2026 年 49 巻 3 号 p. 496-502
Bortezomib, a first-in-class proteasome inhibitor, is widely used to treat multiple myeloma and other hematological malignancies. Despite its therapeutic efficacy, bortezomib causes peripheral neuropathy (PN) in approximately 20–30% of patients, often leading to dose reduction or discontinuation. Preventive or therapeutic approaches to bortezomib-induced PN are currently unavailable, as its precise mechanism remains unclear. In this study, we compared the effects of bortezomib and the second-generation proteasome inhibitor carfilzomib on peripheral nerve cells to identify candidate molecules involved in PN development. Transcriptome profiling of differentiated F11 cells, a hybridoma of a rat embryonic dorsal root ganglion and mouse neuroblastoma cell line N18TG2, revealed that bortezomib selectively upregulated α/β-hydrolase containing domain 4 (Abhd4), whereas carfilzomib did not. This finding was confirmed by quantitative RT-PCR and immunoblotting, which demonstrated consistent increases in Abhd4 mRNA and protein levels following bortezomib treatment. Functional analysis further revealed that Abhd4 overexpression promoted early apoptosis, suggesting a mechanistic link between bortezomib-induced Abhd4 elevation and neuronal vulnerability. Therefore, these results suggest that Abhd4 represents a candidate molecular signature associated with bortezomib-induced PN. Although further in vivo validation is needed, these findings warrant further investigation of Abhd4 as a potential contributor to bortezomib-induced PN.
Bortezomib, a first-generation proteasome inhibitor, has been frequently used to treat multiple myeloma.1) Despite the approval of novel agents, including immunomodulators, anti-CD38 antibodies, anti-B-cell maturation antigen (BCMA)-CD3 bispecific antibodies, and anti-BCMA chimeric antigen receptor T-cell (CAR-T) therapies, bortezomib remains a cornerstone in frontline regimens.2) However, peripheral neuropathy (PN) occurs in 20–30% of patients treated with bortezomib,3) resulting in dose reduction or discontinuation, compromising treatment outcomes. Therefore, the development of methods to prevent and/or treat bortezomib-induced PN is crucial to maintain the QOL of patients with multiple myeloma.
Several potential mechanisms have been implicated in bortezomib-induced PN, including neuronal damage,4) ion channel dysfunction,5) imbalance of tubulin dynamics,6) mitochondrial dysfunction,7) inflammation,8) and macrophage infiltration.9) These phenomena appear independent of proteasome inhibition, which is the primary antimyeloma mechanism of bortezomib.
Carfilzomib, a second-generation proteasome inhibitor with a similar antitumor mechanism, has been associated with a lower incidence of PN than bortezomib.10) Given this difference, a direct comparison of bortezomib and carfilzomib may help elucidate the mechanism underlying bortezomib-induced PN. To date, one report has demonstrated that bortezomib—not carfilzomib—exerts off-target binding to tubulin, thereby inhibiting its α-guanosine triphosphatase activity and causing tubulin hyperpolymerization.11) However, a systematic comparison of bortezomib and carfilzomib regarding alterations in the expression of related genes and their functional consequences is required to understand the mechanisms underlying the dose-limiting adverse events.
In this study, we aimed to identify genes differentially regulated by bortezomib but not by carfilzomib. To this end, we performed transcriptomic analysis of differentiated F11 cell lines, a widely used model for studying peripheral neuronal cells,12) followed by functional validation experiments.
Bortezomib, l-glutamine, retinoic acid (RA), and penicillin/streptomycin were purchased from FUJIFILM (Wako Pure Chemical Corporation, Osaka, Japan). Carfilzomib was obtained from MedChemExpress (Monmouth Junction, NJ, U.S.A.). Antimouse Abhd4 (mAbhd4) and anti-β-actin antibodies were obtained from Sigma-Aldrich (MO, U.S.A.). Anti-Neurofilament H (NfH) antibody was purchased from Novus Biologicals (Centennial, CO, U.S.A.). Rhodamine-conjugated goat antimouse antibody was obtained from Proteintech Group (Rosemont, IL, U.S.A.).
Cell Culture and Drug TreatmentF11 cells, a hybrid of a rat embryonic dorsal root ganglion (DRG) and mouse neuroblastoma cell line N18TG2, were purchased from European Collection of Authenticated Cell Cultures (Salisbury, U.K.; No. 08062601). These cells were cultured and maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific, Waltham, MA, U.S.A.), 1% penicillin/streptomycin, and 2 mM l-glutamine in a humidified atmosphere at 37°C with 5% CO2. A differentiation medium, containing 2% FBS/DMEM with 10 µM RA, was used to induce differentiation of F11 cells. Bortezomib and carfilzomib were diluted in dimethyl sulfoxide (DMSO) and added to the culture medium at 10 nM for drug treatment experiments. The concentration of 10 nM was chosen based on the previously described biochemical IC50 values of 5–10 nM for proteasome inhibition in myeloma cell lines.11) This also represents clinically relevant plasma concentrations achievable in patients.
RNA Extraction and Reverse TranscriptionTotal RNA was obtained from the cells using the QIAshredder and RNeasy Mini kits (QIAGEN GmbH, Hilden, Germany) according to the manufacturer’s instructions. Random-primed cDNA was prepared using 1 µg of total RNA with the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions.
Microarray AnalysisTotal RNA was extracted from cells collected 12 h after bortezomib or carfilzomib treatment. Expression profiling was performed using the Clariom S microarray assay kit (Thermo Fisher Scientific). Microarray analysis was performed as exploratory screening with a single biological sample per condition (n = 1). Expression data were normalized using the Signal Space Transformation-Robust Multi-array Average (SST-RMA) algorithm, and were analyzed and visualized using the Transcriptome Viewer software (KURABO, Tokyo, Japan). A cut-off threshold of a ≥3 log2 difference in signal intensity was chosen to identify genes with substantial expression changes while minimizing false positives. Candidate differentially expressed genes identified through this screening were subsequently validated by quantitative real-time PCR and immunoblotting with multiple biological replicates.
Quantitative Real-Time PCRQuantitative real-time PCR was performed using the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific) with TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) to quantify mRNA. We used probe-primer solutions specific for mouse Abhd4 (Mm00506368_m1) and mouse GAPDH (Mm99999915_g1) mRNA (Thermo Fisher Scientific). GAPDH was used as an internal control to normalize the mRNA expression. To confirm the hybridization capability of the probe-primer solutions with rat Abhd4 and GAPDH, the rat PC-12 cell line was used (data not shown).
ImmunoblottingImmunoblotting was performed as described previously.13) Briefly, cells were rinsed with ice-cold phosphate-buffered saline (PBS) and then treated with ice-cold lysis buffer (20 mM HEPES, 120 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10% glycerol, 10 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM sodium fluoride, and 5 µg/mL leupeptin). After measuring protein concentration using the Bradford assay, equal amounts of total protein were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and then transferred onto nitrocellulose blotting membranes. The membranes were blocked in Tris-buffered saline (TBS) with 0.05% Tween 20 containing 5% skim milk (Yukijirushi, Tokyo, Japan) at room temperature for 30 min. Then, they were incubated overnight at 4°C with primary antibodies against mouse Abhd4 (1 : 1000) or β-actin (1 : 1000) in TBS with 0.05% Tween 20. After incubating with horseradish peroxidase-conjugated anti-rabbit IgG antibody (1 : 5000) for 1 h at room temperature, the proteins were detected using the ECL Prime Western blotting Detection Reagent (Cytiva, Tokyo, Japan). Images of the blots were acquired using ChemiStage CC-16 (KURABO). Abhd4 protein expression was quantified by measuring band intensities using ImageJ and normalized to the respective β-actin intensity.
Cell Transfection and Apoptosis AssayPlasmid vectors expressing enhanced green fluorescent protein (EGFP)-labeled rat Abhd4 (#VB230626-1624wcz, pRP[Exp]-CAG > rABHD4[NM_001108866.1]/EGFP) and negative control EGFP (#VB900088-2245cyx, pRP[Exp]-CAG > EGFP) were purchased from VectorBuilder (Kanagawa, Japan). F11 cells were transfected with these plasmids using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer’s instructions, and then incubated at 37°C for 48 h.
To analyze apoptosis, the transfected cells were collected by pipetting and stained using the APC Annexin V Apoptosis Detection kit with propidium iodide (PI) (BioLegend, San Diego, CA, U.S.A.) according to the manufacturer’s instructions. Then, the stained cells were assessed using the FACS Melody Cell Sorter (BD Biosciences, Franklin Lakes, CA, U.S.A.) to acquire 30000 cells. The acquired data were analyzed using the FlowJo software v10.9 (BD Biosciences).
ImmunocytochemistryCells were fixed and then permeabilized in ice-cold methanol. After washing cells with PBS and blocking with goat serum, mouse monoclonal anti-Neurofilament H antibody was added and incubated for 60 min at 37°C. The cells were washed and incubated with an anti-mouse antibody conjugated to Rhodamine for 60 min, then washed again. Finally, the cells were incubated in Hoechst 33258 for 5 min and washed. The fluorescent-stained cells were placed on a glass-bottomed dish (Matsunami Glass, Osaka, Japan) and visualized under the BZ-X700 microscope (Keyence, Osaka, Japan).
Statistical AnalysisQuantitative data were presented as mean ± standard error of the mean. Comparisons between two groups were performed using an unpaired Student’s t-test. For comparisons between three or more groups, one-way ANOVA followed by Dunnett’s two-tailed post hoc test was used. All experiments were performed independently at least three times; p value of <0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics (version 26; IBM, Armonk, NY, U.S.A.).
First, the F11 peripheral nerve cells were differentiated to identify the molecules upregulated by bortezomib. RA is widely used to induce differentiation in various cells, including F11 cells.14) Furthermore, reports have shown that reducing FBS concentration in the culture medium promotes F11 cell differentiation.15) Therefore, we combined RA stimulation with decreased FBS concentration. After 14 d of incubation with medium changes every 2 d, RA-stimulated F11 cells exhibited elongated neurites and expression of NfH, a nerve cell marker (Figs. 1A, 1B). This differentiation protocol was used for subsequent transcriptome analysis of bortezomib- and carfilzomib-treated cells.
F11 cells were cultured in 10% FBS-DMEM (maintenance medium) or 2% FBS-DMEM supplemented with 10 µM RA (differentiation medium) for 14 d. After incubation, cell morphology was observed using a fluorescent microscope. White arrowheads indicate elongated neurites positive for Neurofilament H. (A) Bright-field images of cells cultured in maintenance (left) or differentiation (right) medium. (B) Fluorescence images after incubation. Nuclei and neurites were stained with Hoechst 33258 (blue) and Neurofilament H (red), respectively. Scale bars, 50 µm.
To identify early transcriptional changes, differentiated F11 cells were treated with bortezomib or carfilzomib at 10 nM, which is close to the reported IC50 value (5–10 nM) for proteasome inhibition in myeloma cell lines11) and relevant plasma levels in patients. Microarray analysis performed 12 h after drug treatment showed selective upregulation of Abhd4 mRNA following bortezomib exposure (Fig. 2A), whereas that induced by carfilzomib was not comparable (Fig. 2B). Notably, no genes met the cut-off criteria for common upregulation by both bortezomib and carfilzomib, suggesting that these proteasome inhibitors induce largely distinct transcriptional responses in differentiated F11 cells. Based on this observation, Abhd4 was chosen for further investigation.
mRNA was extracted from differentiated F11 cells treated with 10 nM bortezomib or carfilzomib for 12 h and subjected to microarray analysis. Signal intensities of individual genes in treated cells were plotted on the vertical axis, while those of untreated control cells were plotted on the horizontal axis. Scatter plots are shown for cells treated with (A) bortezomib and (B) carfilzomib. The cut-off threshold was defined as a ≥3 log2 difference in signal intensity relative to the control.
Quantitative real-time PCR analysis confirmed that the Abhd4 mRNA levels were significantly increased following 12 h of bortezomib treatment, consistent with microarray results (Fig. 3A). By contrast, carfilzomib treatment did not significantly alter Abhd4 mRNA expression (Fig. 3A). Immunoblotting performed 24 h after drug exposure demonstrated increased Abhd4 protein levels in bortezomib-treated cells, but not in carfilzomib-treated cells (Fig. 3B).
Differentiated F11 cells were treated with 10 nM bortezomib or carfilzomib, and Abhd4 expression was analyzed by real-time PCR and immunoblotting. (A) Relative Abhd4 mRNA expression levels from real-time PCR analysis (n = 3 independent experiments). (B) Representative immunoblot showing Abhd4 (approx. 38 kDa) and β-actin (approx. 42 kDa) (upper panel) and quantification of Abhd4 protein levels normalized to β-actin (lower panel) (n = 4 independent experiments). Statistical analysis was performed using one-way ANOVA followed by Dunnett’s test, with the DMSO-treated group as the control. Exact p values are shown; N.S:, not significant.
To examine the effect of Abhd4 upregulation on apoptosis, F11 cells were transfected with expression vectors encoding rAbhd4-EGFP or EGFP-only as a negative control. Overexpression of Abhd4 was confirmed by immunoblotting (Fig. 4A). The cytoplasmic localization of rAbhd4-EGFP was verified using fluorescence microscopy (Fig. 4B). Although the antibody used here is mouse-derived, it showed reactivity with rat. Next, apoptotic status was assessed by Annexin V/PI staining followed by flow cytometry. To identify Abhd4-overexpressing cells, cells with high EGFP fluorescence were gated. Then, quadrant gating was performed based on Annexin V/PI fluorescence intensity (Fig. 4C). The results showed that Abhd4 overexpression significantly increased the proportion of Annexin (+)/PI (−) early apoptotic cells, indicating enhanced susceptibility to early apoptosis (Fig. 4D).
F11 cells were transfected with plasmids encoding EGFP-tagged Abhd4 or EGFP alone as a negative control. Molecular weight markers are indicated on the left. (A) Overexpression of rAbhd4-EGFP (approx. 65 kDa) was confirmed by immunoblotting. (B) Cytoplasmic location of Abhd4-EGFP was assessed by fluorescence microscopy. Scale bars, 20 µm. (C) Representative gating strategy for flow cytometry-based apoptosis analysis. Abhd4-overexpressing cells were identified based on EGFP fluorescence, with gates set using non-transfected cells. (D) Quantification of Annexin V (+)/PI (−) early apoptotic cells among EGFP-positive cells (n = 3 independent experiments). Statistical analysis was performed using Student’s t-test comparing EGFP and Abhd4-EGFP-expressing cells. Exact p values are shown.
In this study, we compared the effect of bortezomib and carfilzomib treatment on the mRNA and protein expression profiles of F11 cells using microarray screening. The upregulation of Abhd4 was identified as a molecular signature distinguishing bortezomib from carfilzomib. We further confirmed that bortezomib upregulated the mRNA and protein levels of Abhd4. Functional analyses demonstrated that Abhd4 overexpression significantly increased the number of early apoptotic cells. These findings provide evidence of an association between Abhd4 and neuronal vulnerability to bortezomib. However, the exact contribution of Abhd4 to bortezomib-induced PN remains to be determined.
Although both bortezomib and carfilzomib inhibit the proteasome, their downstream cellular stress responses differ substantially, particularly in neuronal cells. Bortezomib is a reversible boronic acid-based inhibitor, whereas carfilzomib is an irreversible epoxyketone-based inhibitor, and these structural differences may result in distinct off-target activities. Notably, Malacrida et al. demonstrated that bortezomib, but not carfilzomib, binds to tubulin and induces tubulin hyperpolymerization, leading to cytoskeletal stress.11) Consistent with this phenomenon, our transcriptome screening did not identify genes that were commonly induced by both drugs under the present experimental conditions, suggesting that Abhd4 upregulation reflects a bortezomib-specific cellular response rather than a general consequence of proteasome inhibition. The precise upstream regulatory pathways responsible for Abhd4 induction remain to be elucidated in future studies.
Abhd4 is a lysophospholipase/phospholipase B that selectively hydrolyzes N-acyl phosphatidylethanolamines (NAPEs) and lysoNAPEs to generate N-acyl ethanolamines (NAEs).16) NAEs are bioactive lipids involved in diverse physiological processes, including nociception, cognition, anxiety, appetite regulation, and inflammation. Beyond its enzymatic role in NAE biosynthesis, Abhd4 has been shown to regulate resistance to anoikis, a form of apoptosis triggered by loss of cell attachment. A short hairpin (sh)RNA screening study showed that Abhd4 knockdown promoted anoikis resistance independently of its known role in the anandamide synthesis pathway and glycerophospho (GP)-NAE generation, implying that Abhd4 acts as a regulator of anoikis sensitivity.17) These observations align with our findings, collectively suggesting that Abhd4 upregulation might enhance susceptibility to cell death pathways, particularly in stress conditions such as those induced by bortezomib.
Nevertheless, whether Abhd4 upregulation directly drives early apoptosis or represents a secondary stress response remains unknown. One plausible explanation is that upregulated Abhd4 perturbs lipid homeostasis in neuronal cell membranes, thereby altering phosphatidylserine (PS) dynamics. PS, normally found in the inner leaflet of the plasma membrane, is externalized during early apoptosis, serving as a recognition signal for phagocytes. If Abhd4 activity modifies the availability of NAPEs or other membrane phospholipids, this may destabilize membrane symmetry and promote PS externalization, independent of classical apoptotic cascades. Supporting this hypothesis, macrophages have been shown to infiltrate sciatic nerves, DRG neurons, and the spinal cord during bortezomib-induced PN.9) Thus, Abhd4-dependent lipid remodeling may create a microenvironment that facilitates neuroinflammation and subsequent neuronal dysfunction.
The relevance of Abhd4-induced early apoptosis to bortezomib-induced PN pathology remains unexplored. One hypothesis is that neuronal cell death follows early apoptosis, resulting in structural damage to peripheral nerves and impaired neuronal transmission. Loss of DRG neurons is known to cause sensory neuropathy.4) Malacrida et al. showed that bortezomib, but not carfilzomib, causes tubulin hyperpolymerization via off-target binding to tubulin.11) Furthermore, Pero et al. reported delta-2 tubulin accumulation after bortezomib treatment, resulting in impaired microtubule stability and mitochondria motility.18) These cytoskeletal changes might act synergistically with Abhd4-mediated apoptosis in amplifying neuronal vulnerability.
Alternatively, Abhd4-induced early apoptosis may not necessarily lead to neuronal cell death. PS exposure during early apoptosis induces macrophage recruitment as macrophages recognize PS as an “eat-me” signal.19) Therefore, Abhd4-induced PS exposure could promote inflammatory infiltration, exacerbating PN, indicating a dual role of Abhd4—as an apoptotic regulator and a mediator of neuron–immune interactions in the peripheral nervous system.
This study has several limitations. First, our findings are based on in vitro experiments using differentiated F11 cells. While F11 cells are well-established for studying the functions of peripheral neuronal cells,12) the clinical relevance of Abhd4 upregulation should be validated using primary DRG neurons and in vivo models of bortezomib-induced PN. Second, although we demonstrated that Abhd4 overexpression promotes early apoptosis, further validation using loss-of-function experiments, such as gene silencing, is required to strengthen the causal relationship between Abhd4 and neuronal vulnerability. Third, we did not directly measure the enzymatic activity or downstream lipid products of Abhd4, which is known to act as a lysophospholipase that regulates NAE biosynthesis.16) Future investigations should examine whether bortezomib-induced upregulation of Abhd4 alters cellular lipid profiles and whether these changes contribute to neuronal toxicity. Fourth, the observed association between Abhd4 upregulation and early apoptosis should be further investigated, as Abhd4 upregulation may be a consequence rather than a cause of bortezomib-induced cellular stress. Moreover, early apoptosis does not necessarily lead to cell death and may represent a reversible stress response, warranting loss-of-function experiments to establish a causal relationship.
Taken together, our findings suggest that Abhd4 is a marker that distinguishes the effects of bortezomib from those of carfilzomib on peripheral nerve cells. While the mechanistic link between Abhd4 upregulation and PN pathology remains undetermined, these results provide a foundation for future investigations. Further studies are warranted to determine whether Abhd4 inhibition can mitigate bortezomib-induced PN in vivo and whether Abhd4 expression correlates with neuropathy severity in patients.
In conclusion, we identified Abhd4 as a molecular signature selectively induced by bortezomib, but not carfilzomib in peripheral nerve cells. Abhd4 overexpression was shown to promote early apoptosis, which might be correlated with bortezomib-induced neurotoxicity. However, the causal relationship and clinical relevance should be further validated using in vivo studies with established PN models and patient-derived samples. If the proposed future studies establish that Abhd4 contributes to PN pathogenesis, it may represent a potential therapeutic target.
This work was supported by JSPS KAKENHI (Grant Nos.: JP 20K07154 and JP 23K06278).
Yusuke Konishi is an employee of the for-profit company Sysmex Corp. The other authors declare no conflict of interest.
