Twenty female C57BL/6 mice (15 weeks old) were obtained from Shanghai Slake Laboratory Animal Co., Ltd. The housing temperature for them is between 21 and 25°C, and the relative humidity is 70%, with a 12 h light/dark cycle. They were free to access food and tap water. The experimental animal ethics committee of Ruijin Hospital, Shanghai Jiao Tong university school of medicine approved all procedures.

To avoid interference from the estrous cycle, mice at 16-week-old of age were bilaterally ovariectomized by a dorsolateral approach as previously described (Khaksari et al., 2013). Experimental animals had ovariectomy surgery 2 weeks before the experiments started (Olson and Bruce, 1986). Then they were administered a phytoestrogen-minimal diet (Beijing HFK bioscience Co., Ltd., Beijing, China).

Two weeks after ovariectomies, the animals were randomly divided into four groups (five mice in each group): control group; sham DLBP plus vehicle group; DLBP plus vehicle group; DLBP plus ERβ-specific agonist diarylpropionitrile (DPN) group. DPN were dissolved in DMSO. Vehicle control mouse was only injected with DMSO. DPN dose of 1 mg/kg is roughly equal to what previously used in vivo (Oyola et al., 2012). Single sc injection of DPN (1.0 mg/kg in vehicle) a daily were administered to the mice for 3 weeks.

As for DLBP model, an established disc injury mouse model was used. Mice were surgically induced by annterolateral extraperitoneal approach with previously described methods (Moran et al., 2007). Briefly, by intraperitoneal injection of sodium pentobarbital (0.1 mg/g), animals were anesthetized and placed in a supine position on a heating pad. Mice were anesthetized by intraperitoneal (i.p.) injection of 0.1 mg/g sodium pentobarbital. The mice were fixed on a heating pad in a supine position. Then abdominal skin was shaved and disinfected. An approximately 2 cm incision was made with a ^#15 scalpel at the left lateral abdomen. As shown in Figure 1, an extraperitoneal approach was adopted, but not the conventional transabdominal surgery, to expose the surgical field. After the subcutaneous fat was separated, we retracted the peritoneum to the right side. The pelvic rim was regarded as an anatomic landmark. Then, lumbar spine discs could be identified under a surgical microscope through the region between the left psoas major muscle and the retroperitoneum. Disc injury procedures were similar to Park et al. (2019) previously described. The L4/5 and L5/6 discs were punctured at their center line using a microscalpel with a curved point in turn. The nucleus pulposus tissue was destroyed, and part of them was aspirated via a 26-gauge needle and then removed (Martin et al., 2013). Finally, the abdominal incision was closed with 6–0 vicryl suture suture. The sham-operated mice received the same procedure as the DLBP mice exclusive disc puncture.

Behavioral changes of low back pain are reflected as that is different from those with classical nerve ligation induced neuropathic pain or carrageenan-induced acute inflammatory pain (Miyagi et al., 1976). As for the DLBP model, behavioral tests, including behavioral measures of axial back pain (grip force and tail suspension tests) and radiating hypersensitivity (mechanical sensitivity and cold sensitivity test) were monitored according to previous report (Millecamps and Stone, 2018). Behavioral assays were performed 2 days before surgery to acquire baseline measurements and repeated at weeks 1–3 after initial injury. The animals underwent behavioral testing 4 h post agonist administration as previously adopted (Cao et al., 2012).

A dual-label confocal immunofluorescence examination was used to observe the potential crosstalk between substance P and ERβ in the spinal cord (Li et al., 2013). By boiling in 0.01 M sodium citrate buffer, pH 6.0, slides were blocked with normal 5% goat serum (Vector, S-1000). They were incubated overnight with a mouse anti-substance P antibody (Santa Cruz, sc-58591, 1:100 dilution). Then at room temperature, slides were incubated with Alexa Fluor^§ 594 Goat anti-rabbit (Invitrogen, A11037) for 30 min in the dark. Three times washing in phosphate-buffered saline (PBS) for 5 min each. Room temperature incubation with rabbit anti-ERβ antibody (Affinity, AF6469, 1:100 dilution) for 30 min in the dark, followed by 30 min of fluorescent conjugated secondary antibodies of Alexa Fluor^§ 488 Goat Anti-Rabbit IgG (Invitrogen, Cat:A11034), rinsed three times with PBS. The final step involves mounting slides in the fluorescent mounting medium containing 4’,6-diamidino-2-phenylindole (DAPI) (Invitrogen, P36935). All slides were viewed with a microscope (Nikon Eclipse 50i). Nucleus was visualized as blue by 330–380 nm filter, green mark with 465–495 nm filter, and red mark with 530–600 nm filter.

The RNA isolation protocol was adapted from our previously described procedure (Li et al., 2012). The frozen dorsal root ganglia sample is ground with fine glass powder by using a mortar and pestle. Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. RNA concentration and purity were determined with absorbance measured at OD₂₆₀ and OD₂₈₀. RNA integrity was assessed by microcapillary agarose gel electrophoresis. First-strand cDNA was reverse-transcribed using the TaKaRa Primescript™ RT Reagent Kit (Takara, Tokyo, Japan) according to the manufacturer’s protocol. Substance P gene expression in the dorsal root ganglia was measured by real-time PCR. Gene expression was evaluated with a LightCycler Detection System, supplied by Roche (Roche Diagnostics). PCR amplification was performed using SYBR Premix Ex Taq (Takara, Japan) following the manufacturer’s protocols. The forward and reverse primer sequences for the amplifications were as follows: TAC1 (encoding substance P synthesis): 5′-AAGCGGGATGCTGATTCCTC-3′ and 5′TCTTTCGTAGTTCTGCATTGCG-3′; β-actin: 5′-GGCTGTATT CCCCTCCATCG-3′ and 5′-CCAGTTGGTAACAATGCCATGT-3′. The reaction of qPCR included two steps: first, heating to 95°C for 30 s, second, 20 cycles consisting of 5 s at 95°C followed by 20 s at 60°C. We take the ΔΔCt formula to evaluate the transcript levels. The data were normalized against β-actin and expressed relative to control group.

The spinal cord was placed into a postfixating solution (4% paraformaldehyde) at 4°C overnight and can be further processed. After dehydrated in graded alcohols and xylene, the spinal cord embedded in paraffin for further immunohistochemistry analysis. Sections were baked and were treated with xylene, then rehydrated in graded ethanol to finally rinse with PBS. Slides immunostaining was performed using the streptavidin-biotin peroxidase (SABC) technique. Sections were incubated in 1% H₂O₂ for 15 min in order to block endogenous peroxidase activity, and washed with PBS. After that, we preincubated slides with 5% normal goat serum for 30 min at room temperature. And then, they were incubated with rabbit polyclonal antibody of substance P (Affinity, DF6541, 1:200) at 4°C overnight. Right after washing with PBS, slides were incubated at 25°C for 2 h with peroxidase-conjugated secondary antibody. After incubating with ABC complex (Vectastain ABC kit) for 30 min at 25°C, DAB peroxides substrate solution was used in the detection of staining. After washing thrice with PBS, dehydrating by graded ethanol and clearing in xylene, the sections were mounted with permount medium after 3 min counterstaining with Gill’s hematoxylin solution. The results were analyzed and photographed with a digital microscope (Olympus, Japan). To evaluate substance P’s immunoreactivity intensities, the semiquantitative image-analytical method was used. With ImageJ computer software (Nationwide Institutes of Wellness, Bethesda, MD, USA), semiquantitative image analyses were performed, and the mean results were calculated under 400× magnification (Crowe and Yue, 2019). Briefly, the resulting images were saved as .tiff files, each figure was opened and analyzed using deconvolution methods of ImageJ software. The threshold value was determined using five graphs to reduce the background and unspecific signal. Then, the data value of each group was measured by documenting the “mean grey value,” which represents the quantifies staining intensity. GraphPad prism software (version 9.0 for windows¹) was used to present the collected data.

Data were expressed as mean ± standard deviation. ANOVA methods were used to assess the differences of different groups. Post hoc analysis was performed with Bonferroni’s test. Two-way ANOVA was used for behavioral data and one-way ANOVA was applied to gene expression and immunoreactivity data. The level of statistical significance is often expressed as p-value of less than 0.05. The analysis was done using the SPSS statistical package (SPSS Inc., Chicago, IL, USA).

The effects of ERβ on DLBP were studied by different behavioral pain assessments of axial discomfort (grip force and tail suspension) and radiating discomfort (mechanical sensitivity and cold sensitivity). Baseline pain tests of all behavioral assays showed no differences among all groups of mice (Figure 2).

In order to promote a better understanding of the physiological roles of ERβ in the spinal cord of the mouse model of DLBP, the spatial colocalization of substance P with ERβ in the spinal cord was investigated using dual-label confocal immunofluorescence examination. As shown in Figure 3, positive immunoreactivity for ERβ appears green, and those labeling with substance P appear red. The co-expression of immunoreactivity for substance P and ERβ was observed in the dorsal horn of spinal cord using double immunofluorescence staining, mainly in the laminae I and II, a connection site of pain transmission. Colocalization of substance P and ERβ provided evidence that there may likely be an interaction between them in vivo within the spinal cord of DLBP mouse model.

To investigate the possible action of ERβ on DLBP in mice, we observed the role of ERβ activation on substance P changes in the spinal cord and dorsal root ganglia (Figures 4, 5).

In the dorsal horn of the spinal cord, the content of substance P protein in different groups was determined by immunohistochemistry staining. As shown in Figure 4, the immunopositive staining intensity changes mainly occurred in the posterior horn of the spinal cord. In the DLBP + vehicle group (Figure 4D), the dorsal horn exhibited stronger immunoreactivity density of yellow color compared with the control (Figure 4B) and DLBP sham + vehicle groups (Figure 4C). This indicated that substance P level increased in the connection site of pain transmission in the spinal cord of DLBP mice. While ERβ agonist DPN therapy decreased the intensity of substance P (Figure 4E), which implies that ERβ agonist could regulate substance P content in nociceptive transmission site in the spinal cord dorsal horn. The similar trend of substance P protein level was determined compared with the quantification of substance P in the spinal cord tissue (Figure 4F).

In dorsal root ganglia, the quantitative real-time PCR data demonstrated similar changes of substance P gene to that of the spinal cord (Figure 5). The results showed that the level of substance P in dorsal root ganglia of DLBP mice was significantly increased compared with that of the control and DLBP sham mice. After the application of the specific activation of ERβ, compared with that of vehicle-treated DLBP mice, the level of substance P mRNA in dorsal root ganglia was significantly decreased (Figure 5, p < 0.001). ERβ may regulate substance P in dorsal root ganglia, and then achieve the antinociceptive effect on DLBP.

Animal models have been proven to be essential tools for understanding disease mechanisms and detecting potential treatment methods of DLBP. Ranging from large to small animals, single or multiple-level mechanical lumbar disc injury models have been developed through ventral, dorsal, or lateral punctures approaches using needles or surgical blades (Millecamps and Stone, 2018; Shi et al., 2018b; Tang et al., 2022). We used a previously reported mouse DLBP model in this study, which was set up by puncture and removal of part of the lumbar nucleus pulposus (Shi et al., 2018a). Because the conventional transabdominal approach may lead to gastrointestinal distress and abdominal pain after surgery, which may confound back pain measures (Damle et al., 2013). Therefore, a less invasive extraperitoneal approach was employed to perform disc puncture in the present model. This approach includes fewer eating problems, fewer abdominal organ interruptions, and faster recovery after surgery. Consequently, behavioral measures of DLBP could be improved. Animal models need to consider the emergent behavioral changes following pain induction. Behavioral measures of DLBP are essential for studies using clinically relevant animal models to understand the pain mechanisms (Mosley et al., 2017). In mice, it is difficult to replicate for back pain, furthermore, its measurement and interpretation are much harder (Yang et al., 2018a). Behavioral signs of axial discomfort and radiating pain have been employed for behavioral assays in DLBP models (Millecamps et al., 2012). In our study, the puncture injury DLBP model induced both axial and radiating discomfort for up to 4 weeks after surgery (Figure 2).

Intervertebral disc degeneration is the main contributor to low back pain. Estrogen and its receptors play critical roles in mediating IVD degeneration. Musculoskeletal pain conditions appear to occur far more often in women than in men, and these sex-related differences may be explained, at least in part, by the actions of two major sex hormones.

Although musculoskeletal pain appears to occur far more often in female than in male, however, the exact pathogenesis of the underlying the IVD related pain remains unclear (de Kruijf et al., 2016; Peshkova et al., 2022). Some investigations have demonstrated that ERs exist in the IVD by us (Song et al., 2014, 2017; Jin et al., 2020) and by others (Gruber et al., 2002; Wei et al., 2016). We have observed that ERs are expressed in the human nucleus pulposus and ERs levels downregulated were consistent with the severity of IVD degradation, particularly ERβ level (Song et al., 2014). Estrogen mediated mechanism of pain may partly be induced by ERs in the nucleus pulposus of the elderly woman (Song et al., 2017). These reported results along with the current data of DLBP model point out that activation of ERβ is antinociceptive. In another way, the present study demonstrated a neuropathological mechanism underlying DLBP. ERs might be therapeutic targets in DLBP diseases.

From an anatomical point of view, nociceptive neurons activation in specific dorsal root ganglion produces the nociceptive pain signals, which are conveyed to the dorsal horn neurons of the spinal cord and project to the brain. Then it develops pain sensation (Figure 6). It has been reported that ERs distributed widely throughout the central nervous system (Shughrue and Merchenthaler, 2001). Modulation of pain by estrogen has been implicated by an overwhelming number of studies, and ERs’ effects in mediating estrogen function also have been disclosed (Chen et al., 2021; Kaur et al., 2022). The actions of estrogen are mediated by two classical ER isoforms, ERα and ERβ (Kumar et al., 1987; Gegenhuber et al., 2022). ERα and ERβ may independently exert specific effects on pain modulation. ERβ has been shown to be the preponderant receptor partaken in the spinal cord dorsal horn’s development. ERβ was detected in laminae I–III of the spinal dorsal horn, as indicated in our study. Furthermore, ERα levels were much lower than ERβ levels (Fan et al., 2007). ERβ agonists were effective in reducing chemotherapy pain (Ma et al., 2016). In spinal nerve ligation models, administration of a selective ERβ agonist effectively alleviated allodynia (Piu et al., 2008). In the present study, changes of the pain assessment of axial back pain and radiating discomfort were documented after IVD surgery, which indicated that IVD degeneration could trigger DLBP in mice. Meanwhile, we observed that ERβ agonist DPN could improve both axial and radiating signs of DLBP. At the same time, we also observed that ERβ could regulate substance P, a neuropeptide most known for its role in pain perception, in the spinal cord and dorsal root ganglia. The current results support the assumption that, in the central nervous system, ERβ may play a significant part in the nociceptive sensory processing of DLBP.

There are some limitations in the study. First, only female mice were used in the current study. The possibility of ER β/substance P signaling in the spinal cord mediating antinociceptive effect may also exist in male mice, and further research should be needed. Second, lumbar disc degeneration is a normal aging process in human. IVD injury in mice may not be able to imitate that in humans fully. Third, several pain assessment tests were used to measure DLBP in our study and the literature published in recent years. However, they may still lack specificity for DLBP. More reliable and direct pain assessments should be developed and applied in the future. Meanwhile, we plan to investigate the mechanisms underlying specific activation of ERβ induced modulation of pain transmission in more detail in future studies.