Astrocyte elevated gene-1 (AEG-1) in myeloid cells is a key driver for the development of chemotherapy-induced peripheral neuropathy

Abstract

Chemotherapy-induced peripheral neuropathy (CIPN) is a dose-limiting side effect of chemotherapy treatment, often resulting in the discontinuation of treatment. Paclitaxel activates peripheral macrophages, generating a neuroinflammatory response that contributes to CIPN development and maintenance.

Astrocyte Elevated Gene-1 (AEG-1), also known as Metadherin or LYRIC, is a multifunctional protein that modulates macrophage activity and regulates inflammation through direct interaction with NF-κB, a transcriptional regulator of proinflammatory cytokine/chemokine (PIC) expression.

We aimed to determine whether AEG-1 contributes to the development and maintenance of CIPN pathologies by using both global (AEG-1 KO) and myelocyte-specific knockout (AEG-1ΔMAC) transgenic mouse strains in an animal model of CIPN that replicates specific human clinical phenotypes. We hypothesized that inhibition of AEG1 expression in myeloid cells, such as monocytes and macrophages, would prevent the development and maintenance of CIPN.

Our results showed that global AEG-1 deletion prevented the development of CIPN pathologies induced by PAC, as well as oxaliplatin (OHP). PAC treatment was found to increase AEG-1 and PIC expression in the DRGs of WT mice and in peritoneal macrophages isolated from C57BL/6J mice. However, in the absence of AEG-1 expression, PAC-induced neuroinflammation was completely halted in the DRGs of AEG-1 KO mice. This preventative phenotype and PIC expression profile was mirrored in AEG-1ΔMAC mice, which also displayed reduced NF-κB protein levels and F4/80+ macrophages trafficked to the lumbar DRGs following PAC treatment.

In summary, our results are the first to demonstrate the biological role AEG-1, particularly in myeloid cells, in development of CIPN.

Introduction

Chemotherapy-induced peripheral neuropathy (CIPN) is a serious dose-limiting side effect of many chemodrugs such as platinum-based drugs and taxanes [4], which can lead to dose reductions or the total cessation of treatment [10]. CIPN varies in intensity and duration, with recent studies reporting that 68% of cancer patients display CIPN symptoms one month after completing chemotherapy and up to 30% 6 months or longer after treatment [54]. Currently, there are no FDA approved preventative or therapeutic treatments for CIPN and the current pharmacotherapy is not very effective [9,14,37]. Although the pathogenesis of CIPN has been studied for the last two decades, it is not completely understood. However, accumulated evidence indicates that the initiation and progression of CIPN is associated with oxidative stress, disruption of microtubules, ion channel alterations and the activation of the neuro-immune system [16,17].

Chemotherapeutic drugs including drugs like paclitaxel (Taxol ^®) (PAC) lead to the activation of inflammatory cascades and the release of abundant cytokines and chemokines with pro- and anti-inflammatory characteristics, which play a pivotal role in the development and maintenance of CIPN. The immunomodulatory mechanisms of PAC are likely the cause of elevated NF-κB signaling, downregulation of anti-inflammatory cytokines (such as IL-4, IL-10), and upregulation of pro-inflammatory cytokines (PIC) (such as IL-1β, IL-6, TNFα) in response to PAC administration [2,11,43,60]. In addition, studies indicate that PAC primarily accumulates in peripheral afferent neurons such primary sensory neurons in the DRG [15,63]. The accumulation of PAC in these tissues imposes a greater risk for damage to the host peripheral nervous system. Primary sensory neurons in the DRG receive signals generated from peripheral nerve endings, which integrate and transfer these signals to the spinal cord. Because DRG neurons play a critical role in transducing sensory signals, including pain sensation, many studies have focused on their involvement in the development and maintenance of CIPN. PAC increases macrophage activation and recruitment in the DRGs coinciding with the development of cold hyperalgesia and mechanical hypersensitivity [44,69]. However, our understanding of the neuroimmune mechanisms in CIPN remains modest. Elucidation of the signaling mechanisms involved in the inflammatory crosstalk in CIPN can potentially provide a new basis for development of rational therapy.

Astrocyte Elevated Gene-1 (AEG-1) was discovered as an inducible gene in human primary fetal astrocytes exposed to TNFα or HIV-1 viral peptides. AEG-1 is best described as a scaffold protein, as it lacks enzymatic domains and functions primarily through protein-protein and protein-RNA interactions; its alternative name, Metadherin (MTDH), is derived from these functional properties [49]. It is through these direct interactions that AEG-1 can modulate a wide array of cellular signaling pathways, such as inflammation and lipid metabolism. AEG-1 has been shown to directly interact with and activate NF-κB, a key transcription factor, leading to an enhanced inflammatory response [52]. More importantly, this enhancement in inflammation is inhibited when AEG-1 is deleted, indicating that it too is an important regulator of the inflammatory response system [51]. An affymetrix oligonucleotide microarray analysis of hepatocytes harvested from AEG-1 overexpressing mice showed that AEG-1 activates NF-κB and upregulates pro-inflammatory cytokines/chemokines (PICs) (e.g. Tnfα, IL1β) while downregulating anti-inflammatory nuclear receptor genes (e.g. PPARα, PPARγ) and their downstream targets [55]. Again, this pro-inflammatory state was reversed in hepatocytes from AEG-1 global knockout (KO) mice. In a subsequent study, AEG-1 was shown to be expressed at significantly higher levels in Kupffer cells (liver-specific macrophages) compared to neighboring hepatocytes and a key regulator of inflammation and that its expression is required for macrophage activity [50]. However, most studies on the role of AEG-1 in inflammation have focused on cancer development and chemotherapy drug resistance [12]. In this report, we determined if AEG-1 plays a role in the development of CIPN. We hypothesize that AEG-1 mediates chemotherapy-induced neuroinflammation, via macrophage activation, in the DRGs leading to the development of CIPN. These data will pave the way for the development of novel therapeutic strategies that may be used to treat or even prevent the development of CIPN in cancer patients.

Methods

Animals

All experimental groups consisted of adult (12-28 weeks old) male and female C57BL/6J, AEG-1 global knockout (KO) and Wild-type (WT), AEG-1 myelocyte-specific KO (AEG-1^ΔMAC) and AEG-1^fl/fl mice. Studies were approved by the Institutional Animal Care and Use Committee of Virginia Commonwealth University. All studies were carried out in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals. Other information about experimental animals is described in the Supplementary Materials and Methods.

Drugs

Drugs are described in the Supplementary Materials and Methods.

Paclitaxel-induced peripheral neuropathy

This model of CIPN was induced using similar methods as described in [57]. Mice received a single schedule of systemic (i.p.) vehicle (VEH) or paclitaxel (PAC) (8 mg/kg) injections. One schedule of PAC, consisting of 4 individual injections, was administered on Days 0, 2, 4, and 6.

Oxaliplatin-induced peripheral neuropathy

This model of CIPN was induced using similar methods as described in [62]. Mice received one systemic injection (i.p.) of either VEH or oxaliplatin (OHP) (3 mg/kg) once a day for 5 consecutive days. The mice were then given a 5-day rest period, followed by another 5 consecutive days of single injections.

Behavioral assays

1. Locomotor activity

Spontaneous locomotor activity assay was performed using previously published methods [57].

2. Voluntary wheel running

Voluntary wheel running assay was performed using previously published methods [8].

3. Von Frey testing (evaluation of mechanical sensitivity)

Paw mechanical withdrawal thresholds were determined using von Frey filaments as previously described [57].

4. Acetone test (evaluation of cold sensitivity)

Duration of response to the cold stimuli, acetone, was assessed using previously published methods [57].

Other information about behavioral assays is described in the Supplementary Materials and Methods.

Electrophysiological assays

Caudal Tail nerve conduction

Nerve action potential amplitudes and conduction velocities of stimulated caudal nerves of mice of different experimental groups were recorded using the PowerLab 4/35 electromyography system (AD Instruments, Inc., CO, USA) in a manner as previously published [62]. The additional methods are described in detail in the Supplementary Materials and Methods.

Molecular assays

1. Intra-Epidermal Nerve Fiber (IENF) density

Immunohistochemistry and quantification of IENF density for mouse hind paws from each experimental group was performed using previously published methods [62]. Other information about the assay is described in the Supplementary Materials and Methods

2. FAC sorting for isolated peritoneal macrophages

C57BL/6J mice were euthanized by cervical dislocation 14 days after their initial drug injection and peritoneal cells were collected via peritoneal lavage. Prior to staining with fluorescent antibodies, cell suspensions were incubated with ACK lysis buffer (ThermoFisher Scientific, Waltham, MA) to remove red blood cells. Cell suspensions were then incubated with TruStain FcX (anti-mouse CD16/32, cat# 101320, BioLegend, San Diego, CA, USA) to block Fc receptors and prevent non-specific binding. Next, cell suspensions were then incubated with fluorescent antibodies [CD19 APC (cat# 152410); CD90.2-APC (cat# 105312); CD45-APC FIRE (cat# 103154); F4/80-BV421 (cat# 123132); and CD11b-PE/Cy7 (cat# 101216), BioLegend)] to identify and isolate peritoneal macrophage (CD90.2−/CD19−/CD45+/CD11b+/F4/80+) cell populations. Samples were sorted using a BD FACSAria Fusion SORP High-Speed Cell Sorter (BD Biosciences, San Jose, CA) instrument.

3. qRT-PCR

Snap frozen tissues (spinal cord overlaying the T13-L1 vertebrae, L4-6 lumbar dorsal root ganglia, and isolated peritoneal macrophages) were homogenized and purified RNA was extracted using the Invitrogen^™ PureLink^™ RNA Mini Kit (ThermoFisher scientific). Purified RNA concentrations for each sample were quantified using the QuBit 3.0 fluorimeter (ThermoFisher scientific). An aliquot from each sample was diluted to 20 (DRGs) or 10 (macrophages) ng/μl of RNA and was used to generate cDNA via reverse transcription reaction. Reverse transcription was performed using the iScript^™ cDNA Synthesis Kit (Biorad, Hercules). Each gene was amplified in PCR solution with synthesized primers as described in Supplementary Materials and Methods.

3. Immunofluorescence Staining

AEG-1^ΔMAC and AEG-1^fl/fl mice received four injections of either PAC (8 mg/kg) or vehicle every other day and were euthanized via rapid decapitation 14 days after their initial injection. Lumbar DRGs (L4-6) were dissected from these mice and placed into ice cold phosphate buffered saline (PBS) [pH of 7.4]. These tissues were transferred to 2 mL of 4% paraformaldehyde in PBS and fixed for 1 hour on ice. Tissues were then cryoprotected by incubating in 5 mL of 30% sucrose overnight at 4°C. DRGs were then embedded in OTC medium (cat# 23-730-571, Fisher Scientific, USA) and frozen on dry ice. Frozen tissue blocks were sliced into 14-micron sections via cryostat and mounted on albumin-coated glass slides (made in house). Immunofluorescence staining was performed using a previously published protocol with minor modifications [18]. Tissue sections were washed with PBST (PBS pH 7.4 with 0.3% Triton X-100) at RT, followed by a wash with PBS, and then incubated with TrueBlack^® Lipofuscin Autofluorescence Quencher solution (cat#92401, Cell Signaling Technologies, Danvers, MA, USA) for 3 minutes at RT in the dark (all remaining steps performed in dark setting). After another PBS wash, tissue sections were blocked for 1 hour at RT using fresh blocking buffer (PBS with 0.1% Triton X-100, 2.5% Normal Goat Serum, and 2.5% (w/v) Bovine Serum Albumin). Tissue sections were then incubated with primary antibodies [anti-F4/80 (rat, 1:100, cat #MA1-91124, ThermoFisher scientific, Waltham, MA, USA) or anti-NF-κB p65 (mouse, 1:50, cat#33-9900, ThermoFisher scientific, Waltham, MA, USA)] overnight at 4°C in a humidified chamber. The following day, tissue sections were washed with PBS and then incubated in appropriate secondary antibodies for 1 hour at RT [Goat anti-Rat IgG (H+L) Cross-Adsorbed Secondary Antibody conjugated with Alexa Fluor^™ 488 (1:300, cat# A11006) and Goat anti-Rabbit IgG (H+L) Secondary Antibody conjugated with Alexa Fluor^™ 594 (1:300, cat# A11037, ThermoFisher scientific, Waltham, MA, USA)]. Lastly, sections were again washed with PBS, mounted with Vectashield Antifade Mounting Medium with DAPI (cat# H-1200, Vector Laboratories, Burlingame, CA, USA), topped with a cover slide, and sealed with clear nail polish. Fluorescent images were captured using a Zeiss LSM 710 confocal microscope (Carl Zeiss, AG, Germany) under 40x magnification.

Determination of Paclitaxel by High Performance LC-MS/MS

The identification and quantification paclitaxel in mouse serum was performed using a modified previously published method [39]. Other information about paclitaxel levels assay is described in the Supplementary Materials and Methods.

Statistical analysis

Data obtained were analyzed using GraphPad software, version 9.5.1 (GraphPad Software, Inc., La Jolla, CA) and are expressed as the mean ± S.E.M. An unpaired student’s t-test was used for simple two-group comparisons for baseline values. Data from more than two groups were compared by two-way ANOVA (Holm-šídák multiple comparisons test) or three-way ANOVA (Tuckey’s post hoc test). In all cases, p≤ 0.05 was considered statistically significant. The additional analysis are described in detail in the Supplementary Materials and Methods.

Results

Global deletion of AEG-1 prevents paclitaxel-induced peripheral neuropathy (CIPN).

Impact of global AEG-1 deletion on basal nociceptive responses and spontaneous activity.

Prior to testing AEG-1 KO mice in our studies in the CIPN model, we wanted to determine if global AEG-1 deletion alone altered stimulus-evoked nociceptive responses and spontaneous behavioral activity in mice. Our battery of test did not reveal any significant differences between naïve AEG-1 KO and WT mice in mechanical sensitivity, cold sensitivity, voluntary wheel running distances or spontaneous locomotor activity (Supplementary Fig. 1).

Impact of global AEG-1 deletion on nociceptive responses in CIPN.

Initial studies determined the effect of global AEG-1 deletion on the development and maintenance of CIPN-associated behavioral outcomes in male and female AEG-1 KO and WT mice. We observed that WT mice treated with PAC displayed significant mechanical and cold hypersensitivity, compared to their VEH controls, immediately following their last PAC injection and was maintained throughout the 4-week testing period (Fig. 1A–B). In contrast, PAC treated AEG-1 KO mice displayed no significant differences in mechanical and cold sensitivity compared to their VEH controls at any of the time points tested (Fig. 1). Factorial analysis revealed significant impacts of treatment on PAC-induced mechanical and cold hypersensitivity. Similar to paclitaxel, we did observe a full prevention of OHP, a platinum-based chemotherapeutic agent, -induced mechanical and cold hypersensitivity as well as a reduction in percent body weight loss in AEG-1 KO mice (Supplementary Fig. 2).

Figure. 1: Global AEG-1 deletion prevents PAC-induced mechanical and cold hypersensitivity in mice.

Mechanical (A) and cold (B) sensitivity were assessed in VEH and PAC (8mg/kg, i.p.) treated AEG-1 KO mice (n= 12/group) and WT littermates (n= 12/group) before (BL) and up to 4 weeks after initial injection. Values are expressed as mean ± SEM. Results were compared using RM-three-way ANOVA with post hoc Tukey’s test. Factorial analysis revealed significant impacts of treatment [mechanical: F_treatment (1, 44) = 23.84, p< 0.0001 and cold: F_treatment (1, 44) = 257.9, p< 0.0001], genotype [mechanical: F_genotype (1, 44) = 49.18, p< 0.0001 and cold: F_genotype (1, 44) = 272.0, p< 0.0001], time [mechanical: F_time (3.313, 145.8) = 14.63, p< 0.0001 and cold: F_time (3.121, 137.3) = 59.80, p< 0.0001] and interacting factors [mechanical: F_{time x genotype} (4, 176) = 14.83, p< 0.0001; F_{time x treatment} (4, 176) = 10.93, p< 0.0001; F_{genotype x treatment} (1, 44) = 36.61, p< 0.0001; F_{time x genotype x treatment} (4, 176) = 8.959, p< 0.0001 and cold: F_{time x genotype} (4, 176) = 39.27, p< 0.0001; F_{time x treatment} (4, 176) = 42.84, p< 0.0001; F_{genotype x treatment} (1, 44) = 217.4, p< 0.0001; F_{time x genotype x treatment} (4, 176) = 40.57, p< 0.0001]. *** p< 0.001 vs VEH; ^{# # # #} p< 0.0001 vs. PAC-WT. PAC= paclitaxel, VEH=vehicle, WT = wild type, KO= AEG-1 knockout, BL= baseline.

Impact of global AEG-1 deletion on peripheral nerve fiber electrophysiology and nerve innervation in CIPN.

We continued our investigation into the role of AEG-1 in CIPN by exploring the impact of global AEG-1 deletion on electrophysiological and morphological impairments that occur in peripheral nerve fibers following PAC exposure. Fig. 2A shows that PAC treated WT mice developed significant reductions in peripheral nerve-conduction amplitude at weeks three (p= 0.0007) and four (p= 0.0078) compared to VEH controls. Conversely, PAC treated AEG-1 KO mice (n= 14/group) failed to display any significant differences in amplitude throughout the duration of the testing period when compared to VEH controls and presented significantly higher amplitude at week three (p< 0.0001) compared to PAC treated WT mice. Moreover, PAC treatment did not alter nerve conduction velocity at any of the time points tested in either AEG-1 KO or WT mice (Fig. 2B). To determine if prevention of PAC-induced deficits in nerve-conduction amplitude co-occur with prevention of mechanical and cold sensitivity, mice were tested 3 days after nerve conduction was performed on week 4. We observed significant mechanical (p< 0.0001) and cold (p< 0.0001) hypersensitivity in PAC treated WT mice compared to their VEH controls; while PAC treated AEG-1 KO mice did not and were statistically significant from PAC treated WT mice in both mechanical (p< 0.0001) and cold (p< 0.0001) sensitivity assays (Fig. 2C–D). A portion of these male and female mice were subsequently euthanized after behavioral testing was completed and hind paw skin tissues were collected to assess changes in intraepidermal nerve fiber (IENF) density following PAC treatment. Fig. 2E–F shows that PAC significantly reduced IENF density in both AEG-1 KO (n= 10/group, p= 0.0165) and WT (n= 9-10/group, p= 0.0015) mice compared to their respective VEH controls, however, IENF density was still significantly higher in PAC treated AEG-1 KO mice compared to their WT counterparts (p= 0.0182). Factorial analysis detected significant impacts of treatment, genotype, and time on changes in nerve-conduction amplitude, IENF density, and mechanical and cold sensitivity following PAC treatment.

Figure. 2: Global AEG-1 deletion prevents PAC-induced deficits in peripheral nerve fiber conduction amplitude in mice.

Nerve conduction amplitude (A) and velocity (B) were measured in VEH and PAC (8mg/kg, i.p.) treated AEG-1 KO mice (n= 14/group) and WT (n= 13/group) littermates at BL, and at 3 and 4 weeks after initial after initial injection. Three days after the last neve conduction study was performed (week 4), mechanical (C) and cold (D) sensitivity were assessed in these mice. Hind paw tissue samples were collected from these mice (n= 9-10/group) and IENF density was assessed (E). Representative immunostained sections of VEH treated WT and AEG-1 KO mice showing hind paw IENFs (F). Values are expressed as mean ± SEM. Results were compared using RM-three-way ANOVA with post hoc Tukey’s test (A, B), ordinary two-way ANOVA with post hoc Holm-Šídák test (C-E). Factorial analysis detected significant impacts of treatment [nerve-conduction amplitude: F_treatment (1, 50) = 9.179, p= 0.0039 and mechanical: F_treatment (1, 49) = 40.01, p< 0.0001 and cold: F_treatment (1, 49) = 61.55, p< 0.0001 and IENF: F_treatment (1, 35) = 20.33, p< 0.0001], genotype [nerve-conduction amplitude: F_genotype (1, 50) = 8.777, p= 0.0047 mechanical: F_genotype (1, 49) = 44.21, p< 0.0001 and cold: F_genotype (1, 49) = 59.40, p< 0.0001 and IENF: F_genotype (1, 35) = 10.16, p= 0.0030], time [nerve-conduction amplitude: F_time (2, 100) = 10.91, p< 0.0001], and interacting factors [nerve-conduction amplitude: F_{time x genotype x treatment} (2, 100) = 4.082, p= 0.0198 and mechanical: F_{genotype x treatment} (1, 49) = 19.53, p< 0.0001 and cold: F_{genotype x treatment} (1, 49) = 50.15, p< 0.0001 ]. * p< 0.05, ** p< 0.001, *** p< 0.01, **** p< 0.0001 vs VEH; ^# p< 0.05, ^{# # #} p< 0.001, ^{# # # #} p< 0.0001 vs. PAC-WT. PAC= paclitaxel, VEH=vehicle, WT = wild type, KO= AEG-1 knockout, BL= baseline.

Impact of global AEG-1 deletion on PAC levels.

Next, we investigated the effect of AEG-1 deletion on PAC metabolism in mice. AEG-1 KO (n= 5-9/group) and WT mice received an injection of 8 mg/kg PAC (i.p.) every other day for four days. Serum samples were collected at 1 and 8 hours after the last injection for mass spectrometry analysis. Our data showed that serum levels of PAC decreased over time between the 1 and 8 hr time points in both AEG-1 KO and WT mice, however, there were no significant differences in PAC levels between the AEG-1 KO and WT mice at either time point (Supplementary Fig. 3A).

AEG-1 expression is upregulated in the lumbar DRGs following paclitaxel administration.

Since the genetic deletion of AEG-1 was found to prevent the development of CIPN, we investigated if PAC administration would upregulate the neuronal expression of AEG-1. For that, we assessed the impact of paclitaxel administration on AEG-1 expression in key peripheral (lumbar DRGs) and central (spinal cord) nervous system tissues involved in the neurotransmission of pain/nociceptive sensory perception. WT mice received four injections of either VEH or PAC (8 mg/kg, i.p.) and their lumbar (L4-6) DRGs and spinal cords (T13-L1) were collected 1-week after their initial PAC (8mg/kg, i.p.) injection for qRT-PCR analysis. We observed a significant increase in AEG-1 expression in the lumbar DRGs of PAC treated WT mice (p= 0.0003, t= 3.995, df= 34) compared to their VEH controls (Fig. 3A). However, there was no significant difference in AEG-1 expression in spinal cord tissues (p= 0.8565, t= 0.1824, df= 30) between VEH and PAC treated WT mice (Fig. 3B). These data indicate that PAC treatment rapidly induces AEG-1 upregulation in the lumbar DRGs of mice and may contribute significantly to the onset of CIPN pathologies in mice.

Figure. 3: Paclitaxel administration upregulates AEG-1 expression during the onset of CIPN in mice.

The fold change in AEG-1 gene expression was quantified in lumbar DRGs (A) and spinal cord (B) tissues collected from VEH and PAC (8 mg/kg, i.p.) treated WT mice (n= 16-20) 1 week after their initial injection. Values are expressed as mean ± SEM. Results were compared using unpaired, 2-tailed student T-test. *** p< 0.001 vs. VEH. PAC= paclitaxel, VEH = vehicle, WT = wild type.

Global deletion of AEG-1 prevents paclitaxel-induced neuroinflammation in the lumbar DRGs.

Like the previous study, lumbar DRG samples were collected 1-week after their initial PAC (8mg/kg, i.p.) injection for qRT-PCR analysis. We observed significant increases in the expression levels of TNFα (p< 0.0001), IL1β (p= 0.0171), and IL6 (p< 0.0001) in the lumbar DRGs of PAC treated WT mice compared to VEH controls (Fig. 4A–C). Conversely, lumbar DRGs from PAC treated AEG-1 KO mice showed no significant differences in the expression of these PICs compared to VEH controls (IL1β: p= 0.8621, TNFα: p= 0.3445, IL6: p= 0.7072) and their expression levels were significantly lower compared to PAC treated WT mice (TNFα: p= 0.0019, IL1β: p= 0.0171, IL6: p< 0.0001) (Fig. 4A–C). Additionally, we assessed the expression of CCL2, a chemokine primarily involved in the recruitment of monocytes and macrophages, and C-C motif chemokine receptor 2 (CCR2), its cognate receptor, in the DRGs of PAC treated AEG-1 KO and WT mice. We observed a trend toward increase of CCL2 expression in PAC treated WT mice (p= 0.0533) compared to their VEH controls; however, this increased expression was significant compared to PAC treated AEG-1 KO mice (p= 0.0187) (Fig. 4D). Interestingly, CCR2 expression was increased in both PAC treated WT (p= 0.0185) and AEG-1 KO (p= 0.0123) mice compared to their respective VEH controls (Fig. 4E). These results suggest that AEG-1 expression plays a critical role in the induction of PAC-induced neuroinflammation and may contribute to monocyte recruitment to the lumbar DRGs.

Figure. 4: Global AEG-1 deletion prevents the onset of PAC-induced neuroinflammation in the lumbar DRGs of mice.

The fold change in TNFα (A), IL1β (B), IL6 (C), CCL2 (D), and CCR2 (E) gene expression was quantified in lumbar DRG tissues collected from VEH and PAC (8 mg/kg, i.p.) treated AEG-1 KO mice (n= 13-25) and WT littermates (n= 13-22) that were euthanized 1 week after their initial injection. Values are expressed as mean ± SEM. Results were compared using ordinary two-way ANOVA with Holm-Šídák multiple comparisons test. Factorial analysis revealed significant interactions between PAC treatment and genotype on TNFα [F_{treatment x genotype} (1, 57) = 6.942, p= 0.0108], IL1β [F_{treatment x genotype} (1, 64) = 4.776, p= 0.0325], IL6 [F_{treatment x genotype} (1, 77) = 11.59, p= 0.0011], and CCL2 [F_{treatment x genotype} (1, 79) = 4.484, p= 0.0374] expression in the DRGs. * p< 0.05, **** p< 0.0001 vs. VEH; ^# p< 0.05, ^{# #} p< 0.01, ^{# # # #} p< 0.0001 vs. PAC-WT. PAC= paclitaxel, VEH= vehicle, WT = wild type, KO= AEG-1 knockout.

Myeloid-cell specific deletion of AEG-1 prevents paclitaxel-induced peripheral neuropathy (CIPN).

Impact of PAC on AEG-1 expression in peripheral macrophages.

Based on our results which showed that the lumbar DRGs of PAC treated AEG-1 KO mice expressed lower levels of CCL2 compared to PAC treated WT mice (Fig. 4D), we hypothesized that the protective effects we observed in AEG-1 KO mice may be linked to the activation and recruitment of myeloid cells, such as monocytes and macrophages, to the DRGs in our model of CIPN. We performed a follow up study where we injected male and female C57BL6/J mice with one cycle of PAC (four injections of 8mg/kg, i.p.) or VEH and harvested peritoneal macrophages 2-weeks after their initial injection. Macrophages collected from PAC treated mice displayed significant increases in AEG-1, PICs, and myeloid cell recruitment genes in a sex-dependent manner (Supplementary Fig. 4). Altogether, these results suggest that AEG-1 upregulation may contribute to immune cell mediated inflammation and chemotaxis during the onset of PAC-induced CIPN. Therefore, we hypothesized that targeted deletion of AEG-1 in myeloid immune cells would prevent the development of CIPN pathologies and associated neuroinflammation.

Impact of Myeloid- cell specific deletion of AEG-1 on basal nociceptive responses and spontaneous activity.

For these studies, we utilized a previously validated LysM-CRE mediated conditional knockout (cKO) strain of mice where experimental animals expressed both LysM-CRE and floxed AEG-1 genes, resulting in AEG-1 deletion in myeloid-lineage cells (AEG-1^ΔMAC), while control littermates only expressed floxed AEG-1 genes (AEG-1^fl/fl) [50]. Similar to the global KO mice, we found no significant differences between AEG-1^ΔMAC and AEG-1^fl/fl in mechanical sensitivity, cold sensitivity or spontaneous locomotor basal activity (Supplementary Fig. 5).

Impact of Myeloid- cell specific deletion of AEG-1 on nociceptive responses and spontaneous behavioral outcomes in CIPN.

Next, we investigated the impact of of AEG-1 deletion specifically in myeloid-lineage cells on the development of CIPN-associated behavioral outcomes in male and female AEG-1^ΔMAC and AEG-1^fl/fl mice (n= 11-12/group). Fig. 5A–B shows that PAC treated AEG-1^fl/fl mice developed significant mechanical and cold hypersensitivity, compared to their VEH controls, that last throughout the duration of the four-week testing period. Conversely, PAC treated AEG-1^ΔMAC mice did not display any significant differences at any time point tested in mechanical and cold sensitivity compared to their VEH controls. We also performed the voluntary wheel running assay on AEG-1^fl/fl (n= 10-14/group) and AEG-1^ΔMAC mice (n= 13-16/group) at the 2-weeks post initial PAC injection time point. We observed that PAC treatment led to a significant reduction in voluntary wheel running distances in AEG-1^fl/fl mice compared to both VEH AEG-1^fl/fl controls (p= 0.0416) and PAC treated AEG-1^ΔMAC counterparts (p= 0.0104) (Fig. 5C). Factorial analysis revealed significant impacts of treatment, genotype, time and interacting factors on mechanical and cold sensitivity as well as voluntary wheel running distance in these mice.

Figure. 5: Myelocyte-specific deletion of AEG-1 prevents PAC-induced CIPN behavioral pathologies in mice.

Mechanical (A) and cold (B) sensitivity were assessed in VEH and PAC (8mg/kg, i.p.) treated AEG-1^ΔMAC mice (n= 12/group) and AEG-1^fl/fl littermates (n= 11-12/group) before (BL) and up to 4 weeks after their initial injection. Voluntary wheel running distances were also assessed in VEH and PAC treated AEG-1^ΔMAC (n= 13-16/group) and AEG-1^fl/fl (n= 10-14/group) mice 2-weeks after their initial injection (C). Values are expressed as mean ± SEM. Results were compared using RM-three-way ANOVA with post hoc Tukey’s test. Factorial analysis revealed significant impacts of treatment [mechanical: F_treatment (1, 43) = 21.70, p< 0.0001 and cold: F_treatment (1, 43) = 504.0, p< 0.0001 and wheel running: F_treatment (1, 49) = 4.316, p= 0.0430], genotype [mechanical: F_genotype (1, 43) = 45.96, p< 0.0001 and cold: F_genotype (1, 43) = 404.2, p< 0.0001 and wheel running: F_genotype (1, 49) = 5.646, p= 0.0214], time [mechanical: F_time (4, 172) = 18.11, p< 0.0001 and cold: F_time (4, 172) = 34.35, p< 0.0001], and interacting factors [mechanical: F_{time x genotype} (4, 172) = 8.089, p< 0.0001; F_{time x treatment} (4, 172) = 9.520, p< 0.0001; F_{genotype x treatment} (1, 43) = 18.69, p< 0.0001; F_{time x genotype x treatment} (4, 172) = 4.352, p= 0.0022 and cold: F_{time x genotype} (4, 172) = 24.98, p< 0.0001; F_{time x treatment} (4, 172) = 34.88, p< 0.0001; F_{genotype x treatment} (1, 43) = 378.9, p< 0.0001; F_{time x genotype x treatment} (4, 172) = 23.96, p< 0.0001]. * p< 0.05, **** p< 0.0001 vs VEH; ^# p< 0.05, ^{# # # #} p< 0.0001 vs. PAC-fl/fl. PAC= paclitaxel, VEH=vehicle, fl/fl= AEG-1^fl/fl, ΔMAC= AEG-1^ΔMAC, BL= baseline.

Impact of Myeloid- cell specific deletion of AEG-1 on peripheral nerve fiber electrophysiology and nerve innervation in CIPN.

We further investigated the impact of AEG-1 deletion specifically in myeloid immune cells on CIPN pathologies by examining peripheral nerve fiber electrophysiology and the density of peripheral nerves innervating hind paw skin samples collected from mice following PAC treatment. Caudal tail nerve conduction was performed on PAC and VEH treated AEG-1^fl/fl (n= 14-15/group) and AEG-1^ΔMAC mice (n= 16-17/group) at baseline (BL) and at three and four weeks post initial injection. While we observed no significant differences between VEH and PAC treated AEG-1^ΔMAC mice at either of the time points following treatment, we did observe a significant reduction of peripheral nerve fiber amplitude maintenance in PAC treated AEG-1^fl/fl mice at the four-week time point compared to their VEH controls (p= 0.0025) and PAC treated AEG-1^ΔMAC mice (p= 0.0392) (Fig. 6A). And again, we observed no significant alterations in peripheral nerve fiber velocity maintenance between any of the experimental groups at any of the time points tested (Fig. 6B). IENF density was assessed in hind paw skin samples collected from PAC and VEH treated AEG-1^fl/fl (n= 9-11/group) and AEG-1^ΔMAC mice (n= 10-13/group) immediately following nerve conduction testing on week four. We observed a significant reduction of IENF density in PAC treated AEG-1^fl/fl mice compared to their VEH controls (p= 0.0230) and PAC treated AEG-1^ΔMAC mice (p= 0.0158) (Fig. 6C–D). However, no significant differences in IENF density were observed between VEH and PAC treated AEG-1^ΔMAC mice. Factorial analysis revealed significant impacts of time, treatment, genotype and interacting factors on changes in nerve-conduction amplitude and IENF density following PAC treatment in these mice. These results indicate that the expression of AEG-1 in myeloid cells significantly contributes to the development of CIPN pathologies in mice.

Figure. 6: Myelocyte-specific deletion of AEG-1 prevents PAC-induced deficits in peripheral nerve fiber conduction amplitude and IENF density in mice.

Nerve conduction amplitude (A) and velocity (B) were measured in VEH and PAC (8 mg/kg, i.p.) treated AEG-1^ΔMAC mice (n= 16-17/group) and AEG-1^fl/fl littermates (n= 14-15/group) at BL, and at 3 and 4 weeks after their initial injection. Three days after the last neve conduction study was performed (week 4), AEG-1^fl/fl (n= 9-11/group) and AEG-1^ΔMAC (n= 10-13/group) mice were euthanized and hind paw skin IENF density was assessed (C). Representative immunostained sections of VEH treated AEG-1^fl/fl mice and AEG-1^ΔMAC showing hind paw IENFs (D). Values are expressed as mean ± SEM. Results were compared using RM-three-way ANOVA with post hoc Tukey’s test (A,B) and ordinary two-way ANOVA with post hoc Holm-Šídák test (C). Factorial analysis revealed significant impacts of time [nerve conduction amplitude: F_time (2, 116) = 23.76, p< 0.0001], treatment [nerve conduction amplitude: F_treatment (1, 58) = 5.967, p= 0.0176 and IENF: F_treatment F (1, 39) = 4.995, p= 0.0312], genotype [IENF: F_treatment (1, 39) = 4.364, p= 0.0433], and interacting factors [nerve conduction amplitude: F_{time x treatment} (2, 116) = 3.804, p= 0.0251; F_{treatment x genotype} (1, 58) = 4.476, p= 0.0387]. * p< 0.05, ** p< 0.01 vs. VEH; ^# p< 0.05 vs. PAC-fl/fl. PAC= paclitaxel, VEH=vehicle, fl/fl= AEG-1^fl/fl, ΔMAC= AEG-1^ΔMAC, BL= baseline.

Myeloid-cell specific deletion of AEG-1 prevents paclitaxel-induced neuroinflammation in the lumbar DRGs.

We hypothesized that AEG-1 expression in myeloid cells contributes to the development of CIPN pathologies by facilitating an inflammatory immune response within peripheral neurological tissues, such as the lumbar DRGs. To investigate this hypothesis, we collected lumbar DRGs from male and female AEG-1^fl/fl (n= 11-17/group) and AEG-1^ΔMAC (n= 13-17/group) mice 1-week after their initial PAC (8mg/kg, i.p.) or VEH injection for qRT-PCR analysis. We observed significant upregulation of AEG-1 (p= 0.0324) and target PIC genes, TNFα (p= 0.0242) and IL1β (p= 0.0332) but not IL6 (p= 0.9015), in the DRGs of PAC treated AEG-1^fl/fl mice compared to their VEH controls (Fig. 7A–D). Additionally, PAC treated AEG-1^fl/fl mice displayed a significant increase in NFκB1 (p50) (p= 0.0131) expression compared to their VEH controls (Fig. 7E). Conversely, PAC treated AEG-1^ΔMAC mice showed no significant differences in the expression of these genes compared to their VEH controls [AEG-1: p= 0.1137, TNFα: p= 0.4543, IL1β: p= 0.5726, IL6: p= 0.9015, NFκB1 (p50): p= 0.5127]. However, compared to PAC treated AEG-1^fl/fl mice, DRGs from PAC treated AEG-1^ΔMAC mice displayed significantly lower expression levels of AEG-1 (p= 0.0324), TNFα (p= 0.0059), IL1β (p= 0.0124), and NFκB1 (p50) (p= 0.0029) (Fig. 7). Additionally, we probed tissue samples collected from cKO mice at a later time point, 4 weeks after initial VEH or PAC (8 mg/kg, i.p.) injection, and observed that the loss of AEG-1 expression in myeloid cells continued to block the upregulation of PIC genes in the lumbar DRGs and spinal cords of PAC treated AEG-1^ΔMAC mice compared to their AEG-1^fl/fl counterparts (Supplementary Fig. 6). We also performed a qualitative follow up study using lumbar DRG tissue slices from VEH and PAC treated AEG-1^ΔMAC and AEG-1^fl/fl mice. We observed a clear reduction in F4/80+ macrophages (Supplementary Fig. 7) and total NFκB-p65 protein (Supplementary Fig. 8) in PAC treated AEG-1^ΔMAC compared to PAC treated AEG-1^fl/fl mice. Altogether, these results indicate that AEG-1 expression in myeloid cells significantly contributes to the onset of neuroinflammation and immune cell infiltration of the DRGs following PAC treatment.

Figure. 7: Myelocyte-specific deletion of AEG-1 prevents the onset of PAC-induced neuroinflammation in the lumbar DRGs of mice.

The fold change in AEG-1 (A), TNFα (B), IL1β (C), IL6 (D), and NF-κB (E) gene expression was quantified in lumbar DRG tissues collected from VEH and PAC (8 mg/kg, i.p.) treated AEG-1^fl/fl (n= 11-17/group) and AEG-1^ΔMAC (n= 13-17/group) mice that were euthanized 1 week after their initial injection. Values are expressed as mean ± SEM. Results were compared using ordinary two-way ANOVA with Holm-Šídák multiple comparisons test. Factorial analysis revealed significant interactions between PAC treatment and genotype on AEG-1 [F_{treatment x genotype} (1, 56) = 6.057, p= 0.0170], TNFα [F_{treatment x genotype} (1, 50) = 5.246, p= 0.0262], IL1β [F_{treatment x genotype} (1, 53) = 4.576, p= 0.0372] and NFκB1 (p50) [F_{treatment x genotype} (1, 56) = 6.057, p= 0.0170] expression in the DRGs. * p< 0.05 vs. VEH; ^# p< 0.05, ^{# #} p< 0.01 vs. PAC-fl/fl. PAC= paclitaxel, VEH= vehicle, fl/fl= AEG-1^fl/fl, ΔMAC= AEG-1^ΔMAC.

Discussion

The results of this study indicate that genetic deletion of AEG-1, particularly in myeloid immune cells, is sufficient to prevent the development of CIPN pathologies in an animal model that replicates specific human clinical phenotypes such as nociception, depressions in spontaneous behaviors, as well as reductions in peripheral nerve fiber stimulation and density in skin biopsy sections [5,8,57].

Our assessments of basal behavioral responses determined that global AEG-1deletion has no impact on sensory nociceptive or general motor function in mice (Supplementary Fig. 1) but fully prevented the development of PAC-induced mechanical and cold hypersensitivity (Fig. 1). Furthermore, a similar protective phenotype was observed in OHP-treated AEG-1 KO mice (Supplementary Fig. 2). These results demonstrate that the protective phenotype mediated by AEG-1 deletion may extend to other classes of clinically relevant chemotherapeutic agents. Both PAC and OHP have been shown to induce hypersensitivity through the activation of proinflammatory signaling pathways in both human patients and animals [7,25,38]. AEG-1 KO mice possess a profoundly inhibited inflammatory response, which may explain why they failed to develop CIPN associated hypersensitivity [51].

Global AEG-1 deletion was observed to fully prevent deficits in peripheral nerve fiber stimulation which concurred with the prevention of PAC-induced hypersensitivity in mice (Fig. 2A–D). Additionally, global AEG-1 deletion produced a partial prevention IENF density reduction following PAC treatment (Fig. 2E–F). The prevention of PAC-induced CIPN pathologies in the KO mice did not impact PAC serum levels (Supplementary Fig. 3). Therefore, we believe that AEG-1 inflammatory signaling must play a significant role in a molecular mechanistic pathway activated by PAC that contributes to the development of CIPN. Paclitaxel induces axonal damage through a variety of molecular mechanisms, such as microtubule stabilization and inflammation, which often results in excitotoxicity and nerve fiber loss [1,26].

The induction of neuroinflammation is a key molecular mechanism underlying the development of many PAC-induced CIPN pathologies [38]. PAC has been shown to stimulate the Toll-Like Receptor 4 (TLR4) signaling pathway which leads to the activation of NF-κB, a transcription factor that regulates the transcription of PIC genes, in multiple cancer cell types and myeloid immune cells [40,43,47,60,64–66]. Interestingly, studies have also shown that the TLR4 agonist, LPS, induces the expression of AEG-1 in these cells, and that inhibition of AEG-1 expression significantly impaired LPS/TLR4 mediated NF-κB activation [23,30,51,70]. The distinctive structural composition of the AEG-1 protein allows it to function as a regulator of inflammation through its direct interactions with multiple proteins involved in the NF-κB signaling pathway [49]. Based on these findings and our previous results, we wanted to determine if PAC treatment affects AEG-1 gene expression and if deletion of AEG-1 impacts the transcription of NF-κB-dependent PIC genes that contribute to PAC-induced neuroinflammation. We hypothesized that AEG-1 deletion prevents the onset of CIPN pathologies through the inhibition of proinflammatory signaling pathways. Our results showed that PAC treatment increased the expression levels of AEG-1 and PICs (TNFα, IL1β, IL6) in the DRGs of WT mice (Figs. 3 and 4A–C). Conversely, the upregulation of PIC genes was completely inhibited in the DRGs of PAC treated AEG-1 KO mice. These data indicate that not only does PAC induce AEG-1 upregulation in the lumbar DRGs, but that disruption of AEG-1 expression of fully prevents PAC-induced neuroinflammation as well. We postulate that this may be due to the impairment of the NF-κB signaling pathway, as AEG-1 plays essential roles in the degradation of IκB, an NF-κB regulatory protein, in the cytoplasm, the aggregation of ubiquitinated NF-κB signaling factors at the endoplasmic reticulum, and it also acts as a modulatory cofactor of the NF-κB-p65 transcription factor complex within the nucleus [3,13,27,52].

Myeloid immune cell infiltration is another hallmark of PAC-induced neuroinflammation, and is initiated in peripheral neurological tissues, such as the lumbar DRGs, during the onset of neuropathic pain [67–69]. Therefore, we probed DRG lysates for changes in CCL2, a myeloid cell recruitment chemokine, and CCR2, its cognate receptor. We observed a trend (p= 0.0533) in CCL2 expression in the DRGs of PAC treated WT mice, which was significant compared to PAC treated AEG-1 KO mice (Fig. 4D). While both AEG-1 KO and WT mice displayed a significant upregulation of CCR2 following PAC administration (Fig. 4E). We performed a follow up study where we observed a significant upregulation of AEG-1, our target PICs, and these myeloid recruit genes in circulating macrophages collected from PAC treated C57BL/6J mice (Supplementary Fig. 4). These results indicated that AEG-1 deletion may prevent PAC-induced neuroinflammation by suppressing inflammatory myeloid cell activity in peripheral neurological tissues.

AEG-1 has been shown to be ubiquitously expressed in nearly all mammalian tissues and cell types [20,22,29]. However, previous studies from our lab and others have shown that AEG-1 expression is significantly higher in tissue resident myelocytes compared to their neighboring cells, and that loss of AEG-1 significantly impairs their functional activity and migration during inflammation [23,31,36,50,51]. This expression pattern of elevated AEG-1 in myelocytes is concurrent with recent multi-omic analyses of various human tissues published in the human protein atlas [45,58]. Based on these studies and our previously described gene expression data (Figs. 3–4 and Supplementary Fig. 4), we further refined our hypothesis to AEG-1 deletion in myeloid cells will prevent PAC-induced CIPN pathologies through the inhibition of neuroinflammation. Utilizing a previously characterized LysM-Cre conditional KO (cKO) model developed within our [50], we observed that PAC-induced CIPN behavioral and electrophysiological pathologies, as well as IENF density reduction, were fully prevented in mice lacking AEG-1 expression in myeloid cells (AEG-1^ΔMAC) (Figs. 5–6). Interestingly, the preventative phenotype mediated by myelocyte-specific deletion of AEG-1 closely mirrored that of the global KO, which would indicate that AEG-1 expression in myelocytes represents a key driver of PAC-induced peripheral neuropathy. However, IENF density reduction was only partial prevented in AEG-1 KO mice while AEG-1^ΔMAC mice displayed full prevention. AEG-1 upregulation has been observed in Schwann cells following peripheral nerve injury and contributes to their ability to proliferate and migrate to damaged nerves [61]. Schwann cells play a significant role in axon survival, regeneration, and reinnervation following nerve injury [21]. AEG-1 also impacts inflammatory responses in other neurological cells, such as neurons [34] and astrocytes [22,33,56,59]. This may explain the minor difference in phenotype between the two strains.

We also observed that deletion of AEG-1 in myeloid cells prevented the upregulation of AEG-1, NF-κB (p50), and NF-κB-dependent PIC genes in the DRGs of PAC treated AEG-1^ΔMAC mice in both early (Fig. 7) and late stages of CIPN (Supplementary Fig. 6). These results lead us to perform a follow-up study where we investigated the impact of myeloid cell-specific deletion of AEG-1 on macrophage recruitment and NF-κB-p65 protein production in the DRGs of PAC treated mice. We observed a noticeable reduction in F4/80+ macrophages and NF-κB-p65 protein levels in the lumbar DRGs of PAC treated AEG-1^ΔMAC mice compared to their AEG-1^fl/fl counterparts (Supplementary Fig. 7–8). In addition to being a key binding partner of AEG-1 protein, elevated levels of activated NF-κB have been observed in the tissues of human patients and animals experiencing neuropathic pain [24,28,41]. This may further explain the minor differences in the gene expression profile and CIPN pathology between the AEG-1 KO and AEG-1^ΔMAC mice. Our interpretation of these results is that the deletion of AEG-1 in myeloid cells inhibits PAC-induced neuroinflammation and macrophage recruitment to the lumbar DRGs in an NFκB-dependent manner.

Overall, this is the first study to investigate the role of AEG-1, a well-established oncogene that has been shown to play a key role in multiple cancer hallmarks, expression in CIPN and demonstrates the potential of targeted AEG-1 deletion in myeloid immune cells as a therapeutic strategy for the prevention of CIPN. Furthermore, AEG-1 expression in tumor associated macrophages (TAMs) has been shown to significantly contribute to tumor progression, metastasis, and chemotherapy resistance [32,35,36,50]. Therefore, we believe that targeted inhibition of AEG-1 expression in macrophages may represent a holistic therapeutic strategy that can be utilized to prevent cancer progression and chemotoxicity, which could drastically improve survivorship and the overall quality of life for patients diagnosed with cancer. The work presented in this report also provides significant support for the exploration of targeting AEG-1 in myelocytes in other chronic pain models, such as chronic inflammatory pain [6,19,42,46,48,53].

Abbreviations:

AEG-1

Astrocyte elevated gene-1

CIPN

Chemotherapy-Induced Peripheral Neuropaty

PAC

Paclitaxel

PIPN

Paclitaxel-Induced Peripheral Neuropaty

DRG

Dorsal root ganlia

NF-κB

Nuclear factor kappa-light-chain-enhancer of activated B cells

IL1B

Interleukin 1 beta

TNFα

Tumor necrosis factor alpha

CCL2

C-C Motif Chemokine Ligand 2

CCR2

C-C Motif Chemokine Receptor 2

PIC

Proinflammatory cytokines/chemokines

KO

Knockout

WT

Wild Type

fl/fl

Floxed