Endovascular perforation subarachnoid hemorrhage fails to cause Morris water maze deficits in the mouse

Abstract

Cognitive dysfunction is the primary driver of poor long-term outcome in aneurysmal subarachnoid hemorrhage (SAH) survivors; modeling such deficits preclinically is thus key for mechanistic and translational investigation. Although rat SAH causes long-term deficits in learning and memory, it remains unknown whether similar deficits are seen in the mouse, a species particularly amenable to powerful, targeted genetic manipulation. We thus subjected mice to endovascular perforation SAH and assessed long-term cognitive outcome via the Morris water maze (MWM), the most commonly used metric for rodent neurocognition. No significant differences in MWM performance (by either of two protocols) were seen in SAH versus sham mice. Moreover, SAH caused negligible hippocampal CA1 injury. These results undercut the potential of commonly used methods (of SAH induction and assessment of long-term neurocognitive outcome) for use in targeted molecular studies of SAH-induced cognitive deficits in the mouse.

Introduction

Aneurysmal subarachnoid hemorrhage (SAH) is a neurologically devastating disease that affects 6 to 7 people per 100,000 each year. Mortality remains near 50%¹ while relatively few SAH survivors have significant focal neurologic deficits of the sort that are common after ischemic stroke, long-term cognitive dysfunction is seen in 50% to 60%, allowing only 33% to return to their previous level of employment despite good neurologic outcome.² The long-term cognitive deficits after SAH cross numerous cognitive domains, including memory, executive function, and language (for review, see Al-Khindi,³). Of the memory deficits, most^{4, 5, 6} but not all⁷ have documented visuospatial memory as being particularly affected. Strikingly, no treatment has been shown to improve cognitive outcome after SAH. It is therefore critical to devise experimental methods to elucidate underlying mechanism and develop novel therapeutics for these deficits, as they are a principal driver of the long-term loss of quality of life in SAH survivors.

Experimental models of SAH vary in species used and in method of induction; each has its unique advantages and disadvantages.⁸ A key benefit of mouse models is the ability to incorporate myriad transgenic lines to query mechanism precisely. This is directly relevant to SAH, as apolipoprotein E and haptoglobin genotypes have been identified as genetic risk factors for poor patient outcome, the former being independently associated with poor cognitive outcome after SAH.^{9, 10}

The importance of including long-term outcomes in preclinical studies has been substantiated by disappointing results from recent therapeutic trials. Despite successfully reducing angiographic vasospasm, long-term patient outcome was not affected by administration of the non-glucocorticoid 21-aminosteroid tirilazad¹¹ or the endothelin receptor antagonist clazosentan.¹² It may be that, beyond examining short-term preclinical outcomes (such as early brain injury and cerebral vasospasm), the successful translation of putative therapies would be predicted by evaluation of long-term neurobehavioral outcomes in preclinical models,¹³ as recommended by the Stroke Therapy Academic Industry Roundtable criteria.¹⁴

Experimental SAH research has thus sought to model long-term neurobehavioral outcomes and elucidate the mechanisms responsible. Morris water maze (MWM) deficits have been documented in all three principal rat models 3 to 5 weeks after SAH,^{15, 16, 17, 18, 19, 20} along with T-maze deficits in endovascular perforation SAH;¹⁹ moreover, treatment with statins¹⁷ and minocycline¹⁹ has decreased SAH-induced cognitive deficits. Notably, a strong correlation was seen in rat SAH between MWM performance and neuronal cell counts in hippocampus at 5 weeks.¹⁵

Despite these reports in rat models of SAH, to date there has been no characterization of long-term neurobehavioral deficits in mouse models of SAH. We thus undertook to characterize cognitive deficits in the most commonly used mouse model of SAH, endovascular perforation, using the gold standard of spatial learning, the MWM.

Materials and methods

Animals

All experimental protocols were approved by the Animals Studies Committee at Washington University in St Louis and complied with the Guide for the Care and Use of Laboratory Animals; the Public Health Service Policy on Humane Care and Use of Laboratory Animals; and Washington University Department of Comparative Medicine guidelines. Mice were housed in an AAALAC-accredited facility in temperature- and humidity-controlled rooms with a 12-hour light–dark cycle. Mice were allowed ad libitum access to standard chow and autoclaved tap water. A total of 250 male wild-type C57BL/6 mice (Jackson Labs, Bar Harbor, ME, USA) were used starting at 12 to 14 weeks of age (24 to 30 g). Allocation to experimental group was performed before the beginning of each experiment: tails were numbered by one experimenter and another experimenter assigned mice randomly according to these numbers. All data were collected by experimenters masked to experimental group. The study included five independent experiments (summarized in Table 1): in experiment 1, mice were subjected to SAH, given 3 days of fluid support, and assessed for cerebral vasospasm on post-SAH day 3. In experiment 2, mice were subjected to SAH, given 3 days of fluid support, and assessed via Place MWM. Significant long-term mortality was seen, and so attempts were made to improve survival, first by extending fluid support to postoperative day 7 (experiment 3) and second by not only extending fluid support to 7 days, but also administering ampicillin²¹ (experiments 4 and 5). A more sensitive MWM protocol, the Learning Set task, was employed in a subset of studies (experiment 4, cohort 2, and experiment 5). Traumatic brain injury (TBI) was used as a positive control (experiment 5).

Table 1. Summary of experimental protocols.

Experiment	Surgery	Fluid support	MWM protocol	Duration of experiment	Mortality (%)
1	SAH	Days 0–3: dex-NS	N/A	3 days	7
2	SAH	Days 0–3: dex-NS	Place	23 days	34
3	SAH	Days 0–7: dex-NS	Place	23 days	25
4	SAH	Days 0–3: amp in dex-NS Days 4–7: dex-NS	Place (cohort 1) Learning Set (cohort 2)	23 days 21 days	18
5	TBI	Days 0–3: amp in dex-NS Days 4–7: dex-NS	Learning Set	21 days	0

amp, ampicillin; dex-NS, 10% dextrose in normal saline; MWM, Morris water maze; SAH, subarachnoid hemorrhage; TBI, traumatic brain injury.

Endovascular Perforation Subarachnoid Hemorrhage

For experiments 1 to 4, endovascular perforation SAH was performed as described.^{22, 23} Briefly, mice were anesthetized with isoflurane (4% induction and 1.5% maintenance) and a midline neck incision was made. The external carotid artery was isolated, and through it a 5-0 blunted nylon suture was introduced into the internal carotid artery. The suture was advanced distally until resistance was experienced at the bifurcation of the internal carotid artery into the anterior and middle cerebral arteries. Advancing the suture a further 5 mm caused perforation and SAH. The suture was immediately removed and the external carotid artery was ligated. Sham-operated mice were subjected to all surgical procedures except that, upon feeling resistance at the internal carotid artery bifurcation, the suture was removed without advancement. Cerebral vasospasm was assessed on post-SAH day 3 via pressure-controlled casting with gelatin–India ink solution.

Controlled Cortical Impact Traumatic Brain Injury

In experiment 5, electromagnetically controlled cortical impact model of TBI was performed as previously described.²⁴ Briefly, mice were anesthetized with isoflurane (4% induction and 2% maintenance) and placed in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA). A 1-cm midline scalp incision was made and a left craniotomy was performed over the parietotemporal cortex using a 5-mm trephine (Meisinger, Neuss, Germany) attached to an electric drill (Foredom, Bethel, CT, USA). The impact device was mounted on the stereotaxic frame at an angle of 15° and moved 3.0 mm anterior to lambda and 2.7 mm left of midline, inside the craniotomy. Zero depth was determined, the tip was cocked, and the device was lowered using the stereotaxic arm to set a depth of impact of 2 mm. Traumatic brain injury was triggered via Matlab (The Mathworks, Natick, MA, USA). A 6-mm plastic disc was glued to the skull over the craniotomy and the incision was sutured. Sham-operated mice were subjected to all surgical procedures except for TBI.

Fluid Support

In all cases, mice were administered 0.5 ml of 10% dextrose in normal saline intraperitoneally twice daily. In experiments 1 and 2, support was provided on postoperative days 0 to 3, as previously described.²³ In experiment 3, this regimen was extended to postoperative day 7. For experiments 4 and 5, mice were administered 50 mg/kg ampicillin in dextrose in normal saline twice daily on postoperative days 0 to 3, followed on postoperative days 4 to 7 by dextrose in normal saline.

Neuroscore

For all experiments, gross neurologic function was assessed on postoperative days 0 to 3, 7, and 12 via sensorimotor scoring as previously described.^{22, 23} Briefly, neurologic function was graded based on a motor score (0 to 12) that evaluates spontaneous activity, symmetry of limb movements, climbing, and balance and coordination; and a sensory score (4 to 12) that evaluates body proprioception and vibrissae, visual, and tactile responses. For Experiments 2–4, data presented include sham-operated mice; SAH mice that survived long term and were able to complete MWM testing (survival group, SAH-surv); and mice that survived initially (i.e., at least 3 days after SAH), but either died before MWM testing or were excluded owing to being unable to complete MWM testing (mortality group, SAH-mort).

Morris Water Maze

For all MWM experiments, a 100-cm pool was filled with water rendered opaque with non-toxic white paint. A 10-cm escape platform was made to either protrude 1 cm above the surface of the water (Cued trials) or was submerged 1 cm (Place and Learning Set trials). Mice were released from one of four cardinal drop points. If the mouse found the platform, it was allowed to sit for 15 seconds before removal; if the mouse failed to find the platform, it was placed on the platform for 15 seconds. An overhead camera recorded swim paths. The SMART software (San Diego Instruments, San Diego, CA, USA) analyzed escape latency, swim distances, and swim speeds for all trials. All experiments were conducted by a single investigator (EM) masked to experimental groups.

Two different protocols were used. For the Place MWM,^{24, 25} mice were given four trials per day, one from each drop point. On post-SAH days 13 to 15, mice were subjected to four trials of the Cued task (visible platform) with a single platform location. On days 18 to 22, mice were subjected to the Place task with a single (hidden) platform location. On day 23, a 30-second Probe trial was performed.

For the Learning Set MWM (modified from Hartman et al²⁵), each day mice consisted of eight blocks of two consecutive trials. On post-SAH day 13, mice were subjected to the Cued task: a new (visible) platform location was used for each block, and mice were dropped diametrically opposite the platform. On days 14 to 20, mice were subjected to the Learning Set task with a new (hidden) platform position each day. For block 1, the same distant drop point was used; for blocks 2 to 8, drop points were counter-balanced for near-far. On days 15 to 21, mice were subjected to a 30-second Probe trial (i.e., a probe trial ∼16 hours after completing the previous day's task).

Mice were disqualified from MWM testing if they failed to swim to the visible platform during Cued training. This was owing to either an inability to learn or to complete the procedural aspects involved, including staying above water for 4 or 16 daily trials (Place and Learning Set, respectively) and lifting onto the elevated platform; or because of deficits in vision or motivation to escape the water. Sensorimotor data from these mice are included along with mice that died before MWM testing (mortality group; see Neuroscore section above).

Neuronal Cell Counting

After completing the final Probe trial, mice were killed and transcardially perfused with heparinized phosphate-buffered saline. Brains were postfixed overnight in 4% paraformaldehyde in phosphate-buffered saline, equilibrated in 30% sucrose, sectioned coronally at 50 μm, and stained with cresyl violet. Unbiased stereology was performed using Stereo Investigator (MBF Bioscience, Williston, VT, USA). A single investigator masked to treatment assessed 3 to 5 hippocampal sections taken at 600-μm intervals; morphologically intact neurons were counted in hippocampal CA1 from midline to the lateral extent of the superior limb of the angular gyrus.

Statistical Analysis

Data are expressed as means±s.e.m. Mortality was analyzed by Kaplan-Meier estimator. After testing for normality, Neuroscore and MWM performance were analyzed by repeated measures analysis of variance (ANOVA) followed by Newman-Keuls multiple comparison. Probe performance, CA1 neuronal cell count, and CA1 volume were first assessed for normality and then analyzed by unpaired t-test. Correlation between neuronal cell counts and MWM performance was analyzed by least-squares linear regression. Statistical significance was set at P<0.05.

Results

Short-Term Subarachnoid Hemorrhage Outcome

Acute SAH mortality (0 to 3 days) was similar in experiments 1 to 4 (P=0.80, Kaplan-Meier estimator). Marked cerebral vasospasm was noted 3 days after SAH (arrows) in experiment 1 (Figures 1A and 1B; P<0.05, t-test); cerebral vasospasm was not assessed in experiments 2 to 4 owing to the postmortem nature of this assessment in the mouse. Significant sensorimotor neurologic deficits were noted 1 to 3 days after SAH in experiments 1 to 4 (Figures 1C, 2A, 3A, and 4A), and were similar across SAH groups. Overall, these data are in line with our previous reports^{22, 23} and show the short-term reproducibility of our endovascular perforation SAH mouse model.

Figure 1.

Subarachnoid hemorrhage (SAH) causes short-term neurologic deficits and delayed cerebral vasospasm; long-term survival after SAH is increased by extended fluid support and antibiotic administration. Mice underwent SAH or sham surgery and received fluid support for 3 days (0.5 ml 10% dextrose in normal saline intraperitoneally twice daily). Mice were casted with gelatin–India ink on post surgery day 3; representative images of casted vessels are shown (A). (B) Cerebral vasospasm was assessed via measurement of the proximal middle cerebral artery. Neurobehavioral assessment was performed on post surgery days 1 to 3 via sensorimotor scoring (C). Data indicate mean±s.e.m. *P<0.05 versus sham by t-test (B) or repeated measures analysis of variance and Newman-Keuls multiple comparison test (C). (D) SAH was induced and mortality was recorded with three regimens of postsurgical support: postsurgical injections of dextrose–saline solution for 3 days (dex-NS–3d) or 7 days (dex-NS–7d), or ampicillin in dextrose–saline for 3 days postsurgery followed by another four days of dextrose–saline (amp+dex-NS–7d). No sham-operated control (sham) died. amp, ampicillin; MCA, middle cerebral artery.

Figure 2.

Subarachnoid hemorrhage (SAH) followed by 3 days of fluid support causes deficits in neurologic outcomes but not in the Place Morris water maze. Mice underwent SAH or sham surgery and received fluid support for 3 days (0.5 ml 10% dextrose in normal saline intraperitoneally twice daily). Neurobehavioral assessment was performed on post surgery days 1 to 3, 7, and 12 via sensorimotor scoring (A). Long-term neurocognitive outcome was performed on postsurgical days 13 to 22 by the Place Morris water maze (B). Data indicate mean±s.e.m. *P<0.05 versus sham, ^#P<0.05 versus SAH-surv (survival group) by repeated measures analysis of variance and Newman-Keuls multiple comparison test. (C) A 30-second Probe trial was performed 24 hours after the Place task; time spent in the target quadrant and in the previous location were determined. The dashed line denotes chance performance; data represent 95% confidence interval of the mean. NS, P>0.05 vs. sham by t-test; SAH-mort, SAH mortality group.

Figure 3.

Subarachnoid hemorrhage (SAH) followed by 7 days of fluid support causes deficits in neurologic outcomes but not in the Place Morris water maze. Mice underwent SAH or sham surgery and received fluid support for 7 days (0.5 ml 10% dextrose in normal saline intraperitoneally twice daily). Neurobehavioral assessment was performed on post surgery days 1 to 3, 7, and 12 via sensorimotor scoring (A). Long-term neurocognitive outcome was performed on postsurgical days 13 to 22 by the Place Morris water maze (B). Data indicate mean±s.e.m. *P<0.05 versus sham, ^#P<0.05 versus SAH-surv (survival group) by repeated measures analysis of variance and Newman-Keuls multiple comparison test. (C) A 30-second Probe trial was performed 24 hours after the Place task; time spent in the target quadrant and in its previous location were determined. The dashed line denotes chance performance; data represent 95% confidence interval of the mean. NS, P>0.05 vs. sham by t-test; SAH-mort, SAH mortality group.

Figure 4.

Subarachnoid hemorrhage (SAH) followed by 7 days of fluid support plus antibiotic prophylaxis causes deficits in neurologic outcomes but not in the Place or Learning Set Morris water maze. Mice underwent SAH or sham surgery and received antibiotics for 3 days (50 mg/kg ampicillin intraperitoneally once daily, first dose before surgery) followed by fluid support on days 4 to 7 (0.5 ml 10% dextrose in normal saline intraperitoneally twice daily). Neurobehavioral assessment was performed on post surgery days 1 to 3, 7, and 12 via sensorimotor scoring (A). Long-term neurocognitive outcome was performed on postsurgical days 13 to 22 by the Place Morris water maze (B). A 30-second Probe trial was performed 24 hours after the Place task; time spent in the target quadrant and in the previous location were determined (C). A separate cohort of mice was subjected to the Learning Set Morris water maze on postsurgical days 13 to 21 (D). A 30-second Probe trial was performed 16 hours after the completion of the previous day's task; time spent in the target quadrant and in the previous location were determined (E). (A, B, E) Data indicate mean±s.e.m. *P<0.05 versus sham, ^#P<0.05 versus SAH-surv (survival group) by repeated measures analysis of variance and Newman-Keuls multiple comparison test. (C, E) The dashed line denotes chance performance; data represent 95% confidence interval of the mean. NS, P>0.05 vs. sham by t-test; SAH-mort, SAH mortality group.

Long-Term Subarachnoid Hemorrhage Outcome

Mortality

Delayed SAH mortality (i.e., after day 3) was noted in all long-term experiments (Figure 1D; P=0.0010, P=0.0152, and P=0.0053 for experiments 2 to 4, respectively, Kaplan-Meier estimator). A significant reduction in delayed SAH mortality was noted in experiment 4 versus experiment 2 (Figure 1D; P=0.048, Kaplan-Meier estimator). Together, these data indicate that long-term SAH survival can be increased by extending fluid support from 3 to 7 days and adding antibiotic prophylaxis, but nonetheless long-term SAH survival remains suboptimal.

Neurologic deficits

Mild but statistically significant neurologic deficits were noted in long-term SAH survivors (Figures 2A, 3A, and 4A; P<0.001, P<0.05, and P<0.01 for experiments 2 to 4, respectively, omnibus repeated measures ANOVA). Mice that survived to day 3 (the typical point of assessment for SAH-induced vasospasm-driven delayed cerebral ischemia) but did not complete the MWM (because of either delayed mortality or disqualification) evinced dramatically more severe neurologic deficits than mice that survived and were able to complete the MWM. In total, these data show that SAH survivors had mild sensorimotor deficits, whereas mice that were excluded from long-term assessments (MWM and unbiased stereology) had severe sensorimotor deficits.

Morris water maze

Performance in the Place MWM was similar between SAH and sham groups (Figures 2B and 2C, 3B and 3C, 4B and 4C). No significant differences were seen in the Cued task (visible platform; P=0.36, P=0.46, and P=0.27, experiments 2 to 4, respectively, omnibus repeated measures ANOVA). No significant differences were seen in the Place task (hidden platform; P=0.77, P=0.11, and P=0.85, experiments 2 to 4, respectively, omnibus repeated measures ANOVA). No significant differences were seen in Probe trial performance: time spent in the target quadrant was similar (P=0.15, P=64, and P=0.65 for experiments 2 to 4, respectively, t-test) and time spent in the target area was similar (P=0.61, P=10, and P=0.17 for experiments 2 to 4, respectively, t-test). In all cases, Probe performance exceeded chance levels (i.e., 25% for target quadrant and 1% for target area).

Performance in the more sensitive Learning Set MWM was also similar between SAH and sham groups (Figures 4D and 4E). No significant differences were seen in the Learning Set task (P=0.89, omnibus repeated measures ANOVA). No significant differences were seen in Probe trial performance: time spent in the target quadrant and in the target area was similar (P=0.26 and P=0.52, respectively, t-test). Again, Probe performance exceeded chance levels in both groups.

A total of four SAH mice were disqualified from MWM testing owing to an inability to complete the MWM: one in experiment 2 (unable to learn procedural aspects of Cued training), two in experiment 3 (one unable to learn procedural aspects of Cued training; one unable to swim because of dense hemiparesis), and one in experiment 4 (unable to stay above water for the 16 daily trials).

In summary, these data show that, despite reducing SAH mortality and utilizing multiple MWM protocols, endovascular perforation SAH causes consistent (albeit mild) sensorimotor deficits—but no visuospatial learning or memory deficits—in mice surviving the acute post-SAH period.

Unbiased stereology

Brain sections from mice that completed the MWM were examined for evidence of ischemic infarction: none was found. Specifically, the cortical volume of the ipsilateral versus contralateral hemisphere in SAH mice that completed the MWM were quantitated. In SAH mice ipsilateral volume was 97%±3% of the contralateral volume, essentially the same as what is seen in sham mice—98%±3%. In the four mice that survived but were excluded from MWM testing, cavitation consistent with ischemic infarction was seen in only one (data not shown). This mouse was the only long-term survivor to display a dense hemiparesis and circling behavior that coincides with a large cerebral infarction, providing phenotypic corroboration for the absence of histologic evidence of chronic infarction.

Figure 5.

Experimental traumatic brain injury (TBI) causes significant deficits in neurologic outcomes and in the Learning Set Morris water maze. Mice underwent TBI or sham surgery and received antibiotics for 3 days (50 mg/kg ampicillin intraperitoneally once daily, first dose before surgery) followed by fluid support on days 4 to 7 (0.5 ml 10% dextrose in normal saline intraperitoneally twice daily). Neurobehavioral assessment was performed on post surgery days 1 to 3, 7, and 12 via sensorimotor scoring (A). Long-term neurocognitive outcome was performed on postsurgical days 13 to 21 by the Learning Set Morris water maze (B). A 30-second Probe trial was performed 16 hours after the completion of the previous day's task; time spent in the target quadrant and in the previous location were determined. The dashed line represents chance performance; data represent 95% confidence interval of the mean. *P<0.05 versus sham by t-test (C). (A, B) Data indicate mean±s.e.m. *P<0.05 versus sham by repeated measures analysis of variance and Newman-Keuls multiple comparison test.

Morphologically intact hippocampal CA1 neurons (ipsilateral to side of SAH) were counted via unbiased stereology in mice subjected to the Learning Set MWM in experiment 4. Minimal injury was noted (Figure 6A): neuronal cell counts and spared CA1 volume were similar in SAH and sham groups (Figures 6B and 6C; P=0.88 and P=0.90, respectively, t-test). No correlation was noted between CA1 neuronal cell counts and MWM performance (Figure 6D; P=0.83, Pearson's correlation). In the four mice disqualified from MWM testing, mean neuronal cell counts were 199±47, similar to those completing the MWM. In total, these data show that endovascular perforation SAH fails to cause significant damage to hippocampal CA1.

Figure 6.

Histologic damage in CA1 correlates to performance in the Learning Set task after experimental traumatic brain injury (TBI) but not subarachnoid hemorrhage (SAH). After completion of the Learning Set task, mice from experiments 4 and 5 were killed and their brains were sliced. Representative histologic images of cresyl violet–stained coronal sections are shown (A). Scale bar=500 μm. TBI—but not SAH—caused significant loss of morphologically intact neurons in hippocampal CA1 (B) and in ipsilateral CA1 volume (C). Data represent mean±s.e.m. *P<0.05 versus sham TBI by t-test. NS, P>0.05 vs. sham by t-test. Scatter plots relating neuronal counts to mean distance swum in the Morris water maze after SAH (D) and TBI (E). Dashed lines represent 95% confidence bands. Pearson's correlations are shown, NS, P>0.05.

Long-Term Traumatic Brain Injury Outcome

In experiment 5, included as a positive control, mice were subjected to controlled cortical impact TBI or sham surgery. All mice survived (N=10 in each group). Traumatic brain injury mice evinced significant neurologic deficits (Figure 5A; P<0.001, omnibus repeated measures ANOVA). Traumatic brain injury caused significant deficits in the Learning Set MWM (Figure 5B; P=0.0027, omnibus repeated measures ANOVA). Probe performance exceeded chance levels in sham mice; by contrast, TBI mice performed at or below chance levels (Figure 5C; P<0.00001 versus sham, t-test, for both target quadrant and target area). Traumatic brain injury caused severe injury in ipsilateral CA1 (Figure 6A), ranging from reduction in CA1 thickness (arrow) to near-complete obliteration (arrowhead): significantly fewer neuronal cells were counted in CA1 in TBI than in TBI mice (Figure 6B; P<0.001, t-test), and CA1 volume was strikingly reduced in TBI compared with sham mice (Figure 6C; P<0.001, t-test). A significant correlation was noted between CA1 neuronal cell counts and MWM performance (Figure 6E; P<0.001, Pearson's correlation).

Discussion

Results of the present study provide strong evidence that, in mice, the endovascular perforation SAH model fails to cause long-term learning or memory deficits as assessed by the MWM. This is a major experimental limitation, as it is by far the most commonly utilized mouse model of SAH owing to its relative ease in development, close recapitulation of the physiologic events involved with aneurysmal SAH (e.g., arterial perforation, localization of blood in the basal cisterns, acute rise in intracranial pressure, and transient global ischemia), and applicability to the use of powerful, targeted genetic manipulations. We rigorously examined possible reasons for our finding, including high mortality (albeit comparable to the 20% to 33% reported in rat studies^{16, 18, 19, 20}), lack of sensitivity of the MWM protocol, and extent of neuronal cell loss in hippocampal CA1. We increased survival by modifying postoperative support; moreover, we employed different MWM protocols, the standard Place task and the more sensitive Learning Set task. Neither manipulation identified MWM deficits. We examined the extent of neuronal cell loss in hippocampal CA1, finding that it was inconsistent and did not correlate to MWM deficits. As a positive control, we examined controlled cortical impact TBI, showing reliable CA1 neuronal cell loss and significant MWM deficits, which are consistent with and extend our past results with this model.²⁴ In total, our results indicate that endovascular perforation SAH in mice does not produce demonstrable long-term learning or memory deficits as assessed by the MWM and that the cause of this is likely inconsistent neuronal cell loss in CA1. The latter may have resulted from significant delayed mortality that preferentially affected mice having the most severe neurologic deficits and thus most likely to have had significant CA1 neuronal cell loss and cognitive deficits. As far as we are aware, this is the first report examining long-term neurobehavioral outcome in a mouse model of SAH.

Cognitive deficits afflict half of the SAH survivors⁶ and are discernable as long as 6 years after ictus.²⁶ Such deficits are seen even in patients with ‘good outcome': those with a Glasgow Outcome Scale of 5 nonetheless have demonstrable deficits in multiple assessments of executive function at 6 months.²⁷ These cognitive deficits have major implications with regard to patient-relevant outcomes as they strongly predict long-term morbidity, including poor self-reported quality of life, reduced activities of daily living, impaired instrumental activities of daily living, and the inability to return to work in at least 50% of patients.²⁸

Though the frequency and gravity of long-term cognitive deficits in SAH patients are well documented, their underlying neurophysiologic processes have only recently begun to be elucidated. Multiple lines of evidence implicate hippocampal impairment as playing a central role. First, hippocampal volume is reduced in SAH patients at 1 year, and this reduction correlates to performance on visual memory tests²⁹—a particularly compelling result given the well-known central involvement of the hippocampi and other medial temporal lobe structures in many forms of declarative memory.³⁰ Second, multiple molecular alterations in the hippocampi occur after rat SAH, including loss of synaptic function and long-term potentiation.^{31, 32} Third, cognitive deficits have been strongly correlated to hippocampal CA1 neuronal cell number after experimental SAH.¹⁵ To be sure, other structural areas may also play a role, including left hemispheric lesions,³³ reductions in total gray and white matter volume³⁴ (in patients), and cortical neuronal cell loss (in rats).¹⁵ In total, these data provide substantial evidence that hippocampal damage plays a key role in SAH-induced cognitive deficits, but that alterations in other brain areas also contribute.

We thus examined hippocampus-dependent long-term neurobehavioral outcomes after endovascular perforation SAH in the mouse. Despite attenuating mortality and employing two separate protocols of the MWM—the most commonly used metric for cognitive outcome in rodents—no significant deficit by any measure was seen after SAH. Most likely, this was because of insufficient CA1 neuronal cell loss in long-term SAH survivors. Neuronal cell loss after endovascular perforation SAH in mice has only been examined in two previous reports, both of which assessed cell death at acute time points and only in cerebral cortex (and not in CA1). We documented significant neuronal cell death (as assessed by terminal deoxynucleotidyl transferase dUTP nick end labeling, TUNEL) in the ipsilateral parietal cortex 72 hours post-SAH,²³ and Altay et al³⁵ reported significant TUNEL-positive neuronal cell death in the ipsilateral basal cortex 24 hours post-SAH. When considered in the context of our present results—showing no chronic SAH-induced decrease in morphologically intact CA1 neurons—these results suggest that some combination of severity or location of injury (hippocampus versus cortex and ipsilateral versus bilateral) and timing or method of assessment (acute versus chronic, actively dying versus intact neurons) may account for the lack of MWM deficits in the endovascular perforation mouse model.

Another possible explanation for our negative findings is the long-term mortality associated with this model. Although we succeeded in mitigating delayed mortality by increasing post-SAH support, it nonetheless remained substantial (∼20%). It is possible that significant neuronal cell loss occurred in the mice that died, a contention substantiated by the strikingly greater sensorimotor deficits seen in such mice compared with SAH survivors. If this were the case, by assessing neuronal cell survival ∼1 month after SAH (as opposed to acutely as in previous reports with this model), we would have inadvertently selected against mice with quantifiable neuronal cell loss and attendant MWM deficits.

Others have examined neuronal cell death in alternate mouse models of SAH. Using the prechiasmatic injection model, Sabri et al³⁶ reported significant TUNEL-positive neuronal cell death in ipsilateral middle cerebral artery-territory cortex and hippocampus 48 hours post-SAH. Interestingly, far less neuronal cell death was seen in the hippocampus than in the cortex, but it was not specifically quantified in CA1. Following cisterna magna injection of oxyhemoglobin, Huang et al³⁷ documented TUNEL-positive neuronal cell death in the cortex at 24 and 72 hours, but did not specify what area of the cortex was assessed and provided no quantification. Given the paucity of data regarding the extent and especially the location of neuronal cell injury—particularly in CA1, the brain region most implicated in MWM deficits in rat SAH^{15, 31, 32} and rodent TBI³⁸—it is difficult to predict whether either of these alternative mouse models might better recapitulate the long-term cognitive deficits seen in SAH patients.

In contrast, neuronal cell death and long-term outcomes in rat models of SAH have been more thoroughly characterized. Takata et al¹⁵ were the first to report that rats displayed significant deficits in the Learning Set MWM after cisterna magna double-injection SAH, and that MWM performance correlated with hippocampal and cortical neuronal cell counts. Multiple subsequent reports have documented cognitive deficits in the three most commonly used rat models of SAH: cisterna magna injection,^{15, 17} endovascular perforation,^{16, 19} and prechiasmatic cistern injection.^{18, 20} Each of these studies employed the MWM, widely considered the gold standard for spatial learning in rodent models of central nervous system injury;^{39, 40} Sherchan et al¹⁹ also showed deficits in T-maze performance after rat endovascular perforation SAH. Two reports also show treatment effects, of simvastatin¹⁷ and minocycline,¹⁹ establishing the suitability of the MWM as an outcome measure for therapeutic studies in rat SAH.

Our study has several limitations. First, we assessed only the endovascular perforation mouse model of SAH. We chose to examine this model as it is by far the most commonly used mouse model of SAH and it faithfully recapitulates the arterial damage, basal subarachnoid blood localization, increased intracranial pressure, and transient global ischemia associated with aneurysmal rupture in patients.⁸ However, as delineated above, other mouse models of SAH exist, and one may cause MWM deficits. Second, we examined long-term neurobehavioral deficits using only the MWM, albeit with two distinct protocols. The MWM is considered to be the gold standard for examining spatial deficits after rodent central nervous system injury, and has been reported to show dysfunction in numerous central nervous system disease models, including vascular disease, TBI, developmental disorders, Alzheimer's disease, AIDS dementia complex, and more. However, there are other tests of visuospatial memory, and so we cannot exclude the possibility that deficits in our model may be revealed by another measure. Third, we examined a single mouse strain. C57BL/6 is one of the most commonly used strains in mouse SAH and has been shown by multiple groups to recapitulate numerous short-term sequelae; however, it is possible that, while neuronal cell loss and long-term neurobehavioral deficits were absent in this line, other strains would demonstrate them. Fourth, we examined morphologically intact neurons rather than markers of cell death. This was deliberate, as the processes examined by cell death assays would have ended by the time of our assessment. Nonetheless, it is possible that despite not affecting the number of surviving hippocampal neurons, SAH could have had more subtle molecular effects.^{31, 32} This possibility is distinctly unlikely given the absence of MWM deficits (an assessment critically dependent on hippocampal function) documented herein. Finally, SAH-related morality remained at 18% despite significant post-SAH support. It therefore remains possible that significant neuronal cell loss and MWM deficits would have occurred in the mice that died.

Conclusions

Our results strongly indicate that the endovascular perforation mouse model of SAH does not cause reliable CA1 neuronal cell death and does not produce demonstrable long-term neurobehavioral deficits as assessed by the MWM. This is a major experimental limitation, given that long-term cognitive deficits in SAH patients have such a profound effect on quality of life to the individual and have a similarly significant effect to society in regard to the cost of care and loss of productivity related to these patients. Development of an experimental model of SAH-induced cognitive dysfunction that permits use of genetically modified mice to determine the influence of certain genotypes on SAH outcomes (e.g., apolipoprotein E, haptoglobin, and others), and allow targeted mechanistic inquiry, would represent a major advance forward for the field. Until that time, however, translational studies using long-term assessments of outcome—including cognitive measures—should likely use rat models of SAH, aided by the increasing availability of genetically modified lines.

The authors declare no conflict of interest.