3c, d). The sham-infected mouse tissue sections demonstrated marked staining of the myelin sheath regions (Fig. 3e) and neuronal staining (Fig. 3f). The P. gingivalis-infected mouse tissue sections also demonstrated staining of the myelin sheath (Fig. 3g) and neurons (Fig. 3h), while certain pyramidal neurons appeared to be completely damaged (Fig. 3h, insert).

OPEN IN VIEWER

Fig.3

Immunohistochemical detection of protein carbonyl groups within the thalamus (a) and frontal lobe (b) of wild-type mice generally showed little immuno-staining, whereas TNF-α transgenic mice frontal brains demonstrated weak straining on myelin bundles (c) and occasional cellular staining (d). Sham-infected mice tissue sections demonstrated marked staining of the myelin sheath regions (e) and neuronal staining (f). P. gingivalis-infected mouse tissue sections also demonstrated staining of the myelin sheath (g) and neurons (h), while certain neurons appeared to be completely damaged (h, insert).

DISCUSSION

As atherosclerosis, strokes and AD share a common APOE4 genetic susceptibility, many investigators have used the ApoE^–/– mice as an animal model to assess the influences of co-morbidities of these diseases, including P. gingivalis-mediated infection [32 , 39]. In addition, proinflammatory cytokines, such as TNF-α, IL-1β, and IL-6, are relevant to early stages of AD onset [40]. From animal models, proinflammatory cytokines, such as TNF-α, are known to disturb inter-endothelial cell-cell connections and thereby cause BBB permeability [19]. This makes TNF-α transgenic mice [31] and ApoE^–/– mice brains [27 , 30] ideal test models to investigate if they share common oxidative stress from an intrinsic biological (proinflammatory cytokine) stressor. We first established that the leaky BBB was restricted to the cerebellum and not the cerebral microvessels in both TNF-α transgenic and ApoE^–/– mice brains. We have previously demonstrated P. gingivalis hematogenous incursion from gingival epithelium to aortic vascular tissues in ApoE^–/– mice [35], and into the brain tissues [32, 34]. In addition, innate immune response stimulation was demonstrated by glia via the local synthesis and release of complement cascade proteins following P. gingivalis infection of the brain [34]. The activated complement cascade for non-specific lysing of microbes instead, led to an inappropriate attack of pyramidal neurons [34] and also resulted in accelerated occurrence of age-related granules [32]. Upon entry into the brain following 24 weeks of chronic infection, P. gingivalis infection appeared to be contributing towards hippocampal cerebral tissue injury in which impairment of the BBB was evident. This suggested a preceding negative event was taking place and was contributing to the BBB defect in the infected group. To delineate overlapping cytokine related biological events during infection, inclusion of TNF-α transgenic mice was essential. The rationale for the initiation of the present study was, therefore, to determine if vascular exposure to P. gingivalis and/or the host’s acute phase responses in the form of oxidative stress and corresponding oxidative protein damage, were responsible for the development of a leaky BBB in ApoE^–/– mice at an earlier timepoint of infection (12 weeks), which was not detected at the 24-week infection timepoint. Numerous AD studies have identified the accumulation of carbonylated proteins as a consequence of elevated oxidative stress [41 –44]. There are no studies at present that have attempted to address whether the resulting oxidative stress exposure from the invasion of microbes can lead to the development of leaky BBB. Therefore, to the best of our knowledge, this is the first study that has evaluated the putative role of P. gingivalis infection in mediating BBB deterioration and its contribution to oxidative protein damage in the ApoE^–/– mouse brain model, following bacterial infection.

It is interesting to note that the extent of oxidative protein damage was evident in the wild-type mouse brain tissue lysates, as a prominent band of 43 kDa. This data suggests that even wild-type mice experience intrinsic exposure to basal levels of ROS, due to by-products of normal physiological and metabolic processes. This led to the notion that the gradual accumulation of oxidative stress-related proteins was possible in these mice, as antioxidative and repair mechanisms are not being completely (100%) efficient [45 –49]. Although the extent of oxidative protein damage from the frontal lobe brain region in wild-type mice was detectable by western blotting, this was largely restricted to a single band at 43 kDa. The western blot profiles for the TNF-α transgenic mice differed to those of wild-type mice, particularly by demonstrating enhanced carbonylated proteins at around 25 kDa, 43 kDa, and 68 kDa. This difference may be attributable to the endogenous expression of proinflammatory cytokine and an impaired cerebellar BBB as a consequence of enhanced oxidative stress. Therefore, the endogenous antioxidant and repair systems within the brains of TNF-α transgenic mice were incapable of overcoming the corresponding increases in oxidative stress caused by the enhanced proinflammatory cytokine burden, resulting in increased protein carbonylation and host tissue damage. Differences in western blot profiles were further evident between the sham-infected mice and those infected with P. gingivalis at similar age. P. gingivalis-infected mice demonstrated the highest overall protein carbonyl content among all of the experimental groups analyzed, with numerous prominent bands identified at 25 kDa and 30–43 kDa* (exclusive to the infected group) and at 43 kDa, with lesser intensity bands at 45–70 kDa. In contrast, sham-infected mice only demonstrated a prominent band at around 25 kDa, with additional bands of highly variable degrees of intensities between 29–40 kDa and 45–70 kDa. In addition, these observations are consistent with previous reports that have shown P. gingivalis can not only survive host’s oxidative burst, but also has the ability to subvert its own oxidative stress via rubreythrin gene activation [12]. ApoE^–/– mice are also predisposed to intrinsic stress due to constitutive TNF-α expression [21, 22], oxidative stress-induction, lipid peroxidation [23], impaired immuno-modulatory function in macrophages and innate/adaptive immune responses [24 –26], and an impaired cerebellar BBB [27]. Thus, the prominent band at around 25 kDa, common to TNF-α transgenic mice, sham- and P. gingivalis-infection blots from ApoE^–/– mice, reflects the shared oxidative stress as a consequence of the constitutive expression of TNF-α and cerebellar BBB defects. However, the increased acute phase inflammatory response, as shown by detectable oxidative protein content in P. gingivalis mouse brains, particularly at 30–43 kDa*, are more a consequence of infection. Accordingly, this study supports the host’s phagocytic cell oxidative burst initiated by P. gingivalis infection/pathogenicity to initiate and perpetuate BBB damage [12].

In line with western blot data, immunohistochemical detection of protein carbonyl content in the tissue sections taken from the frontotemporal lobe inclusive of the hippocampus, only low to negligible levels were consistently seen among the wild-type and TNF-α transgenic mice brains. This suggested that these mice were exposed to oxidative stress, but that was below the threshold of detection by immunohistochemistry at the optimal dilution of the primary antibody tested. TNF-α transgenic mouse brain tissue sections showed infrequent immunostaining on myelinated nerve bundles and occasional cellular sites, particularly in non-hippocampal areas of the brain. These findings suggest certain strain and age similarities, but equally imply the existence of differences in the levels and responses to oxidative stress within these anatomical regions of the brain in the TNF-α transgenic mice. As these mice are associated with a persistent low-grade exposure to TNF-α, this alters antioxidant defenses and perpetuates oxidative stress and tissue damage [50, 51]. Therefore, the detection of increased protein carbonyl levels in these mice is expected, as these mice appear to be less efficient in counteracting the elevated levels of oxidative stress induced by TNF-α and the cerebellar BBB defect. In addition, live P. gingivalis infection is likely to exert its effect on the host through other virulence factors, such as its proteolytic activity mediated by gingipains [13 –15]. Concerning the microvascular structure of the brain, it is plausible to extrapolate relevant information from cell-cell junctional protein (catenins, occludin, E-cadherin, and β₁-integrin) disruption from in vitro infection data already documented [13 –15]. Therefore, endothelial tight junction protein disruption by this periodontal pathogen to facilitate self-accessibility across the BBB is highly plausible and confirms IgG leakage into the microvasculature at later timepoints of infection [32].

It is clear that both sham- and P. gingivalis-infection of ApoE^–/– mice at 12 weeks exhibited significant detectable protein carbonyl content by western blotting, but immunohistochemistry showed the major difference in its staining pattern, which was the microvasculature in the hippocampus region of the brain. This observation confirms that P. gingivalis infection plays a significant role in increasing the hosts’ oxidative stress, which subsequently has a significant pathological impact on the cerebral microvasculature. It is reported that the cerebral BBB is breached among the elderly humans and AD subjects [52], but the reasons behind it are poorly understood. One plausible reason is that the episodic nature of periodontal disease, in which patients experience recurrent periods of quiescence and active disease. The biofilm bacteria of the active (recurring) phase of periodontitis are likely to be those that have been newly ‘radicalized’ as suggested by Harding et al., [53] to possess greater inflammophilic virulent properties. It is highly possible that these cyclic exposures of P. gingivalis infections together with its ‘radicalized’ companion species occur during recurrent bacteremias [54], which could contribute to eventual loss of cerebral BBB integrity and function.

These findings imply that P. gingivalis infection significantly increases the overall oxidative stress within the brains of ApoE^{– / –} mice, resulting in enhanced cellular and molecular damage, culminating in an accumulation of protein carbonyl content. P. gingivalis is a well-established mediator of infection and host-derived, tissue damage during periodontitis, where similar to AD, oxidative stress has been implicated in the pathogenesis of disease [1 –7]. Indeed, P. gingivalis levels have been shown to correlate with oxidative stress biomarker levels during periodontal disease and other inflammatory diseases such as atherosclerosis, where this bacterium has been implicated [55, 56]. Therefore, the present findings suggest that P. gingivalis infection is capable of inducing significant immuno-inflammatory responses and associated oxidative stress in the hippocampi of ApoE^{– / –} mice, which is characteristically prominent in the microvascular regions.

In conclusion, this study has revealed a major difference in the hippocampi of P. gingivalis-infected and sham-infected ApoE^{– / –} mice, in terms of increased protein carbonyl/oxidized protein content in the hippocampal micro-vasculature. We, therefore, suggest that hippocampal microvascular structures and the homeostasis of the brain are at risk from elevated oxidative stress and oxidative protein damage, following P. gingivalis infection. This observation has significant implications as it suggests that, following recurrent episodes of active periodontal disease, there exists a possibility for the development of a defective BBB, post neuroinflammation-mediated cerebral parenchymal tissue injury. It is, therefore, highly plausible that the rising levels of intrinsic and extrinsic sources of cytokines, oxidative stress, and developing BBB defects are crucial early modifiers of neurodegeneration and disease severity leading to deteriorating memory. Thus, infection with P. gingivalis should be interpreted as one of the plausible mechanisms by which a susceptible host can develop dementia.