Macrophage Migration Inhibitory Factor (MIF) Makes Complex Contributions to Pain-Related Hyperactivity of Nociceptors after Spinal Cord Injury

Physiologically relevant concentrations of MIF induce hyperactivity in probable nociceptors with a narrow dose dependence

We first asked whether primary sensory neurons isolated from Naive rats and incubated with physiologically relevant concentrations of recombinant human MIF for 15-60 min would produce hyperactivity resembling that observed after SCI (Bedi et al., 2010). Sampling was restricted to neurons with soma diameters of 20-30 µm, ∼70% of which we showed previously to exhibit properties of nociceptors under our culture conditions, specifically, sensitivity to capsaicin and/or binding of isolectin B4 (Bedi et al., 2010; Odem et al., 2018). Probable nociceptors in vitro were subclassified according to their AP accommodation properties when stimulated by a 2 s pulse of depolarizing current at twice rheobase as either RA or NA types (for examples of NA and RA neurons, see also Fig. 3A). Only NA neurons have been found to exhibit SA (defined in vitro as ongoing discharge at RMP in the absence of a source of extrinsic excitation applied by the experimenter) and OA (defined as any ongoing discharge; i.e., SA is a subset of OA) (Odem et al., 2018). Except where otherwise stated, all the data presented in this paper were obtained from NA neurons.

NA neurons were given 10-15 min applications of MIF at doses of 0.1-10 ng/ml (Fig. 1; Table 2) before recording, and MIF remained in the chamber for the duration of the recordings from each coverslip (up to 60 min). These concentrations overlap the range reported in human plasma acutely and chronically after SCI (medians ∼1 ng/ml in each case, total range 0.1-8.6 ng/ml) (Stein et al., 2013; Bank et al., 2015). The 1 ng/ml dose increased the incidence of sampled neurons exhibiting OA at RMP from 0% (vehicle treatment) to 72% (Fig. 1A,B), while during artificial depolarization to −45 mV (Fig. 1B), the 0.5 and 1 ng/ml doses increased the incidence of OA from 8% (vehicle treatment) to 67% and 89%, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Brief treatment (15-60 min) of physiologically relevant concentrations of MIF increases the excitability of NA nociceptors isolated from Naive rats. A, Representative current-clamp recordings (I = 0, 10 s) from neurons during the indicated treatments. Inset, Expanded view of a 4 s segment to highlight DSFs (red arrowheads) during 1 ng/ml MIF treatment. The illustrated APs have been clipped at −20 mV to permit an enlarged view of the underlying DSFs. The actual peaks of the APs were ∼50 mV. B, Effects of different doses of MIF on the incidence of neurons exhibiting OA at RMP or when artificially depolarized to −45 mV. Comparisons of OA at each concentration versus vehicle treatment (0 ng/ml) were performed with Fisher's exact test (*p < 0.012 considered significant after Bonferroni correction for 4 comparisons; **p < 0.0025; ***p < 0.00025). Numbers over each bar represent the number of neurons exhibiting OA/the total number sampled. C–E, Effects of different doses of MIF on excitability properties of NA neurons from Naive rats. C, RMP. D, AP voltage threshold. E, Rheobase. Each circle represents 1 neuron. Bars represent medians. Comparisons between each dose and vehicle were performed by Kruskal–Wallis tests followed by Dunn's multiple comparison tests, with corresponding p values indicated on the figure. F, Lack of effects of different intracellular concentrations of EGTA (0.1 and 3 mm) on the incidence of neurons exhibiting OA at RMP or when artificially depolarized to −45 mV in the absence and presence of MIF (1 ng/ml). Comparisons of OA between basal conditions or with 0.1 mm EGTA with and without MIF were performed with Fisher's exact tests (*p < 0.05; **p < 0.01). Numbers over each bar represent the number of neurons exhibiting OA/total number sampled.

In terms of membrane potential, there are only three possible electrophysiological changes that can drive OA in the absence of transient depolarizing inputs, such as sensory generator potentials or synaptic potentials: depolarization of RMP, hyperpolarization of AP voltage threshold, and enhancement of DSFs (Fig. 1A, red arrowheads) that bridge the gap between RMP and AP threshold (Odem et al., 2018; North et al., 2022). The effects of MIF on RMP and on AP voltage threshold and current threshold (rheobase) are shown in Figure 1 and Table 2, and the effects on DSFs are analyzed in a subsequent section. MIF significantly depolarized RMP to median values of −49.3 mV at 1 ng/ml and −57 mV at 10 ng/ml versus −65.3 mV during vehicle treatment (Fig. 1C). MIF hyperpolarized the AP voltage threshold from −31.6 mV during vehicle treatment to −34.9 mV (0.5 ng/ml) and −36.7 mV (1 ng/ml) (Fig. 1D). MIF (1 ng/ml) also reduced rheobase (the minimum current needed to reach AP threshold) from 115 pA during vehicle treatment to 50 pA (Fig. 1E). No significant effects of MIF treatment were found on RA neurons (Table 3).

An unexpected finding was the unusually steep and narrow dose dependence of MIF's induction of hyperexcitability. As just described, OA, RMP, AP voltage threshold, and rheobase first showed clear changes when exogenous MIF was increased from 0.1 to either 0.5 or 1 ng/ml (Fig. 1). Surprisingly, a further increase in concentration to only 10 ng/ml reversed all of these changes (Fig. 1A-E). Compared with 1 ng/ml MIF, the 10 ng/ml dose significantly hyperpolarized RMP (p = 0.0068, Welch's t test), depolarized AP voltage threshold (p = 0.0062, Welch's t test), and increased rheobase (p = 0.0023, Welch's t test) (Fig. 1C-E).

We found that MIF's effects on nociceptor excitability can be rapidly reversed. Neurons from Naive rats were treated with MIF (1 ng/ml) for 15 min, which was then washed away before recording 5-60 min later. After MIF washout, no OA was recorded at RMP (data not shown, 0 of 10 neurons sampled compared with 13 of 18 exhibiting OA when MIF was not washed out, p = 0.0003, Fisher's exact test). MIF-induced OA recorded at −45 mV was also eliminated by MIF washout (1 of 10 with OA compared with 16 of 18 when MIF remained present, p < 0.0001).

Because our intracellular pipette solution contained 3 mm EGTA, we considered the possibility that some of the observed effects could have been caused by chelation producing an abnormally low intracellular Ca²⁺ concentration. However, when we used a pipette solution with only 0.1 mm EGTA, no significant differences were observed on basal electrophysiological properties (RMP, rheobase, AP voltage threshold, and OA incidence) of NA neurons from Naive rats (n = 14 and 10 neurons for pipette solutions containing 3 and 0.1 mm EGTA, respectively). Furthermore, we observed the same nociceptor hyperactivity triggered by MIF (1 ng/ml) with the low-EGTA pipette solution, expressed as a significant increase in OA incidence measured at rest (50% of neurons with OA) and −45 mV (83% of neurons with OA) (Fig. 1F).

Sham surgery can enhance nociceptor hyperactivity induced by MIF

The sham surgery that exposes the spinal cord for contusive SCI experiments (injuring bone, muscle, and skin) was found to produce modest but significant nociceptor hyperexcitability in the absence of SCI (Odem et al., 2018) and to enhance pain-related behavior (Odem et al., 2019) months after injury. Given these observations and the induction of hyperactivity by MIF in neurons from Naive rats (Fig. 1), we predicted that nociceptors in the Sham group would display greater electrophysiological responsiveness to MIF compared with Naive controls. When tested 3-5 months after surgery at RMP, NA neurons isolated from Sham rats and exposed to doses of MIF that did not elicit OA in neurons from Naive rats exhibited a small but significant increase in OA at 0.5 ng/ml compared with vehicle treatment (0% for vehicle treatment vs 29% for 0.5 ng/ml MIF) (Fig. 2A). When tested at −45 mV, the incidence of OA in the Sham group was 14% for vehicle treatment versus 47% for 0.5 ng/ml MIF (Fig. 2A). The promotion of hyperexcitability in NA neurons by MIF was relatively weak, as it failed to significantly alter RMP, AP voltage threshold, or rheobase (Table 2). The weakness of these effects might be explained by most of the neurons being sampled from DRGs that innervate tissues other than those injured during the sham surgery. The excitability properties of RA neurons showed no significant effects in the Sham group (Table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

NA neurons isolated from injured rats exhibit complex responses to MIF. A, Effects of different doses of MIF (15-60 min pretreatment) on the incidence of NA neurons with OA when isolated 3-5 months after sham surgery. OA was measured at RMP and when artificially depolarized to −45 mV. Comparisons of OA at each concentration versus vehicle treatment (0 ng/ml) were performed with Fisher's exact test (p < 0.025 considered significant after Bonferroni correction for 2 comparisons). B, Effects of MIF on the incidence of neurons with OA 1-4 months after SCI surgery (p < 0.017 considered significant after Bonferroni correction for 3 comparisons). C, Representative recording at RMP of a hyperactive NA neuron isolated from an SCI rat and exposed to MIF starting at t = 0. D, Effect of MIF exposure on mean membrane potential at different time points (n = 11). Comparison between control period (t = −10 s) with subsequent time points was performed with Friedman test followed by Dunn's multiple comparisons test. Data are mean ± SEM. E, Effect of MIF exposure on AP firing frequency measured by 20 s time bins (n = 11). Comparison between control period (−20: 0 s) with subsequent time bins was performed with Friedman test followed by Dunn's multiple comparisons test. Each circle represents 1 neuron. Bars represent medians. MP, Membrane potential.

MIF application after SCI fails to further increase nociceptor hyperactivity and promotes the transition from an NA state to an RA state

Increased sensitivity of nociceptor somata to extrinsic signals following SCI has been implied by the upregulation of growth factor receptors TrkA and TrkB in DRG neurons (Qiao and Vizzard, 2002) and the upregulation of TRPV1 in DRGs and enhanced responsiveness of nociceptors to capsaicin (Wu et al., 2013). As reported previously (Bedi et al., 2010; Bavencoffe et al., 2016; Odem et al., 2018; Berkey et al., 2020; Garza Carbajal et al., 2020), we found 1-4 months after SCI that NA neurons were significantly more likely to exhibit OA at RMP (i.e., SA), with 63% of neurons active in the SCI group (Fig. 2B) versus 0% in the Naive group (Fig. 1B). When tested at −45 mV, 84% of neurons in the SCI group had OA versus 8% in the Naive group (Figs. 1B, 2B). The increase in OA in the SCI group compared with the Naive group was accompanied by a significant depolarization of RMP (p < 0.0001, Brown–Forsythe and Welch ANOVA test followed by Dunnett's T3 multiple comparison test F_(2,41.07) = 20.03, p < 0.0001), hyperpolarization of AP voltage threshold (p = 0.0005, Kruskal–Wallis test followed by Dunn's multiple comparison test), and reduction in rheobase (p < 0.0001, Brown–Forsythe and Welch ANOVA test followed by Dunnett's T3 multiple comparison test, F_(2,38.05) = 15.74, p < 0.0001) (Figs. 1, 2; Table 2). Unexpectedly, these chronic hyperexcitable effects of SCI were not further enhanced by MIF at any of the tested concentrations. Indeed, although our sample size was severely limited at the highest MIF concentration (for reasons described below), OA appeared to decrease (Fig. 2B) and rheobase increased (Table 2) when NA neurons were tested in the presence of 1 ng/ml MIF. This indicates that the neurons had become less excitable.

The failure of MIF to further increase OA (Fig. 2B) or to enhance other measures of excitability in neurons from the SCI group (Table 2) might reflect a ceiling effect or a simple shift in the dose–response relationship for MIF induced by SCI. However, combining persistent depolarization and hyperexcitability induced by SCI with strong acute depolarization of NA neurons by added MIF might induce a hypoexcitable state. To test this possibility, we perfused MIF (1 ng/ml) on NA nociceptors isolated from SCI rats. In the 20 s period following the beginning of MIF application, a pronounced depolarization of membrane potential occurred (Fig. 2C,D) often accompanied by a burst of APs (Fig. 2C,E). This discharge was followed within 60 s by inactivity that persisted during slow recovery toward the initial membrane potential (Fig. 2C-E). This pattern is consistent with a secondary response of long-lasting hypoexcitability induced during prolonged exposure to MIF at concentrations that strongly depolarize nociceptors isolated from injured animals.

In principle, the delayed inactivity observed during prolonged depolarization (Fig. 2C,E) might function in vivo to protect nociceptors and/or their postsynaptic targets from excessive activation and excitotoxicity during prolonged depolarizing conditions, such as inflammation. This idea raised the possibility that the intense depolarization produced by MIF in neurons from the SCI group could trigger a transition of highly excitable NA nociceptors to much less excitable RA nociceptors during the 5-60 min period when the neurons were incubated with MIF before electrophysiological testing. The characteristic excitability differences between these nociceptor types are illustrated by the responses of NA and RA nociceptors in Figure 3A (see also Odem et al., 2018); showing the higher rheobase current, apparently higher AP threshold, and lack of repetitive discharge in the RA neuron but not the NA neuron when each was stimulated at a holding potential of −60 mV by a depolarizing current that was twice the rheobase value. A depolarization-induced shift between nociceptor types would be consistent with the lack of significant differences between NA and RA neurons in somal size, membrane capacitance, capsaicin sensitivity, and IB4 binding, and with observations that, unlike NA neurons, RA neurons never exhibit OA and are significantly less excitable than NA neurons on all tested metrics (Odem et al., 2018).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

MIF causes a shift from NA to RA neurons after SCI. A, Representative traces of recordings from NA and RA neurons during a 2 s pulse of injected current at 2× rheobase. Red arrowhead indicates DSF. The illustrated APs have been clipped at 0 mV to show the DSFs. B, Effects of different doses of MIF on the incidence of RA neurons dissociated from Naive, Sham, and SCI rats compared with vehicle treatment. All neurons sampled were either RA or NA. Comparisons of RA incidence at each concentration versus vehicle treatment were performed with Fisher's exact test. After Bonferroni correction for multiple comparisons, the significance level was *p < 0.012 in the Naive group, p < 0.025 in the Sham group, and p < 0.017 in the SCI group. ***p < 0.00033 in the SCI group. C, Dose–effect relationships between MIF and the incidence of RA neurons sampled from control (Naive and Sham) and SCI groups.

Support for a MIF-induced transition from NA to RA types came from a significant change in the relative distributions of NA and RA neurons triggered by pretreatment with lower concentrations of MIF applied to the SCI group, and also by our highest applied MIF concentration to the Naive group. Among nociceptors with a diameter ≤ 30 µm isolated from SCI rats (all of which were classified as either NA or RA), 17% were the RA type (meaning 83% were NA) in the absence of MIF treatment (Fig. 3B,C), which was similar to the 23%-29% incidence of RA neurons after SCI we found previously in vehicle conditions for Naive and Sham groups or after SCI (Odem et al., 2018). When exposed to 1 ng/ml of MIF, the RA incidence increased to 81% (i.e., 19% were NA) in the SCI group (Fig. 3B,C). In contrast, only 33% of sampled neurons in the Naive group were the RA type at the same MIF concentration (1 ng/ml MIF). The RA incidence increased to 54% in the Naive group treated with 10 ng/ml MIF, which was significantly elevated compared with vehicle (Fig. 3B,C), showing that, in the absence of prior injury to the animal, a sufficiently high concentration of MIF can also increase the proportion of RA neurons. The minimum doses of MIF that significantly increased RA neuron incidence was 0.1 ng/ml in the SCI group and 10 ng/ml in the Naive group (Fig. 3B,C). These results demonstrate that either a high dose of MIF by itself or the combination of a much lower dose of MIF with prior SCI can dramatically increase the incidence of RA neurons (with a corresponding decrease of NA neurons), supporting the hypothesis that the NA and RA electrophysiological signatures in our experimental conditions primarily represent different functional states rather than fixed excitability phenotypes, and that nociceptors can be induced to transition from a highly excitable NA state to a much less excitable RA state by sufficiently prolonged and intense depolarization.

MIF enhances large DSFs without apparent further amplification by SCI or sham surgery

In addition to depolarization of RMP and hyperpolarization of AP voltage threshold, the third logically possible electrophysiological change that can generate OA in isolated neurons is enhancement of DSFs (Fig. 1A, red arrowheads) that transiently exceed AP threshold (Odem et al., 2018). Automated analysis of DSFs in recordings of NA neurons from Naive rats indicated that MIF tends to increase DSF amplitudes measured at RMP at the 1 ng/ml dose, although this effect was not statistically significant (Fig. 4A). When NA neurons were artificially depolarized to −45 mV, DSF amplitudes were significantly increased by treatment with 1 ng/ml MIF in the Naive group, and a trend was seen for larger DSF amplitudes at the 0.5 ng/ml dose (Fig. 4A). While SCI increased DSF amplitudes in NA neurons measured at RMP and at −45 mV in the absence of applied MIF relative to the Naive group (0 ng/ml; Fig. 4A), no significant further increase was produced by any concentration of MIF. This might be because: (1) too few NA neurons could be found in the SCI group after treatment with 0.1 or 1 ng/ml of MIF (Fig. 2B) for sufficient statistical power, (2) a ceiling effect, (3) the few NA neurons remaining after exposure to higher MIF concentrations were a subpopulation that was unresponsive to MIF (e.g., because they did not express MIF receptors), or (4) these few NA neurons were recorded during their transition to the RA state and had become less excitable without yet losing some of their NA properties.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

MIF treatment enhances DSFs in NA neurons from Naive but not SCI rats. A, Effects of different doses of MIF on DSF amplitudes in NA neurons isolated from Naive and SCI rats 1-4 months after surgery. Median DSF amplitudes (bars) are shown when measured at RMP and when artificially depolarized to −45 mV. Each circle represents the mean DSF amplitude measured in a 30 s recording from 1 neuron. Comparisons between each dose and vehicle (0 ng/ml) were performed by Kruskal–Wallis tests followed by Dunn's multiple comparison tests. Comparison among Naive and SCI groups during vehicle treatment was also performed using Kruskal–Wallis tests followed by Dunn's multiple comparison tests, with corresponding p values indicated on the figure. p < 0.05 was considered significant. B, Effects of different doses of MIF on the incidence of sampled neurons (same neurons as in A) with large DSFs (>5 mV) measured at RMP and when depolarized to −45 mV. Comparisons between each dose and vehicle were performed by Fisher's exact test. After Bonferroni correction for multiple comparisons, the significance level was *p < 0.012 in the Naive group and p < 0.017 in the SCI group. ***p < 0.00025 in the Naive group. Comparisons between Naive and SCI vehicle conditions were also performed with Fisher's exact test, with p < 0.025 considered significant after Bonferroni correction. For these comparisons, the significance level was ***p < 0.0005.

The clearest effect of MIF on DSFs was on the incidence of large DSFs, defined as having amplitudes >5 mV, which is sufficient sometimes to elicit APs (Odem et al., 2018). In the Naive group, after treatment with 1 ng/ml MIF at RMP, the incidence of sampled neurons with at least one large DSF increased from 24% in vehicle-treated neurons to 73% in neurons treated with 1 ng/ml MIF (Fig. 4B). At a holding potential of −45 mV, both 0.5 and 1 ng/ml doses of MIF increased the incidence of neurons with large DSFs, from 19% to 64% and 91%, respectively (Fig. 4B), paralleling the increase in OA at −45 mV (Fig. 1B). However, no further increase in the incidence of large DSFs at RMP or −45 mV was found at any MIF dose compared with vehicle in the SCI group (Fig. 4B). Similar to the reversal in effects observed in Naive neurons when comparing the 10 ng/ml with 1 ng/ml doses on RMP (Fig. 1C), AP voltage threshold (Fig. 1D) and rheobase (Fig. 1E), we found a reversal across these doses in the relative DSF amplitudes (Fig. 4A, measured at RMP, p = 0.0098, Welch's t test; at −45 mV, p = 0.0019, Mann–Whitney t test) and in the relative incidence of neurons with large DSFs (Fig. 4B, measured at RMP, p = 0.0075 and −45 mV, p = 0.024, Fisher's exact test).

Treatment with a MIF inhibitor in vitro eliminates SCI-induced hyperexcitability

To assess the necessity of MIF activity in vitro for the occurrence of SCI-induced hyperactivity, we treated nociceptors isolated 1-4 months after surgery with a MIF inhibitor, Iso-1 (40 µm). Iso-1 is reported to be highly specific to MIF, inhibiting MIF tautomerase activity (Lubetsky et al., 2002; Senter et al., 2002; Cheng and Al-Abed, 2006) and blocking the binding of MIF to its receptors, principally CD74, preventing activation of MIF and its downstream signaling (Al-Abed and VanPatten, 2011; Leng et al., 2011; Pantouris et al., 2015; Trivedi-Parmar and Jorgensen, 2018). Pretreatment of cultures in the SCI group with Iso-1 either overnight or beginning 1 h before testing (in both cases, Iso-1 remained present during testing) dramatically decreased the incidence of OA measured at RMP and at −45 mV (Fig. 5A). This reversal of SCI-induced hyperactivity was associated after one or both durations of Iso-1 treatment with significant alterations that would reduce hyperexcitability, including hyperpolarization of RMP (Fig. 5B), hyperpolarization of AP voltage threshold (Fig. 5C), and increased rheobase current (Fig. 5D). Moreover, DSF amplitudes were reduced by Iso-1 treatment when measured at RMP or −45 mV (Fig. 5E), as was the incidence of large DSFs (>5 mV) measured at −45 mV (Fig. 5F). These results show that Iso-1-sensitive signaling (probably through MIF; see Materials and Methods) in the culture dish is required for the expression of nociceptor hyperactivity induced by prior SCI in vivo.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

MIF inhibitor Iso-1 reverses hyperactivity in NA neurons from SCI rats. A, Effect of Iso-1 (40 μm) treatment overnight (ov) or for 1 h before (and during) testing versus vehicle (Veh, 0.1% DMSO in ECS) on the incidence of OA recorded from NA neurons isolated from SCI rats at RMP and when depolarized to −45 mV. Comparisons of OA in each Iso-1 treatment group versus vehicle were performed with Fisher's exact test (*p < 0.025 considered significant after Bonferroni correction for 2 comparisons; **p < 005; ***p < 0005). B, Effects of the same treatments on median RMP. C, Effects on mean (± SEM) AP voltage threshold. D, Effects on median rheobase. E, Effects on median DSF amplitude. F, Effects on incidence of neurons with large (>5 mV) DSFs. Comparisons between vehicle and both Iso-1 treatments were performed by Kruskal–Wallis tests followed by Dunn's multiple comparison tests in B and D, and by Brown–Forsythe and Welch ANOVA test followed by Dunnett's T3 multiple comparison test (F_(2,25.81) = 3.83, p = 0.035) in C. B–D, p < 0.05 was considered significant. E, Effects of Iso-1 on median DSF amplitude measured at RMP and −45 mV. F, Effects of Iso-1 on the incidence of neurons with large DSFs. Comparisons between vehicle and both Iso-1 treatments were performed by Kruskal–Wallis tests followed by Dunn's multiple comparison tests in E and Fisher's exact test in F. E, p < 0.05 was considered significant. F, *p < 0.025 was considered significant after Bonferroni correction; **p < 005; ***p < 0005.

MIF inhibition does not alter basal excitability or acute cAMP-induced hyperexcitability in nociceptors isolated from Naive animals

An important question is how specific the strong suppression of SCI-induced hyperexcitability by MIF antagonist Iso-1 is to the hyperexcitability found after SCI. Arguing against a general neuronal effect of Iso-1 independent of SCI-induced hyperexcitability were three findings. First, while OA and hyperexcitability in neurons from the SCI group were abrogated by coapplication of Iso-1 (Fig. 5), no effects on OA incidence or RMP were found after coapplication of Iso-1 (40 µm, 1 h) with MIF (1 ng/ml) on nociceptors from the Naive group (Fig. 6A,B), suggesting that the inhibitory effects may be specific to the increase in excitability induced by SCI. Second, pretreatment of neurons from the Naive group with Iso-1 in the absence of MIF caused no significant alterations of OA (Fig. 6C), RMP (Fig. 6D), AP voltage threshold (Fig. 6E), or rheobase (Fig. 6F). Third, as part of the experiments in Figure 6C-F, we combined Iso-1 with the adenylyl cyclase activator, forskolin (1 µm, 15 min). As we published previously, the cAMP pathway and its downstream effectors (PKA, EPAC) are important contributors to the maintenance of nociceptor hyperactivity induced 1-8 months earlier by SCI (Bavencoffe et al., 2016; Berkey et al., 2020; Garza Carbajal et al., 2020) or acutely by serotonin (Lopez et al., 2021). Interestingly, Iso-1 failed to prevent forskolin from dramatically increasing OA measured at −45 mV (Fig. 6C) and depolarizing RMP acutely (Fig. 6D) in neurons from Naive rats. While no significant effect of Iso-1 was found on AP voltage threshold after forskolin treatment (Fig. 6E), Iso-1 did not prevent the reduction in rheobase current by forskolin (Fig. 6E). These findings indicate that Iso-1 has selective effects on nociceptors, inhibiting chronic hyperexcitability after SCI but not affecting basal excitability under Naive conditions or acute hyperexcitability induced by the stimulation of cAMP signaling by forskolin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

MIF inhibitor Iso-1 does not alter basal excitability or acute cAMP-induced hyperexcitability in NA neurons dissociated from Naive rats. A, Lack of effect from coincubating NA neurons with MIF (1 ng/ml, 15 min) and inhibitor Iso-1 (40 μm, 1 h) compared with vehicle (0.1% DMSO in ECS) with on OA incidence at RMP and when depolarized to −45 mV (Fisher's exact test). B, Lack of effect on median RMP (Mann–Whitney U test). C, Failure of Iso-1 (40 μm, 1 h) coapplication to prevent increased OA incidence at −45 mV by adenylyl cyclase activator forskolin (1 μm, 15 min) (Fisher's exact test, **p < 0.005 after Bonferroni correction). D, Failure of Iso-1 to prevent depolarization of mean (±SEM) RMP by forskolin (one-way ANOVA and Dunnett's test, F_(2,27) = 0.19, p = 0.82). E, Lack of apparent effect of Iso-1 on mean (±SEM) AP voltage threshold with or without forskolin (Brown–Forsythe and Welch ANOVA test and Dunnett's test, F_(2,24.34) = 1.15, p = 0.33). F, Failure of Iso-1 to prevent reduction of median rheobase by forskolin (Kruskal–Wallis test and Dunn's test). B, D–F, p < 0.05 was considered significant. Fsk, Forskolin; Veh, vehicle.

Conditioned medium from cultures of DRG cells isolated from SCI rats triggers MIF-dependent nociceptor hyperactivity

The inhibitory effects of Iso-1 on SCI-induced hyperactivity in dissociated nociceptors (Fig. 5) indicated that MIF function in vitro is necessary for promoting this hyperactivity. We next looked for evidence that MIF released into the medium from cells dissociated from SCI rats can induce hyperactivity in nociceptors dissociated from Naive rats. Our cultures include various cell types reported to express (and in some cases, release) MIF, including DRG neurons, satellite glial cells, fibroblasts, and macrophages (Abe et al., 2000; Vera and Meyer-Siegler, 2003; Alexander et al., 2012; Lee et al., 2016), and previous studies have detected MIF expression in DRG neurons from uninjured rodents (Vera and Meyer-Siegler, 2003; Alexander et al., 2012). We first exposed nociceptors isolated from Naive rats to culture medium conditioned by overnight (∼18 h) incubation of DRG cells isolated from either Naive or SCI rats (Fig. 7A). After 1 h treatment with conditioned medium, we transferred each coverslip of neurons to ECS and recorded from several sampled neurons over a period of 30-60 min. A significant increase in OA measured at −45 mV but not at RMP occurred following exposure to the SCI conditioned medium (Fig. 7B). Exposure to the Naive conditioned medium had no significant effect on OA at RMP or at −45 mV (Fig. 7B). The SCI conditioned medium produced a trend to depolarize RMP (Fig. 7C), but it had no significant effect on AP voltage threshold (Fig. 7D) or rheobase (Fig. 7E). Incubation with SCI conditioned medium increased DSF amplitudes recorded at −45 mV (Fig. 7F) and caused a large increase in the incidence of neurons with large DSFs recorded at −45 mV (Fig. 7G). Addition of Iso-1 to conditioned medium after removal of the medium from dishes containing cells from SCI animals and before transfer to dishes containing cells from Naive animals prevented the increase in OA at −45 mV (Fig. 7B), weakened the trend for depolarized RMP (Fig. 7C), and eliminated both the increase in DSF amplitude (Fig. 7F) and the increased incidence of neurons with large DSFs recorded at −45 mV (Fig. 7G). These results indicate that after SCI, an Iso-1-sensitive factor (probably MIF), is continuously released by unknown populations of dissociated DRG cells, reaching levels in the medium sufficient to induce hyperactivity lasting at least 30-60 min after washout in nociceptors isolated from rats that had not experienced SCI.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

NA neurons dissociated from Naive rats exhibit MIF-dependent hyperactivity after exposure to conditioned medium bathing DRG cells dissociated from SCI rats. A, Experimental design. B–G, MIF-dependent effects of SCI conditioned media (SCI) versus either Naive conditioned media (N) or SCI conditioned medium plus Iso-1 (40 μm) after 1 h on nociceptors isolated from Naive rats. B, Increased incidence of neurons with OA at −45 mV but not at RMP, which was prevented by Iso-1 (Fisher's exact test, *p < 0.025, **p < 0.005 after Bonferroni correction). C, Lack of strong effect on mean RMP (Brown–Forsythe and Welch ANOVA test followed by Dunnett's T3 multiple comparison test, F_(2,56.49) = 2.5, p = 0.091). D, Lack of effect on mean AP voltage threshold (one-way ANOVA followed by Dunnett's multiple comparison test, F_(2,60) = 0.97, p = 0.38). E, Lack of effect on median rheobase (Kruskal–Wallis test followed by Dunn's multiple comparison test). F, Increased median DSF amplitude measured at −45 mV but not RMP, which was prevented by Iso-1 (Kruskal–Wallis test followed by Dunn's multiple comparison test). C–F, p < 0.05 was considered significant. G, Increased incidence of neurons with large DSFs, which was prevented by Iso-1 (Fisher's exact test, *p < 0.025, **p < 0.005 after Bonferroni correction).

Excitation of probable nociceptors by MIF partially depends on the extracellular signal-regulated kinase (ERK) pathway

MIF stimulation of DRG neurons has been reported to involve Iso-1-sensitive binding to CD74 receptors and transactivation of the ERK pathway (Alexander et al., 2012). We recently showed that activation of the ERK pathway downstream of C-Raf is required to maintain the depolarization of RMP (into the range of ∼−55 to −50 mV) and hyperactivity in nociceptors induced by SCI (Garza Carbajal et al., 2020). Thus, we tested a possible role of the ERK pathway in the excitation of probable nociceptors by MIF shown in Figure 1. Neurons isolated from Naive rats were pretreated with an inhibitor (UO126, 3 µm) of the upstream activator of ERK, MEK (mitogen-activated protein kinase kinase), or with its corresponding vehicle (extracellular recording solution with 0.03% DMSO) for 15 min alone and then for an additional 15 min in combination with MIF (1 ng/ml). Coincubation with UO126 prevented MIF-induced OA at RMP but not at −45 mV (Fig. 8A). This effect was probably from blocking the depolarizing effects of MIF (Fig. 8B) because UO126 treatment did not significantly impact AP voltage threshold (Fig. 8C) or rheobase (Fig. 8D). Because coapplication of the vehicle for UO126 did not alter the effects of MIF (1 ng/ml), we pooled these data with the 1 ng/ml MIF data shown in Figure 1 to increase statistical power for DSF analysis. UO126 significantly decreased DSF amplitudes at rest and at −45 mV (Fig. 8E) but did not decrease the incidence of neurons with large DSFs recorded at rest or when artificially depolarized to −45 mV (Fig. 8F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Partial effects of MEK/ERK inhibition on the hyperexcitable effects of MIF on NA neurons isolated from Naive rats. A, Effect of UO126 (3 μm) 30 min treatment versus vehicle (Veh, 0.03% DMSO in ECS) on the incidence of OA at RMP and when depolarized to −45 mV during treatment with MIF (1 ng/ml) in NA neurons from Naive rats. Neurons were treated first with vehicle or UO126 for 15 min and then coincubated with MIF for 15 min. Comparisons of OA in each treatment group versus vehicle were performed with Fisher's exact test (*p < 0.05; **p < 0.01). B–D, Effects of the same treatments on mean ± SEM RMP (B), AP voltage threshold (C), and rheobase (D). Comparisons were performed by Welch's t test. E, Effects on median DSF amplitude. F, Effects on incidence of neurons with large (>5 mV) DSFs. Numbers over each bar represent the number of neurons exhibiting large DSFs/total number sampled. Comparisons of DSF amplitudes between both conditions were performed by Welch's t test, while comparisons of incidence of neurons with large DSFs were done by Fisher's exact test. G, Effect of UO126 (3 μm) treatment versus vehicle (Veh, 0.03% DMSO in PBS) on pERK levels measured by HCM in control (PBS), MIF (1 ng/ml, 30 min), and NGF (30 ng/ml, 30 min) conditions. UO126 or vehicle was given 30 min beforehand. Comparisons of UO126 effects in each condition were performed by two-way ANOVA with Sidak's multiple comparisons test (F_(2,12) = 25.26, p = 0.0005). Comparisons of pERK levels in vehicle conditions were done by repeated-measures one-way ANOVA with Geisser–Greenhouse correction followed by Dunnett's multiple comparisons test (F_{(1.013,2.026)} = 98.39 p = 0.0096).

The limited impact of UO126 treatment on the hyperexcitable effects of MIF in our conditions was somewhat surprising, given that transactivation of the ERK pathway has been reported in DRGs following MIF treatment. Therefore, we determined by HCM whether MIF treatment (1 ng/ml, 15 min) increases phospho-ERK (pERK) levels acutely in sensory neurons isolated from Naive rat DRGs (Fig. 8G). We used NGF (nerve growth factor, 30 ng/ml, 30 min treatment) as a positive control (Garza Carbajal et al., 2021). While pERK levels were significantly elevated following NGF treatment compared with the control condition, there was no increase following MIF application. As expected, NGF-induced pERK elevation was prevented by 30 min treatment with UO126 (3 µm), and trends were seen for UO126 to decrease pERK in basal and MIF-stimulated conditions. These HCM data extend our electrophysiological observations of limited effects of UO126; together, they suggest that the ERK pathway contributes partially to the hyperexcitable effects of MIF on NA neurons by promoting depolarization without affecting other electrophysiological properties.

MIF can trigger an aversive behavioral state

Our observations of MIF-induced hyperexcitability and OA in probable nociceptors (see also Alexander et al., 2012) plus previous reports of MIF-induced reflex hypersensitivity (Wang et al., 2010, 2011; Alexander et al., 2012) suggested that experimental delivery of MIF in vivo might produce affective pain and that inhibition of MIF activity might reduce ongoing pain. However, complexities in MIF's known actions pose challenges to documenting behavioral consequences of its activity. First, our findings of both positive and negative effects of MIF on nociceptor excitability means that injection of exogenous MIF in vivo might produce either net excitatory or inhibitory effects on nociceptor activity and consequent pain-related behavior, and different effects at different times and distances from an injection site. Second, previously reported behavioral effects of MIF injection in vivo develop over hours (Alexander et al., 2012) and possibly days (Wang et al., 2011). Slow development of complex MIF-dependent effects that might persist in the absence of MIF activity after development makes testing the behavioral roles of MIF difficult.

Our clearest indication that MIF stimulates pain was finding learned aversion to a place associated with MIF injection (conditioned place aversion [CPA]). CPA provides strong evidence for the affective-motivational component of pain in animals (Hummel et al., 2008). On day 1, all Naive rats were exposed to a three-chamber apparatus in a pretest to determine their preferred chamber. On day 2, each rat hindpaw was injected with vehicle and the rat placed later into the nonpreferred chamber as determined on day 1 (vehicle pairing). Four hours later, each rat received hindpaw injections of MIF (50 ng each) and was later transferred to its previously preferred chamber (MIF pairing). On day 3, rats received a post-test identical to the pretest. In one variant of this CPA procedure, the interval between each injection and placement in the paired chamber was 30 min and the duration in the chamber was 30 min. This variant failed to produce CPA (Fig. 9A, left). Importantly, this experiment showed that nonspecific aspects of the protocol, such as where the most recent injection occurred, did not produce place aversion. The second variant of our CPA procedure assumed a longer latency and duration of MIF effects; the interval between each injection and placement in the paired chamber was 1 h, and the duration in the chamber was 2 h. This variant produced significant CPA to the MIF-paired chamber (Fig. 9A, right), revealing that stimulation by MIF can produce an aversive state resembling pain.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

MIF can trigger an aversive behavioral state as indicated by CPA, but other pain-related tests did not reveal significant effects. A, Left, Lack of CPA to a chamber paired for a shorter period (30 min) and shorter time (30 min) after intraplantar injection of MIF (50 ng in each hindpaw). Rats received MIF in their initially preferred chamber as defined on day 1 and vehicle in their nonpreferred chamber (n = 6). Bars represent the median times in the initially preferred chamber. Open circles represent individual values. Differences are tested by paired t test. A, Right, Significant CPA to a chamber paired for a longer period (2 h) and longer time (1 h) after the MIF injections (n = 17). B, C, Lack of effect on the MC test from hindpaw injections of MIF (50 ng, n = 12) or vehicle (n = 12). Results from the last 3 of 6 tests are shown; only the last 2 had the sharp probes elevated to 4 mm. Injections (dashed line) were given 1 h before the first 4 mm test. B, Latency for the first crossing of the connecting chamber either without (0 mm) or with (4 mm) probe elevation. Both groups showed an increase in latency to cross the probes, but there were no significant treatment differences (mixed-effects two-way ANOVA with Sidak's multiple comparisons test, F_(1,22) = 0.27, p = 0.6). C, The number of crossings in 5 min. Both groups showed a decrease in crossings, but there were no significant treatment differences (two-way ANOVA with Sidak's multiple comparisons test, F_(1,22) = 2.38, p = 0.14). Bars represent medians. Circles represent individual animals. D, E, Results of CPP tests on SCI (D) and Naive rats (E) for a chamber paired with intrathecal injection of MIF inhibitor, Iso-1. On day 2, SCI (n = 24) or Naive (n = 6) rats received single-trial differential conditioning with intrathecal injection of Iso-1 (30 µg) in the previously less preferred chamber, and vehicle in the other. Bars represent medians. Comparisons are by paired t test.

We used a different test of voluntary behavior to investigate potential enhancement of evoked pain (hyperalgesia) by MIF injection. A modified version (Odem et al., 2019) of a previously described operant MC test (Harte et al., 2016; Pahng et al., 2017) gave rats a choice to cross sharp probes to explore a relatively novel environment and avoid a bright light. Mechanical hyperalgesia, as indicated by rats voluntarily choosing to spend more time under the light and to cross the probes less frequently, was revealed previously by the MC test in rats given thoracic spinal contusion or sham surgery (Odem et al., 2019). After four 5 min trials in the MC apparatus with the probes retracted (0 mm), both hindpaws of Naive rats were injected with either vehicle or MIF (50 ng). Beginning 60 min later, they were tested with the probes elevated to 4 mm, and this test was repeated after an additional 60 min. MIF injection produced no significant effects compared with vehicle on either the latency for the first crossing or the number of crossings during the 5 min tests with elevated probes (Fig. 9B,C), although a very weak trend to reduce the number of crossings at 0 and 4 mm may have been present.

We also used the MC test to see whether inhibiting MIF function would reduce evoked pain after SCI. Because the pain induced by thoracic SCI is widespread and expressed by hypersensitivity in all four limbs (Bedi et al., 2010), and because, to our knowledge, the only previous study that tested MIF inhibitors on rats' pain behavior used intrathecal injection (Wang et al., 2010), we injected either Iso-1 (30 µg, n = 8) or vehicle (saline with 10% DMSO, n = 3) by lumbar puncture. Neither Iso-1 nor vehicle injected intrathecally after the first 4 mm probe exposure decreased the first crossing latency or increased the number of crossings during the second 4 mm probe exposure (data not shown). These results indicate that MIF injected subcutaneously into the plantar surface of a paw does not enhance affective pain evoked by stepping on sharp probes in Naive rats, and that intrathecal inhibition of MIF does not reduce chronic mechanical hyperalgesia induced by SCI.

To begin to test the hypothesis that continuing MIF activity in the DRG and/or spinal cord is required to maintain chronic ongoing pain after SCI, we used a CPP test with nearly the same schedule as our CPA tests (Fig. 9A). Previous work demonstrated a strong correlation between the incidence of SA (OA at RMP) in dissociated nociceptors and chronic reflex hypersensitivity after SCI (Bedi et al., 2010). In vivo interventions inhibiting SA in nociceptors (antisense knockdown of Nav1.8 channels or delivery of a KCNQ channel opener) had alleviated not only reflex hypersensitivity but also ongoing pain as indicated by a CPP test (Yang et al., 2014; Wu et al., 2017, 2020). After giving SCI rats a pretest on day 1 identical to that used in the CPA test, on day 2 vehicle was injected intrathecally 30 min before placing them into their preferred chamber for 30 min. Four hours later, the procedure was repeated with Iso-1 (30 µg) injected intrathecally before transfer into the nonpreferred chamber. This intrathecal dose of Iso-1 was reported to reduce reflex hypersensitivity in rat models of peripheral inflammation and nerve injury (Wang et al., 2010, 2011). In the post-test on day 3, the SCI rats exhibited a very weak, statistically insignificant trend to spend more time in the chamber previously paired with Iso-1 than the chamber previously paired with vehicle injection (Fig. 9D; p = 0.123). The same CPP procedure on Naive rats revealed no evidence of a conditioned preference for the Iso-1-paired chamber (Fig. 9E, paired t test). Thus, while our CPA results show that MIF application can produce a pain-like aversive state, our MC and CPP tests confirm that the behavioral actions of MIF are complex, and their potential contributions to the maintenance of persistent pain after SCI require further investigation.