Alzheimer's disease (AD) is a devastating neurodegenerative disorder affecting a growing number among the aging population. Although its etiology remains to be elucidated, pathological evidence indicates that, in addition to the characteristic senile plaques and neurofibrillary tangles, significant brain inflammatory changes, featured with exaggerated gliosis and increased proinflammatory cytokines, are an important component of the pathology 12.

Inflammation is a double-edged sword. For instance in AD, an appropriate degree of inflammatory reaction may be beneficial for clearance of aggregated proteins and/or degenerated cells. However, if the inflammation is overly reactive, the secondary inflammatory neuronal damage may become a driving force for the disease, as appears to be the case in AD 1. Therefore, it is important to understand what factors may exaggerate the brain inflammation in AD so that corresponding therapeutic approaches can be developed to ameliorate such changes.

The extent of inflammatory reactions may be regulated by a variety of factors, such as proinflammatory cytokines, chemokines, complement systems, and others. In AD, fibrillar β-amyloid (Aβ) has been shown to directly activate microglial cells and result in increased levels of proinflammatory cytokines, such as interleukin-1 beta (IL-1β) and tumor necrosis factor-α 123. In addition, it can activate the complement system to generate C3a and C5a 4. Moreover, fibrillar Aβ accumulation is associated with increased levels of many chemokines and chemokine receptors, such as monocyte chemoattractant protein-1 (CCR2 ligand), macrophage inflammatory protein 1 (CCR3 ligand), interferon-inducible protein-10 (CXCR3 ligand), IL-8 (ligand for CXCXR1 and CXCR2) and CCR1/3/5. 567. Amongst the inflammatory regulators associated with AD, there is one large category that exerts their cellular effects through heterotrimeric G-protein-coupled receptors (GPCRs). These include all the chemokines 6 and the anaphylatoxins C3a and C5a 89.

In a recent study, we showed that deficiency of membrane GPCR kinase-5 (GRK5) occurs during early AD, and the GRK5 deficiency contributes to augmented microglial activation in vitro via impaired sensitization of relevant GPCRs 10. GRK5 is one of seven GRK family members, whose primary function is to desensitize activated GPCRs by phosphorylating the activated receptor to promote uncoupling of receptor-dependent G-protein activation and by initiating receptor internalization 1112. Therefore, the lack of membrane (functional) GRK5 resulted in prolonged signaling and enhanced microglial activation, which well explained the phenomena that we observed in the cultured microglial cells 10. Moreover, we speculated that the same phenomena might also take place in vivo. However, our later study in GRK5 deficient mice (GRK5KO), an animal model resembling the functional GRK5 deficiency in AD, failed to show any sign of microgliosis, except for moderate increase in axonal defects and synaptic degenerative changes in these animals 13. The negative in vivo data appeared to be inconsistent with our earlier in vitro data. Nonetheless, a hallmark of GRK action is to function only after GPCRs are activated 1112. In other words, whether or not an impact of GRK5 deficiency on a particular stimulus can be revealed depends on whether or not such a stimulus is present. We have reported that, although the GRK5 deficiency indeed caused a mild increase of Aβ level, no fibrillar Aβ deposits or any significant signs of microglial activation were observed in the GRK5KO mice 13. Therefore, we speculated that one reason for the absence of microgliosis in these animals might be due to lack of inflammatory initiators, such as fibrillar Aβ, or any of the downstream chemokines or anaphylatoxins.

In this study, we crossbred the GRK5KO mice with TgAPPsw (Tg2576) mice. TgAPPsw mice overexpress human β-amyloid precursor protein (βAPP) carrying Swedish double mutations, and have been shown to display fibrillar Aβ deposition and moderate levels of inflammation 141516. Therefore, the resultant TgAPPsw mice deficient in GRK5 should provide an ideal model to determine if our previous speculations are correct, and GRK5 deficiency plays a significant role in enhancing brain inflammation in AD.

In order to determine whether GRK5 deficiency has an impact on brain inflammation in AD, we crossbred the TgAPPsw mice with the GRK5KO mice and generated the TgAPPsw mice deficient in GRK5. Only heterozygotes were used in this study, which included four genotype groups: WT control (APPsw^-/-/GRK5^+/+), GRK5KO heterozygote control (APPsw^-/-/GRK5^+/-), TgAPPsw (APPsw^+/-/GRK5^+/+) and the double (APPsw^+/-/GRK5^+/-) mice. As shown in Fig. 1, inactivation of one copy of the GRK5 gene resulted in at least 50% down-regulation of the GRK5 expression at the protein level in the heterozygote GRK5KO and the double mice.

Evaluation of potential changes in brain inflammation was performed in 18-month old female mice, due to the known gender impact in these mice 181920 [Li, 2008, BRB-submitted]. Immunofluorescent staining with antibodies against microglial and astrocyte markers was employed to reveal changes in microgliosis and astrogliosis. Our initial examination with antibody against CD45, a surface marker of microglia 15, revealed an increase of microgliosis in both hippocampal (Fig. 2A–D) and cortical (Fig. 2E–H) areas of the double mice as compared to the TgAPPsw mice. Subsequent characterization of the microgliosis showed that, consistent with previous findings 15, CD45⁺ microglial cells were largely located surrounding or near Aβ deposits (Fig. 3A–C). Staining with a rat mAb to CD11b (Mac-1), another well-established microglial and macrophage marker 1521, revealed similar changes of microglia in these animals (Fig. 3D–F). In addition, a mouse mAb OX-42 to CD11b/c also decorated the activated microglial cells in a way similar to anti-CD45 (Fig. 3G). However, at least in our experiments, OX-42 appeared to non-specifically stain the Aβ plaques, in addition to its positive staining for microglial cells. It has been previously established that activated microglial cells, at least at the early phase, produce IL-1β, an important proinflammatory cytokine that participates in a self-sustained vicious cycle to amplify the inflammatory neuronal damage 2. Our examination revealed that, although less profound than CD45 immunoreactivity (IR), IL-1β-IR was indeed found near or surrounding the Aβ plaques (Fig. 3H) and was colocalized with CD45⁺ microglial cells (Fig. 3I). Therefore, the microgliosis in the double mice displayed similar traits to that observed in other AD transgenic models, except for the exaggerated extent.

To better understand the extent of the exaggerated microgliosis in these animals, we quantified CD45⁺ microglial cells per mm² in hippocampus and cortex. Our quantitative image analysis (Fig. 4A &4B) showed that, as compared to the WT mice, both the TgAPPsw and the double mice displayed significantly more CD45⁺ microglial cells in both of the examined brain regions (p < 0.001 for all); when compared to the TgAPPsw mice, the double mice had approximately double the number of CD45⁺ microglial cells, in both hippocampus and cortex (P < 0.001). In addition, the heterozygote GRK5KO mice did not show any significant changes in the number of CD45⁺ microglial cells, as compared to the WT, which is consistent with our previous report for the negative finding of brain inflammation in the GRK5KO mice 13. When two-way ANOVA was used to analyze interactions between the two gene modifications (over-expression of human APPsw and inactivation of one copy of the murine GRK5 gene), a significance of p < 0.001 was revealed. Meanwhile, the changes caused by both gene modifications in the double mice were significantly higher than the sum of the changes caused by APPsw over-expression in TgAPPsw and by GRK5 gene inactivation in GRK5KO mice (p = 0.003 for hippocampus and p < 0.001 for cortex). These results indicate that the two gene modifications synergistically interact with each other, and both contribute the microgliosis observed in the double mice.

In parallel to the microglial analysis, potential changes of astrogliosis in these animals were also assessed using IF staining with antibody to GFAP, an astrocyte-specific intermediate filament protein. We found more hypertrophic astrocytes brightly stained with anti-GFAP in both hippocampus and cortex from the double mice as compared to the TgAPPsw mice, and both the double mice and the TgAPPsw mice showed more hypertrophic astrocytes than the WT or heterozygous GRK5KO mice (Fig. 5A–H). Further characterization of the astrogliosis in the double mice revealed that although some of the hypertrophic astrocytes located near Aβ plaques (Fig. 6A–C), compared to the relation of microglia to Aβ plaques, hypertrophic astrocytes were more widespread in the entire brain region. In addition, at least some of the reactive astrocytes were IL-6 positive (Fig. 6D–F). Quantitative image analysis (Fig. 7A &7B) indicated that both the TgAPPsw and the double mice showed significantly more GFAP⁺ astrocytes in the hippocampus and cortex (p < 0.001 for all) as compared to either the WT or the heterozygous GRK5KO mice; furthermore, the double mice displayed greater numbers of astrocytes than that in the TgAPPsw mice (p < 0.01 for hippocampus and p < 0.001 for cortex). Consistent with our previous report 13, the number of GFAP⁺ astrocytes was not significantly different between the heterozygous GRK5KO and the WT mice. When two-way ANOVA was employed, significant interactions (p < 0.05 for hippocampus, and p < 0.01 for cortex) were revealed between the two gene modifications. The comparisons of effects of the double gene modifications with the sum of the effects of the each single gene modification alone revealed significant difference for the cortex (p = 0.02) but not for the hippocampus (p = 0.10).

It is known that both resting and reactive astrocytes express GFAP, but the latter usually expresses higher level of GFAP, and therefore stain brighter. Because of this, counting the number of GFAP⁺ astrocytes may not accurately reflect the extent of the astrogliosis. Therefore, a quantitative image analysis of GFAP-IR intensity was also performed in parallel to the GFAP⁺ cell counting. The results were similar to those from simple cell number counting, except that the significance was more striking (Fig. 7C &7D). The double gene modification effects were significantly higher than the sum of the each single gene alone for both hippocampus (p = 0.006) and cortex (p = 0.009). The interactions between the two gene modifications were also highly significant for both brain regions (p = 0.002 for hippocampus, and p = 0.005 for cortex). Therefore, these results suggest that Swedish mutated APP gene over-expression and GRK5 deficiency synergistically interact with each other to promote the astrogliosis observed in the double mice.

By analyzing the extent of gliosis in the brains of TgAPPsw mice with normal or reduced GRK5 expression, we demonstrated that the TgAPPsw mice with reduced GRK5 displayed significantly exaggerated microgliosis and astrogliosis. Consistent with our previous report 13, GRK5 deficiency alone does not lead to any significant inflammatory changes in either the heterozygote or homozygote GRK5KO mice. However, when an inflammation is present, as evoked by overexpression of the Swedish mutated APP gene, the GRK5 deficiency synergistically interacts with the active inflammatory processes and leads to amplified inflammatory responses. Such phenomena are consistent with the manner of GRK action, as GRKs are not thought to induce signals directly, but rather, they primarily modify the activity of activated GPCRs by uncoupling the receptor from G-protein, and initialing receptor internalization 1112. In the GRK5KO mice, the reason why we did not observe a significant inflammation is likely because there was no an active inflammatory signaling for the GRK5 to act on. In the double mice, however, when Swedish mutated APP was overexpressed, the accumulation of fibrillar Aβ initiated a complex of active inflammatory processes, which provided an opportunity to reveal impact of GRK5 deficiency. As demonstrated in this study, inactivation of one copy of the native GRK5 gene led to approximately doubled extent of microgliosis and significantly exaggerated astrogliosis. Therefore, these in vivo results strongly support the model that GRK5 deficiency indeed plays a significant role in amplifying the brain inflammatory responses in AD, as we previously speculated.

Since the primary role of GRKs is to desensitize GPCRs and dampen the activated GPCRs signaling, deficiency of GRK5 may lead to hyperactivity of those GPCRs that are specifically regulated by GRK5. As previously demonstrated, the fibrillar Aβ can initiate a complicated inflammatory process, which may include increased levels of proinflammatory cytokines, chemokines, and activated complement components 1234567. It is known that receptors for all chemokines and the anaphylatoxin C3a and C5a are GPCRs 689. In fact, even fibrillar Aβ itself has been shown to activate GPCRs, such as macrophage scavenger receptor, or formyl chemotactic receptor 2 (FPR2 or FPRL-1) 222324. Therefore, theoretically, all these GPCRs could be potential substrates for GRK5. On the other hand, however, the substrate spectrums for different GRK members may overlap, and the lost function due to deficiency of one GRK member may be largely compensated by other GRK members 1112. Only when the lost function cannot be fully compensated, a phenotype relevant to the specific GPCR signaling can then be revealed. For example, GRK2, GRK5, and GRK6 KO mice have been shown to display selectively impaired desensitization of adrenergic, muscarinic and dopaminergic receptors, respectively 172526. Unfortunately, to the best of our knowledge, information regarding the specific regulation of GRK5 on most, if not all, of the GPCRs involved in the fibrillar Aβ initiated inflammatory processes, particularly the in vivo data, is unavailable. Therefore, further investigations are warranted to understand the detailed molecular mechanisms by which the GRK5 deficiency exaggerates the brain inflammation in the TgAPPsw mice.

The main purpose of this study was to determine whether or not GRK5 deficiency in vivo contributes to exaggerate brain inflammation, and the results confirmed the positive role of GRK5 deficiency in the TgAPPsw mice. In the meantime, the fact that the double mice displayed significantly stronger inflammatory changes than those in the TgAPPsw mice makes the double mice a distinct model from the original TgAPPsw mice. It has been suggested that inflammatory responses in currently available APP transgenic mouse models, including the TgAPPsw mice, are not as strong as those in human AD patients 16. Therefore, at least in resembling the strength of the inflammatory responses in AD, the TgAPPsw mice with reduced GRK5 may be a better model than the single TgAPPsw mice, and should serve as an additional AD animal model for relevant studies.

This study has demonstrated that GRK5 deficiency in vivo significantly exaggerates the microgliosis and astrogliosis in the presence of an inflammatory initiator, the excess fibrillar Aβ and the subsequent active brain inflammatory reactions in the TgAPPsw mice. In the absence of an active inflammatory response, however, the effects of the GRK5 deficiency on inflammation in vivo remain silent. These findings, along with our previous findings for the role of GRK5 deficiency in promoting axonal defects and synaptic degeneration 13 [Li, 2008, BRB-submitted], strongly support that the hypothesis that GRK5 deficiency plays a significant role in AD pathogenesis. In addition, the TgAPPsw mouse combined with deficient GRK5 is a better model than the single TgAPPsw mouse, due to a stronger resemblance to the strength of the inflammatory responses in AD.

AD: Alzheimer's disease; Aβ: β-amyloid; GPCRs: G-protein coupled receptors; GRK5: GPCR kinase-5; GRK5KO: GRK5 knockout; βAPP: β-amyloid precursor protein; PCR: polymerase chain reaction; PBS: phosphate buffered saline; IF: Immunofluorescent; mAb: monoclonal antibody; pAb: polyclonal antibody; GFAP: glial fibrillary acidic protein; IL-1β: interleukin-1beta; IL-6: interleukin-6; IR, immunoreactivity.

The authors declare that they have no competing interests.

LL carried out the pathological studies, performed the statistical analysis, and participated in the maintenance of the transgenic mice and initial drafting of the manuscript, JL participated in the establishment and maintenance of the transgenic mouse colonies, WZS designed the study, supervised and coordinated the experimental conduction, and helped to draft the manuscript. All authors read and approved the final manuscript.