N-Adamantyl Phthalimidine: A New Thalidomide-like Drug That Lacks Cereblon Binding and Mitigates Neuronal and Synaptic Loss, Neuroinflammation, and Behavioral Deficits in Traumatic Brain Injury and LPS Challenge

Abstract

Neuroinflammation contributes to delayed secondary cell death following traumatic brain injury (TBI), has the potential to chronically exacerbate the initial insult, and represents a therapeutic target that has largely failed to translate into human efficacy. Thalidomide-like drugs have effectively mitigated neuroinflammation across cellular and animal models of TBI and neurodegeneration but are complicated by adverse actions in humans. We hence developed N-adamantyl phthalimidine (NAP) as a new thalidomide-like drug to mitigate inflammation without binding to cereblon, a key target associated with the antiproliferative, antiangiogenic, and teratogenic actions seen in this drug class. We utilized a phenotypic drug discovery approach that employed multiple cellular and animal models and ultimately examined immunohistochemical, biochemical, and behavioral measures following controlled cortical impact (CCI) TBI in mice. NAP mitigated LPS-induced inflammation across cellular and rodent models and reduced oligomeric α-synuclein and amyloid-β mediated inflammation. Following CCI TBI, NAP mitigated neuronal and synaptic loss, neuroinflammation, and behavioral deficits, and is unencumbered by cereblon binding, a key protein underpinning the teratogenic and adverse actions of thalidomide-like drugs in humans. In summary, NAP represents a new class of thalidomide-like drugs with anti-inflammatory actions for promising efficacy in the treatment of TBI and potentially longer-term neurodegenerative disorders.

Traumatic brain injury (TBI), a major cause of mortality and morbidity, affects an estimated 69 million people worldwide with limited treatment options. From all cases, mild TBI accounts for approximately 81% of brain injuries, with moderate TBI accounting for 11% and severe TBI accounting for 8%.¹ In addition to the acute damage and accompanying functional impairments caused by TBI, it acts as a conduit to the development of chronic neurodegenerative disorders, in particular, Parkinson’s disease (PD) and Alzheimer’s dementia (AD).²⁰⁰⁻⁴

TBI is caused by primary mechanical damage during the initial impact, which is followed by secondary neurochemical and intracellular processes that give rise to complex neurological deficits underpinned by imbalances of intracellular and extracellular ions, neuronal loss, microglial activation, astrocytic gliosis, and the infiltration of blood leukocytes.^5,6 An immune response is activated by pro-inflammatory cytokines and chemokines released from various immunocompetent cells. The parenchymal and peripheral immune cells that are attracted by these cytokines and chemokines further amplify the immune response and attract yet more immune cells to the local injury site.⁷

Tumor necrosis factor-α (TNF-α) is a pro-inflammatory cytokine that is rapidly produced in response to various deleterious stimuli, predominately by monocytes in the periphery and by microglia, neurons, and astrocytes within the central nervous system (CNS).⁸ The potent pro-inflammatory properties of TNF-α act as key regulators of acute stage inflammation, triggering inflammatory cytokine signaling cascades.⁹ TNF-α generation and release initiates reparative mechanisms, but in excess, it has been found to be associated with neurodegenerative processes and disorders, including TBI,^10,11 PD,^12,13 AD,^14,15 multiple sclerosis (MS),^16,17 and amyotrophic lateral sclerosis (ALS).^18,19

Within minutes to hours of a TBI, intracerebral mRNA and protein levels of TNF-α are significantly elevated in patients.²⁰ In TBI rodent models, elevated TNF-α mRNA can be detected prior to the appearance of other cytokine proteins, and the process of leukocyte infiltration to the injury site is enhanced by upregulation of TNF-α.²¹⁻²³ Changes in TNF-α expression levels reflect the severity of TBI, since reports have shown that a substantial elevation of TNF-α can be detected in moderate to severe TBI models, whereas there is only modest change in mild TBI models.²⁴ In response to TBI, the prompt induction of TNF-α can be harmful,²⁵⁻²⁷ although the temporal and spatial control of its production and activity may be subsequently essential for long-term recovery.²⁶⁻²⁸

Numerous reports suggest that neuroinflammation is a key feature in TBI. Initial neuronal death induced by TBI may result in neuroinflammatory responses that lead to subsequent delayed cellular damage.^23,24,29 Hence, reducing the initial production of TNF-α may provide a potential treatment strategy for TBI.

Thalidomide (α-phthalimidoglutarimide), initially developed as a sedative alternative to barbiturates in the 1950s, was later discovered to be a potent inhibitor of TNF-α synthesis.³⁰ However, because of its adverse teratogenic effects,^31,32 novel analogs have since been synthesized and evaluated aimed at increasing drug potency and reducing adverse effects. In this regard, thalidomide and analogs have been found to possess pharmacologically valuable antiangiogenic, antiproliferative, and anti-inflammatory properties and are effectively used in the treatment of multiple myeloma and other cancers.^33,34 A key feature of thalidomide, its metabolites, and clinically useful analogs is a glutarimide ring conjoined to a phthalimide moiety (Figure 1). A primary target for which teratogenicity and key pharmacological actions are considered to be mediated is the protein cereblon, a substrate receptor for the ubiquitin E3 ligase complex Cullin-RING ligase 4 (CRL4).³⁵ CRLs play an essential role in targeting proteins for ubiquitin-mediated proteolysis.³⁶ Interaction of thalidomide and clinically used analogs with CRL4 is considered mediated by glutarimide ring interactions within a restrictive shallow binding pocket on the surface of cereblon.³⁷ The flat, planar glutarimide moiety hydrogen bonds with key amino acids within the pocket to align the phthalimide group to protrude beyond the surface of cereblon to support its interaction with substrates, such as zinc finger transcription factors ikaros (IKZF1), aiolos (IKZF3), and Sal-like protein 4 (SALL4), as well as targets that lack zinc fingers like Casein kinase CK-1α, to promote their proteosomal degradation.³⁶

Figure 1.

Biological activity of NAP is not mediated by Cereblon. The binding of thalidomide analogs to cereblon was examined by using a cereblon/BRD3 binding FRET assay. (A) Chemical structures of Immunomodulatory drugs (IMiDs), thalidomide (Thal), pomalidomide (Pom), lenalidomide (Len), and NAP evaluated. (B) An initial concentration-dependent evaluation of binding between Pom and cereblon provided an IC₅₀ value of approximately 2.8 μM. Based on this, (C) a single concentration of 5 μM was evaluated for Thal, Len, Pom and NAP, vs vehicle Control (Con), which was set at 0% cereblon inhibition (***, p < 0.001 vs NAP (Tukey’s multiple comparisons test), NAP was not statistically different from the control value). Thereafter, the thalidomide analog-mediated (10 μM) degradation of ikaros and aiolos in the MM1.S cell line (D) and SALL4 in H9 hES cells (with 20 μM Thal and analogs) (E) was evaluated by Western blotting. (F) Replication of the actions of Thal, Pom and NAP on SALL4 levels was evaluated in H9 hES cells, with (G) quantification relative to β actin expression. ***, p < 0.001 vs control (Con) value; ^###, p < 0.001 vs NAP value (Tukey’s multiple comparisons test).

Our recent medicinal chemistry efforts utilized a “phenotypic drug discovery” strategy⁶⁰ and resulted in the generation of 15 novel and previously uncharacterized adamantly and noradamantly phthalimidines that, like thalidomide and its clinically used analogs, possess anti-inflammatory action when evaluated in a phenotypic anti-inflammatory screen focused toward lowering TNF-α and nitrite generation.³⁸ These agents possess a caged adamantane structure comprised of three condensed cyclohexane rings fused in an armchair conformation that provides a diamond-like three-dimensional structure, in lieu of a glutarimide ring (Figure 1). The adamantane moiety can provide unique pharmacological properties when incorporated into the structure of an already active drug,^39,40 such as thalidomide, to augment lipophilicity and potentiate blood-brain barrier (BBB) uptake, as well as to provide steric bulk to potentially reduce metabolic liability and hinder/modify drug-target binding interactions, as occurs between thalidomide and cereblon.

In the current study, we evaluated a particularly promising adamantyl-based phthalimidine across cellular and animal models of inflammation, and in a mouse model of moderate TBI. Specifically, using phenotypic based screens,^60,59 we found that N-adamantyl phthalimidine (2,3-dihydro-2-[1-(1-tricyclo[3.3.1.1^3,7]dec-1-yl)ethyl]-1H-isoindol-1-one) (NAP) significantly decreased multiple inflammatory markers in cellular and animal models of lipopolysaccharide (LPS)-induced inflammation, reduced α-synuclein-mediated cell death and inflammation in primary cultures of dopaminergic neurons and microglia, and mitigated behavioral, biochemical, and histological deficits caused by controlled cortical impact (CCI) TBI. Use of a fluorescence resonance energy transfer (FRET)-based assay⁴¹ to evaluate drug–target interactions demonstrated a lack of NAP binding to cereblon and no subsequent depletion of aiolos or SALL4.

Results

Cellular Studies

NAP Cereblon Binding and Downstream Action on Ikaros, Aiolos, and SALL4

A substantial and significant liability in the clinical use of thalidomide analogs is their potential to induce tertatogenesis.¹¹² To investigate whether or not NAP biological activity is mediated by cereblon within the CRL4 complex, similar to other thalidomide analogs (Figure 1A), a cereblon/BRD3 binding FRET assay was employed to evaluate the inhibitory effects (Figure 1B). As illustrated in Figure 1C, NAP lacked competitive inhibition of cereblon binding to BRD3, whereas pomalidomide, lenalidomide, and thalidomide all inhibited the interaction between cereblon and BRD3. This suggests that NAP does not bind to cereblon.

The binding of thalidomide analogs to cereblon induces degradation of neosubstrates of cereblon, such as aiolos, ikaros, and SALL4, which are considered to underpin the antiproliferative and teratogenic actions of this class of drugs.³⁶ Whereas the expressions of aiolos, ikaros, and SALL4 were decreased by thalidomide, lenalidomide, and pomalidomide, they were unaffected by NAP (Figure 1D–G).

NAP Mitigates Markers of Inflammation in RAW 264.7 Cells Challenged with LPS and Primary Neuronal Cultures Challenged with α-Synuclein

As an initial phenotypic screen to suggest potential anti-inflammatory action, NAP was evaluated in a concentration-dependent manner in RAW 264.7 cells challenged with a submaximal concentration of LPS. This immortal mouse macrophage cell line possesses some features common to microglia and responds to LPS challenge by eliciting a classical inflammatory response.²⁰⁶ Pretreatment with NAP induced a substantial and statistically significant 48% decline in nitrite levels, a stable end product and surrogate measure of nitric oxide (NO) metabolism,²⁰³ and lowered TNF-α generation by 45% (p < 0.001 and p < 0.05, respectively; Figure 2A).

Figure 2.

NAP mitigates LPS and α-synuclein-induced neuroinflammation in cultured RAW 264.7 cells and primary neuronal and microglial cell cultures. (A) Cultured RAW 264.7 cells were pretreated with either vehicle or NAP (10–100 μM) and challenged with LPS (60 ng/mL) 1 h later. At 24 h following LPS challenge, cellular viability, nitrite (a stable marker of NO generation), and TNF-α levels were quantified. NAP was well-tolerated and maintained cellular viability at >90% across all concentrations evaluated and significantly lowered LPS-induced elevations in nitrite and TNF-α levels. ***, p < 0.001 vs the control (LPS + Veh) group. Mixed primary cultures of neurons and microglia were challenged with oligomeric forms of α-synuclein (250 nM) (B) or Aβ (5 uM) (C) for 72 h, and their survival was quantified in the presence and absence of NAP in a concentration-dependent manner. The addition of oligomeric forms of α-synuclein (250 nM) to a primary coculture of dopaminergic neurons and microglia resulted in (D) a reduction in neurite number per cell, (E) an elevated expression of OX-41-positive microglial cells, indicative of microglial activation, and (F) an elevated release of TNF-α. Pretreatment with NAP significantly mitigated these α-synuclein-induced effects. Mean ± SEM (n = 5–6 per group). *, p < 0.05 vs the α-synuclein alone (red bar) group.

In the light of reports of TBI-induced elevations in brain α-synuclein^200,201,106 and Aβ levels,⁴² we evaluated the ability of NAP to mitigate the toxicity of aggregated α-synuclein and Aβ in mixed primary neuronal cultures. This is additionally pertinent given that an increasing number of epidemiological reports describe an increased incidence of PD and Alzheimer’s dementia following TBI.²⁻⁴ As illustrated in Figure 2B,C, NAP significantly mitigated neuronal cell death (33 and 29% cell loss) induced by oligomeric α-synuclein and Aβ, respectively. In evaluating mechanisms underpinning cellular dysfunction, neurite number was quantified in dopaminergic cultures by TH immunostaining and was found to be significantly reduced by α-synuclein (50%, Figure 2D). In contrast, microglia activation, evaluated by OX-41 immunostaining, was significantly elevated (141%, Figure 2E) and resulted in a rise in TNF-α levels in these same cultured cells (175% of controls, Figure 2F). Each of these changes was significantly inhibited by NAP.

Together, these studies demonstrate across phenotypic models of neuroinflammation that NAP, which does not appear to bind cereblon and cause aiolos, ikaros, and SALL4 degradation, possesses proinflammatory cytokine lowering actions, in line with other thalidomide analogs.¹⁸ Such actions can result in improved neuronal cell survival following a pathologically relevant challenge (i.e., with α-synuclein or Aβ).

Animal Studies

NAP Lowers TNF-α Levels Induced by LPS Administration in Rodents

As an initial evaluation of whether the anti-inflammatory actions evident in cellular studies translate in vivo, the ability of NAP to mitigate LPS-induced elevations in TNF-α was assessed in rats. In line with prior studies,¹⁰⁸ systemic administration of LPS resulted in significant and marked elevations in plasma and brain levels of TNF-α and IL-6 (Figure 3A,B), accompanied by significant and smaller increases in IFN-γ (Figure 3C) and anti-inflammatory IL-10 and IL-13 levels (Figure 3D,E). Systemic administration of NAP (13.3 and 26.6 mg/kg intraperitoneally (i.p.)), significantly lowered LPS-induced plasma and brain TNF-α levels and brain IL-6 levels and demonstrated a trend to lower IFN-γ, notably, without altering anti-inflammatory IL-10 or IL-13.

Figure 3.

NAP significantly reduces LPS-induced proinflammatory (TNF-α, IL-6) cytokines without altering anti-inflammatory (IL-10 and IL-13) cytokine levels in plasma and brain. Treatment of animals with LPS (1 mg/kg i.p.) markedly elevated levels of proinflammatory TNF-α (A), IL-6 (B), and IFN-γ (C) and anti-inflammatory IL-10 (D) and IL-13 (E) in the plasma and brain. Pretreatment of the animals with NAP (13.3 and 26.6 mg/kg i.p.) mitigated the LPS-induced increases in pro-inflammatory proteins, without affecting anti-inflammatory IL-10 or IL-13. *, p < 0.05; **, p < 0.01; ***, p < 0.001; and ****, p < 0.0001, refer to the effects of LPS compared to the control value (i.e., Veh or NAP 0). ^#, p < 0.05, and ^##, p < 0.01, refer to the effect of drug treatments vs the LPS + Veh. Values are presented as mean ± S.E.M., of n observations (Veh, n = 4; LPS + Veh, n = 5; LPS + NAP, n = 4 each dose).

NAP Mitigates Multiple Behavioral Impairments Induced by TBI

To evaluate whether NAP-mediated mitigation in proinflammatory status is of biological relevance, NAP was administered following an acute moderate cerebral injury in mice. As illustrated in Figure 4, following a baseline behavioral evaluation 7 days earlier, mice were challenged with CCI TBI (day 0), received either NAP or vehicle 5 h later and a further 24 h thereafter, and were again behaviorally evaluated 1 and 2 weeks later. The EBST protocol was performed to investigate impairments in motor asymmetry in TBI mice. As shown in Figure 5A, both TBI–saline and TBI–NAP groups demonstrated a higher swing preference toward their contralateral side, as compared with their performance prior to injury (PRE). However, TBI mice subsequently treated with NAP showed significantly less contralateral turns at 1 week after TBI (p < 0.001). This difference between TBI–saline and TBI–NAP groups was no longer evident at 2 weeks after TBI, in large part due to the slow gradual recovery of the TBI–saline group in this measure. NAP can thus be considered to quicken recovery of TBI-induced motor symmetry impairment.

Figure 4.

Experimental design/time line of NAP efficacy evaluation in TBI. Mice were assessed for their baseline values of sensory/motor behavior, motor coordination/balance, and postural asymmetry functions using an adhesive removal test (ART), a beam walking test (BWT), and an elevated body swing test (EBST) 1 week (wk) pre-CCI TBI injury. On day 0, mice were challenged with TBI or a sham procedure, and 5 h later, the mice received their first injection of NAP (30 mg/kg body weight i.p.) or saline. The next day (24 h following TBI), a second injection was administered. Behavioral tests were performed at 7 and 14 days post-TBI; thereafter, mice were euthanized for assessment of contusion volume and histology/immunocytochemistry.

Figure 5.

NAP treatment improved TBI functional recovery, as evaluated by behavioral measures. (A) Postural asymmetry was evaluated by elevated body swing test (EBST). CCI TBI-induced deficits were mitigated by NAP treatment (30 mg/kg body weight i.p.) one and 2 weeks after injury. (B) Sensory/motor function was measured by the adhesive removal test. Mice spent a greater time to remove an adhesive sticker from their contralateral front paw compared to the ipsilateral side after TBI injury. Treatment with NAP significantly reduced this behavioral impairment. (C, D) TBI induced deficits in motor coordination and balance were evaluated by a beam walking test. TBI-challenged mice took a longer time to traverse the beam (C) and, additionally, had more contralateral foot falls (D), compared with their ipsilateral side, at 1 week post injury. Those treated with NAP showed significantly less behavioral abnormalities in beam transit time and foot falls, compared with the TBI+saline group. Analysis by two-way ANOVA followed by Bonferroni’s test: **, p < 0.01; ***, p < 0.001. Data are expressed as mean ± SEM; n = 9 (TBI+Saline), n = 24 (TBI + NAP).

The ART paradigm was used to examine somatosensory function in the same mice following TBI + saline or TBI + NAP treatments. Both groups of mice spent a longer time removing the sticker from their contralateral front paw as compared with their performance at the PRE time point. However, NAP treatment of TBI mice resulted in a significantly shorter time removing stickers from their contralateral paw when compared with the TBI + saline group at 2 weeks after TBI (p < 0.001), although there was no difference between TBI + saline and TBI + NAP groups at 1 week after TBI (Figure 5B).

The BWT paradigm was used to evaluate motor coordination in TBI mice and involved two independent measures (average transit time (Figure 5C) and number of contralateral foot falls (Figure 5D) while traversing the beam). The average transit time and contralateral foot falls were largely increased after TBI, compared with the PRE time point, and reached statistical significance in TBI + saline mice at 1 week post injury. Notably, NAP treatment significantly alleviated these TBI-induced motor deficits at 1 week (p < 0.001) (Figure 5C,D). A slower but gradual recovery of these parameters in TBI + saline mice was evident at 2 weeks postinjury. In summary, across multiple behavioral measures, NAP administration post-TBI resulted in either more rapid or a significantly greater mitigation of impairments.

NAP Treatment Attenuates TBI-Induced Lateral Ventricle Size Enlargements

Histological evaluation was performed to measure the contusion size and lateral ventricle enlargement induced by TBI. The contusion size is presented as a percentage of the contralateral hemisphere for each group at 2 weeks after TBI. Following CCI TBI, there was a marked brain tissue loss in the ipsilateral cortex (Figure 6A). As illustrated in Figure 6B, TBI induced a significant lesion volume, as compared to the sham group (p < 0.001) in both saline and NAP post-treated animals, and there was no difference between these two groups.

Figure 6.

Postinjury treatment with NAP (30 mg/kg body weight i.p.) mitigates CCI TBI-induced swelling of the lateral ventricle (LV) (C), but the contusion volume (B) is not affected. The LV size and contusion volume were quantified using ImageJ 1.52q software. (A) Representative Giemsa-stained coronal brain sections in Sham (control without TBI), TBI+Saline and TBI + NAP mice at 2 weeks post-TBI. (B) A significant elevation of lesion size was observed in TBI+Saline and TBI + NAP groups. ***, p < 0.001 compared with the sham group. (C) Significant differences in LV size ratio of ipsilateral and contralateral sides were seen between NAP-treated and saline-treated groups (**, p < 0.01; ***, p < 0.001). Analysis by one-way repeated measure ANOVA followed by Holm-Sidak method. Data are expressed as mean ± SEM; n = 4 (SHAM), 7 (TBI+Saline), 11 (TBI + NAP).

Lateral ventricle size was measured to evaluate intracranial CSF fluid in the mice following TBI saline or NAP treatments. Changes in intracranial CSF dynamics is a common clinical problem after TBI.⁴³ TBI + saline mice showed a significant enlargement of ipsilateral lateral ventricle size (p < 0.001), whereas mice subjected to TBI that received NAP demonstrated significantly decreased ventricular enlargement, as compared to the TBI + saline group (p < 0.001); (Figure 6C).

NAP Treatment Attenuated TBI-Induced Microglial and Astroglial Activation in Cortex and Thalamus

To evaluate the role of neuroinflammation on TBI-mediated brain changes, the microglial activation state was assessed by quantifying the morphology of ionized calcium binding adaptor molecule 1 (Iba1)-positive cells with the caveat that although most Iba1-positive cells in brain are microglia, monocyte subsets are also Iba1-positive. Microglial morphology is typically classified as ramified in the “resting” condition with smaller somata and large process extensions. This is a simplification corresponding to a low-activity, quiescent status whereby they maintain a physiological role of surveying their microenvironment and optimizing neuronal cell function.⁴⁴ After neuronal injury, microglia began to swell and reduce the size of their processes, with transformation into an “active” form (Figure 7A). This is a simplification corresponding to a reactive state displaying inflammatory and phagocytic features.²⁰⁴ On the basis of the morphologies of Iba1 immunoreactivity, we classified Iba1-positive cells into four different cell types: ramified and intermediate types as the resting/surveillance state and amoeboid and round types as the activated/reactive state (Figure 7A,B). At 2 weeks following TBI, the ratio of the amoeboid type was significantly increased and that of the ramified type was largely reduced in the ipsilateral thalamus, as compared with the sham group (p < 0.001; Figure 7B). The number of the activated forms of microglia per 200 μm² was significantly reduced in ipsilateral thalamus in NAP treated injured mice compared with the TBI + saline group (p < 0.001; Figure 7C). Notably, the total number of microglia within the ipsilateral thalamus was also significantly increased by TBI (p < 0.001; Figure 8A,B), and NAP treatment likewise significantly reduced this increment (p < 0.01; Figure 8A,B).

Figure 7.

Effect of NAP treatment at 2 weeks after CCI TBI (30 mg/kg body weight i.p.) on the microglial morphology ratio in the ipsilateral (injured) thalamus. (A) Iba1 immunofluorescence staining showing microglial cells with ramified, intermediate, amoeboid, and round morphology. The morphological stages are classified as follows: (1) activated-amoeboid and round forms; (2) resting/quiescent: ramified long branching processes with a small cell body; and (3) intermediate transition forms. (B) Quantification of the proportions of microglia of the four different phenotypes. (C) Quantification of the cell numbers of microglia in resting/quiescent and activated/reactive forms. NAP treatment significantly reduced the activated forms of microglia compared with the TBI-saline group. ***, p < 0.001, resting vs active form within TBI + NAP group; ^###, p < 0.001, active form of microglia in TBI + saline vs TBI + NAP groups. Two-way ANOVA with Bonferroni’s t-test for multiple comparisons. Scale bar, 5 μm.

Figure 8.

Postinjury treatment with NAP (30 mg/kg body weight i.p.) reduced Iba1-positive microglia and GFAP-positive astrocytes at 2 weeks after TBI. (A) Immunofluorescence of Iba1 and GFAP in thalamus. Iba1, a marker for microglia, is shown in green. GFAP, a marker for astrocytes, is shown in red. (B) TBI injury significantly increased the number of microglia in the ipsilateral thalamus, and this increment was reduced in NAP-treated mice. ^##, p < 0.01, ipsilateral TBI + saline vs ipsilateral TBI + NAP; ***, p < 0.001 ipsilateral vs contralateral in both groups. Two-way ANOVA with Bonferroni’s t-test for multiple comparisons. (C) TBI injury significantly increased the number of astrocytes in the ipsilateral thalamus, and this increment was reduced in NAP-treated mice. Mean ± SEM (n = 3 in TBI + saline group; n = 7 in TBI + NAP group). ***, p < 0.001, ipsilateral sham vs ipsilateral in both TBI + saline and TBI + NAP groups. ^##, p < 0.01; ^###, p < 0.001, ipsilateral TBI + saline vs ipsilateral TBI + NAP; ***, p < 0.001 ipsilateral vs contralateral in both groups. Two-way ANOVA with Bonferroni t-test for multiple comparisons. Scale bar, 20 μm.

Astroglial activation was assessed by quantifying the total cell number of glial fibrillary acidic protein (GFAP)-positive cells per 200 μm² within thalamus. In the TBI + saline group, GFAP-positive cell number on the ipsilateral side was increased, as compared to the contralateral side (p < 0.001; Figure 8A,C), and an increase was also observed in the TBI + NAP group, comparing ipsilateral and contralateral sides. However, NAP treatment significantly reduced TBI-induced astroglial activation. Data are presented as percentage of GFAP-positive cells in thalamus (p < 0.01; Figure 8A,C).

NAP Treatment of TBI Mice Reduced the Production of TNF-α within Microglial Cells

Microglia expressing the proinflammatory cytokine TNF-α were elevated in number within the ipsilateral cortex adjacent to the contusion region, as compared to the contralateral side (Figure 9A). Notably, administration of NAP reduced the number of TNF-α and TNF-α/Iba1-positive cells in cortical areas, with the results expressed as a percentage of colocalization of TNF-α in Iba1-positive cells (p < 0.05; Figure 9B,C).

Figure 9.

NAP reduces TBI-induced TNF-α generation within activated microglia in the ipsilateral hemisphere in TBI mice. (A) TBI induced the proinflammatory cytokine TNF-α positive cells, shown in green, in the ipsilateral cortex. Iba1, a marker for microglia, is shown in red. Postinjury treatment with NAP (30 mg/kg body weight i.p.) decreased the increase in TNF-α-positive (B) and TNF-α/Iba1 doubly labeled (C) cells at 2 weeks after TBI. ***, p < 0.001, ipsilateral vs contralateral in both groups. Mean ± SEM (n = 3 in TBI + saline; n = 7 in TBI + NAP). ^#, p < 0.05, ipsilateral TBI + saline vs ipsilateral TBI + NAP. Two-way ANOVA with Bonferroni’s t-test for multiple comparisons. Scale bar, 20 μm.

TBI Induced the Expression of Ramified P2RY12 Microglial Cells

The expression of purinergic adenosine diphosphate/triphosphate (ADP) receptor purinergic receptor P2Y, G protein coupled, 12 (P2RY12) on microglial cells has been described as a nonactivated/homeostatic state marker.⁴⁵ P2RY12-expressing cells showed two different forms, rounded and ramified, when evaluated in sham and TBI-challenged mice (Figure 10A,B). The rounded P2RY12-positive cell type was mainly expressed within the cortex and thalamus, two key inflammatory sites in the TBI mice. However, the ramified type was observed in almost all areas of the brain. Notably, ramified P2RY12-positive cells all colocalized with Iba1-positive cells (Figure 10A,B) and were significantly reduced by TBI in ipsilateral cortex (Figure 10C) and thalamus (p < 0.001) (Figure 10D), suggesting a reduced resting cell number.

Figure 10.

TBI reduced ramified microglia in the ipsilateral hemisphere of mice. (A) Adenosine diphosphate (ADP) receptor P2RY12 is shown in green; Iba1, a marker for microglia, is shown in red. Cortex and thalamus (B) of both TBI + saline and TBI + NAP groups were analyzed by immunostaining. TBI significantly decreased the ramified form of P2RY12-positive cells in (C) cortex and (D) thalamus. ***, p < 0.001, ipsilateral vs contralateral in both groups. Mean ± SEM (n = 3 in both TBI + saline and TBI + NAP); two-way ANOVA with Bonferroni’s t-test for multiple comparisons. Postinjury treatment with NAP (30 mg/kg body weight i.p.) did not affect the expression changes of P2RY12-postive cells at 2 weeks after TBI. Scale bar, 20 μm.

NAP Treatment Mitigated the Reduced Expression of PSD-95 Induced by TBI in CA1, Dentate Gyrus, and Cortex Regions of TBI-Challenged Mice

Postsynaptic density-95 (PSD-95) is a key scaffold protein present within the excitatory postsynaptic density (PSD) of neuronal dendritic spines. PSD-95 regulates the trafficking and localization of receptors within the PSD, particularly glutamate receptors; its elevated expression augments receptor-mediated synaptic current and its loss is observed in neurodegenerative disorders (NDs). It has hence been implicated in synaptic development, stability, and plasticity.⁴⁶⁻⁴⁸

To investigate the impact of CCI TBI on the functional plasticity of neurons and whether NAP can modify it, we evaluated PSD-95 within the CA1 region of the hippocampus, in the light of studies by O’Keefe and colleagues on identifying place cells within the CA1 region and the importance of this site in event-related cognition.²⁰⁵ As illustrated in Figure 11, the expression level of PSD-95 was substantially decreased in the ipsilateral CA1 region after TBI, as compared with the contralateral side. Notably, NAP treatment fully ameliorated this TBI-mediated effect (Figure 11A,B). A similar effect was observed in cortex, albeit of less magnitude (not shown).

Figure 11.

NAP mitigates TBI-induced postsynaptic density protein (PSD-95) reduced expression in the ipsilateral hemisphere of TBI-challenged mice. (A) TBI reduced PSD-95 expression levels, shown in green, in the ipsilateral CA1 of vehicle (saline)-treated animals, as compared to PSD-95 expression levels evident within the same area of the contralateral CA1 of the same coronal slice from the same animal. Postinjury treatment with NAP (30 mg/kg body weight i.p.) mitigated this TBI-induced reduction in expression level of PSD-95 in the ipsilateral CA1. (B) Quantification of PSD-95 levels in TBI + saline and TBI + NAP groups (a ratio of ipsilateral to contralateral expression of PSD-95 immunoreactivity was evaluated for each animal). Analysis by unpaired t-test; *, p < 0.05 (p = 0.0277) (TBI + NAP vs TBI + saline). Specifically, the ratio of PSD-95 expression in ipsi- vs contralateral CA1 was significantly less in the (TBI + saline)-treated group (and was less than unity (i.e., 1.0)) than in the (TBI + NAP)-treated group (which was close to unity). There was no significant difference in CA1 contralateral PSD-95 expression between different groups of animals. (C) Diagram of the region selected for analysis. Scale bar, 20 μm.

Discussion

A large body of research indicates that neuroinflammation is a key feature across acute and chronic NDs^19,49−51 in which persistently elevated brain levels of cytokines interfere with signaling pathways and thereby impair neurotransmission, mitochondrial function, and cell viability. Although epidemiological data show a potential protective effect of prior medication with nonsteroidal anti-inflammatory drugs (NSAIDs) against the development of NDs and gene polymorphisms of select inflammatory cytokines appear to alter AD and PD risk,⁵²⁻⁵⁴ randomized clinical trials of anti-inflammatory agents in acute and chronic NDs have largely failed.⁵⁵ Whereas it is possible that neuroinflammation may represent a response to pathologic features of the disorder rather than a causative element in neurodegeneration, GWAS reveals that in excess of 60% of the genes linked to sporadic AD are allied to inflammation, and changes in inflammatory pathway genes and proteins are, likewise, reported as over-represented in TBI and PD.^56,57 It is thus possible that NSAIDs, which cover a broad range of agents that selectively inhibit cyclooxygenase (COX-1 and/or COX-2) and lead to a decline in levels of prostaglandins, prostacyclin, and thromboxanes, may not represent the optimal drugs to mitigate the neuroinflammatory element of TBI and other NDs.⁵⁸ In the light of the chronic microglial activation that occurs in many NDs, including TBI, and accompanying elevated levels of pro-inflammatory mediators, such as TNF-α, prostaglandins, and reactive oxygen and nitrogen species, cytokine-suppressive anti-inflammatory drugs (CSAIDs) may represent a preferable treatment approach.⁵⁵ We therefore developed a “phenotypic”-based screening approach^59,60 to evaluate potential compounds to mitigate the neuroinflammatory component associated with NDs, as such an approach is largely agnostic to the underpinning mechanisms and has been demonstrated to result in more first-of-class successful drugs, as compared to “target-based” approaches.⁶⁰ This is important in complex neurological disorders, as multiple, separate, and parallel mechanisms likely are involved in disease progression, and a target-based drug development strategy would not be expected to engage multiple mechanisms.^59,60 Such a phenotypic drug development approach has previously resulted in multiple successful drugs, including medicines listed as essential by the World Health Organization such as aspirin and diazepam⁶¹ that were widely and successfully used well before their ultimate targets were characterized. The present study demonstrates that the CSAID NAP can decrease the production of cytotoxic cytokines such as TNF-α and free radicals such as nitric oxide, attenuate microglial and astrocyte activation in a well-characterized animal model of TBI, and augment brain recovery.

Depending on the phenotypic expression of microglia and their level of TNF-α synthesis and release, microglial-generated TNF-α has both physiological and pathophysiological actions in the brain. As reviewed by Clark and Vissel⁶² and Olmos and Llado,⁶³ TNF-α is key in the homeostatic maintenance of physiological neuronal activity. Physiological TNF-α concentrations modulate excitatory neurotransmission, the trafficking of AMPA receptors, homeostatic synaptic scaling, and long-term potentiation, as well as having a role in neurogenesis. TNF-α levels can also impact mitochondrial function as well as specific neurotransmitters, such as orexin, neuronal Ca²⁺ homeostasis, and their associated signaling cascades.⁶⁴⁻⁶⁶ In contrast, the presence of pathogen- or damage-associated molecular patterns (PAMPs and DAMPs) allied to pathogens like LPS and trauma as in TBI, respectively, can trigger the activation of microglia via their agonist action at nonspecific pattern recognition receptors (PPRs) that include the toll-like receptors (TLRs) and instigate elevated TNF-α generation and its release in pathophysiological concentrations.⁵⁰ Excessive or unregulated chronic activation can induce a “cytokine storm” in the brain and result in neurotoxic levels of cytokines and free radicals (e.g., nitric oxide and superoxide), which can damage neurons and initiate a self-perpetuating cycle of inflammatory neurotoxicity.^19,49,51,62

Consistent with work published by others, our CCI-induced moderate TBI provoked activation of microglia and astrocytes in surrounding ipsilateral brain areas (Figures 7 and 8), with a resulting elevation in microglial TNF-α generation (Figure 9). In line with this, significantly lower levels of ramified (resting/surveillance) microglia expressing P2RY12 were evident in brain ipsilateral to TBI, as compared to the contralateral side (Figure 10C,D). P2RY12 is highly expressed on resting/quiescent microglia and often used to discriminate them from macrophages.⁶⁷ Activation of microglial P2RY12 by ADP/ATP induces microglial chemotaxis toward sites of release, which is often associated with apoptotic and/or necrotic cells.⁶⁸ Once in the vicinity of an inflammatory microenvironment, microglial P2RY12 expression becomes downregulated.^69,70 Microglia have often been categorized into functional M1 and M2 groups to simplistically define their activated versus quiescent states, although that this classification does not account for the complexities of microglial phenotypes in the diseased brain.⁷¹ In the light of the complex pattern of P2RY12-expressing microglial in AD brain samples described recently by Walker et al.,⁶⁷ we evaluated the phenotype of such P2RY12-positive cells into a ramified versus round form and noted differential actions of TBI on these two P2RY12-expressing populations. We found the round form of P2RY12-positive cells was slightly increased in the ipsilateral cortex, as compared to the contralateral cortex (data not shown), but this pattern was not evident in the thalamus. A previous study reported that microglial processes rapidly form a dense plexiform aggregate at the site of a capillary injury. P2RY12 is not only expressed in microglia but also in platelets.⁷² Unlike P2RY12-positive ramified cells that colocalized with Iba1-positive cells, P2RY12-positive round ones did not. We hence hypothesize that the chemotaxic potential of high-P2RY12-expressing ramified microglia supports their ability to move to areas of local brain injury/inflammation, and the presence of round lower-intensity-P2RY12 cells are blood-borne platelets indicative of their infiltration at BBB leakage sites.

Prior studies have demonstrated that microglial activation is an early event, occurring within 24 h^21,24,73 and potentially extending from weeks to months after TBI.⁷⁴ A marked rise in TNF-α occurs almost immediately, with reports of elevated intracerebral mRNA and protein levels in patients and animal models within minutes to hours.^20,75 In general, its rise is detected prior to the appearance of other cytokine proteins, which subsequently become upregulated via the actions of TNF-α on the transcriptional regulator NF-κB.⁷⁶ These then drive the immune response that sequentially underpins the acute, subacute, and subsequently chronic neuroinflammation that ensues.⁷⁷ Hence, depending on its concentration, time of release, and the signaling pathway engaged, TNF-α can intensify oxidative stress, contribute to glutamate release, and exacerbate BBB dysfunction, each of which may promote further dysfunction and cellular loss.^19,78 However, a delayed, smaller TNF-α rise may augment recovery following TBI, in line with its purported neuroprotective functions, by decreasing oxidative stress and augmenting neurotrophic factor synthesis.^28,79,80 We found that NAP effectively lowers TNF-α generation across the cellular and animal models evaluated, reduces microglial and astrocyte activation, and these result in reduced intracranial CSF pressure, reflected in lateral ventricle size (sometimes termed clinically as “hydrocephalus ex vacuo”, an enlargement of the ventricle because of tissue changes but not necessarily loss), and attenuation of TBI-induced behavioral impairments.

Our NAP study confirms that thalidomide-like drugs can effectively mitigate TBI-induced impairments across cellular and animal models and achieve this by acting as CSAIDs to mitigate neuroinflammation.^23,73,81 Importantly, however, NAP lacks a glutarimide ring that is considered critical for the binding of thalidomide and active analogs within cereblon, a substrate receptor subunit for the ubiquitin E3 ligase complex CRL4.^35,82 Cereblon creates an E3 ubiquitin ligase complex by interacting with the protein damaged DNA binding protein 1 (DDB1), Cullin-4A (CUL4A), and regulator of cullins 1 (Roc1). This drug–protein complex tags specific proteins/transcription factors with ubiquitin and, thereby, targets them for proteolysis. Although dissimilar, such protein/transcription factors share a characteristic β-hairpin loop structure with a glycine at a conserved position when observed in X-ray crystal structures.⁸³⁻⁸⁵ Among these, proteolysis of Ikaros zinc finger proteins ikaros and aiolos^86,87 leads to the downregulation of transcription factors such as interferon regulatory factor 4 (IRF4) and c-Myc, resulting in the efficacy of thalidomide-like drugs in multiple myeloma. SALL4 docks at the same “hot spot” on the cereblon surface as the other neosubstrates in the presence of thalidomide-like drugs, and its degradation mediates teratogenicity.^84,85

The molecular mechanism of action for thalidomide-like drugs was originally inadequately understood, with early studies detailing clinically valuable pleiotropic actions on both myeloma cells as well as on the immune system.⁸⁸ Once considered to be mediated via multiple different pathways,⁸⁹ the seminal studies of Ito at al.,^35,90 demonstrating binding to cereblon, brought mechanisms under a unifying site, potentially accounting for thalidomide’s antiproliferative, antiangiogenic, and immunomodulatory activity, as well as its teratogenic poly pharmacological actions (Figure 12).^90,91 Fewer reports have addressed the anti-inflammatory properties of thalidomide-like drugs.⁹² A recent study involved the humanization of cereblon in mice, in which a single amino acid within the binding domain for thalidomide and analogs (valine for isoleucine at positions 391 and 387 in murine and human cereblon, respectively)⁹³ results in mouse cereblon being resistant to thalidomide-based anticancer treatments and teratogenicity. This change prevented the degradation of ikaros, aiolos, CK-1α, and other neosubstrates in mouse.^3,94 Whereas the humanization of cereblon in mice re-established active E3 ubiquitin ligase degradative activity of thalidomide analogs, equipotent anti-inflammatory actions were evident in both wild-type and humanized cereblon mice, suggesting a separation of the pathway(s) underpinning inflammation to be independent of cereblon. Our study supports separation of these pathways as NAP, which lacks cereblon binding consequent to its adamantly moiety, possesses cellular and in vivo anti-inflammatory activity of biological relevance in a mouse model of TBI and both rats and cells challenged with LPS (Figure 12).

Figure 12.

Scheme depicting models of our increasing knowledge of how thalidomide-like drugs target both their pharmacologically useful and their detrimental actions. (A) Original model in which there are multiple different targets that account for the pleotropic actions of thalidomide and analogs.⁸⁹ (B) With the discovery of thalidomide’s interaction with cereblon and its role as a substrate receptor for the ubiquitin E3 ligase complex, cereblon became a primary target via which pleotropic actions are achieved through the breakdown of specific neosubstrates.^35,90 (C) Current model: The interaction of thalidomide-like drugs with cereblon changes the stability of key neosubstrates that account for numerous both positive and toxic actions. Notably, independent direct actions on other targets account for other pharmacological effects: in particular, anti-inflammatory actions to lower TNF-α,⁹⁴ as supported by NAP in our current study, which possesses anti-inflammatory activity without binding cereblon and degrading ikaros (IKZF1), aiolos (IKZF3), SALL4, and others key teratogenic proteins.

Crystal structures of thalidomide-like drugs bound to cereblon in complex with DDB1 show that the glutarimide ring binds within a shallow pocket comprised of three tryptophan residues, with a phenylalanine side-chain as the base. The phthalimide portion of the drug protrudes out of this pocket to the surface of cereblon, allowing interactions to selectively target protein degradation.³⁶ A minimal cereblon-binding requirement in relation to the glutarimide ring is the presence of at least one carbonyl on a 5- or 6-membered ring; however, 7- or more membered rings appear too large to bind.³⁷ NAP is in line with this structure–activity relationship, appears unable to bind cereblon, and hence is not capable to tag neosubstrates for degradation (Figure 1). The TNF-α lowering action of thalidomide-like drugs is considered as mediated by enhancing its mRNA degradation⁹⁵ via interacting with key elements within its 3′-untranslated region^18,96 and thus appears not to involve cereblon.

There are several important considerations of long-term effects of TBI in terms of subsequent development of dementia and PD, even should short-term symptoms spontaneously resolve. Although the link between TBI and neurodegeneration remains debatable,⁹⁷ an increasing number of epidemiological studies suggest a strong association between TBI and later development of several types of dementia.^{2−4,98−103} Data from our gene array studies in rodents support a link between both concussive and blast TBI and dementia, illustrated by the identification of upregulated gene sets associated with AD and PD observed as early as 3 and 14 days after injury.^104,105 It is hence possible that TBI injury predisposes individuals toward neuropathology leading to cognitive impairment by mechanisms similar to those seen in AD and PD, whereby TBI acts as a disorder that ultimately manifests as a proteinopathy via multiple potential mechanisms including seeding and increasing α-synuclein or Aβ generation and oligomerization.^106,107 In this light, the action of NAP in AD and PD cellular models is promising (Figure 2C–G), and as thalidomide-like drugs have demonstrated efficacy in AD and PD animal models,^{108,109−111} it could prove valuable to evaluate NAP across further models of TBI as well as the chronic disorders of AD and PD. It remains unclear how many of the adverse and dose-limiting actions of thalidomide and clinical analogs are mediated via cereblon versus alternative independent pathways,¹¹² but NAP provides a framework for the development of CSAIDs with reduced teratogenic risk. Its promising anti-inflammatory action in the current studies supports its further evaluation in relation to toxicity, teratogenicity, and targeted mechanism(s) of action.

In summary, neuroinflammation is a promising drug target in the treatment of TBI and a broad spectrum of other neurological disorders, with TNF-α being a key element. The drug class of immunomodulatory drugs (IMiDs) that are based on the backbone of thalidomide and its clinically available analogs, lenalidomide and pomalidomide, can substantially mitigate inflammation at the level of TNF-α synthesis and hold potential as a new treatment strategy but are compromised by their adverse actions.^31,112 NAP represents a new class of IMiDs that potently mitigates neuroinflammation and TBI-induced deficits in a well-characterized rodent CCI model. Notably, NAP does not bind cereblon, a key target associated with the antiproliferative, antiangiogenic, and teratogenic actions of the IMiD drug class,^36,82,90 and NAP hence warrants further investigation as a new candidate drug for neurodegenerative and systemic disorders with an inflammatory element.

Materials and Methods

Synthesis of 2,3-Dihydro-2-[1-(1-tricyclo[3.3.1.1^3,7]dec-1-yl)ethyl]-1H-isoindol-1-one) (NAP)

A mixture of (i) phthaldialdehyde (310.0 mg; 2.311 mmol), (ii) 1-(1-adamantyl)ethylamine hydrochloride (498.7 mg; 2.311 mmol), and (iii) potassium carbonate (176.0 mg; 1.273 mmol) in 160 mL of tetrahydrofuran was stirred for 7 days under a nitrogen atmosphere at room temperature. After removing solvent, the residue was purified by silica gel chromatography (EtOAc/Hex = 1/3) to provide 97.0 mg of 2,3-dihydro-2-[1-(1-tricyclo[3.3.1.1^3,7]dec-1-yl)ethyl]-1H-isoindol-1-one) (NAP) as white needle crystals. Chemical characterization confirmed the structure and high purity of NAP, illustrated in Figure 1A.

Cellular Studies

NAP Activity in LPS Stimulated RAW 264.7 Cells

Mouse RAW 264.7 cells, originally acquired from ATCC (Manassas, VA), were grown in Dulbecco’s modified Eagle’s medium (DMEM) media supplemented with 10% fetal calf serum (FCS), penicillin 100 U/mL and streptomycin 100 μg/mL, and were maintained at 37 °C and 5% CO₂. Cells were grown in accord with ATCC guidelines as previously described.²⁰⁶ On the day of study, RAW 264.7 cells were challenged with LPS (Sigma, St Louis, MO: serotype 055:B5) at a final concentration of 60 ng/mL. This LPS concentration routinely induces a submaximal rise in both TNF-α and nitrite levels without a loss of cell viability. Such a submaximal rise is useful for assessing whether the addition of an experimental drug can either lower or further elevate levels of TNF-α and nitrite. In a drug pretreatment paradigm, either NAP (10–100 μM) or vehicle (Veh), n = 3–4, was administered 60 min prior to LPS challenge. At 24 h following the addition of LPS, conditioned media was harvested and both secreted TNF-α protein (Mesoscale Discovery) and nitrite levels (Griess Reagent System, (Promega, Madison, WI, catalog no. G2930) and Nitrate/Nitrite Fluorometric Assay Kit (Abnova, catalog no. KA1344)) were quantified as recommended by the manufacturers. Fresh media was replaced in the wells, and cell viability was thereafter evaluated by a CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI). For cell culture studies, NAP was prepared immediately prior to use in 100% DMSO and then added to cell culture media at a dilution of greater than 200-fold to provide the desired concentrations; control-/veh-treated cells were subjected to the exact same procedure, but without the addition of NAP.

NAP Activity in Primary Dopaminergic Neurons and Microglia Challenged with α-Synuclein

Rat dopaminergic neurons and microglia were maintained in culture as described by Zhang et al. and Callizot et al.^113,114 with modifications. In general, the cellular composition at the time of treatment was approximately 48% astrocytes, 11% microglia, and the remains were neurons, of which 1% was tyrosine hydroxylase (TH)-immunoreactive.¹¹⁴ Fetuses (14–15 days gestation), removed aseptically from euthanized, timed-pregnant female rats, were immediately placed in ice-cold L15 Leibovitz medium. This contained 2% penicillin (10 000 U/mL), streptomycin (10 mg/mL), and 1% bovine serum albumin (BSA). The ventral portion of the mesencephalic flexure, a midbrain region rich in dopaminergic neurons, was utilized as a basis for cell preparation and dissected portions were subjected for 20 min at 37 °C to a trypsin/EDTA solution (0.05% trypsin and 0.02% EDTA). DMEM containing DNAase I grade II (0.5 mg/mL) and 10% FCS was then added, and cells were mechanically dissociated by 3 passages through a 10 mL pipet. Thereafter, cells were centrifuged for 10 min at 180g and 4 °C on a layer of BSA (3.5%) in L15 medium. The cell pellet was resuspended in culture medium comprising Neurobasal medium with a 2% solution of B27 supplement, 2 mmol/L l-glutamine, 2% PS solution, 10 ng/mL brain-derived neurotrophic factor (BDNF), 1 ng/mL glial cell line-derived neurotrophic factor (GDNF), 4% heat-inactivated FCS, 1 g/L of glucose, 1 mM sodium pyruvate, and 100 μM nonessential amino acids. Cells were subsequently seeded to a density of 80 000 cells/well in poly-l-lysine-precoated 96-well plates, and maintained in a humidified incubator (37 °C, 5% CO₂, 95% air). Half of the medium was exchanged with fresh medium every 2 days.

On day 7 in culture, cells were preincubated for 1 h with culture medium freshly spiked with known NAP concentrations or vehicle, and were then exposed to α-synuclein (250 nM α-synuclein, 72 h, n = 6 per group: human recombinant α-synuclein amino acids (AAs) 1–140 from rPetide, Watkinsville, GA), which a prior study reported contains <1.3 U endotoxin/mg of peptide, a concentration incapable of producing any significant effects in terms of neurotoxicity or induction of reactive oxygen species (ROS) production.¹¹⁴ This α-synuclein was previously prepared as a 4 μM solution and slowly shaken at 37 °C for 72 h in the dark to induce oligomerization. Thereafter, cell culture supernatants were removed and frozen for later assays, and cells were washed in physiological buffered saline (PBS).

Cellular Assays

Cells were fixed in 4% paraformaldehyde (PFA) in PBS (pH 7.3, 21 °C, 20 min), washed twice in PBS, and then permeabilized. PBS containing 0.1% saponin and 1% FCS was added (21 °C, 15 min) to abolish nonspecific binding. Cells were then incubated with either (i) rabbit polyclonal anti-TH (dilution: 1:2000 in PBS containing 1% FCS, 0.1% saponin, 21 °C, 2 h) to visually quantify dopaminergic neurons and neurites or (ii) mouse monoclonal anti-OX-41 (diluted 1:500 in PBS containing 1% FCS and 0.1% of saponin, 21 °C, 2 h) to visualize microglia. After washing the sections, they were incubated with Alexa Fluor 488 goat anti-rabbit IgG (Jackson, 1:400) in PBS containing 1% FCS, 0.1% saponin (21 °C) for 1 h. Photomicrographs (20/well at 10× magnification) were automatically acquired by ImageXpress (Molecular Devices, San Jose, CA) and were analyzed by Custom Module Editor (Molecular Devices). The following parameters were quantified: (i) dopaminergic neuron survival (determined by counting the number of TH-positive neurons across conditions, in comparison to the control condition, i.e., without α-synuclein challenge), (ii) total neurite network of dopaminergic neurons (determined from the length of TH-positive neurites), (iii) total microglia activation (determined from the area of microglial cells, μm2 of OX-41 staining), and (iv) TNF-α release, determined by quantifying the level of TNF-α protein in cell culture medium samples by ELISA (rat TNF-α ELISA kit, Abcam, ab46070), in line with the manufacturer’s instructions.

NAP Activity in Primary Cortical Neuron Cultures Challenged with Aβ

Following the same procedures as described above for dopaminergic primary cultures, primary cortical neurons were prepared from brain cortex samples that were aseptically dissected from 14–15 day old rat (Wistar) fetuses, as described by Callizot et al.¹¹⁵ with modifications. The cells were eventually seeded to a density of 45 000 cells/well in poly-l-lysine precoated 96-well plates and maintained in a humidified incubator (37 °C, 5% CO₂, 95% air). Half of the medium was exchanged with fresh medium every 2 days.

On day 11 of culture, cells were preincubated for 1 h with culture medium freshly spiked with known NAP concentrations or veh, and were then exposed to amyloid-β peptide (Aβ; AAs 1–42, 5 uM, 72 h, n = 6 per group: Bachem, Torrance, CA). The Aβ was previously prepared as a 40 μM solution that was slowly shaken at 37 °C for 72 h in the dark to induce its oligomerization.¹¹⁵ Thereafter, cells were washed in PBS and then fixed by addition of a cold solution of ethanol (95%) and acetic acid (5%) for 5 min at −20 °C. They were then twice again washed in PBS and permeabilized. Finally, cells were incubated in PBS containing 0.1% saponin and 1% FCS (21 °C, 15 min) to abolish nonspecific binding and then immuno-stained with an anti-microtubule-associated-protein 2 (MAP-2) chicken polyclonal antibody (2 h, dilution 1/1000 in PBS, with 1% fetal calf serum and 0.1% of saponin). This antibody specifically binds to neurons and their neurites, to allow quantification of neuronal cell survival. To accomplish this, photomicrographs (20/well at 10× magnification) were acquired by ImageXpress (Molecular Devices, San Jose, CA) and analyzed by Custom Module Editor (Molecular Devices).

Cereblon Binding Assay in Human SH-SY5Y Neuronal Cultures

A bead-based AlphaScreen technology was adopted for cereblon binding assays with minimal modifications from the manufacturer’s protocol (BPS Bioscience catalog no. 79770). NAP, pomalidomide, thalidomide or lenalidomide was incubated with reaction mixtures including cereblon/DNA damage-binding protein 1–Cullin 4a–ring-box protein 1 complex (CRBN/DDB1–CUL4A–Rbx1, 12.5 ng) and bromodomain-containing protein 3 (BRD3) (6.25 ng) in an Optiplate 384-well plate (PerkinElmer catalog no. 6007290). After 30 min of incubation with shaking at room temperature, AlphaLISA anti-FLAG Acceptor and Alpha Glutathione Donor beads (PerkinElmer) were sequentially added and then incubated for 1 h at room temperature for each of the added chemicals. Alpha counts were thereafter read on a Synergy Neo2 (BioTek) for the analysis. Relative activity or inhibition was calculated as the highest value, which was set to100%, and the lowest value was set to 0% after subtraction of the “blank value” from all readings.

NAP activity on neosubstrates was evaluated in both MM1.S (myeloma) and H9 hESC (human embryonic stem) cell lines. Specifically, MM1.S cells were obtained from ATCC (Manassas, VA), grown in RPMI media supplemented with 10% FBS, penicillin 100 U/mL, and streptomycin 100 mg/mL, and were maintained at 37 °C and 5% CO₂. MM1.S cells were treated with 10 μM of thalidomide analogs (thalidomide, lenalidomide, pomalidomide, and NAP) for 24 h; thereafter, their cell lysates were prepared for Western blot analysis, as described previously.¹¹⁶ In contrast, H9 hESC lines were obtained from WiCell Research Institute (catalog no. WA09; Madison, WI) and grown on growth factor reduced matrigel-coated dishes in mTeSR1 media (STEMCELL Technologies, Vancouver, Canada), supplemented with 5 ng/mL bFGF, penicillin 100 U/mL, and streptomycin 100 μg/mL and maintained at 37 °C and 5% CO₂. H9 hESC cells were treated with 20 μM of thalidomide analogs (thalidomide, pomalidomide, and NAP) for 24 h, and their cell lysates prepared for the Western blot analysis, as described previously.¹¹⁶

For Western blot analysis, total proteins were extracted using RIPA buffer (ThermoFisher Scientific, Waltham, MA) containing Halt Protease Inhibitor Cocktail (ThermoFisher Scientific). Thereafter, proteins were separated by gel electrophoresis and then transferred to polyvinylidene difluoride (PVDF) membranes (ThermFisher Scientific), as described previously.¹¹⁶ The following primary antibodies were used: (i) anti-Ikaros antibody (CST catalog no. 9034; 1:1000; Cell Signaling Technology, Danvers, MA), (ii) anti-Aiolos antibody (CST catalog no. 15103; 1:1000; Cell Signaling Technology), (iii) anti-SALL4 antibody (SC101147; 1:1000; Santa Cruz Biotechnology, Dallas, TX), and (iv) anti-β-actin antibody (CST catalog no. 3700; 1:5000; Cell Signaling Technology). After incubation at 4 °C overnight, the following HRP conjugated secondary antibodies were used: (i) goat anti-rabbit IgG (ThermoFisher Scientific) for ikaros and aiolos, and (ii) goat anti-mouse IgG (ThermoFisher Scientific) for SALL4 and β-actin. β-actin, a protein that is generally expressed in all eukaryotic cells, was used as an internal control against which the other protein expression levels were compared. Antigen–antibody complexes were detected using enhanced chemiluminescence (Thermo, iBright CL1500).

Animal Studies

All rodents were housed at 25 °C in a 12 h/12 h light/dark cycle with continuous access to food and water. All efforts were made to reduce animal suffering and to minimize the number of animals used by incorporating the outcome measures from our prior studies¹¹⁷ and a power analysis.¹¹⁸ The procedures used in this study were fully approved by following the Institutional Animal Care and Use Committees (Intramural Research Program, National Institute on Aging, NIH (protocol no. 331-TGB-2021); Case Western Reserve University (protocol no. 2016-0209).

Systemic and Brain LPS Anti-Inflammatory Studies

Male Fischer 344 rats (approximately 150 g weight) were randomly assigned across groups, and then administered NAP (13.3 or 26.6 mg/kg (equimolar to 12.5 and 25 mg/kg thalidomide i.p.), suspended in 1% carboxymethyl cellulose (CMC) in PBS) or vehicle (the same, but without NAP) 60 min prior to either LPS (1 mg/kg, Sigma, St Louis, MO, E. coli O55:B5 in PBS, 0.1 mL/kg i.p.) or vehicle. These doses were selected for the initial in vivo evaluation of NAP as they are equimolar to, or are less than, doses of thalidomide and analogs that have proven to be well-tolerated in prior rodent studies, and are of translational relevance to humans.^{23,108,109,119} At 4 h after LPS, animals were euthanized, and blood and brain tissue were immediately harvested and placed on wet ice. Plasma was rapidly obtained, and sections of cerebral cortical tissue and hippocampus were quickly dissected into separate vials on wet ice. All samples stored at −80 °C. Brain samples were sonicated in a TRIS based lysis buffer (Mesoscale Discovery) with 3× protease/phosphatase inhibitors (Halt Protease and Phosphatase Inhibitor Cocktail, ThermoFisher Scientific), were then centrifuged for 10 min at 10 000g and 4 °C, and protein concentrations were determined using the Bicinchoninic acid assay (BCA, ThermoFisher Scientific). An ELISA for TNF-α, IL-6, IL-10, 1L-13, and IFN-γ was subsequently performed (Mesoscale Discovery) on the rat plasma, hippocampal, and cerebral cortical samples, following the manufacturer’s protocol. Choice of the LPS and drug doses was made on the basis of our former study.¹²⁰

In Vivo Model of TBI

TBI studies were conducted in 8 week old male C57/BL6 mice (25–30g, body weight, Jackson Laboratory, Bar Harbor, ME). Forty-one mice were randomly assigned across three groups (sham (8 mice), CCI-saline (9 mice), and CCI-NAP (24 mice, prepared in 1% CMC in PBS)) to evaluate the effects of NAP on TBI. Mice were assessed for sensory/motor activity, balance function, motor asymmetry, motor coordination, and lesion size. Animals were subsequently evaluated for cellular changes using histology and immunohistochemistry.

Animal Model of TBI and Drug Administration

Mice were anesthetized with 2.5% tribromoethanol (Avertin: 250 mg/kg; Sigma, St. Louis, MO)) and placed in a stereotaxic frame (Kopf Instruments, Tujunga, CA). Using sterile procedures, the skin was retracted, and a 4 mm craniotomy was performed at a point midway between the lambda and bregma sutures and laterally midway between the central suture and the temporalis muscle. The skull was carefully removed without disruption of the underlying dura. The CCI instrument consists of an electromagnetic impactor, Impact One (Leica Biosystems Inc., Buffalo Grove, IL) that allows alteration of injury severity by controlling contact velocity and the level of cortical deformation independently. Prior to injury induction, the tip of the impactor was angled and kept perpendicular to the exposed cortical surface. In these experiments, the contact velocity was set at 5.4 m/s, dwell time was set at 0.2 s, and deformation depth was set at 2 mm to produce moderate TBI. The injury site was allowed to dry prior to suturing the wound. During surgery and recovery, a heating pad was used to maintain the core body temperature of the animals at 36–37 °C.

Mice were treated with NAP (30 mg/kg i.p. in 0.1 mL/10 g body weight) or saline vehicle, with the first injection administered 5 h after injury and the second injection administered the next day (Figure 1). Sham animals were anesthetized, placed in the stereotaxic instrument, and subjected to craniotomy but had no CCI injury.

Behavioral Assessments

Asymmetrical Motor Function

Body asymmetry was quantitatively analyzed by the use of the elevated body swing test (EBST), as initially described by Borlongan and co-workers.¹²¹ Briefly, animals were examined for lateral movement/turning when their bodies were suspended 10 cm above the testing table. The animals were lifted from the table while held by the base of the tail. A left/right swing was counted when the head/torso of the animal moved more than a 10° angle from its vertical axis after elevation. The frequency of the left/right swings was scored across 20 consecutive trials. An uninjured animal shows an equal frequency to swing to either the left or right side. The number of contralateral swings was determined and used to generate a mean number of swings for each treatment group, which then was statistically analyzed.

Somatosensory Function Assessment

A tactile adhesive removal test (ART) was used to evaluate somatosensory function; this test measures the ability of the animal to perform sensitive paw-to-mouth movements, mouth-to-paw dexterity, and sensory input. Essentially, two small adhesive stickers were used as bilateral tactile stimuli that were placed on the distal–radial region on the wrist of each forelimb. Animals were pretrained daily for 3 days before CCI, and the time required (no longer than 2 min) for the animal to remove each sticker from the forelimb was recorded 4 days before CCI and at 1 and 2 weeks after CCI. The times taken to remove the stickers were used to generate a plot displaying the latency time of the sticker removal from each paw; the times were then used for statistical analysis.

Fine Motor Coordination

CCI-induced deficits in fine motor coordination were assessed by the use of a beam walk test (BWT). Mice have an inherent preference for a darkened enclosed environment, as compared to an open illuminated one. Each animal was placed in darkened goal box for a 2 min habituation and then the trial began from the other (light) end of the beam. The beam was constructed with the following dimensions: 1.2 cm (width) × 91 cm (length). The time taken for each animal to traverse the beam to reach the dark goal box and the number of ipsilateral and contralateral foot falls were documented, with the caveat that total time was not to exceed 30 s. Five trials were recorded for each animal before CCI and at 1 and 2 weeks after CCI. The mean times to traverse the beam were calculated, and a plot was generated to evaluate treatment effects on beam walk times and foot falls; these times were then used for statistical analysis.

Histological Analysis

Fixation and Sectioning

Animals were anesthetized with 2.5% tribromoethanol, Avertin (Sigma, St. Louis, MO) and perfused transcardially with 0.9% saline and 4% PFA in 0.1 M phosphate buffer (PB, pH 7.2). Brains were removed and postfixed for 1 day in 4% PFA and sequentially transferred to 20 and 30% sucrose in 0.1 M PB until the brain sank. The brains were cut into 25 um sections on a cryostat (Leica Biosystems Inc., Buffalo Grove, IL). Every seventh section was selected from a region spanning from striatum to hippocampus.

Quantification of Brain Lesion and Lateral Ventricle Size in TBI Animals

One set of post-TBI 2-week brain sections (25 μm) were mounted on slides. The sections were then stained in 10% Giemsa KH₂PO₄ buffered solution (pH 4.5) for 30 min at 40 °C. After a brief rinse, slides were destained, differentiated, and dehydrated in absolute ethanol. Thereafter, the sections were cleared in xylene and then coverslipped. Slides were scanned in an All-in-One Fluorescence Microscope BZ-X710 (Keyence Corporation of America, Itasca, IL), and brain image areas were quantified using ImageJ 1.52q software (National Institutes of Health, Bethesda, MD). The calculation formula for contusion volume size and lateral ventricle size were as follows: Σ(area of contralateral hemisphere – area of ipsilateral hemisphere)/Σ(area of contralateral hemisphere); Σ(area of ipsilateral lateral ventricle)/Σ(area of contralateral lateral ventricle). There were 9 brain sections from each mouse counted, with regions starting from bregma at 0.86 to −1.46 mm.

Immunofluorescence

Twenty-four brain sections per mouse were incubated with blocking buffer (4% Bovine Serum Albumin, Sigma, St. Louis, MO) for 1 h. A series of primary antibodies were prepared in the blocking buffer, and the sections were incubated in the solution overnight. The antibodies used were goat anti-GFAP (1:500; Abcam, Cambridge, MA), rabbit anti-TNF-α (1:500; Abbiotec, Escondido, CA), rat anti-P2RY12 (1:200; Biolegend, San Diego, CA), rabbit anti-PSD-95 (1:200; Invitrogen, Carlsbad,CA), mouse anti-MAP2 (1:500; Abcam, Cambridge, MA) or mouse anti-Iba1 (1:500; Abcam, Cambridge, MA). After incubation with primary antibody, the sections were washed and incubated for 3 h at room temperature in diluted secondary antibody prepared with blocking solution ((secondary antibody conjugated with Alexa 488 or 594 (1:500; Life Technologies, Grand Island, NY)). The sections were then washed with 0.1 M PB (pH 7.2), mounted with Antifade Mounting Medium with DAPI (Vector, Burlingame, CA), and cover-slipped. A series of 4 images per mouse brain were taken using a Laser Scanning Microscope (Zeiss 710, Oberkochen, Germany). Cell numbers of each image were counted using ImageJ 1.52q software (National Institutes of Health, Bethesda, MD).

Immunofluorescence Analysis and Quantification

Iba1, GFAP, TNF-α, P2RY12, and PSD-95-positive cells were identified with a 40×or 100× oil magnification objective. For each mouse, 4–6 fields of cortex, hippocampus, or thalamus were captured from both ipsilateral and contralateral hemispheres and analyzed. The number of immunoreactive (IR) cells in each field was quantified using NIH software ImageJ 1.52q. The results are presented as percentages of change compared with the contralateral hemisphere and/or the control group. Omission of the primary antibody was used as a further control, and sections from the different animal groups were reacted in the same well. Observers were blinded as to treatment group analysis.

With regard to Iba1 immunostaining, microglial cells were subclassified into morphological subtypes in line with prior studies,^122,123 as microglia morphology is considered highly representative of their functional state.¹²⁴ These subtypes included ramified- and intermediate-type microglial cells as well as amoeboid- and round-type cells.

Statistical Analysis

For statistical analysis of behavioral measurements, a two-way repeated measure analysis of variance (ANOVA) was used to evaluate both group and time factors. Multiple within-subject comparisons were undertaken with the Bonferroni’s correction post hoc test when the main effect of time was significant. For quantification of contusion volume size and lateral ventricle size, a one-factor analysis repeated measures ANOVA was used to compare the 3 groups of data followed by a Bonferroni’s correction post hoc test on data from 2 weeks postlesion. Data were analyzed using SigmaPlot version 12.5 (Systat Software Inc., San Jose, CA) with the significance level set at p < 0.05 for each assessment. All data are presented as the average ± standard error of the mean (SEM). The timeline for the histochemical and behavioral experiments with NAP is shown in Figure 4.

Glossary

Abbreviations:

ART

adhesive removal test

IKZF3

Aiolos

Aβ

amyloid-β peptide

AD

Alzheimer’s disease

ALS

Amyotrophic lateral sclerosis

ANOVA

analysis of variance

BWT

beam walk test

BBB

blood–brain barrier

BDNF

brain-derived neurotrophic factor

BSA

bovine serum albumin

BRD3

bromodomain-containing protein 3

CMC

carboxymethyl cellulose

CK1α

Casein kinase 1 alpha

CNS

central nervous system

CRBN/DDB1-CUL4A-Rbx1

cereblon/DNA damage-binding protein 1–Cullin 4a–Ring-box protein 1

CCI

controlled cortical impact

DAMPs

damage-associated molecular patterns

CRL4

E3 ligase complex Cullin-RING ligase 4

COX

cyclooxygenase

CSAIDs

cytokine-suppressive anti-inflammatory drugs

EBST

elevated body swing test

FCS

fetal calf serum

FRET

fluorescence resonance energy transfer

GDNF

glial cell line-derived neurotrophic factor

GFAP

glial fibrillary acidic protein

hESC

human embryonic stem

IKZF1

ikaros

IR

immunoreactive

IFN

interferon

IL

interleukin

Iba1

Ionized calcium binding adaptor molecule 1

LPS

lipopolysaccharide

MAP-2

microtubule- associated-protein 2

MS

multiple sclerosis

NAP

N-adamantyl phthalimidine

ND

neurodegenerative disorder

NSAIDs

nonsteroidal anti-inflammatory drugs

PFA

paraformaldehyde

PD

Parkinson’s disease

PAMPs

pathogen-associated molecular patterns

PPRs

pattern recognition receptors

PBS

Physiological buffered saline

PSD-95

Postsynaptic Density protein 95

P2RY12

purinergic receptor P2Y12

ROS

reactive oxygen species

SALL4

Sal-like protein 4

SEM

standard error of the mean

TLRs

toll-like receptors

TBI

traumatic brain injury

TNF-α

tumor necrosis factor-alpha

Author Contributions

S.C.H.: TBI studies, IHC, biochemistry, data analysis, editing; W.L.: synthetic chemistry/characterization, editing; D.T.: biochemistry/LPS studies, data analysis, editing; D.S.K., Y.K.K., I.H., J.-E.G., and B.-S.H.: cereblon binding/cellular studies, data analysis, editing; Y.-H.C., W.S., B.J.H., and N.H.G.: conceptualization, funding, supervision/mentoring, data analysis, manuscript writing.

This research was supported in part by the following: (i) The Intramural Research Program, National Institute on Aging, National Institutes of Health, United States; (ii) Grants from (a) the Ministry of Science and Technology, Taiwan (MOST 104-2923-B-038-001-MY3 and MOST 108-2321-B-038-008), (b) DP2-107-21121-01-N-05, Taipei Medical University, Taipei, Taiwan, and (c) National Institutes of Health R56 AG057028; and (iii) The Duane and Joyce Collins Neurosurgery Fund, Department of Neurological Surgery, Case Western Reserve University, Cleveland, OH.

The authors declare the following competing financial interest(s): NAP is protected under U.S. Patent Application No. 15,764,193 that is assigned to the NIH (US Government) and is being developed for the treatment of neurodegenerative disorders. N.H.G., W.L., and D.T. are named inventors on the patent application but have no rights to any agents within it, having assigned them in entirety to the NIH.

Notes

All primary data are available on request.