**3. Results**

### *3.1. Effects of SCI, TUDCA, bmSC and TUDCA/bmSC Combinatorial Treatment on the General Health Status and Body Weight of the Animals*

The general health of the rats was not affected by either treatment with TUDCA alone or in combination with human bmSC, and no adverse effects such as sickness behavior or urinary infections were observed. Five weeks after SCI, two animals, one treated with 5 × 300 mg/mL TUDCA and one in the saline control group, had wounds on their flanks due to biting and were treated topically with antibiotic ointment. During the first four days following SCI surgery, the animals lost 9–11% of their weight, which they recovered with a weight gain of 9.5% in the second week (W1–2), subsequently 8.2% (W2–3), 7.3% (W3–4), 4.9% (W4–5), and 2.2% (W5–6). Changes in body weight over time were significant [two-factor ANOVA; effect of time: F (8, 374) = 370, *p* < 0.001; treatment: F (5, 374) = 37, *p* < 0.001], but post hoc tests showed no significant differences in weight loss or gain between the control group and TUDCA-treated (Figure 1b) or bmSC-treated rats (Figure 1c). Laminectomy without SCI caused an initial weight loss of 4%.

### *3.2. Expression of TGR5 after SCI*

In the rat brain, astrocytes and neurons are immunoreactive for the bile acid receptor TGR5 [12]. We confirmed expression in the spinal cord with quantitative RT-PCR and found the level of TGR5 mRNA at 0.6 ± 0.2 ‰ (mean ± SD, n = 5) compared to the ribosomal gene 36B4, which was approximately the same as in the cerebral cortex (0.7 ± 0.2 ‰). It did not change significantly after SCI with or without treatment.

### *3.3. Faster Recovery of Bladder Control with TUDCA Treatment*

Normal micturition requires coordinated activation of the detrusor muscle of the bladder and relaxation of the external urethral sphincter, which are controlled by spinal and supraspinal centers [28]. Since these connections were affected by the spinal cord contusion, the animals with SCI were unable to urinate spontaneously and needed manual assistance with bladder voiding during the first days after surgery. Laminectomy alone did not cause urinary retention. The volume of manually expelled urine was evaluated to assess the recovery of bladder control (Figure 2). One positive effect of TUDCA treatment, not reported before, was that recovery of this function occurred faster than in rats with saline injection (Figure 2a). The total volume of urine retained during the post-acute period was not significantly different between groups (Figure 2b), but the average duration of compromised bladder control was shorter for animals treated with TUDCA (Figure 2c). The recovery of the areflexive bladder in the group treated with additional bmSC was similar to that of the group treated with TUDCA only. Eventually, however, all rats recovered autonomic control of micturition.

### *3.4. Effect of TUDCA and bmSC Treatment on Allodynia/Hyperalgesia*

Spinal cord injury can lead to neuropathic pain. This was assessed at five weeks after SCI by determining the withdrawal threshold (PWT) to mechanical stimulation of the hind paws using an automated von Frey test. Although we found no significant differences between experimental groups (Figure 2d; ANOVA, F (6, 82) = 2.5, *p* < 0.05, put post-hoc Dunnett's test vs. SCI n.s. for all groups), the number of rats that had a PWT below 20 g, which we considered indicative of allodynia or hyperalgesia, was higher after SCI than after the laminectomy operation. This measure of pain sensitivity appeared to be worse with high doses of TUDCA and lower with treatment of bmSC alone (Figure 2e), but variability and group size do not permit a conclusive interpretation.

### *3.5. Effect of TUDCA and bmSC Treatment on Recovery of Sensory-Motor Functions*

One day after SCI, the rats' ability to use their hind legs was assessed in the open field. The SCI caused paralysis, as indicated by no or only slight movement of joints. Laminectomy without contusion resulted in temporary gait instability in some cases. Two TUDCA-treated SCI animals, which presented a BBB score above 2 at 1 dpo, were excluded from the analysis because we considered this an indication of a less severe SCI rather than an effect of treatment (indicated in Figure 1a). With time after injury, sensory-motor functions of the rats improved significantly, and different treatments were effective [two-factor ANOVA, time after SCI: F (9, 470) = 121, *p* < 0.0001 (pre-SCI data not included); treatment effect: F (5, 470) = 3.1, *p* < 0.01 (laminectomy group not included); interaction: F (45, 470) = 0.7, n.s.]. During the first 4 dpo, the majority of treated animals recovered the ability to move their hind legs. After treatment with 300 mg/kg TUDCA, this recovery occurred faster, such that BBB scores were significantly higher at 2 and 4 dpo compared to saline treatment (Figure 3a,b, subacute phase of these groups expanded in c,d). Improvements after two injections of 100 mg/kg TUDCA were not distinguishable from results after saline injections. Recovery of motor function improved further between one and three weeks after lesion, when all rats showed plantar stepping and weight support at least in stance but usually no coordination between fore and hind limbs (BBB 9–11). No improvement beyond this level was observed in the chronic phase (Figure 3a,b). Combinatorial therapy with TUDCA and bmSC or bmSC alone achieved higher BBB scores

at 1–6 weeks (ΔBBB 0.7–1.9 higher than SCI-control, ΔBBB 0.5–1.2 higher than SCI-T200; *p* < 0.05), but there were no significant differences between treatment with bmSC and combinatorial treatment. The experiment confirmed previous results with bmSC injection only [8].

**Figure 2.** Autonomic functions and neuropathic pain. (**a**) Percentage of rats that required manual bladder voiding each day in the post treatment period. Bladders were voided manually every 12 h. (**b**) Average volume of retained urine per day and rat. Data shown here are for the SCI rats treated with saline and with five injections of TUDCA (mean ± SEM). (**c**) Time after SCI that passed until the animals no longer required manual voiding of the bladder [ANOVA, F (5, 43) = 2.7, *p* < 0.05; post hoc Dunnett's test vs. SCI control \* *p* < 0.05]. (**d**) To assess neuropathic pain we measured the PWT to mechanical stimulation (von Frey test; median ± 25–75 percentile, range; ANOVA n.s.). (**e**) Percentage of rats that had a PWT of below 20 g, which was considered as an indication of allodynia. Experimental groups are abbreviated as follows. SCI-control: two injections of saline; SCI-T200: two injections of 100 mg/kg TUDCA at t0 and 24 h later; SCI-T600: two injections of 300 mg/kg TUDCA at t0, 24 h; SCI-T1500: five injections of 300 mg/kg TUDCA at t0, 24 h, 2 dpo, 4 dpo and 6 dpo; SCI-bmSC: one injection of bone marrow-derived stromal cells at t0 + 2 h; SCI-T + bmSC: combinatorial treatment with bmSC and two injections of 100 mg/kg TUDCA; sham: laminectomy only.

In addition to the assessment in the open field, rats were subjected to the Rotarod test, which measures their ability to maintain equilibrium on a rotating bar. Before SCI, all rats had been trained to perform this task for at least 300 s, and at 4 dpo, none of the animals that met the BBB inclusion criterion was able to do so. Spontaneous recovery caused a significant increase in Rotarod score during the first four weeks in all animals (Figure 4a,b). No further improvement occurred subsequently [ANOVA, effect of time after SCI: F (3, 352) = 22.0, *p* < 0.001; treatment effect: F (5, 352) = 1.5, n.s., interaction: F (15, 352) = 0.8, n.s.]. Some animals which showed weight supported plantar steps in the open field refused to hold on to the bar. In the absence of an independent criterion to distinguish between voluntary refusal and inability to perform the task, no data were excluded from the evaluation, resulting in a high variability in this assay. An additional assessment using the percentage of rats that at 6 W maintained themselves for longer than 30 s on the Rotarod showed no improvement at all following TUDCA treatment but revealed that more of the rats in the groups treated with bmSC or TUDCA + bmSC were able to do this (Figure 4c). With eight animals per group, the statistical power did not suffice for this effect to be significant.

### *3.6. Effect of TUDCA Treatment on Neuroinflammation*

To a large degree, the devastating effects of SCI are due to the neuroinflammatory response of microglia and blood-derived macrophages. Since TUDCA is known for its anti-inflammatory effect on microglia, we confirmed this for our study by measuring the expression of marker genes in the spinal cord at 4 dpo. This experiment was done for the high dose of TUDCA. As in previous experiments, SCI caused mRNA upregulation of the inflammatory cytokine IL-6 and chemokine CCL-2 in spinal cord extracts. This appeared to be a local response and was not found in the cerebral cortex (Figure 5a,b). Treatment with TUDCA significantly reduced the expression of these and other (Table S1) inflammatory genes in the spinal cord. The lesion also caused alternative activation of microglia/macrophages, which was demonstrated by a highly significant increase in transcription of arginase-1 and IL-4R (Figure 5c,d). Contrary to our expectation, the bile acid treatment also reduced this response, indicating that it did not promote differentiation of an M2 phenotype. The cytokine IL-10 promotes alternative activation of macrophages and is produced by a subpopulation of these cells. At 4 dpo, we found a significant reduction of IL-10 transcripts in the spinal cord. This also occurred after TUDCA treatment but, compared to the SCI-controls, was no longer significant (Figure 5e).

The inflammatory activation of microglia or macrophages was also reflected by the expression of the complement receptor 3A (integrin αM, CD11b), which increased 7-fold after SCI. Similar to the other pro-inflammatory markers, its expression was significantly reduced after treatment with TUDCA (Figure 6a). Platelet endothelial cell adhesion molecule-1 (CD31) is found on endothelial cells, macrophages, and various lymphocytes. Its expression was detected in the control tissue, increased after SCI, and also was much reduced after TUDCA treatment (Figure 6b). Activation of astrocytes in neuropathologies is associated by an upregulation of the glial fibrillary acidic protein (GFAP). Transcripts of this gene, which was highly expressed in the spinal cord, increased 4-fold after SCI, and the response was inhibited by bile acid injections (Figure 6c). The validity of these effects on gene regulation can be appreciated by comparison with marker genes of other cell types, such as B lymphocytes (CD20; Figure 6d) and T lymphocytes (CD3ζ; Figure 6e). Indicators of these cells did not significantly change at 4 dpo with or without TUDCA injections. A marker gene of regulatory T-cells (FoxP3; Figure 6f) was expressed at a lower level after SCI and not influenced by treatment.

**Figure 3.** Recovery of motor functions after SCI assessed in the open field. Mean BBB scores were monitored before surgery, daily during the first 4 dpo after surgery, and then once per week during 6 weeks of evaluation. (**a**) Results for SCI-control, treated with saline, and the TUDCA treated groups. (**b**) Results for SCI-control and the SCI groups treated with bmSC and TUDCA + bmSC. (**<sup>c</sup>**,**d**) Changes in the acute/subacute phase; effects of treatment with 2 × 300 mg/kg TUDCA or with bmSC were significant at 2 dpo and/or 4 dpo [see main text for statistical evaluation; \* (SCI-T600), + (SCI-T1500), # (SCT-bmSC) indicate *p* < 0.05; error bars indicate SEM]. Recovery of motor function improved in all treatment groups until a plateau was reached after three weeks; experimental groups are indicated as for Figure 2.

The lesion-induced activation of microglia and astrocytes was confirmed by microscopical inspection using double-staining IF for Iba-1, CD68, and GFAP (Figures 7 and 8). As expected from previous experiments, SCI disrupted the blood spinal cord barrier, caused an influx of hematogenous macrophages, and activated resident microglia and astrocytes. At 4 dpo and 6 weeks, under all treatment conditions, the center of the lesion was filled with cellular debris and Iba-1/CD68 positive cells. These made up 8.3 ± 1.1% of the cells in the center of the lesion (ROI 0.075 mm2; Figure 7a–c). The majority of CD68 positive cells were likely to be of hematogenous origin, and these were absent in the spinal cords of animals that had received only the laminectomy. Already at 4 dpo, the lesion center was devoid of neurons (NeuN IR) and astrocytes (GFAP IR). After TUDCA treatment, 5.7 ± 1.0% of the cells in this area were macrophages, as identified by morphology and CD68 IR. With increasing distance from the lesion center, the number of CD68 positive cells decreased such that at 8 mm posterior and anterior distances, only a few macrophages were observed in the SCI animals. Therefore, distance to the lesion significantly affected the strength of the CD68 signal, and a treatment effect was only significant at the central location (see quantification below).

**Figure 4.** Recovery of motor functions after SCI assessed with the Rotarod assay. At 4 dpo, we confirmed that no SCI treated animals showed weight supported steps. Beginning at 7 dpo, some rats in all groups showed gradual improvement. (**<sup>a</sup>**,**b**) Time that rats were able to keep their balance on the rotating bar; mean +/− SEM, no significant differences between groups were detected. (**c**) Percentage of rats that performed the Rotarod task for more than 30 s at 6 W after SCI (n = 7–8 rats/group). Experimental groups are indicated as in Figure 2.

**Figure 5.** Expression of marker genes of inflammation. At 4 dpo, RNA extracts from spinal cord and cerebral cortex were analyzed with quantitative RT-PCR (treatment conditions abbreviated as in Figure 2). (**a**) Gene expression of IL-6 [ANOVA, spinal cord: F (2, 12) = 71.6, *p* < 0.0001; cortex: n.s.]; (**b**) CCL-2 [ANOVA, spinal cord: F (2, 12) = 8.9, *p* < 0.01; cortex: n.s.]; (**c**) arginase-1 [ANOVA, F (2, 12) = 56.0, *p* < 0.0001], (**d**) IL-4Rα [ANOVA, F (2, 12) = 66.0, *p* < 0.0001], (**e**) IL-10 [ANOVA, F (2, 12) = 18.7, *p* < 0.001]. Significant differences detected with post hoc Dunnett's tests (*p* < 0.05, *p* < 0.001, *p* < 0.001) are indicated with \*\*, \*\*\* symbols for SCI vs. laminectomy and #, ### for SCI vs. SCI-T600.

### *3.7. Effect of TUDCA Treatment on Glial Activation after SCI*

Increasing the IR of GFAP (Figure 8a–c,g–i) and Iba-1 (Figure 8d–i) indicated activation of astrocytes and microglia at 4 dpo. In the anterior and posterior of the SCI site, both signals were stronger than in animals that had only received the laminectomy. Contusion injury caused a 15-fold increase in Iba-1 IR in and close to the lesion center compared to levels in sham-operated animals. After treatment with 2 × 300 mg/kg TUDCA, Iba-1 and CD68 signals were significantly reduced in the lesion center (Figure 9a,b). Morphological changes of GFAP positive cells around the lesion site indicated the activation of astrocytes at 4 dpo (Figure 8j,k). Quantification of the GFAP IR confirmed this observation for the white matter posterior of the lesion center. Astrocyte activation was less pronounced (i.e., not significantly different from laminectomy treatment) after bile acid treatment (Figure 9c).

At 6 W after SCI, Iba-1 IR was no longer as elevated as at 4 dpo but still 3- to 4- fold higher than in non-lesioned tissue. No significant differences between TUDCA-, bmSC- or TUDCA/bmSC-treated and saline-treated animals were observed in the chronic phase (Figure 9d). At this time, a prominent glial scar had formed in all SCI animals. Quantification of GFAP IR in this area (Figure 9e) confirmed a strong increase from the acute to the chronic phase after SCI (note different scales of *y*-axis in panels Figure 9c,e). The scar area appeared to be reduced by bile acid/bmSC treatment, but the differences in GFAP intensity did not reach significance.

Responding to an inflammatory environment, microglia cells are known to respond with a morphological transformation from a branched appearance to an ameboid shape. We confirmed this using a Sholl analysis of Iba-1 IR cells in gray matter at an 8 mm distance from the lesion center (Figure 10a–c). The morphological change associated with SCI was highly significant. Microglia cells in SCI injured animals at 4 dpo had fewer branches

than in rats with laminectomy. This change was less pronounced after TUDCA treatment (*p* < 0.001), corroborating the hypothesis that the bile acid affected microglial activation.

**Figure 6.** Expression of marker genes of cell type activation. At 4 dpo, RNA extracts from the spinal cord and cerebral cortex were analyzed with quantitative RT-PCR (treatment conditions abbreviated as in Figure 2). (**a**) Gene expression of CD11b, marker of microglia and macrophage [ANOVA, F (2, 12) = 46.3, *p* < 0.0001]; (**b**) CD31, endothelial cells, macrophages and lymphocytes [ANOVA, F (2, 12) = 23.9, *p* < 0.0001]; (**c**) GFAP, astrocytes [ANOVA, F (2, 12) = 8.6, *p* < 0.01]; (**d**) CD20, B-cells [ANOVA, F (2, 12) = 1.6, n.s.]; (**e**) CD3ζ, T-cells [ANOVA, F (2, 12) = 2.2, n.s.]; (**f**) FoxP3, regulatory T-cells [ANOVA, F (2, 12) = 7.7, *p* < 0.01]. Significant differences detected with post hoc Dunnett's tests (*p* < 0.05, *p* < 0.001, *p* < 0.001) are indicated with \*\*, \*\*\* symbols for SCI vs. laminectomy and ##, ### for SCI vs. SCI-T600.

**Figure 7.** Activation of macrophages after SCI. (**<sup>a</sup>**–**<sup>c</sup>**) Confocal microscopy images show examples of macrophages in the SCI lesion center at 4 dpo. All cells are IR for Iba-1 (**<sup>a</sup>**, green) and most also for CD68 (**b**, red), shown in combination with nuclear staining (**b**,**c**, Hoechst 33342, blue). (**d**–**f**) CD68 IR macrophages in the spinal cord after laminectomy (**d**, sham), SCI (**e**), and SCI 2 × 300 mg/kg TUDCA treatment (**f**). Same magnifications are used in (**<sup>a</sup>**–**<sup>c</sup>**) and in (**d**–**f**) (scale bars in (**<sup>c</sup>**,**f**)).

### *3.8. Effect of TUDCA Treatment on SCI-Induced Apoptosis*

Several groups reported that TUDCA treatment reduced cell death after SCI [22,24–26]. We found previously that the bmSC preparation used here was also cytoprotective [8]. To allow comparison with these publications, we evaluated cellular apoptosis in the lesion center and at 4 mm anterior and posterior to this position (Figure S1). Four days after SCI, we found 11.3 ± 3.1% (mean ± SD) of the cell nuclei in the lesion area to be TUNEL positive, whereas sham operated rats had only 0.40 ± 0.35% apoptosis in the white matter at the respective position (*p* < 0.001). Treatment with two injections of 300 mg/kg TUDCA significantly reduced TUNEL staining in the lesion area to 5.9 ± 4.2% (*p* < 0.05). In the ventral white matter at positions anterior and posterior of the lesion center, we found a slight, non-significant increase in the number of apoptotic cells at 4 dpo (control: 0.2 ± 0.3%, SCI: 1.1 ± 0.7%), which was not affected by treatment (SCI-T600: 1.1 ± 0.3%). The analysis of TUNEL staining in the chronic phase after SCI revealed continuing high levels of apoptosis without (SCI: 7.6 ± 18.6%) and with TUDCA treatment (SCI-T600: 16.4 ± 17.1%), indicating that bile acid injections had no lasting effect on cell survival (*p* > 0.1). In rats treated with bmSC, the proportion of apoptotic cells at 6 W was lower (SCI-bmSC: 4.1 ± 7.8%, *p* < 0.05), but there was no significant effect due to additional treatment with TUDCA (SCI-T + bmSC: 1.4 ± 1.0%, *p* < 0.05 vs. SCI; *p* > 0.1 vs. SCI-bmSC).

**Figure 8.** Activation of microglia and astrocytes after SCI. The IF photographs show examples of grey matter at 4 dpo and a distance of ca. 0.4 mm from the lesion center, which were double-stained for astrocytes (GFAP; **<sup>a</sup>**–**<sup>c</sup>**) and microglia (Iba-1; **d**–**f**). Photographs were superimposed with additional nuclear staining (Hoechst 33342; **g**–**i**). (**<sup>a</sup>**,**d**,**g**) Section from laminectomy-operated animal. (**b**,**e**,**h**) SCI with saline injection; (**<sup>c</sup>**,**f**,**i**) SCI treated with two injections of TUDCA 300 mg/kg. (**j**,**k**) At 4 dpo under all SCI treatment conditions, astrocytes showed increased GFAP IR but a glial scar had not ye<sup>t</sup> developed. Scale = 50 μm in (**i**) (for (**<sup>a</sup>**–**i**), 40× objective) and 100 μm in (**j**) (for (**j**,**k**), 20× objective). Macrophages fill the lesion center.

**Figure 9.** Activation of macrophages, microglia and astrocytes after SCI. Data from spinal cord areas were normalized for IR in the anterior position of sham-operated animals and analyzed with 2-factor ANOVA followed by post hoc Sidak tests. (**a**) Quantification of Iba-1 IR (fluorescence integrated density) at 4 dpo in white matter at 8 mm anterior, 8 mm posterior of the lesion site and in the lesion center [2-factor ANOVA, location effect: F (2, 27) = 11.7, *p* < 0.01; treatment effect: F (2, 27) = 29.9, *p* < 0.001; interaction: F (4, 27) = 3.2, *p* < 0.05]. (**b**) Quantification of CD68 IR at 4 dpo in areas 8 mm anterior and posterior of the lesion site and in the lesion center [2-factor ANOVA, location effect: F (2, 27) = 6.7, *p* < 0.01; treatment effect: F (2, 27) = 3.3, *p* = 0.05; interaction: *p* < 0.05; data were normalized to SCI, lesion center, as almost no CD68 cells are found outside this area]. (**c**) Quantification of GFAP at 4 dpo in white matter at 8 mm anterior, 8 mm posterior of the lesion site [location effect: F (1, 18) = 14.2, *p* < 0.01; treatment effect: F (2, 18) = 6.6, *p* < 0.01; interaction: F (2, 18) = 3.4, n.s.]. (**d**) Quantification of Iba-1 IR at 6 W in white matter 8 mm anterior, posterior and in the lesion center site [location effect: F (2, 98) = 5.9, *p* < 0.01; treatment effect: F (4, 98) = 5.4, *p* < 0.001; interaction: F (8, 98) = 0.3, n.s.]. (**e**) Quantification of GFAP IR at 6 W in white matter 8 mm anterior and posterior of the lesion center and in the glial scar [location effect: F (2, 97) = 36.1, *p* < 0.001; treatment effect: F (4, 97) = 4.1, *p* < 0.01; interaction: F (8, 97) = 2.5, *p* < 0.001]. Significant differences detected with post hoc comparisons tests (*p* < 0.05, *p* < 0.001, *p* < 0.001) are indicated with \*, \*\*, \*\*\* for SCI vs. control and # SCI vs. SCI-T600].

**Figure 10.** Effect of SCI and TUDCA treatment on microglia morphology. (**<sup>a</sup>**–**<sup>c</sup>**) Examples of Iba-1 IR cells in the spinal cord gray matter at 8 mm distance from the injury site at 4 days after sham operation (**a**), SCI control (**b**) and SCI with TUDCA treatment (**c**) are shown with superimposed rings for Sholl analysis. (**d**) Number of intersections of cellular processes at 5 μm intervals from the cell soma [two-factor ANOVA, distance from soma F (7, 335) = 78.6, *p* < 0.0001; treatment effect: F (2, 335) = 107, *p* < 0.001; interaction: F (14, 335) = 12.9, *p* < 0.001, results of Bonferroni multiple comparison tests are indicated with \*\*\* (*p* < 0.001) for sham vs. SCI-control and sham vs. SCI-T600 and with ## (*p* < 0.01) for SCI vs. SCI-T600]. All panels have the same magnification (scale bar in **b**).
