Mouse Model of Spinal Cord Hypoperfusion with Immediate Paralysis Caused by Endovascular Repair of Thoracic Aortic Aneurysm

Adult C57BL/6 mice (age range between 16 and 20 weeks for both males and females) were purchased from Jackson Laboratories (USA). The mice were randomly assigned to the sham and ligation groups in each of the pilot and final configurations: pilot using three double ligations (sham, n = 5; ligation, n = 5); pilot using five double ligations (sham, n = 17; ligation, n = 22); pilot using six double ligations (sham, n = 2; ligation, n = 2); and final configuration using three double ligations with two single ligations (sham, n = 48; ligation, n = 55). In the final configuration, one sham mouse was excluded from the study due to a hind limb injury unrelated to the surgery. Supplemental Digital Content 1 (https://links.lww.com/ALN/D36) illustrates the experimental design animal tree with the total number of animals and the detailed number of animals in each group. Neither body weight nor distribution of each sex differed significantly across the groups. All male mice (n = 58 of 58) weighed more than 23 g; 59% (n = 31 of 53) of the female mice weighed 23 g or less, and 42% (n = 22 of 53) weighed more than 23 g. The mice were randomly chosen for either the sham or ligation surgery using the procedure described under Random Selection and Blind Testing.

All procedures used aseptic technique, and all mice were housed in HEPA-filtered Bio-clean units. All procedures complied with the National Institute of Health Guide for the Care and Use of Laboratory Animals and were approved by the Ohio State University’s Animal Care and Use Committee (approval No. 2007A0195).

The core temperature for all mice was maintained between 33.0 and 33.5°C on a temperature-controlled surgical platform (World Precision Instruments, USA; see Supplemental Digital Content 2, https://links.lww.com/ALN/D37, discussing the choice of core body temperature during anesthesia preparation). The mice were anesthetized using 4% isoflurane (100 ml/min O₂) in 96% O₂ for induction and then maintained at 2% isoflurane (100 ml/min O₂) in 98% O₂ throughout all surgical procedures. Two layers of disinfection, first with liquid surgical scrub and second with 70% ethanol, were performed to clean the surgical site after shaving the hair. A lubricated mouse tracheal intubation cannula was inserted into the trachea (Hugo Sachs VK32) through the mouth after exposing the trachea with a vertical ventral midline incision in the neck extending slightly past the ear pinna. Then the larynx and trachea were exposed by retracting the overlaying fat, muscle, and submaxillary gland. A mouse ventilator (Hugo Sachs–Harvard Apparatus Minivent, USA; tidal/stroke volume, 250 µL; rate, 230 ventilations/min) was used for ventilation. The throat incision was closed with surgical glue and a single suture, and the tracheal cannula was fixed in place to the lip of the mouse with a surgical clip. Then the mouse was injected subcutaneously at the site of the incision with bupivacaine (0.1 mg/mouse in 0.02 ml) and gentamycin (0.1 mg/mouse in 0.1 ml). Then 5 to 10 min later, the chest wall was opened.¹² 

Intubated mice were placed in a right lateral position, and the left forelimb was extended beneath the mandible and secured to the surgical platform with adhesive tape to expose the lateral thoracic cage beneath the left scapula. A small transverse incision was made underneath the left shoulder; then a blunt dissection of the subcutaneous fat to expose the underlying rib cage was performed. Using scissors, skin was cut longitudinally (rostral to caudal) to the lower costal cartilage; then the fat overlying the rib cage was bluntly dissected and moved to either side, exposing the entire left thoracic cage. The intercostal muscles between the 8th and 9th ribs were cut with scissors, exposing the lateral pleura. With the incision wide open, the lower lobe of the lung was pushed out of the field using retractors while the skin and fat on each side were held away from the incision site using lateral hooks. The subcutaneous fat was kept moist throughout the procedure using normal saline. Using forceps while viewing through a 10× power microscope lens, the fully exposed descending aorta was bluntly dissected in a rostral to caudal direction while carefully exposing the intercostal arteries (8th to 10th). A second incision was made in the muscle between the 6th and 7th rib, and retractors were used to hold it wide open and to push the lungs away from the field. While viewing through a 16× power lens, the exposed descending aorta was bluntly dissected a second time in a rostral to caudal direction to increase the exposure and visibility of the intercostal arteries (6th through 10th).

Continuing with the 16× power lens, the intercostal arteries were sutured (ligated) bilaterally using 9-0 nonabsorbable nylon sutures (ETHILON) starting with the lower-most vessel (10th intercostal pair) and ending with the 6th intercostal pair; the 8th, 9th, and 10th artery pairs were double ligated, and the 6th and 7th pairs were single ligated. This configuration was chosen after conducting a pilot study where we first tried double ligating the 6th through the 11th (in the six double-ligation group), the 6th through the 10th (in the five double ligation group), and the 8th through the 10th (in the three double ligation group; see Supplemental Digital Content 3, https://links.lww.com/ALN/D38, discussing the rationale for the use of double ligation). The rib cage and intercostal muscles were then sutured at the two incision sites using 6-0 polypropylene after absorbing any fluid or blood leak at the incision site. Finally, the subcutaneous fat was put back after being moistened, and the skin was sutured using 5-0 polypropylene sutures. Figure 1A illustrates the intercostal ligation and ligature. Sham control mice underwent the exact same surgical procedures without ligation of the intercostal arteries.

All animals recovered spontaneously from anesthesia within 10 to 20 min on the surgical platform while their core temperatures were maintained. The mice were extubated and placed into new clean cages maintained on a warmer set at 34.5°C for 2 to 5 days. Bladder care by a gentle manual evacuation of the bladder was performed every 12 h for the duration of the experiment to prevent urine retention and infection. The mice were given 0.5 ml of 10% dextrose in water subcutaneously twice daily and were also given prophylactic antibiotics—gentamycin (0.1 mg/mouse in 0.1 ml) and Baytril (0.2 mg/mouse in 0.2 ml) twice daily subcutaneously through the first 7 days postsurgery. The mice were kept on a regular diet supplemented by a Stat high caloric diet (PRN Pharmaceutical, USA) throughout the study. Animal welfare, postoperative pain control, and humane endpoints were ensured per the protocols established by the Animal Care and Use Committee at the Ohio State University (see Supplemental Digital Content 4, https://links.lww.com/ALN/D39, discussing postsurgery pain management and postoperative pain control agents, concentration, dose, and frequency).

Tissue perfusion data were obtained from five male mice using a Transonic Type N24 needle probe connected to a BLF22 laser Doppler system (Transonic Systems Inc., USA). Twenty-four hours before measuring tissue perfusion, the mice underwent dorsal laminectomy of T8 to T10, the skin was stapled, and the mice were left to recover after an injection of gentamycin (0.1 mg/mouse in 0.1 ml). The next day, the mice were anesthetized with 2% isoflurane in 100% oxygen, the spinal cord was exposed, and the laser Doppler probe was stabilized over a marked location on the spinal cord using a micromanipulator (fig. 1B). Spinal cord perfusion measurements were obtained before ligation. The mice were then placed in a left lateral position, the chest was opened, and the intercostal arteries were surgically ligated (as discussed supra under Surgical Procedures).

Intercostal artery blood flow measurements were taken before and after ligation by applying the laser Doppler probe tip on the surface of the vessels. The mouse was then switched back to the prone position, and the tip of the laser Doppler probe was placed on the spinal cord at the same marked location to acquire postligation spinal cord tissue perfusion measurements. In accordance with the protocol approved by the Ohio State University Institutional Lab Animal Care and Use Committee, all mice were sacrificed at the end of the tissue perfusion measurement. Powerlab data acquisition system and LabChart software (ADInstruments Inc., USA) were used to record the output signals from the laser Doppler system with output signals set to the tissue perfusion flow unit (mL · min⁻¹ · 100 g tissue⁻¹) and blood velocity unit (m · s⁻¹ · 100 g tissue⁻¹).

Several factors were addressed before use of the laser Doppler tissue perfusion probe. The probe was immobilized during the entire measurement to ensure proper measurement of reflected light. This was ensured by surgical implantation of the probe via micromanipulator. A mark for Doppler measurement location ensured consistent pre- and postligation probe placement. Care was taken to avoid arterial occlusion by the probe, and duplicate readings were taken to ensure consistent measurement upon placement. Additionally, the same ambient room lighting was maintained during the experiment to prevent confounding the results.

Bilateral hindlimb function was blindly assessed with the Basso Mouse Scale for Locomotion¹⁹  at 6 h; 1, 2, 3, 5, and 8 days; and 2 weeks after injury (see Random Selection and Blind Testing for a discussion of the blind procedure used for the behavioral assessment). The Basso Mouse Scale, a 10-point scale (0 to 9) with operational definitions of key locomotor features, quantifies the rate and extent of functional impairments in mice with neurovascular injury¹⁹  (see Supplemental Digital Content 5, https://links.lww.com/ALN/D40, explaining the Basso Mouse Scale assessment method).

The mice were anesthetized with a mixture of 120 to 150 mg/kg ketamine (which is 1.5 times the regular dose of ketamine for sedation) and 5 to 10 mg/kg of xylazine (which is the regular dose for sedation) and then transcardially perfused with 25 ml of 0.1 M phosphate-buffered saline, followed by 50 ml of 10% formalin (see Random Selection and Blind Testing for a discussion of the randomization procedure used to select mice for tissue harvesting). Spinal cords were removed and postfixed in 10% formalin for 5 days.

Spinal cords from an equal number of male and female mice subjected to sham surgery or surgical ligation were collected at 2 days (sham, n = 6; ligated, n = 6) and 8 days (sham, n = 6; ligated, n = 6). The Comparative Pathology and Digital Imaging Shared Resource of the Ohio State University Comprehensive Cancer Center performed all pathology procedures. An experienced prosection technician trimmed spinal cords in the area of experimental interest (lower thoracic and upper lumbar segments where the hypoperfusion was expected) to identify regions of ischemia. Additional spinal cord tissue immediately proximal to the areas of expected ischemia was also trimmed and processed. Spinal cords were further postfixed for 48 h in 10% formalin. The tissues were routinely processed for histopathology on a Leica Peloris 3 tissue processor (Leica Biosystems, USA) and embedded in paraffin. The spinal cords were sectioned at an approximate thickness of 4 to 5 μm to produce multiple cross-sections per region of interest and batch stained with hematoxylin and eosin on a Leica ST5020 autostainer (Leica Biosystems) using a routine and quality-controlled protocol.

Immunohistochemistry to detect the presence of astrocytes positive for glial fibrillary acidic protein (Dako product No. Z0334, Agilent, USA; 1:5,000 = 0.58 μg/mL), microglia positive for ionized calcium binding adaptor molecule 1 (Iba1, product No. NB100-1028, Novus Biologicals, USA; 1:1,000 = 0.5 μg/mL), oligodendrocytes positive for oligodendrocyte transcription factor (Olig2, product No. 109186, Abcam, USA; 1:400 = 0.325 μg/mL), and apoptotic cells and bodies positive for cleaved caspase-3 (product No. 9661, Cell Signaling Technologies, USA; 1:180 = 0.26 μg/mL) was performed using a Lab Vision 360 automatic immunostainer (Thermo Scientific, USA) with optimized and quality-controlled protocols specific to each primary antibody. All immunohistochemistry procedures followed optimized, validated protocols for paraffin-embedded tissues with confirmed quality control for antibody specificity, including positive, negative, no primary, and isotype controls for all antibodies.

All slides were evaluated by an American College of Veterinary Pathologists board-certified comparative pathologist (Dr. Corps) using a Nikon Eclipse Ci-L upright microscope (Nikon Instruments, Inc., USA). Representative photomicrographs were taken using an 18-megapixel Olympus SC180 microscope-mounted digital camera and cellSens imaging software (Olympus Life Science, USA).

Fresh spinal cord samples (n = 3) were fixed for a minimum of 2 h at room temperature using 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, in a volume of fixative that was greater than 10 times the total volume of the tissue sample being fixed. The samples were stained using a LYNX tissue processor (Electron Microscopy Sciences, USA) according to the following protocol: rinsed three times for 5 min each in 0.1 M phosphate buffer, stained for 1 h in 1% osmium tetroxide in 0.1 M phosphate buffer, rinsed three times in distilled water, stained in aqueous 1% uranyl acetate for 1 h, rinsed three times in distilled water, and then dehydrated in an ethanol series (50%, 70%, 80%, 95%, 95%, 100%, 100%, and 100%) in which each step from 50 to 95% was 10 min long, and each 100% step was 15 min.²⁰  The samples were then changed to acetone for 10 min, and the resin infiltration series began according to the following schedule: 1:1 acetone:resin mix without accelerant for 1 h, 1:2 acetone:resin mix without accelerant for 1 h, two changes of 100% resin mix with accelerant over 2 to 6 h. The resin mix used was mixed according to the manufacturer’s recommendations using the Eponate 12 kit (Ted Pella, Inc., USA).

The samples were transferred into silicone molds containing freshly made resin mix and cured for 24 to 48 h at 60°C. The samples were then sectioned using an LC7 ultramicrotome (Leica, Germany) at 90-nm thickness and poststained using aqueous 1% uranyl acetate for 3 min, rinsed in water, and stained with Reynold’s lead citrate²¹  for 2 min. The samples were then rinsed in water a second time and imaged using a Spirit G2 Biotwin transmission electron microscope (FEI, USA).

To study changes in the blood pressure and blood gases during surgery, 15 adult C57BL/6 mice were randomly assigned to three groups (baseline group, n = 5; ligation group, n = 5; and sham group, n = 5), with three males and two females in each group. The baseline group underwent intubation and ventilation using the same anesthesia procedure and the same ventilator settings as the ligation and sham groups but did not receive any surgical intervention.

A CODA high-throughput noninvasive blood pressure system (product No. CODA-HT8, Kent Scientific Corporation, USA) was used with the following settings: maximum occlusion pressure of 250 mmHg; deflation time at 20 s; minimum volume at 15 μl; display style “One Channel per Graph”; acclimation cycles at 5; number of sets at 1; time between sets at 30 s; cycles/set at 20; and time between cycles at 5 s.

A CODA dual-channel blood pressure system (Kent Scientific Corporation) was used with the following setting: maximum occlusion pressure at 250 mmHg and deflation time at 20 s. The animals were fixed on the heating surgery stage set at 33.5°C. Before the start of surgery, 5 to 15 acclimation cycles were performed until reaching tail volume of 15 µl or more. During the procedure, the room temperature was stable (20 to 23°C), and the experiment took place well away from air conditioning vents or fans. Eight intraoperative blood pressure readings for each mouse were taken: two readings at the beginning of the experiment (after intubation and ventilation in the baseline group; after opening the chest in the ligation and sham groups); four readings midexperiment (1 h after intubation and ventilation in the baseline group; during ligation of the intercostal vessels in the ligation group; and 30 min after opening the chest in the sham group); and two readings at the end of the experiment (90 min after intubation and ventilation in the baseline group; after complete ligation of the intercostal vessels in the ligation group; and 1 h after opening the chest in the sham group). An example intraoperative blood pressure reading from this experiment is shown in Supplemental Digital Content 6 (https://links.lww.com/ALN/D41).

At the end of the experiment and after recording the last two blood pressure readings, a cardiac puncture was performed by advancing a 25-gauge needle into the apex of the heart for blood sampling. Thus, 0.2 ml of blood was collected and used to measure pH, Pao₂, Paco₂, bicarbonate concentration, hematocrit, and hemoglobin concentration using an Abbott i-Stat hematology blood gas analyzer.

All mice were subjected to preoperative behavioral assessment using the Basso Mouse Scale 5 days after arrival to the Ohio State University from Jackson Laboratory. Any mouse with preoperative abnormal behavior was not included in the study. Then mice were randomly given numbers, and all mice were subjected to the same preoperative care.

Each surgery day, four mice were randomly selected to undergo surgery as follows: one cage of females and one cage of males were taken to the surgery station. Then one mouse (of either sex) was randomly selected and then underwent surgery (ligation or sham), and then another mouse of the opposite sex was subjected to the same surgery. This pattern was then repeated for two additional mice to receive the other type of surgery, such that four mice were operated upon on the same day under the same conditions to have two ligated mice (one male and one female) and two sham mice (one male and one female).

The behavioral assessments were performed blindly. Before the performance of each behavioral assessment, all living mice to be tested that date (ligated and sham) at all time points (e.g., 1 day, 3 days, 1 week, and so forth) were randomly assigned alphabetical letters (A, B, C, and so forth). The behavioral assessments were performed, and Basso Mouse Scale scores were assigned with the observers blind and unaware of the status of each mouse. Once all assessments were complete at the end of the testing session, the blinding was resolved by matching each mouse’s number to the randomized letter and the corresponding Basso Mouse Scale score.

A random selection process was followed when sacrificing the mice and for tissue harvesting and processing. Histopathology, electron microscopy, and statistical analysis were each performed by different team members that were blindly given the samples labeled as “group A” and “group B,” and were therefore unaware of which group was the sham or the ligation.

For surgical results, categorical variables are summarized as frequencies (percentages) and compared between groups using chi-square tests or Fisher’s exact tests where relevant. Continuous variables are summarized as means (SD) and compared between groups using Student’s independent t tests. Mouse blood flow and velocity changes were compared between time points using paired t tests. Overall survival was compared between study groups using a Kaplan–Meier survival plot and log-rank test. Hypothesis testing was two-sided, and P values <0.05 were considered statistically significant. SAS version 9.4 (SAS Institute, USA) was used to conduct the statistical analysis.

For the Basso Mouse Scale data, the statistics could not be run in a single repeated measure, as the animals were deducted at planned endpoints as the study progressed. Using an established analysis strategy for multiple endpoints,²²  we applied a repeated-measures ANOVA for the intervals of 6 h, 1 days, and 2 days and a second ANOVA for 3, 5, and 8 days. Tukey’s post hoc test was used to compare the sham versus ligand at each time point. The figures discussed below in the Results display all animals tested at each time point. The use of parametric statistics for the Basso Mouse Scale and similar locomotor rating scales is supported by Abelson and Tukey²³  and confirmed to be statistically appropriate for spinal cord injury.²⁴ 

For laser Doppler tissue perfusion, all statistics were calculated by paired t test. All data were calculated using GraphPad Prism 8.4.3. One mouse and all associated data were excluded (mouse 3) from the study due to abnormal velocity detected with Doppler ultrasound before ligation, resulting in a positive outlier using the Grubbs’ test.

In the pilot study, there was a significant difference in paralysis outcomes and mortality among the three ligation strategies at day 1 (table 1). Increasing the number of arteries ligated and/or the number of ligations on each artery increased the severity of paralysis, but at a cost of survivability (table 1). Double ligation of three pairs of intercostal arteries produced only mild paralysis (Basso Mouse Scale score of 7.5 or higher) at 24 h with 0% mortality (n = 0 of 5; P < 0.001). Double ligation of six pairs of intercostal arteries caused 100% severe paralysis (Basso Mouse Scale score of 4 or lower) at 24 h with 100% mortality (n = 2 of 2; P < 0.001). Double-ligation of five pairs of intercostal arteries produced 59% (n = 13 of 22) of mice with severe paralysis at 24 h, but the mortality rate was still very high at 72% (n = 16 of 22; P < 0.001).

Based on these results, our final model used the configuration of double ligation of three pairs of intercostals (8th to 10th) and single ligation of two pairs of intercostals (6th and 7th), which resulted in three stratified groups of behavioral deficits, 9% (n = 3 of 32) severe paralysis, 38% (n = 12 of 32) moderate paralysis, and 53% (n = 17 of 32) mild paralysis and an overall mortality rate of 28% (n = 9 of 32) at day 1 (P < 0.001; table 1).

Overall, considering both the pilot and final configurations, we found that severely paralyzed mice (Basso Mouse Scale score of 4 or lower) had the highest mortality rate (83%; n = 15 of 18) compared to moderately paralyzed mice (Basso Mouse Scale score between 4 and 7.5; 33%; n = 6 of 18) and mildly paralyzed mice (Basso Mouse Scale score of 7.5 or higher; 24%; n = 6 of 25) with P < 0.001 (table 1).

In the final model configuration, we also found a significantly lower overall survival rate for the ligation group (84%; n = 46 of 55) compared to the sham group (100%; n = 48 of 48) with P = 0.003. No significant difference in survival probability between the ligation and sham groups occurred at 6-h and 12-h postoperation but was apparent at postoperative days 2, 8, and 14 (see Supplemental Digital Content 7, https://links.lww.com/ALN/D42, table showing survival probability by study group and time). Figure 2 illustrates the overall survival comparison between ligation and sham groups over a 2-week period. Mice that died in the first 3 days postsurgery expired due to respiratory difficulties. Mice that died after 3 days postsurgery expired due to excessive weight loss or infection.

Intercostal artery and spinal cord blood flow (ml · min⁻¹ · 100 g tissue⁻¹) and erythrocyte velocity (m · s⁻¹ · 100 g tissue⁻¹) were measured in male C57BL/6 mice by laser Doppler tissue perfusion (n = 4). Baseline intercostal artery blood flow (ml · min⁻¹ · 100 g tissue⁻¹; mean = 67.66; 95% CI, 59.32 to 76, P ≤ 0.001) and velocity (m · s⁻¹ · 100 g tissue⁻¹; mean = 4.57; 95% CI, 2.52 to 6.62; P = 0.030) were higher than the thoracic spinal cord blood flow (ml · min⁻¹ · 100 g tissue⁻¹; mean = 31.83; 95% CI, 27.13 to 36.53; P ≤ 0.001) and velocity (m · s⁻¹ · 100 g tissue−¹; mean = 1.30; 95% CI, 1.19 to 1.41; P ≤ 0.001). After intercostal arteries ligation, a significant and instantaneous drop in blood flow (% change; mean = −63.81; 95% CI, −72.28 to −55.34) and velocity (% change; mean = −78.14; 95% CI, −91.15 to −65.13) occurred in the arteries (fig. 3, A to D). The intercostal artery occlusion also caused significant instantaneous drop in blood flow in the thoracic spinal cord (% change; mean = −68.55; 95% CI, −80.23 to −56.87) and an instantaneous reduction in blood flow velocity (m · s⁻¹ · 100 g tissue⁻¹) in the thoracic spinal cord (mean = −62.43; 95% CI, −78.2 to −46.66; fig. 3, E to H; see Supplemental Digital Content 8, https://links.lww.com/ALN/D43, table showing mean blood flow and velocity in the intercostal arteries and spinal cord pre- and postligation and the percentage of change in the intercostal arteries and spinal cord blood flow pre- and postligation).

Ligation of five pairs of intercostal arteries induced spinal cord hypoperfusion resulting from a variable but significant drop in spinal cord blood flow. Figure 3 and Supplemental Digital Content 8 (https://links.lww.com/ALN/D43) show the mean blood flow and velocity in the intercostal arteries and spinal cord pre- and postligation and the percentage of change in the intercostal and spinal cord blood flow and velocity pre- and postligation (n = 4). Figure 4A shows an example of the change in the intercostal artery blood flow and velocity pre- and postligation in one mouse. Figure 4 (B and C) shows examples of the change in the spinal cord blood flow and velocity pre- and postligation in two other mice. For example, one mouse experienced an 83.6% decrease in spinal cord blood flow postligation largely related to low spinal cord baseline blood flow levels (fig. 4B, mouse 4). In contrast, another mouse had comparatively higher spinal cord baseline blood flow and experienced only a 55.1% decrease in spinal cord blood flow postligation (fig. 4C, mouse 2).

Spinal cords from both male and female mice undergoing ligation were clearly distinguished from those undergoing sham surgical procedures without ligation. Within the 2- and 8-day ligation groups, there was variability in the severity and distribution of lesions, and changes in the white matter were less pronounced than those in the gray matter unless necrosis was present (fig. 5).

Prominent hypercellularity was present in both dorsal and ventral horn gray matter in ligated mice, particularly at 8 days postligation. The cells contributing to hypercellularity included large numbers of Iba-1 + microglia (fig. 5N), glial fibrillary acidic protein + astrocytes (fig. 5O), and fewer Olig2 + oligodendrocytes (fig. 5P). Microglia frequently had small, dense, round to oval, peripheralized nuclei with large, round to oval cytoplasmic surface area. Microglia in sham animals exhibited the expected morphology of thin arborizations and a small cell body. Astrocytes in and around affected gray matter had large, open oval nuclei with increased cytoplasm and plump, shorter cytoplasmic processes rather than the small soma and small, central round nuclei observed in spinal cords from mice subjected to the sham procedure. Frequently, increased numbers of microglia and astrocytes exhibiting these respective morphological changes were present in viable tissue along the margins of foci of cell death and tissue loss. These changes were more prominent in spinal cords examined at 8 days postligation compared to 2 days postligation (fig. 5, E to H). Interestingly, in three spinal cords collected 2 days after ligation, the dorsal horns were markedly hypocellular, with almost no neurons, microglia, astrocytes, or oligodendrocytes compared to sham spinal cords at the same time point (fig. 5, E to H). Unique in the ligated spinal cords examined at 8 days postprocedure were multifocal clusters of numerous small, proliferative capillaries with variable luminal diameter. These capillaries were frequently found in the dorsal horn in areas of marked hypercellularity and glial neuroinflammation (fig. 5, Q and R).

Lesions in the white matter were highly variable in spinal cords examined at 2 or 8 days postligation but were less pronounced than lesions in the gray matter. These changes included swollen axon cylinders (spheroids) in dilated myelin sheaths (fig. 5, E and M) and occasional small foci of neuroinflammation (including increased microglia and astrocytes; fig. 5, F, N, G, and O) adjacent to blood vessels or foci of cell death. Spheroids were observed in all white matter tracts but often present in only one to three funiculi in an individual spinal cord. Variability in the size and severity of axon swelling was also common, with spheroids of varying size present in individual spinal cords at both time points.

Evaluation of cleaved caspase-3 immunohistochemistry at the 2- and 8-day time points revealed rare, individualized cells with positive punctate cytoplasmic or defined nuclear staining. Positively labeled cells were rare, with a single positive cell present per spinal cord section (data not shown).

In each sample, including the control, some fixation and expansion artifacts were present; therefore, we did not include those criteria in our assessment of ischemia-related damage.²⁵  Rather, we relied on the relative severity of extracellular space and the presence of broken cell membranes relative to the control sample as our criteria for damage assessment. Because standard fixation induces cellular swelling,²⁶  and extracellular space here was not preserved, we expect to see very little extracellular space. After intercostal ligation, extracellular space was substantially increased relative to the control 24-h postischemia (fig. 6, A and B). In contrast, by 48 h, less extracellular space was present (fig. 6C), but more cellular damage, i.e., ruptured membranes, was observed.

In all ligated mice, hind limb motor deficit of variable degrees was observed immediately after recovery from anesthesia. The 6-h time point showed the lowest Basso Mouse Scale score. Over the 2-week period, there was gradual increase of locomotor function of the mice that began 24 h postligation and continuously improved from 3 to 14 days (fig. 7). The ligation group had significantly lower Basso Mouse Scale scores relative to the sham group through 8 days (P < 0.01; fig. 7A).

After ligation, a large range of locomotor deficits occurred at each time point. The greatest variability occurred early after ligation such that at 6 h, Basso Mouse Scale scores ranged from 0 (no hind limb movement) to 8 (nearly normal locomotion; fig. 7B). The large Basso Mouse Scale score drop in our ligated mice at 6 h is not attributable to anesthesia effects because the sham mice returned to normal or near-normal behavior at this time point (fig. 7A).

By 14 days, the variability had narrowed substantially. Variability occurred both as underperforming and outperforming the group mean. A single mouse scored 0 on the Basso Mouse Scale at the initial test and showed no improvement over time, maintaining full hind limb paralysis until sacrificed at the 8-day time point. The relationship between severity of deficits and mortality was examined for high severity with Basso Mouse Scale scores of 4 or lower, moderate severity with Basso Mouse Scale scores between 4 and 7.5, and mild severity with Basso Mouse Scale scores of 7.5 or higher. The Fisher’s exact test showed that Basso Mouse Scale scores were significantly different across the four configurations (P < 0.001; table 1). For example, 100% of the six double-ligation group (n = 2 of 2), 59% of the five double-ligation group (n = 13 of 22), 9% of the three double-ligation and two single-ligation group (n = 3 of 32), and 0% of the three double-ligation group (n = 0 of 5) had Basso Mouse Scale scores of 4 or lower at day 1 with P < 0.001.

In contrast to the blood pressure and blood gases results of the aortic cross clamp mouse model of Awad et al.,¹²  there was no significant difference in the intraoperative blood pressure or blood gases among the baseline, ligation, and sham groups. The parameter of significance to determine whether the ligation surgery has any effect on blood pressure is the mean arterial pressure (MAP). Between the three groups, there was no significant difference in the mean intraoperative MAP. The mean intraoperative MAP was 76.2 mmHg (95% CI, 72.8 to 79.5) for the baseline group (n = 5), 78.2 mmHg (95% CI, 74.8 to 81.7) for the ligation group (n = 5), and 79.6 mmHg (95% CI, 76.1 to 83.0) for the sham group (n = 5). The baseline–ligation difference was −2.0 mmHg (−6.8, 2.8), nonsignificant with P = 0.40. The baseline–sham difference was −3.4 mmHg (−8.1, 1.4), nonsignificant with P = 0.17. The ligation–sham difference was −1.3 mmHg (−6.2, 3.5), nonsignificant with P = 0.59.

No significant difference was observed among the three groups in the mean pH, mean Paco₂, mean bicarbonate concentration, mean Pao₂, or the mean hemoglobin concentration (see Supplemental Digital Content 9, https://links.lww.com/ALN/D44, table showing the intraoperative blood gas measurements from the baseline, ligation, and sham groups.). The mean pH was 7.4 (95% CI, 7.4 to 7.4) for the baseline group (n = 5), 7.4 (95% CI, 7.4 to 7.4) for the ligation group (n = 5), and 7.4 (95% CI, 7.4 to 7.5) for the sham group (n = 5). The baseline–ligation difference was −0.0 (95% CI, −0.0 to 0.0) nonsignificant, with P = 0.97. The baseline–sham difference was −0.0 (−0.1, 0.0), nonsignificant with P = 0.16. The ligation–sham difference was −0.0 (95% CI, −0.1 to 0.0), nonsignificant, with P = 0.17.

The mean Paco₂ was 35.9 mmHg (95% CI, 34.0 to 37.8) for the baseline group (n = 5), 35.2 mmHg (95% CI, 33.3 to 37.2) for the ligation group (n = 5), and 33.8 mmHg (95% CI, 31.8 to 35.7) for the sham group (n = 5). The baseline–ligation difference was 0.7 mmHg (95% CI, −2.1 to 3.4), nonsignificant with P = 0.60. The baseline–sham difference was 2.1 mmHg (−0.6, 4.9), nonsignificant with P = 0.12. The ligation–sham difference was 1.4 mmHg (95% CI, −1.3 to 4.2), nonsignificant with P = 0.27.

The mean bicarbonate concentration was 21.9 mM (95% CI, 20.7, 23.1) for the baseline group (n = 5), 21.5 mM (95% CI, 20.3 to 22.7) for the ligation group (n = 5), and 21.9 mM (95% CI, 20.7 to 23.1) for the sham group (n = 5). The baseline–ligation difference was 0.4 mM (−1.3, 2.1), nonsignificant with P = 0.62. The baseline–sham difference was −0.0 mM (−1.7, 1.7), nonsignificant with P = 1.00. The ligation–sham difference was −0.4 mM (95% CI, −2.1 to 1.3), nonsignificant with P = 0.62.

The mean Pao₂ was 300.2 mmHg (95% CI, 260.3 to 340.1) for the baseline group (n = 5), 315.0 mmHg (95% CI, 275 to 354.9) for the ligation group (n = 5), and 326.8 mmHg (95% CI, 286.9 to 366.7) for the sham group (n = 5). The baseline–ligation difference was −14.8 mmHg (95% CI, −71.3 to 41.7), nonsignificant with P = 0.60. The baseline–sham difference was −26.6 mmHg (95% CI, −83.1 to 29.9), nonsignificant with P = 0.33. The ligation–sham difference was −11.8 mmHg (95% CI, −68.3 to 44.7), nonsignificant with P = 0.66.

The mean hemoglobin concentration was 13.8 g/dl (95% CI, 13.4 to 14.2) for the baseline group (n = 5), 13.5 g/dl (95% CI, 13.1 to 14.0) for the ligation group (n = 5), and 12.9 g/dl (95% CI, 12.5 to 13.3) for the sham group (n = 5). The baseline–ligation difference was 0.3 g/dl (95% CI, −0.3 to 0.9), nonsignificant with P = 0.35. The baseline–sham difference was 0.9 g/dl (95% CI, 0.3 to 1.5), nonsignificant with P = 0.009. The ligation–sham difference was 0.6 g/dl (95% CI, 0.0 to 1.2), nonsignificant with P = 0.05.

We have demonstrated that the number and degree of ligated intercostal arteries is directly proportional to the severity of paralysis and inversely proportional to survivability. Our final configuration—three double and two single ligations—caused clear motor deficits below the ligation level without high mortality for 14 days or more. Survival rate ranged between 100% (6- and 12-h groups) and 63% (2-week group).

Our model demonstrates four behavioral similarities with humans. First, the immediate paralysis occurring in our model is like the immediate paralysis occurring in humans after thoracic endovascular aortic repair²⁷  and contrasts with the delayed paralysis occurring in the open repair mouse model⁹  and humans.²⁷  Second, the wide behavioral deficit variability (severe to mild) occurring in response to the same ligation procedure replicates the clinical profile in patients post-thoracic endovascular aortic repair (fig. 7).⁶ 

Third, gradual paralysis improvement occurred throughout the 2-week follow-up period with sustained improvement starting at day 3 (fig. 7A). The varying degrees of deficit and gradual improvement mimic the recovery process in humans who develop thoracic endovascular aortic repair–induced paralysis, in which mild paralysis is transient and improves with time, while severe paralysis is permanent.⁶  Deficit variability showing improvement may be attributed to extensive collateral circulation and adaptation to blockage. Etz et al.²⁸  proved that in humans, spinal cord perfusion pressure drops markedly then recovers gradually during the first several hours postextensive segmental artery sacrifice. In the same study, they showed that all but one patient, who had remarkably low postoperative spinal cord perfusion pressure and experienced paraparesis, regained normal spinal cord function. Etz et al.²⁹  also proved the existence of an extensive collateral network around pig spinal cords. Griepp et al.³⁰  demonstrated that the collateral network around the pig spinal cord enlarges within 24 h after extensive ligation of segmental arteries, showed that maximum collateral circulation expansion is achieved by the fifth day postligation, and correlated these results to Etz et al.’s findings in humans. All three studies demonstrate similarities between human and pig spinal cord collaterals and human and pig temporal response to segmental artery ligation or sacrifice. We found similar behavioral or locomotor findings as those in humans with gradual improvement of motor function over the first few days in mildly and moderately paralyzed mice and permanent motor function loss in severely paralyzed mice.^6,28  No known study demonstrates the existence of collateral circulation around the mouse spinal cord. However, similar dysfunctional findings in our mice with human and pig studies led us to conclude that anatomical similarities—i.e., a collateral network around the spinal cord that gradually improves after intercostal artery ligation—should also exist in mice.

A critical foundation of our model included restricted spinal cord blood flow confirmation. We extended laser Doppler measurement of spinal cord blood flow from rats to mice³¹  to collect in vivo blood flow measurements from the exposed spinal cord and intercostal arteries. The percentage of change variability in spinal cord blood flow among our ligated mice is explained by the existence of collaterals around the spinal cord and the degree of patency of these collaterals that varies among the ligated mice.^29,30  Further, the degree of the collateral’s patency and the preligation baseline spinal cord blood flow variability explain the variability in the postligation percent decrease in spinal cord blood flow, which explains spinal cord tissue damage variability and the resultant behavioral paresis among the ligated mice. We therefore postulate that baseline spinal cord blood flow evidences the extent of collateral network development around the spinal cord, where a relatively low baseline blood flow demonstrates a less developed network, and a comparatively higher baseline blood flow demonstrates a more extensively developed network.

Thus, mice with more developed and patent collaterals will be associated with a lesser drop in blood flow, mild behavioral deficit, and progressive recovery over time, whereas mice with less developed and patent collaterals will be associated with a marked drop in blood flow and severe behavioral deficit. Therefore, the pre–thoracic endovascular aortic repair degree of patency and development of collaterals around the spinal cord may predict postprocedure paralysis severity and permanence, which may prove to be important clinical screening characteristics.

Fourth, severely paralyzed mice had the highest mortality rate (83%; n = 15 of 18) compared to moderately paralyzed (33%; n = 6 of 18) and mildly paralyzed mice (24%; n = 6 of 25; P < 0.001). This relationship aligns with the 2020 national incidence and mortality data reported for thoracic endovascular aortic repair.⁶  Patients who developed reversible spinal cord injury symptoms had significantly worse 1-yr survival rates than those without spinal cord injury (transient spinal cord injury, 80% [95% CI, 73 to 87%]; no spinal cord injury, 87% [95% CI, 86 to 88%]; log-rank P = 0.1) but significantly better survival than those with permanent spinal cord injury symptoms (permanent spinal cord injury, 54% [95% CI, 47 to 61%]; transient spinal cord injury, 80% [95% CI, 73 to 87%]; P < 0.0001).⁶ 

Histopathological spinal cord changes in ligated mice also reflect ischemic change, severity variability, and precise localization of lesions. Apparent activation of astrocytes and microglia occurred postligation based on classic morphological changes in other sterile injury models.^32–34  A more modest increase in Olig2 + oligodendrocytes was present in spinal cords examined 8 days postligation compared to sham control spinal cords, but no appreciable change in Olig2 + cells was present in spinal cords examined 2 days postligation (fig. 5). Electron microscopic images differentiated abnormal cellular processes—broken cellular membranes and diffuse tissue damage after ligation. This supports our conclusion that the ligated animal’s behavioral deficit is due to spinal cord ischemic damage and is not a mere side effect of the surgical approach.

Despite lesion variability, a pattern of changes suggesting a temporal continuum emerged. In the gray matter of spinal cords at 2 days postligation, there is a paucity of cells, particularly in the dorsal horn but also extending multifocally into the ventral horns. These areas are surrounded by activated microglia and astrocytes and are multifocally accompanied by white matter lesions including histological evidence of axonal injury. In spinal cords 8 days postligation, there is marked gliosis diffusely in the gray matter and in multifocal perivascular regions in the white matter composed predominantly of reactive microglia and astrocytes. This is frequently accompanied by proliferation of small capillaries (fig. 5). A single spinal cord was characterized by loss of architecture and paucity of cells with replacement by cellular debris in approximately 60% of the examined section. This spinal cord corresponded to severe paralysis and Basso Mouse Scale score of 0 at 8 days postprocedure. Of interest is the lack of cleaved caspase-3 immunolabeling. It is possible that apoptosis is a component of the model but is missed via immunohistochemistry at the examined time points, or based on the immunohistochemistry findings at 2 days, apoptosis may not represent the predominant mechanism of cell death in this ischemia model.

These spatially variable lesions are consistent with magnetic resonance imaging findings in humans with thoracic endovascular aortic repair–induced spinal cord injury that showed variability in location of the lesion in the spinal cord across the anterior gray matter and anterior portion of posterior white column, central cord lesion, and posterior gray matter, and posterior white column.³⁵  This difference in the location and extent of infarction area can be explained by collateral conception.^29,30  The blood flow decrease in the anterior spinal artery caused by ligation will be partially compensated by flow from the collaterals in and around the spinal cord.^29,30  The area of the critical drop in tissue perfusion within the spinal cord varies among ligated mice due to the variability in degree of collateral development and patency. Collateral conception together with ischemia-induced inflammation can explain the proliferation of small capillaries at the infarct area margins.

Electron microscopy further demonstrates ischemic spinal cord tissue damage. Because hypoperfusion-induced tissue damage results in tissue that might contain abnormal (swollen or shrunken processes) or a diffuse type of damage not limited to cracking or breaks in the sample, we differentiated this type of damage from the mechanically based cracking damage (fig. 6). Abnormal cellular processes, broken cellular membranes, and diffuse tissue damage were more prevalent in the ligated samples. These findings further support our conclusion that the ligated animal’s behavioral deficit is caused by spinal cord ischemic damage and is unrelated to the surgical approach.

Our study has three limitations. Our initial focus was development of a ligation strategy to achieve the desired paralysis phenotype with an acceptable mortality rate, precluding a pre-experiment sample size calculation (see Supplemental Digital Content 10, https://links.lww.com/ALN/D45). Histopathological results showed posterior spinal cord lesions, suggesting that sensory tests may have revealed additional mice demonstrating deficit; multimodal sensory testing is recommended for future studies. Finally, no autopsies were performed on expired mice to establish actual cause of death.

We presented a mouse model with confirmed spinal cord hypoperfusion and spinal cord lesion histological topography caused by ligation of the intercostal arteries, which causes immediate paralysis that mimics in both severity and reversibility human paralysis after thoracic endovascular aortic repair. This model has good translational potential to support development of neuroprotective drugs and therapeutics for ischemic spinal cord injury after thoracic endovascular aortic repair.

The authors thank Cole McLarty, B.M.Sc. (Transonic Systems Inc., Ithaca, New York), for his assistance with the flow instrumentation and Jeff Tonniges, Ph.D., and resources (Campus Microscopy and Imaging Facility and the Ohio State University Comprehensive Cancer Center Microscopy Shared Resource at Ohio State University, Columbus, Ohio) for sample preparation and imaging assistance.

Supported in part by institutional and/or departmental resources. The Comparative Pathology and Digital Imaging Shared Resource, Department of Veterinary Biosciences and the Comprehensive Cancer Center at the Ohio State University (Columbus, Ohio) was supported in part by grant No. P30 CA016058 from the National Cancer Institute (Bethesda, Maryland). Funding for the shared surgical space in the Neuroscience Department Surgical Core at the Ohio State University was supported in part through grant No. P30NS104177 from the National Institutes of Health, NINDS (Bethesda, Maryland). Finally, funding for support from the Basso Lab at the Ohio State University was provided in part by grant No. RO1NS074882 from the National Institutes of Health.

The authors declare no competing interests.

Supplemental Digital Content 1, Experimental design animal tree, https://links.lww.com/ALN/D36

Supplemental Digital Content 2, Choice of core body temperature during anesthesia preparation, https://links.lww.com/ALN/D37

Supplemental Digital Content 3, Rationale for the use of double ligation, https://links.lww.com/ALN/D38

Supplemental Digital Content 4, Postsurgery pain management, https://links.lww.com/ALN/D39

Supplemental Digital Content 5, Explanation of Basso Mouse Scale assessment method, https://links.lww.com/ALN/D40

Supplemental Digital Content 6, Example of an intraoperative blood pressure reading, https://links.lww.com/ALN/D41

Supplemental Digital Content 7, Survival probability by study group and time, https://links.lww.com/ALN/D42

Supplemental Digital Content 8, Mean blood flow and velocity in the intercostal arteries and spinal cord pre- and postligation and the percentage of change in the intercostal artery and spinal cord blood flow pre-and postligation, https://links.lww.com/ALN/D43

Supplemental Digital Content 9, Intraoperative blood pressure and blood gas measurements from the baseline, ligation, and sham groups, https://links.lww.com/ALN/D44

Supplemental Digital Content 10, Explanation of no pre-experiment sample size calculation, https://links.lww.com/ALN/D45