Vincristine- and bortezomib-induced neuropathies – from bedside to bench and back

Abstract

Vincristine and bortezomib are effective chemotherapeutics widely used to treat hematological cancers. Vincristine blocks tubulin polymerization, whereas bortezomib is a proteasome inhibitor. Despite different mechanisms of action, the main non-hematological side effect of both is peripheral neuropathy that can last long after treatment has ended and cause permanent disability. Many different cellular and animal models of various aspects of vincristine and bortezomib-induced neuropathies have been generated to investigate underlying molecular mechanisms and serve as platforms to develop new therapeutics. These models revealed that bortezomib induces several transcriptional programs in dorsal root ganglia that result in the activation of different neuroinflammatory pathways and secondary central sensitization. In contrast, vincristine has direct toxic effects on the axon, which are accompanied by changes similar to those observed after nerve cut. Axon degeneration following both vincristine and bortezomib is mediated by a phylogenetically ancient, genetically encoded axon destruction program that leads to the activation of the Toll-like receptor adaptor SARM1 (sterile alpha and TIR motif containing protein 1) and local decrease of nicotinamide dinucleotide (NAD⁺). Here, I describe current in vitro and in vivo models of vincristine- and bortezomib induced neuropathies, present discoveries resulting from these models in the context of clinical findings and discuss how increased understanding of molecular mechanisms underlying different aspects of neuropathies can be translated to effective treatments to prevent, attenuate or reverse vincristine- and bortezomib-induced neuropathies. Such treatments could improve the quality of life of patients both during and after cancer therapy and, accordingly, have enormous societal impact.

Introduction

Vincristine (VCR) and bortezomib (BTZ) are powerful chemotherapeutic drugs widely used to treat several hematological cancers. The discovery of VCR dates back to 1958, when two research teams, one from the University of Western Ontario and the other from the pharmaceutical company Eli Lilly, independently tested the biological effects of an extract from the Madagascar rosy periwinkle (Catharanthus roseus, previously named Vinca rosea), a plant with beautiful deep pink flowers, in the hope of identifying new drugs to treat diabetes (Duffin, 2002; Johnson et al., 1963; Noble, 1990). This plant has enjoyed a popular reputation as a medicinal among indigenous peoples in various parts of the world, e.g., as a hypoglycemic by natives of Madagascar, the Philippines and South Africa. Although it turned out that the periwinkle extract was not very effective at lowering the blood sugar levels of mice, it dramatically decreased the number of leukocytes and prolonged survival of mice with leukemia. Subsequent fractionation of the vinca plant extract led to the isolation of an indole alkaloid that was named vincristine (Johnson et al., 1963). Further research showed that VCR inhibits the polymerization of tubulin and its incorporation into microtubules, thereby preventing mitotic spindle assembly and leading to an attenuation of mitosis and consequent apoptosis (Himes et al., 1976; Owellen et al., 1976; Smith et al., 2016). In 1963, VCR was approved by the FDA as Oncovin, and it is still widely used for treating leukemias, lymphomas, brain tumors, and solid tumors in adults (Gidding et al., 1999). It is also administered as part of treatment protocols for children with acute lymphoblastic leukemia, neuroblastoma, Wilms tumor, rhabdomyosarcoma, Ewing sarcoma, and retinoblastoma (Diouf et al., 2015; Gidding et al., 1999; Ness et al., 2013). After VCR came to the market, the survival rates of acute lymphoblastic leukemia in children increased from 10% to 90% (Gatta et al., 2005; Pui et al., 2009; Vora et al., 2013). VCR is now among the most commonly used anticancer drugs worldwide.

Disrupting tubulin polymerization also induces apoptosis in other rapidly dividing cell types and, in addition, blocks anterograde and retrograde axonal transport, resulting in multiple side effects, including myelosuppression, hair loss and neurotoxicity. Neurotoxicity is the most common dose limiting side effect of VCR, manifesting mainly as axonal sensorimotor and autonomic neuropathies, and, less commonly, cranial neuropathies (Deangelis et al., 1991; Lavoie Smith et al., 2015; Madsen et al., 2019). The severity of VCR-induced peripheral neuropathy (VIPN) in patients depends on dose intensity and total cumulative dose wherein higher individual and cumulative doses are associated with more severe neuropathy (Casey et al., 1973a; Sandler et al., 1969; Verstappen et al., 2005). Most adult patients treated with VCR develop sensory symptoms beginning at a cumulative dose of 4–5 mg and motor symptoms at doses around 30–50 mg (Casey et al., 1973). Loss of ankle jerks is usually the first manifestation produced by VCR, followed by paraesthesias (numbness and tingling) in the fingers and a little later in the toes (Casey et al., 1973; Verstappen et al., 2005). In a study of patients receiving high intensity chemotherapy with VCR (1.33 mg VCR weekly) 70% of patients experienced numbness, 62% pain and 60% tingling sensations, whereas low intensity VCR (0.67 mg VCR per week) resulted in numbness, pain and tingling in 43%, 14%, and 34% of patients, respectively. If weakness develops, it affects initially finger and wrist extensor muscles, and later toe and foot dorsiflexion. Patients frequently complain of difficulties writing, buttoning, walking and walking stairs. On neurological examination, patients have an abnormal Romberg sign, tandem walking, decreased sensation to pin prick and vibration and reduced strength of wrist and finger extension and toe- and ankle dorsiflexion (Casey et al., 1973b; Verstappen et al., 2005). Nerve conduction studies demonstrate decreased amplitude of distal muscle and sensory nerve potentials with only slight variations in conduction velocities, consistent with an axonal neuropathy (Casey, 1973; Guiheneuc et al., 1980) In addition, VCR can cause autonomic dysfunction, which leads to constipation, paralytic ileus, urinary retention and orthostatic hypotension (Hirvonen et al., 1989; Roca et al., 1985; Sandler et al., 1969). Cranial neuropathies, described mainly in children, are associated with vocal cord paralysis, ocular muscle paresis and tongue weakness (Dixit et al., 2012; Latiff et al., 2010; Naithani et al., 2009; Ryan et al., 1999). VIPN typically improves in patients after VCR treatment has ended, but further deterioration of neurologic signs continues for about 4 weeks after the last dose of VCR (“coasting” phenomenon) in about 30% of patients (Casey et al., 1973; Verstappen et al., 2005).

While VCR was discovered somewhat serendipitously, BTZ was identified in a targeted screen for inhibition of the proteasome (Adams, 2002). The proteasome plays a critical role in the regulated degradation of more than 80% of cellular proteins and is required for normal function of the cell. Thus, scientists surmised that dysregulating the degradation of such proteins should have profound effects on tumor growth and cause cells to undergo apoptosis. Indeed, compound PS-341 (later called bortezomib) caused cell death in several tumor cell lines maintained by the National Cancer Institute and decreased tumor growth in mice (Adams, 2002; Adams et al., 1999). Importantly, cancerous cells were much more sensitive to blockade of the proteasome than normal cells. BTZ inhibits the 20S core proteasome, resulting in cancer cell death via multiple mechanisms, including suppression of the unfolded protein response, accumulation of ubiquitinated proteins, stabilization of tumor suppressor proteins, such as p21, p27, Bax and p54, and induction of reactive oxygen species (Arkwright et al., 2017; Bladé et al., 2005; Utecht and Kolesar, 2008). A prominent theory regarding the cancer-killing properties of BTZ is the decreased activation of nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB), a transcription factor bound in the cytoplasm to its inhibitory regulatory protein IκB. Cellular stress and stimulation with pro-inflammatory signals induce the ubiquitination and subsequent degradation of IκB, leading to release and translocation of NFκB to the nucleus, where it induces the expression of specific proinflammatory and antiapoptotic genes. BTZ prevents the degradation of IκB and the consequent activation of NFκB, resulting in cell cycle arrest and induction of the apoptotic cascade (Adams, 2002; Arkwright et al., 2017). BTZ was approved by the FDA in 2003 for treatment of multiple myeloma and in 2007 to treat mantle cell lymphoma (Kane et al., 2006). Following FDA approval of BTZ, it has been used in the clinic for a large number of hematologic cancer patients.

Peripheral neuropathy is considered the main non-hematological toxicity of BTZ, and can result in dose modifications depending on its severity (Richardson et al., 2009, 2006). The incidence of bortezomib-induced neuropathy (BIPN) varies depending on the study, assessment method and patient population. A meta-analysis of 25 clinical trials including 3459 patients treated with intravenous BTZ revealed an incidence of all grade peripheral neuropathies of 33.9% and that of high-grade events of 8.1% (Peng et al., 2015). BIPN typically starts in the feet and is characterized by numbness and tingling, moderate to severe burning pain, and imbalance (Bilińska et al., 2013; Chaudhry et al., 2008; Expósito Vizcaíno et al., 2018; Zaroulis et al., 2014). Rarely, BTZ can also cause severe autonomic neuropathy manifesting as orthostatic hypotension (Giannoccaro et al., 2011; Stratogianni et al., 2012). On neurological examination, patients have loss of sensation to pinprick, vibration and proprioception in a stocking-glove distribution and decreased tendon stretch reflexes in their legs (Chaudhry et al., 2008; Richardson et al., 2009, 2006; Zaroulis et al., 2014). Sensory nerve action potential (SNAP) amplitudes of the sural nerve were decreased in 86% of patients with BIPN, indicating degeneration of large myelinated axons in sensory nerves, while a reduction of compound motor action potentials has been observed in up to 50% of patients (Bechakra et al., 2018; Chaudhry et al., 2008; Richardson et al., 2009; Stubblefield et al., 2006). A decrease of unmyelinated intraepidermal nerve fiber density (IENFD) on skin biopsies was noticed in some studies (Giannoccaro et al., 2011), but not others (Bechakra et al., 2020, 2018; Chaudhry et al., 2008; Richardson et al., 2009), which, instead found no change (Chaudhry et al., 2008) or increased IENFD (Bechakra et al., 2020; Richardson et al., 2009) and a decrease of nerve fibers in the subdermal plexus (Bechakra et al., 2020, 2018). In addition to the toxic neuropathy and in contrast to many other chemotherapy-inducing agents, BTZ can cause a severe, immune-mediated, demyelinating neuropathy (Pitarokoili et al., 2017; Ravaglia et al., 2008; Saifee et al., 2010; Schmitt et al., 2011; Xu et al., 2019).

VCR- and BTZ-induced neuropathies can last long after chemotherapy has ended and cause permanent disability and reduced quality of life. This is an especially significant problem in children with a long expected lifespan after treatment with VCR (Goldsby et al., 2010; Kandula et al., 2018; Kirchhoff et al., 2011; Lieber et al., 2018; Wright et al., 2017). Indeed, the clinical significance of chemotherapy-induced neuropathy (CIPN) is increasing proportionately with the increasing number of cancer survivors. While the mechanisms of action of both VCR and BTZ on cancer cells are well established, the molecular mechanisms that lead to peripheral neuropathy represent an area of intense research. Cellular (in vitro) and animal (in vivo) models have been generated to reveal mechanisms and screen for therapeutics (Bruna et al., 2020; Höke and Ray, 2014; Lehmann et al., 2020). More recently, these models have been complemented by genetic studies in patients (Campo et al., 2018; Diouf et al., 2015; Mahmoudpour et al., 2018), which generate new hypotheses that can then be taken back to the lab and tested in appropriate in vitro and in vivo models.

Here, I describe key discoveries that emerged from consideration of different in vitro and in vivo models of VCR- and BTZ induced neuropathies in the context of clinical findings. I then summarize recent findings pertinent to a common axon degeneration program that underlies VCR- and BTZ induced neuropathies and discuss how our understanding of molecular mechanisms of neuropathies can be translated to effective treatments for VIPN and BIPN.

Modeling BTZ-induced neurotoxicity

In vitro models of BTZ-induced neurotoxicity

Cellular models provide simple, well controlled and cost-effective systems that can be used as screening platforms to investigate cell-autonomous processes (Lehmann et al., 2020). Several in vitro model systems have been employed to evaluate BTZ-induced neurotoxicity, including primary neurons, murine neural stem cells, neurons derived from human induced pluripotent stem cells and neurons differentiated from rat pheochromocytoma- and human neuroblastoma cell lines (Meregalli et al., 2014; Poruchynsky et al., 2008; Staff et al., 2013; Wing et al., 2017). BIPN is sensory predominant and affects sensory dorsal root ganglion neurons (DRGs) and their processes early in the disease. Neurons derived from DRGs of embryonic, postnatal and adult animals can be easily grown in culture. Embryonic rodent DRG neuron cultures have the advantage of higher cell yields and produce greater proportions of neurons compared to DRGs cultures obtained from adults, but the cells are more immature and require neurotrophin supplementation during the first week of culture and, thus, may not fully represent the changes seen in adult neurons. In embryonic neuron cultures of mouse and rat DRGs, 50–100 nM BTZ, but not 10 nM, induces a dose-dependent, long-lasting inhibition of the proteasome (Meregalli et al., 2014), which leads to differential expression of 92 proteins related to cellular component organization (protein processing in the endoplasmic reticulum), macromolecular complex subunit organization, protein complex subunit organization, and cellular component assembly (Karademir et al., 2018). Proteasome inhibition in DRGs is associated with axon injury, as shown either by fragmented neurites (Geisler et al., 2019a) or decreased neurite outgrowth (Staff et al., 2013), indicating that BTZ affects neurites, which is similar to the clinical phenotype of axonal sensory neuropathy seen in patients treated with BTZ. The dose of BTZ in these in vitro models is clinically relevant, as the standard dose of BTZ (1.3 mg/m²) given to patients yields a median peak blood concentration that corresponds to a free concentration of 110 nM BTZ at peak (Staff et al., 2013).

BIPN symptoms in patients start in the feet, which may indicate perturbed axonal transport possibly due to interference with tubulin dynamics. Indeed, increased tubulin polymerization has been observed within 24 hours after adding 100 nM BTZ to embryonic DRGs (Meregalli et al., 2014; Staff et al., 2013). Similarly, neurons derived from human SH-SY5Y and KCNR neuroblastoma cells treated with 100nM BTZ show an increase of the polymerized tubulin fraction (Poruchynsky et al., 2008). The magnitude of tubulin polymerization is comparable to the effect of the microtubule stabilizer paclitaxel, suggesting that this might contribute to BTZ neurotoxicity (Poruchynsky et al., 2008). In contrast to paclitaxel, however, BTZ does not lead to significant microtubule bundling or polymerization of purified microtubule protein in solution, which suggests that the effect of BTZ is indirect and not a result of direct binding to microtubulin as occurs with other microtubulin stabilizing agents, such as paclitaxel (Poruchynsky et al., 2008). In a human cortical cell line (HCN2), but not embryonic DRGs, increased acetylation of alpha-tubulin was observed (Poruchynsky et al., 2008; Staff et al., 2013). Acetylation of α-tubulin is a post-translational modification indicative of stabilized microtubules. The tubulin polymerization is a class effect of proteasome inhibitors, because six different proteasome inhibitors induce tubulin polymerization in three different cell lines and embryonic rat DRGs (Meregalli et al., 2014; Poruchynsky et al., 2008; Staff et al., 2013). Increased tubulin polymerization and acetylated tubulin may cause decreased axonal transport. Indeed, decreased mitochondrial transport and increased histone deacetylase 6 (HDAC6) aggresomes in the soma were observed within the first 24 hours after BTZ was added to embryonic DRGs and HCN2 neurons (Poruchynsky et al., 2008; Staff et al., 2013).

Embryonic DRG sensory neurons are particularly suitable to be cultured in compartmentalized chambers, which allows for the isolation of axon and soma compartments to determine the subcellular site of action of the drug. Surprisingly, although BTZ induces an axonal neuropathy with symptoms starting in the feet, axon fragmentation was not induced by adding BTZ to the axon compartment (Geisler et al., 2019a). However, BTZ applied to the soma produced axon degeneration within 24 hours. These results indicate that the cell body is necessary for BTZ-induced axon degeneration, perhaps by transcribing a factor that is transported from the soma into the axon to activate a degeneration program, which has been shown for degeneration induced by local deprivation of nerve growth factors (NGF; Simon et al., 2016). In this regard it may be interesting that serum NGF is significantly decreased in patients who developed BIPN (Yan et al., 2020; Youk et al., 2017), while NGF levels were unchanged in patients who did not develop BIPN (Youk et al., 2017), suggesting that low NGF contributes to BIPN (Yan et al., 2020). Intramuscular injection of rat NGF in combination with oral vitamin B supplements for two months led to alleviation of symptoms and improvement of electromyography in patients with BIPN (Yan et al., 2020). However, whether NGF can be used therapeutically to decrease BIPN will require further investigation as a large phase III clinical trial of 1019 patients with diabetic neuropathy randomized to receive either recombinant human NGF or placebo for 48 weeks failed to show efficacy and was associated with high toxicity manifesting as hyperalgesia and allodynia at the injection site (Apfel, 2002a, 2002b).

DRG neuron cultures derived from adult animals (eight weeks) also exhibit proteasome inhibition, decreased neurite outgrowth and tubulin polymerization, but with greater vulnerability than embryonic neurons, because as little as 10 nM BTZ is sufficient to induce these changes in adult DRG culture (Meregalli et al., 2014).

BTZ also induces dose-dependent reduction of axon outgrowth and cell viability in human iPSC derived neurons, which may be especially useful to screen for therapeutics (Wing et al., 2017).

In vivo models of BTZ-neurotoxicity

While mechanistic studies of BTZ neurotoxicity performed in vitro are valuable, their inability to capture the inherent complexity of organ systems is a major drawback. Accordingly, results obtained from in vitro studies frequently benefit from in vivo verification. While animal models address many of the shortcomings of in vitro studies, they are time and resource-intensive, require advanced personnel training and must adhere to strict ethical guidelines (Bruna et al., 2020). To increase translatability to clinical practice, outcome measures of preclinical studies should focus on parameters and biomarkers with proven relevance in clinical settings (Cavaletti, 2014). Most animal models use intraperitoneal (i.p.) or intravenous (i.v.) injections of the chemotherapeutic drug. However, i.p. administration is not a mode of delivery in patients because of local toxic effects. Intravenous injections in animal models more closely resembles the situation in human patients. While it is difficult to evaluate spontaneous pain and tingling in rodents, the withdrawal of the hind paw in response to touch by a graded series of filaments (von Frey testing and its variations) is frequently used as a proxy to test for mechanical hyper- and hypoalgesia (Chaplan et al., 1994). More recently, a place escape avoidance paradigm and place preference tests were introduced to better reflect the affective/aversive component of neuropathy pain (Hamity et al., 2020, 2017). As an objective measure for axon loss and demyelination, nerve conduction studies are performed in patients and can also be done in mice and rats using the same set up and similar parameters (Argyriou et al., 2019). Similarly, skin biopsies from the ankle and distal thigh are obtained in patients to evaluate for loss of unmyelinated or thinly myelinated fibers to diagnose small fiber neuropathy (Lauria et al., 2005; McArthur et al., 1998; McCarthy et al., 1995). Using the footpad and the same stains and counting rules, these measures are also easily obtained in rodents and can be compared to findings in humans. If the animal models mirror key features observed in human patients, they can be exploited to provide information that is difficult to obtain in patients, including, particularly, insights into disease mechanisms.

An overview of different in vivo models to investigate BIPN is provided in table 1. In rats, commonly employed treatment regimens include: i) BTZ given daily over five days intraperitoneally (i.p.) or intravenously (i.v.) at doses of 0.2 or 0.4 mg/kg/day to a cumulative dose of 1 or 2 mg/kg, or ii) 0.2 mg/kg BTZ administered three times a week for 8 weeks to a cumulative dose of 4.8 mg/kg. The latter more reflects doses given to human patients in whom BTZ is infused twice weekly for two weeks (days 1, 4, 8 and 11), which is repeated after a 10-day rest period to complete one treatment cycle. Rats that receive BTZ for a short duration and with lower cumulative doses (1–2 mg/kg) show mechanical hyperalgesia that starts around day five and continues to at least day 35 (Stockstill et al., 2018). Similarly, these rats develop cold, but not heat, allodynia (Duggett and Flatters, 2017; Yamamoto et al., 2015). Administering BTZ over an eight week period, in addition, results in a decrease of the nerve conduction velocity of the caudal (tail) nerve and degeneration of myelinated and unmyelinated axons of the caudal and sciatic nerves, cytoplasmic vacuolization in satellite glia cells that envelope DRG neurons, detachment of satellite glia cells from the nerve cell body, and occasional morphological alterations in DRG neurons (Meregalli et al., 2010, 2014, 2018b).

Table 1:

Animal models of bortezomib-induced neuropathy

Strain	Gender	Age	Regimen	Route	Cumulative dose	Findings	References
Mouse
ICR	male	3–4 weeks	0.2 mg/kg/d × 5	i.p.	1 mg/kg	mechanical hyperalgesia transcription in DRG neurons	(Ludman and Melemedjian, 2019)
C57Bl/6	male	9 weeks	0.4 mg/kg 3x/week × 4	i.p.	4.8 mg/kg	mechanical and thermal hyperalgesia decreased sensory nerve action potential axon loss in the sciatic nerve neuroinflammation in DRGs, sciatic nerve, and spinal cord satellite glia cell degeneration, mitochondrial swelling in DRG neurons	(Boehmerle et al., 2015; Moschetti et al., 2019)
C57Bl/6	male	8 weeks	0.8 mg/kg twice weekly × 4	i.v.	6.4 mg/kg	decreased IENF density	(Geisler et al., 2019)
OF1	female	10–12 weeks	1 mg/kg twice weekly × 6	i.p.	12 mg/kg	mechanical hyperalgesia abnormalities on rotarod testing decreased sensory nerve amplitude potentials and conduction velocity decreased IENF density neuroinflammation	(Alé et al., 2014; Bruna et al., 2011)
OF1	n/a	10 weeks	1 mg/kg 2x/week × 8 weeks	i.p.	16 mg/kg	mechanical and thermal hyperalgesia decreased sensory nerve action potential axon loss in the sciatic nerve neuroinflammation in DRGs	(Alé et al., 2016)
Rat
SD	male	300–350g	0.15 mg/kg days 0,3,5,7	i.p.	0.6 mg/kg	mechanical hyperalgesia, no change to cold or heat allodynia increased neuronal firing in the dorsal horn of the spinal cord	(Caleb R. Robinson et al., 2014; C. R. Robinson et al., 2014; Robinson and Dougherty, 2015) (Duggett and Flatters, 2017)
SD	both	180–200 g	0.2 mg/kg days 0,3,7,10	i.p.	0.8 mg/kg	normal weight gain, mechanical and cold hyperalgesia no change to heat stimulation no effect on motor coordination	(Duggett and Flatters, 2017)
Wistar	male	200–250 g	0.2 mg/kg twice weekly × 2 weeks (days 1,4,8,11)	ip	0.8 mg/kg	mechanical hyperalgesia and cold allodynia decreased circularity index of the axons of the sciatic nerve	(Yamamoto et al., 2015)
SD	male	200–220	0.2 mg/kg/day × 5 days	i.p.	1 mg/kg	mechanical hyperalgesia astrocyte activation and neuroinflammation dorsal horn spinal cord	(Stockstill et al., 2018; Zhang et al., 2014)
SD or Wistar	male only or female only	220–250	0.4 mg/kg/day × 5	i.p.	2 mg/kg	mechanical and cold allodynia activation of transcriptional programs in DRG neurons neuroinflammation in DRGs	(Duan et al., 2020; Li et al., 2018; Liu et al., 2016, 2019; Zhang et al., 2014)
Wistar	female	Adult	0.2 mg/kg 3x/wk × 4 weeks	i.v.	2.4 mg/kg	decreased sensory nerve conduction velocity vacuolization of Schwann cells and myelin sheath degeneration, no change in unmyelinated axons in sciatic nerve damage satellite glia cells	(Cavaletti et al., 2007)
Wistar	female	175–200g	0.2 mg/kg 3x/wk × 8 weeks	i.v.	4.8 mg/kg	mechanical hyperalgesia, no change in heat sensitivity decreased sensory nerve conduction velocity loss of myelinated axons and morphological alterations of unmyelinated axons in the sciatic nerve, cytoplasmic vacuolization of satellite glia cells, shrinkage of DRG neurons persistent proteasome inhibition	(Chiorazzi et al., 2013; Meregalli et al., 2014, 2012, 2010)

Similar to rat, many different models are employed in mice with cumulative doses ranging from 1 mg/kg to 16 mg/kg. The findings in mice are similar to those obtained in rats and include mechanical hyperalgesia in shorter treatment protocols (Ludman and Melemedjian, 2019a), whereas longer protocols, in addition, report decrease of sensory nerve amplitudes and degeneration of myelinated axons in the sciatic nerve and of unmyelinated/thinly myelinated intraepidermal nerve fibers (Alé et al., 2015; Boehmerle et al., 2015; Bruna et al., 2011; Carozzi et al., 2010; Carozzi et al., 2013; Geisler et al., 2019a).

Another important consideration is that cancer patients given BTZ, particularly multiple myeloma patients, frequently have neuropathy at baseline, likely due to the hematological malignancy (Stubblefield et al., 2006). While most protocols evaluate BTZ in non-cancerous mice, Meregalli et al. established a protocol in which immunocompromised (SCID) mice were injected with human multiple myeloma cells and treated with 1 mg/kg BTZ i.v. weekly for 5 weeks (Meregalli et al., 2015), a dose that effectively decreased tumor growth in these mice. Multiple myeloma bearing mice had signs of mild axon degeneration in the sciatic nerve and a decreased soma size of DRG neurons at baseline. Despite these changes, the neurotoxic effects induced by BTZ were similar between tumor-bearing and non-tumor bearing mice and included decreased nerve conduction velocity, mechanical hyperalgesia, mild axon degeneration in the sciatic nerve and vacuolization of satellite glia cells (Meregalli et al., 2015).

Thus, different treatment regimens are being utilized to evaluate mechanisms for BTZ-induced neurotoxicity, some of which are discussed below.

BTZ-induced neuroinflammation and pain

Bortezomib-induced neurotoxicity is peculiar among different types of chemotherapy-induced neuropathy as it can be extremely painful (Bilińska et al., 2013). Many patients with BIPN suffer from intense, persistent burning pain that significantly reduces quality of life long after chemotherapy has ended. Because currently available drugs treat only symptoms with only partial success (Dorsey et al., 2019; Majithia et al., 2016), many studies on BIPN aim to reveal molecular mechanisms that lead to BTZ-induced pain in order to develop more effective therapies. Much progress has been made in recent years elucidating mechanisms underlying neuropathic pain after peripheral nerve injury, and it has become clear that BTZ activates some of the same pathways. Pain results from the detection of noxious stimuli by sensory neurons in the DRG, relay of action potentials from DRGs to the spinal cord, and transmission of the warning signal to the brain (Jain et al., 2020; Scholz and Woolf, 2007; von Hehn et al., 2012). As discussed in more detail below, the development of neuropathic pain after BTZ involves dysregulation of transcriptional programs in DRGs, activation of the peripheral immune system, altered synaptic transmission in the dorsal horn of the spinal cord, and transformation of spinal astrocytes (Fig. 1).

Figure 1: Molecular mechanisms of BIPN and sites of therapeutic intervention based on results obtained from rodent models.

BTZ leads to activation of transcriptional programs and channel overexpression in DRG neurons, which results in neuroinflammation and increased neuronal firing. In the dorsal horn of the spinal cord, BTZ induces astrocyte, but not microglia, activation, dysregulation of sphingolipid metabolism and neuroinflammation, which is associated with increased firing of wide dynamic range neurons and pain sensitization. Several pharmacological interventions indicated in green have been shown to reduce BTZ-induced hyperalgesia in rodent models, including TNFalpha and NLRP3 antagonists, Fingolimod (blocks sphingosine-1-phosphate receptor signaling in astrocytes), PC1, (inhibits prokineticin receptors), metformin (blocks the binding of HIF1alpha to the HIF1-alpha responsive binding element), and IVIg (blocks neuroinflammation). Mechanisms that lead to BTZ-induced axon degeneration and treatment strategies are shown in more detail in figure 2. Abbreviations: ATF3 – activating transcription factor 3; Ccl2 – chemokine ligand 2; Ca_v3.2 – voltage gated calcium channel 3.2; HIF1alpha – Hypoxia inducible factor 1 subunit alpha; IL – interleukin; M – macrophage; NLRP3 – nod like receptor protein 3; PC1 – prokineticin receptor antagonist; PK – prokineticin; S1P – sphingosine 1 phosphate, STAT3 – signal transducer and activator of transcription 3; TNF-alpha – tumor necrosis factor alpha; TNFR – TNF receptor; TRPA - transient receptor potential ankyrin 1 ion channel.

BTZ does not penetrate into the central nervous system, but crosses the blood-nerve barrier to accumulate in DRGs and peripheral nerves, such that these are the main anatomical sites of direct neurotoxicity (Adams et al., 1999; Meregalli et al., 2014). In line with in vitro findings indicating that BTZ-induced neurotoxicity is at least in part transcriptionally regulated (Geisler et al., 2019a), thus requiring the cell body involvement to induce axon degeneration, in vivo studies have revealed induction of several transcription programs underlying BTZ-induced pain in DRG neurons (Fig. 1).

One of the earliest activated transcription factors following BTZ administration is the cytosolic Hypoxia-Inducible Factor 1 Subunit alpha (HIF1α; Ludman and Melemedjian, 2019b). Upon hypoxia and metabolic stressors, HIF1α translocates to the nucleus where it binds to the hypoxia response element (HRE) to regulate the expression of mitochondrial and metabolic factors. In the absence of these stressors, HIF1α remains in the cytosol where it undergoes continual proteasomal degradation. Five days after initiation of BTZ in mice, HIF1α protein levels are increased in lumbar (L4–6) DRG neurons. The increase of HIF1α is associated with mechanical hyperalgesia, whereas suppression of HIF1α with siRNAs or blocking the binding of HIF1α to HRE with the antibiotic echinomycin prevents BTZ-induced hyperalgesia (Ludman and Melemedjian, 2019b). These results suggest that transcription of genes that are under HIF1α control is crucial for the development of BTZ-induced neuropathic pain. Importantly, the commonly used anti-diabetic drug metformin suppresses the translation of HIF1α in DRGs and thereby prevents the development of BTZ-induced pain in this mouse model (Ludman and Melemedjian, 2019b). Blockade of HIF1α has no effect once pain is established, indicating that the binding of HIF1α to HRE plays a role in the development of BTZ-induced hyperalgesia, but not in its maintenance.

A hallmark of neuropathic pain after nerve cut is a robust immune response that includes infiltration by activated immune cells, activation of glial cells and increased production of proinflammatory mediators at the injury site, which, collectively, have been shown to sensitize nociceptors and induce pain (Ji et al., 2016; Scholz and Woolf, 2007; Shamash et al., 2002). The pro-inflammatory cytokine tumor necrosis factor alpha (TNFα), in particular, has a lead role in activating a cascade of other cytokines, notably interleukin (IL) 1β, IL-6, IL-8 and transforming growth factor (TGF) β. Similar to nerve injury, BTZ leads to increased expression of mRNAs encoding several cytokines. However, in contrast to nerve injury that is associated with a pronounced immune response in the peripheral nerve (Shamash et al., 2002), the immune response after BTZ occurs first and most potently in DRGs and only subsequently and to a lesser extent in the peripheral nerve and dorsal horn of the spinal cord (Alé et al., 2014).

During a course of administration of 1mg/kg BTZ i.p. twice weekly for 5 weeks, expression of TNFα mRNA increased in DRGs after only two BTZ injections (Alé et al., 2014). This was followed by a peak in IL-6 mRNA and an increase of the TNF receptor 1 in the second week, suggesting auto-induction of TNFα at later time points (Alé et al., 2014). In contrast to nerve injury, in which TNFα is released at the injury site by Schwann cells and macrophages (Shamash et al., 2002), high TNFα immune reactivity was especially abundant in small DRG neurons, whereas DRG satellite glia cells did not exhibit increased TNFα expression following BTZ administration (Alé et al., 2014). At later time points, the levels of TNFα and IL-6 decreased in DRGs, but to levels still higher than observed in control animals. In the peripheral nerve and dorsal horn of the spinal cord, TNFα and IL-6 mRNA expression peaked later and to a lesser extent than in DRGs. Co-administration of antibodies against TNFα prevented BTZ-induced hyperalgesia, sensory-motor dysfunction and even reduction in sensory nerve amplitude (SNAP; Alé et al., 2014; Chiorazzi et al., 2013; Li et al., 2018; Zhang et al., 2014). While the increase in TNFα after nerve cut is a consequence of the axon injury, the preservation of the SNAP amplitude following block of TNFα after BTZ suggests that axon loss after BTZ treatment is, at least in part, secondary to TNFα activation and consequent neuroinflammation.

The binding of TNFα to its cell surface receptors engages multiple signal transduction pathways, including activation of mitogen-activated protein (MAP) kinases, such as cJun NH₂-terminal kinases (JNKs). Accordingly, phosphorylated JNK1 and 2 are found in TNFα positive DRG neurons on days 5 and 10 (but not day 1), during daily i.p. administration of BTZ for 5 days at a dose of 0.2 mg/kg (Zhang et al., 2014). Knockout of TNF receptor 1 and 2 or injection of thalidomide, an inhibitor of TNFα synthesis, significantly blocks BTZ-induced JNK activation and the development of allodynia (Zhang et al., 2014). JNK is a critical upstream messenger of the transcription factor c-jun, which can bind to the promoter of the small chemokine ligand 2 (ccl2), a potent monocyte-attracting chemokine that greatly contributes to the recruitment of blood monocytes into sites of inflammatory responses. Following BTZ, c-jun and ccl2 mRNA are upregulated in DRG neurons, but not local glia cells, and remain elevated for at least 5 days after discontinuation of BTZ treatment (Liu et al., 2016). The binding of c-jun to the ccl2 promoter is enhanced by the Activating Transcription Factor (ATF) 3, which is also increased in DRG neurons at that time (Liu et al., 2016). Blocking c-jun prevents BTZ-induced ccl2 upregulation and attenuates macrophage infiltration into DRGs and mechanical hyperalgesia. However, blocking ccl2 does not completely rescue BTZ-induced hyperalgesia, indicating that other mechanisms are also involved in BTZ-induced pain (Liu et al., 2016).

TNFα also activates p38 MAP kinase, which leads to increased expression of the non-selective cation channel transient receptor potential ankyrin 1 (TRPA1). Expressed by nociceptors, TRPA1 is activated by various irritants, including mustard oil and capsaicin, and integrates a variety of noxious stimuli. Blocking p38 MAP kinase blocks BTZ-induced increased expression of TRPA1 and hyperalgesia (Guo et al., 2020; Liu et al., 2019). Similarly, the IL-6 inhibitor SC144 reduces BTZ-induced upregulation of pJNK, p38 MAP kinase and TRPA1 and mechanical hyperalgesia and block or genetic deletion of TRPA1 prevents BTZ-induced mechanical and cold hyperalgesia, highlighting the importance of this pathway in the development of BTZ-induced pain (Liu et al., 2019; Trevisan et al., 2013; Wang et al., 2017). The expression of TNFα appears to be partially regulated by micro-RNA155 (miR155), because inhibition of miR155 decreases BTZ-induced expression of TNFα, p38MAPK and TRPA1 and decreases mechanical hyperalgesia (Duan et al., 2020).

Thus, several animal models revealed early and robust BTZ-induced enhancement of TNFα and its effect on downstream signaling cascades that lead to mechanical hyperalgesia, whereas inhibiting TNFα signaling at different points in the cascade attenuates BTZ-induced pain. Interestingly, patients who develop grade 2 or higher peripheral neuropathy during BTZ therapy show a distinct increase of serum TNFα level, which is observed before development of peripheral neuropathy and increases further after peripheral neuropathy is detected (W. Zhao et al., 2019). Because TNFα levels correlate so closely to neuropathy symptoms, the early increase in TNFα level can be used as an effective biomarker to predict the development of peripheral neuropathy, which would allow intervention in the disease process early, possibly by using one of several FDA-approved TNFα inhibitors. As BTZ decreases TNFα in cancer cells (Hideshima et al., 2001), but increases it in nervous tissue (Alé et al., 2014; Li et al., 2018; Zhang et al., 2014), treating BIPN with TNFα antagonist would produce synergistic effect on BIPN and cancer.

Similar to its effect on cancer cells, BTZ leads to activation and upregulation of Signal Transducer and Activator of Transcription 3 (STAT3; C.-C. Liu et al., 2018). STAT3 is a transcription factor that modulates the expression of a variety of genes, including those that control cell growth, survival, and immune response. Phosphorylated STAT3 binds to the Nod like receptor protein 3 (NLRP3) promoter leading to increased NLRP3 mRNA and protein expression in neurons and macrophages in DRGs following BTZ (C.-C. Liu et al., 2018). NLRP3 belongs to the inflammasome family and has been shown to be critical in driving the immune response to sterile tissue damage (Cowie et al., 2019). Intrathecal activation of NLRP3 induces hyperalgesia, whereas inhibition of STAT3 or NLRP3 leads to attenuation of BTZ-induced hyperalgesia (C.-C. Liu et al., 2018). NLRP3 expression also contributes to oxaliplatin and paclitaxel-induced neuropathies (Jia et al., 2017; Wahlman et al., 2018). Given the central role of NLRP3 activation to different forms of chemotherapy-induced neuropathies and many other diseases, many direct and indirect inhibitors are being evaluated in preclinical and clinical studies and may be useful in blocking BTZ-induced pain (Zahid et al., 2019).

STAT 3 signaling also plays a crucial role in the activation of prokineticin 2 (Qu et al., 2012; Xin et al., 2013). Prokineticins belong to a new family of cytokines first shown to cause profound GI contractions, hence their names (Y. Zhao et al., 2019). Prokineticins induce a pro-inflammatory macrophage profile, stimulate chemotaxis and prompt the release of other pro-inflammatory cytokines. Comprising of prokineticin 1 and 2, they act through two G-protein coupled receptors (prokineticin receptor 1 and 2), are widely expressed in many tissues and involved in a large spectrum of biological activities, including inflammation and nociceptive neurotransmission (Y. Zhao et al., 2019). In DRGs, prokineticin 1 receptor is mainly expressed in small nociceptor neurons also expressing the transient receptor potential vanilloid receptor channel TRPV1, while PRK2 is expressed in medium/large sized neurons co-localized with TRPA (Maftei et al., 2020). Prokineticin 2 protein and Prokineticin 1 and 2 receptor expression are increased following BTZ, but, surprisingly, only late in the course, after macrophage infiltration into DRGs was observed and after mechanical hyperalgesia had developed (Moschetti et al., 2019a). However, treating the mice with the PKR antagonist PC1 at a time when BTZ-induced mechanical hyperalgesia was well developed reversed the allodynia, decreased a late proinflammatory response in DRGs, sciatic nerve and spinal cord and lessened BTZ-induced structural damage of satellite glia cells (Moschetti et al., 2019a). Importantly, PC1 was also effective in decreasing BTZ induced allodynia after BTZ administration was finished. Thus, while the prokineticin system has no role in the initial development of BTZ-induced pain, PC1 may be an effective drug for patients treated with BTZ after symptoms have appeared. The protection from BTZ-induced pain may be due to reversal of neuroinflammation in the spinal cord and prevention of secondary sensitization necessary for sustained pain.

Experiencing pain depends on the efficient transmission of nociceptive information from peripheral nociceptor neurons to second-order neurons in the spinal cord. The spinal cord superficial dorsal horn is the site of synapses between DRG and second order neurons and where the intensity of pain is critically controlled (von Hehn et al., 2012). Although it does not cross the blood brain barrier, BTZ induces increased firing and persistent after discharges of wide dynamic range neurons in the dorsal horn of the spinal cord in rats (Robinson et al., 2014; Robinson and Dougherty, 2015), an effect mediated, at least partially, by tonic activation of presynaptic NMDA receptors (Carozzi et al., 2010; Xie et al., 2017) and astrocytes. Astrocytes are activated and express increased gap junction protein 43 early in the course of BIPN, well before proinflammatory cytokines are increased and invasion of macrophages into the dorsal horn of the spinal cord is observed (Robinson and Dougherty, 2015). Interestingly, and in contrast to pain induced by spinal nerve ligation, microglia are not activated in the dorsal horn of the spinal cord after BTZ treatment (Robinson et al., 2014). Astrocyte activation contributes to mechanical hyperalgesia, as concurrent treatment of rats with the anti-inflammatory minocycline or gap junction decoupler carbenoxolone prevented the development of mechanical hyperalgesia and the increased expression of connexin 43 and the astrocyte marker glial fibrillary acidic protein (Robinson et al., 2014; Robinson and Dougherty, 2015).

One mechanism by which astrocytes contribute to sensitization and pain behavior in BIPN is via sphingosine-1 phosphate signaling. Sphingolipids are crucial components of cellular membranes (Singh and Spiegel, 2020). The metabolism of sphingolipids generates signaling molecules such as ceramide, sphingosine and sphingosine 1 phosphate (S1P) that are essential for many biological functions. Bortezomib caused an increase of sphingosine 1 phosphate receptor (S1PR) ligands, including S1P, and this dysregulation was associated with time-dependent development of neuropathic pain (Stockstill et al., 2018). Activation of S1PR via intrathecal administration of the S1PR1 agonist SEW2871 caused mechanical allodynia, increased levels of NfκB, phosphorylated p38 and NLRP3 (Doyle et al., 2019). In the central nervous system, including the spinal cord, S1PR is most highly expressed in astrocytes. Astrocyte specific deletion of S1PR1 or treatment with the S1P1R antagonist Fingolimod (trade name Gilenya®, Novartis) prevented the development of BTZ-induced neuropathic pain, astrocyte activation and neuroinflammation in the dorsal horn of the spinal cord, including the increase of TNFα, IL-6 and NLRP3 expression. (Fig. 1) Astrocytic S1PR also increased presynaptic glutamate release in the dorsal horn of the spinal cord following BTZ, which may contribute to increased firing of dorsal horn neurons and sensitization, whereas blockade of S1PR prevented increased firing (Stockstill et al., 2018). Importantly, 20 days after conclusion of BTZ treatment, at a time when pain was fully developed, infusion of Fingolimod completely reversed the neuropathic pain (Stockstill et al., 2018). Fingolimod is an immunomodulating drug used in patients mostly for the treatment of multiple sclerosis. Thus, the preclinical findings of attenuating BTZ-induced pain by Fingolimod should be easily translatable to the bedside. Indeed, a clinical trial currently evaluates Fingolimod for treatment of CIPN once it has occurred (clinicaltrials.gov #NCT03943498). Fingolimod may be especially useful for patients treated with BTZ, as it is not only efficacious in treating BTZ-induced hyperalgesia, but also bone pain (Grenald et al., 2017) that is frequently observed in multiple myeloma patients. In addition, Fingolimod effectively targets BTZ-resistant cells, increasing their sensitivity to BTZ and leading to superior tumor growth inhibition (Beider et al., 2017), further supporting the use of Fingolimod to treat BIPN in multiple myeloma patients. One caveat may be that Fingolimod was effective in reversing BTZ-induced pain only in male, but not female rats (Stockstill et al., 2020), which may be due to gender-specific differences in the innate immune system that are also observed, e.g., for NLRP3 activation (Cowie et al., 2019).

Intravenous immunoglobulins (IVIg) are commonly used in clinical praxis as immunomodulatory therapy in several immune-mediated diseases, including inflammatory neuropathies. Given the pronounced neuroinflammation in different models of BIPN, Meregalli et al. (2018) evaluated whether a clinically relevant treatment regimen can prevent BIPN in a rat model (Meregalli et al., 2018b). Administering IVIg at a dose of 1g/kg every 2 weeks during BTZ administration, they observed less macrophage infiltration in the sciatic and caudal (tail) nerves, improved mechanical and heat allodynia, and protection from loss of IENF compared to vehicle treated rats. However, the BTZ-induced decrease of nerve conduction velocity reflecting function of myelinated axons did not improve with IVIg treatment, suggesting that IVIg protected especially small nerve fibers.

Surprisingly, despite the prominent role of neuroinflammation in BIPN, immunodeficient animals treated with BTZ develop a painful peripheral neuropathy with the same features observed in immunocompetent mice (Carozzi et al., 2013). This finding may be due to release of proinflammatory cytokines from DRG neurons (in addition to infiltrating macrophages) or because other mechanisms also have important roles in the development of BIPN. For instance, by inhibiting the proteasome, BTZ directly increases protein levels of the calcium channel Cav3.2 in nociceptors of the DRG, which leads to increased neuronal excitability and pain behavior (Tomita et al., 2019). In addition, axon degeneration is an important component of BIPN that contributes to pain. The mechanism of axon degeneration after BTZ and VCR are similar and will be described together further below.

Models of VCR-induced neurotoxicity

By binding to microtubule ends VCR blocks the addition of new tubulin subunits, leading to microtubule depolymerization and inhibition of microtubule dynamic instability (Himes et al., 1976; Owellen et al., 1976; Smith et al., 2016). Microtubules are critical for intracellular transport, serving as tracks for “motor” proteins that carry vesicular cargo. Of the vinca-alkaloids vinblastine, VCR and vinorelbine, VCR has the highest affinity for tubulin and the highest incidence of peripheral neuropathy (Lobert et al., 1996). Furthermore, in isolated squid axoplasm vesicle motility assays, 1 μM VCR inhibits fast anterograde and retrograde axonal transport with the highest potency when compared to paclitaxel, erubulin, and ixabepilone (LaPointe et al., 2013). The inhibition of axonal transport by VCR was confirmed in in vivo models in some (Green et al., 1977) but not other studies (Bradley and Williams, 1973). The health of the neuron depends on both anterograde transport of proteins, RNA, mitochondria and other organelles from the cell body to the synapse and retrograde transport of toxic components (misfolded proteins, damaged organelles) from the synapse to cell body. Disruption in axonal transport is thought to affect the longest axons first, causing a dying back peripheral neuropathy in which the most distal parts of nerves are affected first in the feet and hands in a “stocking-glove” distribution. The dying back phenomenon of VCR-induced neuropathy has been modeled in in vitro and in vivo studies.

In vitro models of VCR-induced neurotoxicity

The estimated chronic level of VCR during cancer chemotherapy in humans ranges between 10–100 nM (Silva et al., 2006). Adding 100 nM VCR to plated embryonic DRG neurons is neurotoxic and results in rapid degeneration of axons and soma, whereas 20 nM has no effect (Silva et al., 2006). Exposure of DRG neurons to 40 or 50 nM leads to selective degeneration of neurites while the soma remains intact (Geisler et al., 2019a; Silva et al., 2006). In contrast to BTZ, which induces axon degeneration when added to the soma compartment (Geisler et al., 2019a), adding 40 or 50 nM of VCR to the soma or proximal axon compartments has no effect on the integrity or growth of DRG neuron axons (Geisler et al., 2019a; Silva et al., 2006). However, when the distal axon compartment is exposed to the same dose of VCR, axons became fragmented and their lengths decreased in a dying back pattern, suggesting a distal to proximal gradient of axon degeneration (Geisler et al., 2019a; Silva et al., 2006). These data were confirmed by combining a microfluidic divider with a multielectrode array. Exposure of axons, but not cell bodies, to VCR caused electrophysiologic failure that preceded morphological signs of axonal degeneration (Ravula et al., 2007). The amplitudes of the degenerating axons declined, beginning in the outer third of the axon compartment. Similarly, when VCR was added to the soma compartment, potassium chloride mediated depolarization remained constant throughout the 2 day period, whereas 6 hours after addition of VCR to the axonal compartment spike activity increased in the soma compartment, suggesting that an injury signal is transmitted from the axon to the cell body (Ravula et al., 2007). Degenerative morphological changes were first noticed in neurites 18 hours after adding VCR, after which the rate of degeneration rapidly progressed (Ravula et al., 2007). This observation led Jonathan Glass and colleagues to suggest almost 20 years ago that VCR-induced neurotoxicity is due to the activation of a local axon degeneration program (Silva et al., 2006; Wang et al., 2000). This visionary hypothesis has now been confirmed and molecular mechanisms of this pathway identified (Essuman et al., 2017; Geisler et al., 2019b, 2016; Gerdts et al., 2013; Summers et al., 2018; Yang et al., 2015) as we shall see further below.

In vivo models of VCR-induced neurotoxicity

The dose-dependent VCR toxicity in patients described earlier is also observed in animal models, in which VCR at higher doses is associated not only with more signs of neuropathy, but also higher mortality (Table 2). The most commonly used rodent model of VIPN consists of five daily intraperitoneal or intravenous injections of 0.1 mg/kg VCR, which is repeated after two days rest, resulting in a cumulative dose of 1 mg/kg (Table 2). This regimen results in mild weight loss (~7%), and typically no mortality. Under this treatment schedule, mice and rats developed mechanical and thermal hyperalgesia, hyper-responsiveness of C-fibers, neuroinflammation in the sciatic nerve, DRGs and spinal cord and gastrointestinal changes, including delayed gastric emptying, decreased gastrointestinal motility and damage to the intestinal wall (Table 2). Myelinated and unmyelinated axons were swollen and exhibited cytoskeletal changes consisting of disturbed orientation of microtubules, decreased microtubule density and abnormal clustering of neurofilaments in the central portion of the axoplasm (Tanner et al., 1998; Topp et al., 2000). In conjunction with the pale toluidine blue staining of DRG neuron cell bodies and proximal axons due to build-up of neurofilaments, these findings are morphological signs of impaired axonal transport. A loss of intra-epidermal nerve fibers in the skin and of myelinated and unmyelinated axons in the sensory saphenous nerve was not observed using this treatment schedule (Topp et al., 2000).

Table 2:

Animal models of vincristine-induced neuropathy

Strain	Gender	Age	Regimen	Route	Cumulative dose	Findings	References
Mouse
CD1	male	30–35g	1.7 mg/kg twice weekly × 10 weeks	i.p.	34 mg/kg	74% mortality thermal hypoalgesia, gait disturbance axon degeneration in the sciatic nerve	(Contreras et al., 1997)
Swiss OF1	female	12 weeks	1.5 mg/kg twice weekly × 4 weeks	i.p.	12 mg/kg	11.5% mortality abnormalities on rotarod testing decreased sensory and motor nerve amplitudes and conduction velocities decreased number of myelinated axons, macrophage infiltration in the sciatic nerve, decreased IENF density	(Bruna et al., 2011)
C57Bl/6	male & female	12–20 weeks	1.5 mg/kg twice weekly × 4 weeks	i.p.	12 mg/kg	17.5% mortality mechanical and heat hyperalgesia, decreased tail nerve amplitude without change in conduction velocity axon degeneration in toe nerves, axon loss of the sural nerve decreased IENF density	(Geisler et al., 2016)
Swiss OF1	female	12 weeks	1 mg/kg twice weekly × 4 weeks	i.p.	8 mg/kg	5% mortality decreased sensory nerve amplitude and conduction velocity abnormalities on rotarod testing	(Bruna et al., 2011)
C57Bl/6	male & female	3–4 months	0.75 mg/kg twice weekly × 4 weeks	i.p.	6 mg/kg	mechanical hyperalgesia, no axon degeneration	(Chen et al., 2020)
C57Bl/6	male	8–10 weeks	0.5 mg/kg/day × 5 × 2	i.p.	5 mg/kg	mechanical hyperalgesia	(Starobova et al., 2019)
C57Bl/6	male	9 weeks	0.1 mg/kg/day × 14	i.p.	1.4 mg/kg	mechanical and thermal hyperalgesia	(Moschetti et al., 2019)
C57Bl/6	male, female	10–17 weeks	0.1 mg/kg × 5 × 2	i.p.	1 mg/kg	mechanical allodynia monocyte/macrophage infiltration in the sciatic nerve no ATF3 upregulation in DRGs no microglia activation in the spinal cord no loss of IENF	(Lee et al., 2020; Montague et al., 2018; Old et al., 2014; Shi et al., 2018)
C57Bl/6		8–10 weeks	10 μg daily × 4, 5 day rest, additional 2 injections	i.pl.	40 μg	mechanical and heat hyperalgesia inflammation of paw inflammation DRGs	(Starobova et al., 2020, 2019)
Rat
SD	male	250–300g	0.2 mg/kg daily × 5 × 2	i.v.	2 mg/kg	weight loss, mechanical and heat hyperalgesia; abnormalities on rotarod testing	(Aley et al., 1996)
SD	male	220–240g	0.2 mg/kg twice weekly × 4 weeks	i.p.	1.6 mg/kg	no mortality decreased motor nerve amplitude smaller axonal diameter	(Erdoğan et al., 2020)
SD Wistar	male	150–350g	0.1 mg/kg/day × 5× 2	i.p.	1 mg/kg	no weight loss, no mortality hot and cold hyperalgesia neuroinflammation in the sciatic nerve hyperresponsive C fibers in saphenous nerve, cytoskeletal abnormalities in unmyelinated and myelinated sensory axons, no axon loss; structural changes and activation of DRG neurons; increased synaptic activity and neuroinflammation in the dorsal horn of the spinal cord	(Aley et al., 1996; Kim et al., 2020; Singh et al., 2019; Tanner et al., 2003, 1998, 1998; Thibault et al., 2013; Topp et al., 2000; Weng et al., 2003)
Wistar	male	275–300g	0.1 mg/kg/day × 5× 2	i.p.	1 mg/kg	weight loss, decreased food intake mechanical hyperalgesia delayed gastric emptying, decreased intestinal motility, damage to digestive wall	(López-Gómez et al., 2018)
Rattus norvegicus albinus	male	250–450 g	0.2 mg/kg weekly × 5	i.p.	1 mg/kg	gait disturbance and limping decreased motor and sensory nerve amplitudes axon shrinkage and myelin sheath alterations in the sciatic nerve	(Ja’afer et al., 2006)
Wistar	female	175–200	0.2 mg/kg weekly × 4	i.v.	0.8 mg/kg	decreased sensory and motor nerve amplitudes, axon degeneration tail nerve, decreased IENF density, increased serum neurofilament light chain	(Meregalli et al., 2018a)
SD	n/a	160–180g	0.1 mg/kg, 0.15 mg/kg QOD × 5	i.v.	0.5 mg/kg, 0.75 mg/kg	decreased body weight mechanical and cold hyperalgesia, thermal hypoalgesia axon degeneration in the sciatic nerve	(Authier et al., 2003, 1999)
SD	male	200–300	0.05 mg/kg × daily × 10	i.p.	0.5 mg/kg	decreased IENF density	(Siau et al., 2006; Siau and Bennett, 2006)

Another treatment regimen consists of intraperitoneal injections of 0.75 mg/kg to 1.7 mg/kg VCR twice weekly for four weeks in mice or 0.2 mg/kg intravenously weekly for four weeks in rats (Table 2). Mice receiving 0.75 mg/kg twice weekly had about 10% weight loss, but no mortality and developed pronounced mechanical- and mild thermal hyperalgesia (Chen et al., 2020). They did not exhibit morphological signs of axon degeneration in the toe, sural, tibial and sciatic nerves and had no loss of IENF. Groups of mice receiving 1 mg/kg VCR twice weekly for four weeks had 5% mortality and developed decreased sensory nerve action potentials and conduction velocities of the tail, but no changes in the IENF density (Bruna et al., 2011). Such change was observed, instead, when 1.5 mg/kg VCR was given i.p. twice weekly for 4 weeks (Bruna et al., 2011; Geisler et al., 2016). This treatment regimen reflected many signs seen in patients, including axon loss in the distal, but not proximal sensory sural nerve, axon degeneration of distal toe nerves, decrease of IENF density in the skin, decreased compound nerve amplitude with preserved conduction velocity and mechanical and heat hyperalgesia. There was no axon loss in the motor predominant tibial nerve (Geisler et al., 2016). Therefore, this model mimics moderately severe sensory predominant peripheral neuropathy commonly seen in patients. The weight loss in this model over the four week period was 6% and the mortality rate 12– 18 % (Bruna et al., 2011; Geisler et al., 2016). Giving VCR to mice at a dose of 1.7 mg/kg twice weekly had a mortality rate of 74% in one study (Contreras et al., 1997) and 50% in another (Bruna et al., 2011), indicating this as a dose limit for VCR in mice. In rats, the administration of 0.2 mg/kg VCR i.v. weekly for four weeks results in both decreased sensory and motor nerve amplitude, axon degeneration in the caudal nerve, loss of intraepidermal nerve fibers, and a dose and time-dependent increase of serum neurofilament levels indicative of axon injury (Meregalli et al., 2018a).

VCR-induced neuroinflammation and pain

VCR-induced neuropathy characteristically causes numbness, tingling and weakness in the hands and feet, but is also frequently associated with neuropathic pain (Lieber et al., 2018; Liew et al., 2013; Pal, 1999). Similar to BTZ, a neuroinflammatory response is observed in rodent models following administration of VCR. However, while BTZ causes early neuroinflammation in DRGs, VCR leads to early macrophage infiltration in the sciatic nerve. Administering two cycles of VCR to mice at a dose of 0.2 mg/kg daily for 5 days with a 2 day rest period in between, produced mechanical hyperalgesia and increased macrophages in the sciatic nerves in both cycles, but no microglia activation in the spinal cord, no increase of the neuronal activation marker ATF3 in DRGs and no loss of cutaneous IENFs (Old et al., 2014). It was shown further that macrophages expressing the chemokine receptor Cx₃CR₁ transmigrate from the blood vessel into the sciatic nerve where they produce reactive oxygen species, which leads to activation of local axonal TRPA1 channels, which evoked pain. Deletion of the macrophage chemokine receptor CX₃CR₁ blocked hyperalgesia and macrophage migration during the first cycle of VCR administration, but had no effect on macrophage invasion or mechanical hyperalgesia during the second. Hyperalgesia during the 2^nd cycle was modulated, in addition, by the chemokine receptor 2 (CCR2; Montague et al., 2018). Inhibiting CCR2 in Cx3CR1 KO mice (but not in wildtype mice) blocks hyperalgesia in the 2^nd cycle, suggesting that in the absence of Cx3CR1, CCR2 mediates some of the VCR-induced hyperalgesia observed in the 2^nd cycle. Macrophages infiltrating the sciatic nerve following VCR administration express IL-6 (Kiguchi et al., 2008) and perineural injection of IL-6 antibodies or genetic deletion of IL-6 prevents VCR-induced allodynia (Kiguchi et al., 2008).

A local inflammatory response in the paw was observed when 10 μg VCR was administered subcutaneously into the hindpaw of mice daily for four days, followed by a five day break and further two injections (Starobova et al., 2019). Seven days after initiating this treatment, a local inflammation of the paw included full thickness epidermal necrosis with edema of underlying soft tissue and leukocyte infiltration into the dermis of the paw. Co-treatment with the anti-inflammatory tetracycline antibiotic minocycline reduced paw swelling, mechanical hypersensitivity and leukocytic perivascular infiltration (Starobova et al., 2019). In rat models, mincocycline also attenuated neurotoxicity induced by bortezomib and the chemotherapeutic oxaliplatin (Boyette-Davis and Dougherty, 2011; Robinson et al., 2014; Y.-Q. Zhou et al., 2018) suggesting that it could be an effective treatment of different forms of CIPN. However, in a phase II randomized clinical study of patients with locally advanced or metastatic colorectal cancer, 100 mg minocycline twice a day from start of chemotherapy over four months did not reduce numbness, tingling, fatigue or inflammatory biomarkers in patients receiving oxaliplatin (Wang et al., 2019), dampening enthusiasm regarding use of mincocycline to treat chemotherapy-induced neuropathy.

DRGs also exhibit an immune response following administration of VCR. After four daily VCR injections subcutaneously into the paw (Starobova et al., 2020), RNA sequencing of lumbar DRGs and subsequent cell type enrichment analysis of the 232 upregulated and 45 downregulated genes revealed a predominant effect on genes associated with the immune system (Starobova et al., 2020). Similar to BTZ, VCR also leads to activation of the prokineticin system in DRGs and spinal cord after mechanical hyperalgesia is established (Moschetti et al., 2019b). Administering VCR 0.1 mg/kg i.p. daily for 14 days, resulted in mechanical hyperalgesia starting on day 3. On day 7, high levels of prokineticin receptors 1 and 2, toll-like receptor 4, ATF3 and IL-1β and the macrophage marker CD68 were observed in DRGs without relevant alterations in the spinal cord. At day 14, an upregulation of prokineticin system and neuroinflammation were evident both in DRGs and the spinal cord. Similar to BIPN, treatment with the prokineticin antagonist PC1 reversed VCR-induced pain and neuroinflammation in mice (Moschetti et al., 2019b).

In contrast to BTZ, which showed prominent astrocyte-, but not microglia activation in the DHSC, both astrocytes and microglia are activated in the spinal cord and express high levels of TNFα following VCR (Qin et al., 2020; Shen et al., 2015). The increased TNFα is associated with the development of hyperalgesia, as VCR-induced hyperalgesia is prevented by intrathecal administration of antibodies to TNFα. In addition to prokineticin, the C-X-C motif chemokine ligand 1 (CXCL1) is elevated following VCR and associated with the expression of pain (L. Zhou et al., 2018). Weakly expressed in naïve animals, CXCL-1 is increased in astrocytes following VCR and acts on c-c chemokine receptor type 2 (CCR2) in spinal neurons to increase excitatory synaptic transmission (Zhang et al., 2017; L. Zhou et al., 2018). Indeed, increased neuronal activity in the deep layers III and IV of the dorsal horn as shown by increased staining of the immediate early gene c-fos accompanied by increased staining for Piccolo, a marker of active presynaptic elements, was observed (Thibault et al., 2013). Similarly, wide dynamic range neurons express higher rates of spontaneous activity at baseline and a greater A and C fiber response following VCR, which may lead to manifestation and sensitization of pain (Weng et al., 2003). Interestingly, levo-coryldamine (lCDL), an analgesic with sedative hypnotic properties that has been used in China for more than 40 years, decreases CXLC1 and TNFα expression in astrocytes and reverses the associated hyperalgesia in a rodent model and thus, might be useful to decrease pain in VIPN (L. Zhou et al., 2018; Zhou et al., 2020).

In addition to inducing neuroinflammation, VCR causes increased tetrodotoxin sensitive sodium current density in medium, but not small, DRG neurons, which leads to mechanical hyperalgesia in a mouse model (Chen et al., 2020). Mice lacking Na_v1.6 in nociceptors expressed partially attenuated mechanical allodynia in the absence of axon degeneration, indicating that VCR-induced axon degeneration is not a requirement for VCR-induced pain (Chen et al., 2020). How VCR leads to changes in Na_v1.6 expression in medium sized DRG neurons requires further investigation but it is possible that inflammatory cytokines upregulate Na_v1.6 channel expression and that Na_v1.6 in turn promotes inflammation and axon degeneration (Alrashdi et al., 2019; Xie et al., 2013). Considering the emerging role of Na_v1.6 as an important contributor to chronic pain (Chen et al., 2018; Xie et al., 2015, 2013) and VCR-induced neuropathy (Chen et al., 2020), blocking Na_v1.6 with a selective sodium channel blocker may be another valuable treatment approach to decrease VCR-induced pain.

Mechanisms that lead to VCR and BTZ-induced axon degeneration

Axon degeneration is a prominent feature of both VCR and BTZ-induced neuropathies and is associated with numbness, weakness and pain. While some medications decrease pain and tingling, there are currently no treatments that reduce numbness or weakness (Dorsey et al., 2019; Majithia et al., 2016). However, recent discoveries described below inspire hope that this is about to change.

Much progress has been made in recent years to elucidate mechanisms of axon degeneration (Coleman and Höke, 2020; Figley and DiAntonio, 2020; Gerdts et al., 2016; Simon and Watkins, 2018). Our deep mechanistic understanding stems largely from axon cut injury studies, but regulatory genes identified in these studies also influence axon survival in other circumstances, including VCR- and BTZ-induced neuropathies. The field emerged from the serendipitous discovery of Wallerian degeneration slow (WLDs) mice in which axonal degeneration is greatly delayed after nerve transection (Lunn et al., 1989), which indicated that axons distal to a nerve cut do not die passively because of lack of nutrients, but, akin to apoptosis, engage a genetically encoded axon degeneration program. Already 20 years ago, scientists noticed the similarities of axon degeneration and VIPN and hypothesized that VCR-induced neurotoxicity is also due to the activation of a local axon degeneration program (Silva et al., 2006; Wang et al., 2001, 2000). By treating plated DRGs with different concentrations of VCR, they showed that axon degeneration is delayed in DRG neurons obtained from WLDs mice (Wang et al., 2001, 2001), thus demonstrating for the first time that VCR-induced axonal degeneration can, indeed, be blocked. The spontaneous WLDs mutation results in a chimeric protein (Coleman et al., 1998; Mack et al., 2001) of which the NAD-synthesizing enzyme nicotinamide mononucleotide adenylyltransferase 1 (NMNAT1) activity is responsible for the axon-sparing activity of the Wlds phenotype (Araki et al., 2004; Sasaki et al., 2009). Accordingly, overexpressing cytNMNAT1, a stable form of NMNAT, also leads to long-lasting protection from VCR-induced neurite degeneration in vitro (Press and Milbrandt, 2008; Sasaki et al., 2009; Vohra et al., 2010). However, the Wlds gene product results in the expression of a gain-of-function protein that does not normally exist, limiting the therapeutic utility of this discovery. Until recently, no loss-of-function mutants were identified that had similar axon protective properties. However, it was discovered recently that sterile alpha and TIR motif-containing protein 1 (SARM1) is an essential component of this endogenous axonal self-destruction pathway (Gerdts et al., 2013; Osterloh et al., 2012). Whereas axons of wildtype mice degenerate within 3 days after sciatic nerve cut, severed axons of SARM1 knockout mice remain intact for more than 2 weeks (Gerdts et al., 2013; Osterloh et al., 2012).

SARM1 is a multidomain Toll-like receptor adapter whose Toll/interleukin-1 receptor (TIR) domain exerts pro-neurodegenerative action through its intrinsic NADase activity (Essuman et al., 2017; Gerdts et al., 2015). Once activated, SARM1’s NADase cleaves NAD⁺ into nicotinamide (NaM) and adenosine diphosphate-ribose (ADPR), which results in metabolic collapse, influx of calcium, activation of calcium-dependent proteases and axon fragmentation. Recent discoveries revealed that NAD⁺ binds directly to the ARM domain of SARM1 (Jiang et al., 2020). While high concentrations of NAD⁺ inhibit SARM1, low NAD⁺ concentrations enhance SARM1’s NADase activity (Jiang et al., 2020) thereby further decreasing NAD⁺ levels, which may establish a feed-forward loop along the entire length of the axon. Cytosolic NAD+ is produced by the short-lived axonal survival factor nicotinamide mononucleotide adenylyltransferase 2 (NMNAT2), which is continually transported from the cell body into the axon (Gilley et al., 2015; Gilley and Coleman, 2010). Reduced NMNAT2 levels, e.g., after nerve cut, lead to decrease of NAD⁺ and activation of SARM1. Thus, Wlds is likely axon protective because the long-lived NMNAT1 is “mis”localized to the axon where it substitutes for the labile NMNAT2, maintains NAD⁺ levels and inhibits SARM1.

Axon loss in most neuropathies, including those induced by VCR and BTZ, differ from that caused by axotomy in that i) fewer axons are degenerating at any given point in time, ii) the degeneration occurs in short segments from distal to proximal and iii) the axon remains connected to the cell body. Despite these differences, genetic deletion of SARM1 protects from loss of both myelinated and unmyelinated axons in a mouse model of VIPN (Geisler et al., 2016). Genetic deletion of SARM1 also prevents VCR-induced mechanical and heat hyperalgesia, indicating that blocking SARM1 improves functional outcomes, including pain (Geisler et al., 2016). VCR is somewhat similar to nerve cut in that axonal transport is impaired, but not blocked. This likely leads to a more profound loss of NMNAT2 distally, activating the SARM1 pathway locally. In this model, residual axonal transport is sufficient to maintain NMNAT2 adequate to block SARM1 activation in proximal axons thereby degeneration to the distal axon.

Excitingly, although BTZ has a different mechanism of action, BTZ-induced axon degeneration is also blocked in SARM1 KO mice (Geisler et al., 2019a), indicating that SARM1 is part of a fundamental axon degeneration program. Both VCR and BTZ lead to a decrease of axonal NMNAT2 in cultured DRGs (Geisler et al., 2019b), which activates SARM1 and causes a drop of NAD⁺ followed by axon fragmentation (Essuman et al., 2017; Geisler et al., 2019a; H. Liu et al., 2018). The upstream pathways engaged by VCR and BTZ differ, involving MAP-kinases after VCR and transcription and apoptotic enzymes after BTZ (Geisler et al., 2019a; Summers et al., 2018; Yang et al., 2015). Both pathways then converge on a core axon degeneration program consisting of NMNAT2 - SARM1- and NAD⁺ (Fig. 2). The proposition that NMNAT2 decreases following VCR due to decreased transport is supported by the concomitant increase of NMNAT 2 in the cell body (unpublished observation), but whether a similar mechanism leads to a decline of NMNAT2 following BTZ and whether other mechanisms are involved in the activation of SARM1 by BTZ require further investigation. Although axonal transport is decreased after BTZ (Meregalli et al., 2014; Poruchynsky et al., 2008; Staff et al., 2013), the loss of NMNAT2 in the axon after BTZ administration is not accompanied by an increase in the cell body (unpublished observation) and the kinetics of the axonal NMNAT2 decrease are different than following VCR (Geisler et al., 2019a), suggesting that mechanisms other than or in addition to axonal transport may contribute. It will also be important to elucidate whether the profound protection from VCR-induced allodynia by SARM1 knock-out (Geisler et al., 2016) is due to decreased axon degeneration or if SARM1 acts on additional pathways that modulate pain behavior. For instance, SARM1 has been shown to regulate the neuroinflammatory response in DRG neurons after nerve cut (Wang et al., 2018) and TNFα can trigger SARM1-dependent axon degeneration in sensory neurons via an interesting noncanonical necroptotic signaling mechanism (Ko et al., 2020). Although additional exciting research lies ahead, the identification of a core axon degeneration program already provides several avenues for the development of new therapeutics (Fig. 2; Coleman and Höke, 2020; DiAntonio, 2019; Loring and Thompson, 2020).

Figure 2: Model of a common final axon degeneration pathway activated by vincristine and bortezomib showing prospective therapeutic strategies (left, green) and candidate therapeutics (right).

Vincristine and bortezomib engage distinct upstream pathways that lead to loss of axonal NMNAT2. NMNAT2 is an axonal survival factor with short half-life that must be constantly transported from the cell body. SARM1 is inhibited by NMNAT2 and loss of NMNAT2 leads to activation of SARM1, which triggers depletion of the metabolic co-factor NAD+ leading to local metabolic collapse and axon fragmentation. SARM1 activity can be blocked by using pharmacological inhibitors of the SARM1 NADase, dominant/negative mutants of SARM1 or antisenseoligonucleotides/siRNAs. NAD⁺ levels can be maintained by supplementation with the NAD⁺ precursor NR, which leads to an increase of both NMN and NAD⁺ or a combination therapy of NaR and the NAMPT blocker FK866, which decreases NMN and increases NAD+. NMNAT2 levels can be maintained by compounds that modulate NMNAT2 stability. Abbreviations: cADPR – cyclic ADP-ribose; NaAD – nicotinic acid adenine dinucleotide; NaR – nicotinic acid riboside; NaM - nicotinamide; NaMN – nicotinic acid mononucleotide, NAMPT – nicotinamide phosphoribosyltransferase; NMN – nicotinamide mononucleotide; NMNAT – nicotinamide nucleotide adenylyltransferase; NR – nicotinamide riboside; NRK – NR kinase

Insofar as SARM1 is activated by low NAD⁺ concentration (Jiang et al., 2020) and acts by decreasing NAD⁺ (Essuman et al., 2017; Gerdts et al., 2015; Sasaki et al., 2016a), countering the NAD⁺ loss may be one possibility to ameliorate VIPN and BIPN. Indeed, maintaining NAD⁺ by overexpressing NAD⁺ synthesizing enzymes NMNAT1, nicotinamide phosphoribosyltransferase (NAMPT) and nicotiamide riboside kinase (NRK) 1 or supplementing with the NAD⁺ precursor nicotinamide riboside (NR) has been shown to potently protect axons from degeneration following VCR and BTZ in vitro (Geisler et al., 2019a; Gerdts et al., 2015; Sasaki et al., 2009, 2016b). NR is a form of vitamin B3 that readily enters the cell and has no known adverse effects. Accordingly, NR is currently in clinical trials to treat paclitaxel-induced neuropathy, another SARM1-dependent CIPN (Turkiew et al., 2017). Thus, NR or compounds that increase NMNAT2 levels (Ali et al., 2017), may be a promising therapeutic avenue to attenuate VIPN and BIPN (Fig. 2). Recently, it was revealed that SARM1 is allosterically activated upon increase of the direct NAD⁺ precursor nicotinamide mononucleotide (NMN; Bratkowski et al., 2020; Z. Y. Zhao et al., 2019), which led to the hypothesis that the ratio between NAD⁺ and NMN may be an important determinant of axon degeneration, as both low NAD⁺ and high NMN can induce axon degeneration (Coleman and Höke, 2020; Di Stefano et al., 2017, 2015). Therefore, Liu et al. developed a combination therapy that increases NAD⁺ and at the same time decreases NMN, and showed that it potently blocked VCR-induced axon degeneration in vitro (H. Liu et al., 2018), which represents another promising strategy for translation to the clinic.

After the enzymatic pocket of the NADase of the SARM1 molecule was identified (Bratkowski et al., 2020; Essuman et al., 2017; Sporny et al., 2019), the first small molecule noncompetitive inhibitors, berberine chloride and zinc chloride, which block the SARM1 NAD⁺ hydrolase were developed (Loring et al., 2020). These inhibitors are moderately potent, but represent a first step toward the development of more efficacious inhibitors, which is currently pursued by at least two pharmaceutical companies. Such NADase inhibitors, however, may also target other NAD-utilizing enzymes in neurons and non-neuronal cells resulting in undesired side effects (Coleman and Höke, 2020). Another attractive therapeutic approach is to utilize a dominant-negative SARM1 mutant that avoids potential side effects of inhibiting other NADases. SARM1 mutants have been developed that bind wildtype SARM1 as dominant/negatives and potently inhibit axon degeneration (Geisler et al., 2019b). Using adeno-associated virus (AAV) and a neuron-specific promoter, SARM1 mutants can be expressed selectively in peripheral somatosensory neuron populations, thereby avoiding possible side effects and interference with systemic chemotherapy or other drug treatments. Alternatively, SARM1 could be targeted with siRNA or antisense oligonucleotides. Such an approach has been shown to be highly efficacious in treating and reversing neuropathy due to transthyretin mutations (Adams et al., 2018; Benson et al., 2018). Thus, the identification of a core axon degeneration program that underlies different forms of axon degeneration has opened the gate to the development of therapeutics to prevent VCR and BTZ-induced neuropathies (Fig. 2).

Concluding remarks

New and improved chemotherapy regimens led to a dramatic and fortunate increase in the number of cancer survivors, but also to an urgent need to ameliorate the long term sequelae of treatments. Chemotherapy-induced neuropathy, in particular, causes long-lasting impairments and reduced quality of life. A plethora of in vitro and in vivo models of VCR and BTZ-induced neuropathies developed in recent years expanded and sharpened our understanding of molecular mechanisms underlying different aspects of neuropathies and also serve as a testing ground for new treatments. Some of the preclinical discoveries have recently been moved into clinical trials while others are being developed further in close collaborations between academia and the pharmaceutical industry. Numerous recent exciting discoveries in the fields of neuroinflammation, pain sensitization and axon degeneration encourage optimism that treatments to prevent, attenuate or reverse VCR- and BTZ-induced neuropathies are at our door step. Such treatments may improve the quality of life of patients both during and after cancer therapy and have enormous societal impact.

Highlights:

• Peripheral neuropathy is a common side effect of the chemotherapeutic drugs vincristine and bortezomib

• Bortezomib activates transcriptional programs in dorsal root ganglion neurons leading to neuroinflammation and pain

• Vincristine inhibits transport of axonal survival factors

• Both bortezomib and vincristine activate a common axon degeneration program that can be targeted therapeutically

Abbreviations:

BIPN

bortezomib-induced peripheral neuropathy

BTZ

bortezomib

CIPN

chemotherapy-induced peripheral neuropathy

DHSC

dorsal horn of the spinal cord

DRG

dorsal root ganglia

GFAP

glial fibrillary acidic protein

HCN2

human cortical cell line 2

IENF

intraepidermal nerve fibers

MAPK 38

mitogen activated protein kinase 38

NAD⁺

nicotinamide dinucleotide

NFkB

nuclear factor kappa B

NGF

nerve growth factor

IENF

intraepidermal nerve fibers

IENFD

intraepidermal nerve fiber density

i.p.

intraperitoneal

i.pl.

intraplantar

i.v.

intravenous

JNK

cJun NH2 terminal kinase

QOD

every other day

SARM1

Sterile Alpha and TIR motif containing 1

SCID

severe combined immune deficiency spontaneous mutation

S1P

sphingosine 1 phosphate

S1PR

S1P recepto

TGFβ

transforming growth factor beta

TNFα

tumor necrosis factor alpha

TRPA

transient receptor potential ankyrin 1 ion channel

TRPV

transient receptor potential vanilloid receptor channel

VCR

vincristine

VIPN

vincristine-induced neuropathy