The predominant morphine metabolite, M3G, is devoid of analgesic effects whether it is injected through the s.c. or i.c.v. routes or directly into the PAG (Shimomura et al., 1971; Pasternak et al., 1987). However, the first M3G administrations in rats have elicited neuroexcitatory effects that could oppose morphine and M6G antinociception (Table 1). For instance, i.t. or i.c.v. administration of M3G induces robust behavioral excitation in rodents characterized by spontaneous agitation, hyperalgesia and allodynia, epileptic episodes and death following high doses of M3G (Labella et al., 1979; Woolf, 1981; Yaksh et al., 1986; Yaksh and Harty, 1988; Bartlett et al., 1994a; Bian and Bhargava, 1996). Following these observations, M3G’s pronociceptive effects were evaluated after direct injection or when coadministered with morphine or M6G. In rodents, i.p., s.c. and i.t. injections of M3G alone clearly induce thermal hyperalgesia and mechanical allodynia (Juni et al., 2006; Komatsu et al., 2009, 2016; Lewis et al., 2010; Due et al., 2012; Arout et al., 2014; Allette et al., 2017; Roeckel et al., 2017; Blomqvist et al., 2020). Additionally, morphine and M6G analgesic effects are markedly reduced by M3G (Smith et al., 1990; Qian-Ling et al., 1992; Ekblom et al., 1993; Faura et al., 1996, 1997; Gardmark et al., 1998). Hence, whether it is injected alone or with morphine or M6G, several studies have demonstrated that M3G has pronociceptive properties in rodents (Table 1).

Interestingly, Smith and Smith (1995) observed that, when morphine is infused continuously in rats, the higher the plasmatic metabolic ratio M3G/morphine is, the lower the antinociception is, independently of the M3G or morphine plasmatic concentrations. Similar observations were made in the extracellular cortical fluid following s.c. administration of morphine (Barjavel et al., 1995). Consequently, M3G was proposed to counteract morphine-induced analgesia and to produce neuroexcitatory effects responsible for some morphine side effects (Gong et al., 1991; Smith and Smith, 1995; Faura et al., 1997; Roeckel et al., 2017; Blomqvist et al., 2020).

Although a considerable number of studies have indicated that M3G possesses pronociceptive properties, some studies did not observe pronociceptive effects following M3G administration or when it was coadministered with morphine or M6G (Table 2). For instance, Ouellet and Pollack (1997) observed that, when M3G was infused for 12 h in rats, there was no hyperalgesia or modulation of morphine analgesia (Suzuki et al., 1993). In another study, it was even noted that the i.v. coadministration of morphine and M3G improved morphine analgesia (Lipkowski et al., 1994).

Interestingly, in a MRP3^–/– mouse model, the antinociception and hyperalgesia induced by an injection of morphine remained intact (Swartjes et al., 2012). In these mice, although morphine is still metabolized into M3G, M3G has been shown to remain trapped in hepatocytes due to the lack of the MRP3 efflux transporter. Therefore, plasma levels of M3G were extremely low in these transgenic animals compared to control animals (Zelcer et al., 2005; Swartjes et al., 2012). These data indicate that hyperalgesia may occur without significant contribution of hepatic M3G. However, it is worth noting that, although M3G is not found in the blood of these animals, morphine might be directly metabolized into M3G within the CNS and could still elicit its central effects (Gabel et al., 2022).

In humans, there have been few reports of the pronociceptive effects of M3G (Table 1). Smith and collaborators observed in 14 cancer patients improved pain relief, which was corroborated by a decrease in the M3G/(morphine + M6G) ratios. These results indirectly suggest a pronociceptive role of M3G by reducing morphine analgesia (Smith et al., 1999). In a pharmacokinetic-pharmacodynamic study involving 50 patients with moderate to severe pain, M3G effects seemed to oppose morphine analgesia (Mazoit et al., 2007). Several case reports have also suggested that M3G might play a role in morphine’s side effects such as morphine-induced hyperalgesia and seizures following high dose of morphine. However, these observations have shown considerable heterogeneity and do not demonstrate a pronociceptive role of M3G in humans (Morley et al., 1992; Sjogren et al., 1993, 1998; Rozan et al., 1995; Hagen and Swanson, 1997; Kronenberg et al., 1998).

Contrastingly, several reports have not observed any correlation between analgesia and plasmatic concentrations of M3G or the metabolic M3G/(morphine + M6G) ratios (Table 2; Samuelsson et al., 1993; Goucke et al., 1994; Wolff et al., 1995; Andersen et al., 2002; Toce et al., 2019). In addition, there have been two studies, published by the same group, in which healthy volunteers were administered M3G to evaluate its effects in humans (Penson et al., 2000, 2001). The first study was a randomized, double-blind, placebo-controlled trial in which M3G was infused in 10 healthy volunteers. Analgesia was assessed with numerical and visual analog scales in a submaximal ischemic pain model. No M3G-induced hyperalgesia or dysphoria was observed. In addition, the coadministration of M3G along with morphine or M6G did not affect analgesia (Penson et al., 2000). In the second study, which was blinded, but not controlled, three concentrations of M3G were used, but no effect was observed (Penson et al., 2001). These two studies are extremely valuable, although the number of subjects used was relatively small for obvious reasons.

Considered together, the few studies in humans are matter of debate, whereas, in rodents, reports have shown much more consistency toward the pronociceptive effects of M3G, even though these effects are not always observed. The origin of the behavioral effect of M3G might rely on its glucuronide moiety. Indeed, M3G is not the only “3-glucuronide” metabolite displaying pronociceptive effects. Several studies published by Lewis et al. (2013, 2015) showed that estradiol-3-glucuronide, as well as ethyl-glucuronide, produces hyperalgesia after i.t. administration. Interestingly, glucuronic acid injected alone also triggered a similar effect, demonstrating the importance of the glucuronide moiety in the pronociceptive effects of these molecules (Lewis et al., 2013).

Supporting this idea, other 3-glucuronide metabolites of morphine-derived compounds, such as normorphine, noroxymorphone and hydromorphone, display pronociceptive properties (Yaksh and Harty, 1988; Smith et al., 1997; Wright et al., 2001). Consistently, Peckham and Traynor (2006) showed robust sex differences in analgesia only with morphine derivatives that are conjugated into a 3-glucuronide metabolite. Importantly, these observations were not related to binding affinity, ability to activate the MOR or lipophilicity. We also recently observed that sex differences in morphine analgesia could have their origin in morphine metabolism. Indeed, morphine metabolism is higher in the female brain, resulting in higher levels of M3G in pain-related brain regions (Gabel et al., 2022).

The molecular mechanisms underlying the effects of M3G remain a matter of debate (Table 3). On the one hand, one study published observations in MOR^–/– mice suggesting its requirement for M3G pronociceptive effects (Roeckel et al., 2017). In this valuable study, i.p. administration of M3G induces thermal hyperalgesia and tactile allodynia in WT but not MOR^–/– animals. In addition, M3G binds MOR on brain membranes from WT mice, although with low affinity (∼1.4 μM), and induces a weak Gi-dependent activity but no β-arrestin2 recruitment (Figure 2). This activity is not observed neither in brain membranes from MOR^–/– mice, nor in the presence of naloxone (Roeckel et al., 2017).

On the other hand, M3G showed only low (>μM) affinity for MOR in several binding studies employing radio-labeled molecules, such as [D-Ala2, N-MePhe4, Gly-ol]-enkephalin (DAMGO) or naloxone (Labella et al., 1979; Christensen and Jorgensen, 1987; Pasternak et al., 1987; Coimbra-Farges et al., 1990; Chen et al., 1991; Roeckel et al., 2017). It was even proposed that the apparent affinity of M3G for MORs results from residual morphine contamination in M3G stock solutions (Bartlett and Smith, 1995). In addition, several in vivo studies have demonstrated that M3G’s pronociceptive effects persist in the presence of naloxone or naltrexone, two non-selective antagonists of MORs, whether they are injected systematically or directly into the CNS (Labella et al., 1979; Woolf, 1981; Yaksh et al., 1986; Yaksh and Harty, 1988; Halliday et al., 1999). Altogether, these pieces of evidence indicate that M3G might not bind to MOR, although one study suggested that this receptor appears to be mandatory for M3G effects.

One interesting hypothesis suggests the existence of an alternative non-opioid receptor that could mediate M3G effects (Table 3). More precisely, in silico studies have indicated that M3G is able to bind the Toll-like receptor 4 (TLR4) and myeloid differentiation factor 2 (MD-2) complex through an interaction with the lipopolysaccharide (LPS) binding pocket of MD-2 (Hutchinson et al., 2010; Lewis et al., 2010; Grace et al., 2014) (for a review of opioid interactions with TLR4, see Gabr et al., 2021). TLR4 downstream signaling involves the activation of 3 parallel intracellular pathways: the NF-κB, the MAPK and the PI3K/AKT pathway. In agreement with the initial reports, it has been shown in vitro that the reporter cell line HEK-Blue™ hTLR4 exhibits significant activation upon M3G stimulation, which is inhibited by LPS from Rhodobacter sphaeroides (LPS-RS), a selective TLR4 antagonist (Lewis et al., 2010; Xie et al., 2017). This reporter cell line expresses the human TLR4 and a reporter gene under the control of a promoter inducible by NF-κB and AP-1, two transcription factors involved in TLR4 signaling cascade and proinflammatory cytokines release. In addition, it has been shown that the PI3K/AKT pathway, the third TLR4 intra-cellular signaling pathway, is also activated following M3G stimulation (Figure 2; Hutchinson et al., 2010; Wang et al., 2021). In human cancer cell lines, the activation of the AKT pathway by M3G results in upregulation of programmed death ligand 1 (PD-L1), which promotes tumor growth (Wang et al., 2021). It is, however, worth noting that, although Wang et al. (2021) observed activation of the AKT and NF-κB pathways in the A549 cell line (a human lung cancer cell line), they did not observe activation of the MAPK pathway in their model.

In vivo, M3G-induced hyperalgesia following i.p. administration in rodents is abolished by administration of TLR4 antagonists, as well as in a TLR4^–/– mouse model (Figure 2; Due et al., 2012; Allette et al., 2017). Consistently, M3G seems to display proinflammatory properties through upregulation of NF-κB and proinflammatory cytokines, including interleukin 1β (IL-1β), interleukin 6 (IL-6), and tumor necrosis factor α (TNFα), such that it was proposed to be involved in the modulation of morphine properties (Figure 2; Lewis et al., 2010; Grace et al., 2014; Doyle and Murphy, 2018; Iqbal et al., 2020; Wang et al., 2021). These interesting findings take into account that a considerable number of studies have described the immunomodulatory effects of morphine and M3G (Wybran et al., 1979; Shavit et al., 1986; Freier and Fuchs, 1994; Thomas et al., 1995; Wang et al., 2012; Eisenstein, 2019). Considered together, these data suggest that TLR4 could be responsible for the inflammation triggered by M3G, which would thwart morphine’s analgesic effects.

Several studies have implicated TLR4 in dampening morphine antinociceptive effects or in some side effects, such as antinociceptive tolerance (Hutchinson et al., 2010; Liu et al., 2011; Wang et al., 2012, 2020, 2021; Eidson and Murphy, 2013; Bai et al., 2014; Grace et al., 2014; Eidson et al., 2017; Thomas et al., 2022). For instance, a recent study revealed that antinociceptive tolerance was prevented in TLR2 and TLR4 null mutants, but not in MyD88^–/– animals (Thomas et al., 2022). Since several studies suggested that TLR4 could be the receptor mediating M3G effects, M3G has been proposed to play a role in morphine side effects, especially in antinociceptive tolerance (Juni et al., 2006; Blomqvist et al., 2020). However, one should note that two studies invalidate the implication of TLR4 in morphine’s effects (Fukagawa et al., 2013; Mattioli et al., 2014). The TLR4 mutant mouse strain C3H/HeJ, which expresses a non-functional TLR4, a TLR4^–/– mouse strain on a C57BL/6 background and the B10ScNJ mouse strain, which has a spontaneous mutation that completely removes the TLR4 coding sequence, were used. In the first study, after repeated injection of morphine, CD11b (a marker of microglial activation) mRNA expression was increased in the spinal cord of control mice. Minocycline, a microglial inhibitor, attenuated the development of morphine tolerance in these mice. Conversely, in the C3H/HeJ mutant mouse strain and in a TLR4^–/– mouse strain, neither the increase of CD11b mRNA expression, nor the antinociceptive tolerance was affected by TLR4 invalidation (Fukagawa et al., 2013). In the second study, neither acute antinociceptive response to a single dose of morphine, nor the development of antinociceptive tolerance was affected by TLR4 invalidation in the C3H/HeJ and B10ScNJ mouse strains (Mattioli et al., 2014). These results suggest that, in these models, TLR4 is not involved in the modulation of the antinociceptive effect of morphine, in its side effects or in the microglial activation observed during morphine tolerance. This evidence is interesting and provides insight into the complexity of M3G physiology.

On the one hand, M3G-induced hyperalgesia is abolished in a MOR^–/– mouse model (Roeckel et al., 2017). On the other hand, the same effect was observed in a TLR4^–/– mouse model (Due et al., 2012). This piece of evidence raises the possibility that the hyperalgesia observed following M3G administration might depend on the cross-talk between MORs and TLR4s within the CNS (Figure 2), for which both receptors are mandatory (for review, see Zhang et al., 2020). To support this idea, both receptors are expressed in microglia, astrocytes and even neurons under pathological conditions (Aicher et al., 2000; Lehnardt et al., 2003; Calvo-Rodriguez et al., 2017; Maduna et al., 2018; Zhang et al., 2018; Nam et al., 2019). Secondly, the mitogen-activated protein kinase (MAPK) pathway is recruited following both MOR and TLR4 stimulation. This pathway seems to be involved in morphine-induced hyperalgesia, as well as in the inflammatory response following TLR4 activation (Zhang et al., 2020). Finally, different studies have reported that M3G effects were abolished in presence of MAPK pathway inhibitors (Figure 2; Komatsu et al., 2009; Wang et al., 2021). Taken together, the MAPK pathway represents an interesting target to assess to better understand M3G effects.

Several studies have also suggested that, although M3G alone does not induce hyperalgesia, its coadministration with morphine decreases analgesia (Ekblom et al., 1993; Faura et al., 1996, 1997). In these studies, relatively low concentrations of M3G were injected through the i.p. route, while most of the studies in which direct hyperalgesia was observed injected high concentrations of M3G directly into the CNS (Table 1). Hence, it could be possible that, following CNS administration, M3G reaches sufficient CNS concentrations to activate both MOR and TLR4 on its own and produce hyperalgesia, although it has a low apparent affinity for MOR. In contrast, after peripheral injection of low dose of M3G alone, M3G would not reach sufficient CNS concentrations for MOR activation even though TLR4 might be activated. The presence of morphine along with M3G would then allow MOR and TLR4 activation and thus hyperalgesia. Nonetheless, this hypothesis remains to be investigated. Interestingly, in humans, M3G plasmatic and cerebrospinal fluid (CSF) concentrations following morphine administration show significant variation according to administration types, doses and patients (Hand et al., 1987b; Osborne et al., 1990; Hasselstrom and Sawe, 1993; Goucke et al., 1994; Westerling et al., 1995; Wolff et al., 1995, 1996; Hoffman et al., 1997; Christrup et al., 1999; Smith et al., 1999; Sarton et al., 2000; Meineke et al., 2002). For instance, after i.v. injection of 5 mg of morphine in healthy volunteers, M3G maximal plasmatic concentration reaches approximatively 100 nM, whereas it reaches 2 μM after a 30 min infusion of 0.5 mg/kg of morphine in neurosurgical patients (Hasselstrom and Sawe, 1993; Meineke et al., 2002). In the CSF, M3G concentrations range approximatively from 4 nM in patients that were given 30 mg of morphine orally to 0.7 μM in patients receiving chronic oral morphine therapy (Hand et al., 1987b; Goucke et al., 1994; Wolff et al., 1995, 1996; Smith et al., 1999). Depending on dose and treatment duration, M3G might reach the required CNS concentrations to induce MOR and TLR4 activation.

It is also worth noting that, although numerous studies have proposed pieces of evidence that TLR4 is involved in M3G’s effects, there is few data regarding the direct binding of M3G to TLR4. In a biophysical binding assay, M3G has been shown to bind the accessory protein MD-2 with a relatively low dissociation constant of approximatively 1.5 μM (Grace et al., 2014). However, there is no study in which radiolabeled molecules were used to investigate whether M3G can bind TLR4 or not. Therefore, one should consider an additional assumption that suggests the existence of an alternative receptor that could trigger a TLR4-dependent signaling pathway (Figure 2). In addition, to our knowledge, TLR4/MOR heteromers have not yet been described, although such association might participate in the complex response to M3G.