Role of Sulfonylurea Receptor 1 and Glibenclamide in Traumatic Brain Injury: A Review of the Evidence

Cerebral edema and contusion expansion are major determinants of morbidity and mortality after TBI. Current treatment options are reactive, suboptimal and associated with significant side effects. First discovered in models of focal cerebral ischemia, there is increasing evidence that the sulfonylurea receptor 1 (SUR1)—Transient receptor potential melastatin 4 (TRPM4) channel plays a key role in these critical secondary injury processes after TBI. Targeted SUR1-TRPM4 channel inhibition with glibenclamide has been shown to reduce edema and progression of hemorrhage, particularly in preclinical models of contusional TBI. Results from small clinical trials evaluating glibenclamide in TBI have been encouraging. A Phase-2 study evaluating the safety and efficacy of intravenous glibenclamide (BIIB093) in brain contusion is actively enrolling subjects. In this comprehensive narrative review, we summarize the molecular basis of SUR1-TRPM4 related pathology and discuss TBI-specific expression patterns, biomarker potential, genetic variation, preclinical experiments, and clinical studies evaluating the utility of treatment with glibenclamide in this disease.


Introduction
The public health impact of traumatic brain injury (TBI) on death and disability is immense. With more than 50 million cases annually, TBI is projected to remain the leading global cause of This is a proposed model of a SUR1-TRPM4-AQP4 hetero-multimeric complex showing AQP4 tetramers (blue channels) interposed within the SUR1-TRPM4 octamer (top view on the left, coronal view on the right). This model demonstrates SUR1-TRPM4 activation by increased intracellular Ca 2+ (green arrow), which results in depolarization from Na + influx and raised intracellular oncotic pressure. This, in turn, drives intracellular water influx (blue arrows) via AQP4 channels causing astrocyte swelling. Adapted with permission from Stokum et al., 2018. (C) Schematic showing the spectrum of ionic edema, vasogenic edema, and blood brain barrier destruction resulting in hemorrhage. The role of SUR1-TRPM4 is highlighted. Ionic edema, driven by osmotic forces, involves transcapillary flux of ions across the capillary membrane into the interstitium as demonstrated by luminal and abluminal SUR1-TRPM4 (orange) and water (blue) channels. Cytotoxic edema/cellular swelling (shown in the endothelial cell) involves movement of ions into cells creating an osmotic gradient that requires water to enter (in neurons, astrocytes, or endothelial cells). Vasogenic edema results from water and proteinaceous fluid traversing across a disrupted blood brain barrier that may occur in the setting of tight-junction degradation as well as oncotic edema of endothelial cells. In its extreme forms, when oncotic edema and cell death occurs in endothelial cells (shown) and astrocytes podocytes (not shown), complete breakdown of the blood-brain barrier results in erythrocyte extravasation, and hemorrhage into the interstitium. All these processes have been shown to involve SUR1-TRPM4, and thus may be inhibited by glibenclamide (also known as glyburide). (D)Schematic diagram of the two SUR1 regulated heteromeric channels expressed in the CNS after TBI: SUR1-KIR6.2 vs. SUR1-TRPM4. There are four subunits of SUR1 (pink) and four subunits of the pore-forming subunits KIR6.2 (lavender, left) or TRPM4 (green, right). ATP depletion activates both channels. SUR1-TRPM4 activation results in cell depolarization; the opposite effect (hyperpolarization) is achieved with SUR1-KIR6.

Neuroinflammation
The complex role of post-traumatic neuroinflammation involves both deleterious secondary injury as well as repair/recovery, thus making it a nuanced but important target for therapeutic modulation [103].
While not studied specifically in TBI, in vitro and in vivo models of lipopolysaccharide induced neuroinflammation demonstrate microglial activation of toll-like receptor 4 (TLR4), and de novo upregulation of SUR1-TRPM4, initiating a signaling cascade that increases nitric oxide (NO) production and contributes to harmful peroxynitrite-mediated protein radical formation [100,104,105]. TLR4 induced cellular depolarization via SUR1-TRPM4 decreases the inward driving force for intracellular Ca 2+ influx. This results in low-amplitude repetitive oscillations in Ca 2+ , which activates calcineurin (a phosphatase). Calcineurin dephosphorylates nuclear factor of activated T-cells (NFATc1), allowing nuclear translocation and NFATc1-dependent upregulation of nitric oxide synthase-2 (Nos2) mRNA, NOS2 protein, and NO ( Figure 2B, green) [100]. In this model, SUR1-TRPM4 inhibition (via GLI) increased intracellular Ca 2+ , as predicted, but reduced NFATc1, likely via persistent activation of calmodulin dependent protein kinase-II (CAMKII) ( Figure 2B, red). Activated CAMKII phosphorylates and inactivates calcineurin, thereby preventing calcineurin dependent NFATc1 dephosphorylation and nuclear translocation. Here, GLI reduced microglial NOS2 expression/NO production, suggesting a mechanism by which it may be protective in other neuroinflammatory conditions such as TBI. SUR1-TRPM4 inhibition (via GLI) increased intracellular Ca 2+ , as predicted, but reduced NFATc1, likely via persistent activation of calmodulin dependent protein kinase-II (CAMKII) ( Figure 2B, red). Activated CAMKII phosphorylates and inactivates calcineurin, thereby preventing calcineurin dependent NFATc1 dephosphorylation and nuclear translocation. Here, GLI reduced microglial NOS2 expression/NO production, suggesting a mechanism by which it may be protective in other neuroinflammatory conditions such as TBI. Some key Sur1 signaling pathways after brain injury. This schematic illustrates upstream regulators of Sur1 (Abcc8) transcription such as Hif1-α, SP1 and TNF-α. The precipitating injury (e.g., hypoxia, TBI, mechanical stimulus) is listed in green. Downstream consequences of inhibiting upregulated Figure 2. Putative SUR1 molecular signaling pathways beyond cerebral edema in brain injury. (A) Some key Sur1 signaling pathways after brain injury. This schematic illustrates upstream regulators of Sur1 (Abcc8) transcription such as Hif1-α, SP1 and TNF-α. The precipitating injury (e.g., hypoxia, TBI, mechanical stimulus) is listed in green. Downstream consequences of inhibiting upregulated Sur1 (via glyburide/glibenclamide or genetic knockout) on three major categories (blood brain barrier integrity, neuroinflammation, and cell death) are shown in red. Specific references that have demonstrated these effects are listed contextually and categorized by disease model. (B) Neuroinflammation, Sur1-Trpm4, and nitric oxide synthase-2 (NOS2). This illustration shows effects of Sur1-Trpm4 activation on nitric oxide (NO) production in microglial cells (pathway in green). Toll like receptor-4 (TLR4) activation induces de novo upregulation and activation of Sur1-Trpm4. Resultant depolarization, decreases the inward driving force for Ca 2+ , yielding low-amplitude repetitive oscillations of Ca 2+ . This activates calcineurin (CN), which dephosphorylates cytoplasmic NFAT and allows nuclear translocation. Nuclear NFAT is a transcription factor that upregulates transcription of Nos2, and thereby increases expression of NOS2 and production of NO. Inhibition of Sur1-Trpm4 by glyburide (glibenclamide) shown in the red pathway, increases intracellular calcium which activates calmodulin dependent protein kinase-II (CAMKII). CAMKII phosphorylates CN thereby inactivating it and preventing dephosphorylation and nuclear translocation of NFAT. Thus, Nos2 transcription is inhibited and NO production reduced.

BBB Integrity
Ongoing BBB breakdown perpetuates hemorrhage progression and contusion expansion after TBI. Although SUR1-TRPM4 contributes to BBB permeability via oncotic death of endothelial cells [14,15,76], it has also been implicated in this process in connection with other mechanisms involving ZO-1 [67,83], MMP-9 and/or tissue plasminogen activator (tPA) [91,95,102] (Figure 2A). MMP-9 is a zinc-dependent endopeptidase (collagenase) zymogen, widely implicated in extracellular matrix degradation and BBB disruption in many CNS diseases. However, its overlap with SUR1-TRPM4 requires further investigation, particularly in TBI [14,106]. While neutrophils are a major source of MMP-9 [107], it is also secreted from neurons, microglia, pericytes, and endothelial cells; indeed, it is prominently expressed in the microvascular endothelium during the first 24 h after focal ischemia [102,108]. In a model of I/R, tPA induced SUR1-TRPM4 channel opening and MMP-9 secretion from activated brain endothelial cells ( Figure 2C) [102]. Conversion of plasminogen to plasmin (a serine protease) by tPA then induced canonical activation of protease activated receptor 1 (PAR1) by cleavage at Arg41. This triggered phasic MMP-9 secretion, and also opened SUR1-TRPM4 channels. The latter, via Na + influx and a negative feedback loop, decreases the electrochemical driving force for intracellular Ca 2+ influx. SUR1-TRPM4 inhibition by GLI (and shRNA against Abcc8) unexpectedly reduced phasic MMP-9 secretion from activated endothelium. Although the precise mechanism of the benefit remains unestablished, it is thought to be related to increased phosphorylated-CamKII and subsequent desensitization and internalization of PAR1 [102]. Nonetheless, it remains to be determined whether/to what extent this mechanism and GLI modulates BBB permeability after TBI. While other protease-mediated interactions (such as calpains) are also indisputably important contributors to BBB permeability after TBI, unlike MMP-9, they have not been evaluated in the context of SUR1-TRPM4 interactions. The link between caspases and SUR1-TRPM4 ± GLI have been evaluated in the context of apoptosis, not BBB permeability (Section 2.2).

SUR1-TRPM4 Expression in TBI
Patterns of increased SUR1±TRPM4 expression in post-traumatic human contusions are similar to those described in rat TBI models (Table 1) [50,51,[60][61][62]. Scientific research understanding the complexity of SUR1's association with its pore-forming subunit partners in TBI is limited, but rapidly evolving. As discussed in this section, previous studies have identified overexpression of KIR6.2 and SUR1 in post-traumatic human contusions (particularly astrocytes), however TRPM4 was not evaluated by this group [51,62]. In a separate study, TRPM4 overexpression was reported in a rodent model of TBI [109]. Determining precise temporal, spatial, and cellular expression patterns of both SUR1-TRPM4 and SUR1-KIR6.2 (and their functional implications if expressed simultaneously) after TBI is particularly important given the opposite effects of the two channels ( Figure 1D). This will also impact their potential utility as biomarkers. In 2019, Gerzanich et al. evaluated SUR1, TRPM4, and KIR6.2 protein expression patterns in human contusion as well as rodent models of TBI; their findings are discussed in detail below (summarized in Table 1) [50]. Based on current information from these studies, SUR1-TRPM4 and SUR1-KIR6.2 multimers are not always expressed together after TBI, however simultaneous expression of SUR1, TRPM4, and KIR6.2 have been reported within certain cell-types [50].
In general, SUR1 colocalization (by immunofluorescence) and coassembly (by FRET imaging) with TRPM4 but not KIR6.2 has been noted in microvessels. Microglial cells and astrocyte tend to express all three channel subunits-SUR1, TRPM4, and KIR6.2-the functional implications of which are currently unestablished. Astrocyte expression has been most prominently noted in penumbral tissue. Increased endothelial expression of channel subunits typically preceded expression in microglia or astrocytes; this may impact the timing of edema endotypes (vasogenic±contusion expansion vs. cellular/cytotoxic), neuroinflammatory responses and repair mechanisms.
In the following subsections, expression of post-traumatic SUR1, TRPM4, and KIR6.2 is discussed-including 'penumbral' expression. A traditional understanding of traumatic 'penumbra' typically refers to brain tissue surrounding the primary lesion (e.g., contusion) that is damaged, ischemic, or susceptible to various forms of secondary injury but retains the potential for recovery i.e., the impact of insult on the 'penumbra' is not irreversible. However, this definition is likely to evolve and develop molecular sophistication. In early work on the SUR1-TRPM4 pathway, Patel et al., defined penumbra (at time = 0) as the region of tissue outside primary contusion where NFκB and SP1 were upregulated and secondary hemorrhage occurred. In this work, a parapenumbra was also identified where SP1 was upregulated but NFκB was absent-here, cell death may occur however hemorrhage progression was absent [60]. Recent work by the same group classifying human tissue (not at time = 0, but rather after secondary hemorrhage was completed), operationally defined the penumbra as GFAP+ and TUNEL− tissue [50]. Strictly speaking, since secondary hemorrhage had already occurred, this tissue would be more consistent with parapenumbra based on the definition used in the earlier study.

Increased SUR1±TRPM4 Expression in Rodent Models of TBI
To date, evaluation of SUR1±TRPM4 expression in preclinical TBI has occurred exclusively in rats and predominantly in models of focal cortical impact, with only a single study in diffuse central fluid percussion injury (FPI) [50,60,61,109]. Despite different TBI severity models in these studies, SUR1 [50,60,61] and TRPM4 [50,109] are consistently upregulated in different cell types of the neurovascular unit.
In focal contusion (10 g weight, 5 cm drop, 1 m/s impact velocity), SUR1 expression was observed in tissue immediately beneath the impact as early as 3 h, predominantly in microvessels [61]. By 24 h, this became prominent in both neurons and capillaries in regions beyond the ipsilateral cortex, including the hippocampus, thalamus, and contralateral dorsomedial thalamus [61]. Further work in contusion by the same group (controlled cortical impact, CCI, 5 mm impactor, 4.5 mm depth, 1 m/s impact velocity) parsed out temporal and cellular differences in SUR1-TRPM4 expression in contusion core (GFAP-, TUNEL+) vs. penumbra (GFAP+, TUNEL-) as illustrated in Figure 3A [50]. In the core, SUR1 expression peaked at 6 h, remained elevated at 12 h, declined at 24 h and increased again at 72 h. During the first 24 h, SUR1 expression colocalized with TRPM4, and was predominantly noted in microvessels, which may have important implications for the treatment of vasogenic edema (see Section 3.5). There was no evidence of microvascular KIR6.2 expression. By 72 h, SUR1 expression was primarily noted in small round cells thought to be microglia/macrophages. Here, double labelling with TRPM4 or KIR6.2 demonstrated that these cells expressed all three channel subunits. In contrast, penumbral expression of SUR1 at all timepoints was noted in GFAP+ stellate cells thought to be astrocytes. These cells also expressed all three channel subunits: SUR1, TRPM4, and KIR6.2. Additional research is warranted to investigate the functional interactions and implications of this co-expression in both astrocytes and microglia/macrophages post-TBI, particularly since similar findings have been noted in human post-traumatic tissue (Section 3.2).  There is minimal increase in extravasated blood in the GLI treated group at 24 h vs. baseline. However, extravasated blood volume almost doubles by 24 h in vehicle rats. Differences between SUR1 protein and Abcc8 mRNA expression in a milder model of focal impact were also elevated vs. sham [60]. Here, it was preceded by transcription factor specificity protein-1 (Sp1) expression, and was predominantly noted in hippocampal neurons at 6 h, peaking at 12 h, and declining by~24 h post injury. While SUR1 expression has not been quantified in FPI, TRPM4 results were surprising in that increased expression persisted for up to 8 weeks after injury [109]. In this model, expression was noted in astrocytes, with differences noted between hippocampal grey vs. white matter (fissure). Higher TRPM4 levels were observed in the latter. TRPM4 expression was correlated with astrocyte swelling: TRPM4+ astrocytes had nearly double the soma size (p = 1.2 × 10 −12 ) but did not undergo cell death. Future work exploring physiologic consequences of TRPM4 and/or KIR6.2 colocalization with SUR1 in this model could be valuable to guide therapeutic studies of channel inhibition in diffuse injury.

Increased Human SUR1±TRPM4 Expression in TBI
In a cohort of 26 patients with contusion, neuronal SUR1 expression was almost threefold greater in contusions vs. controls (meningioma, schwanomma, rhabdoid tumor, p < 0.001) [62]. While detected as early as 6 h post injury, neuronal SUR1 expression peaked at 24 h after which it stabilized. Early SUR1 expression was also detected in CD31 + endothelial cells (again, approximately threefold greater expression vs. controls). However, no temporal trend was noted in these cells; microvascular SUR1 expression remained elevated up to 100 h post-injury ( Figure 3B). Semiquantitative immunohistochemistry did not detect any SUR1 expression in control GFAP+ glial cells (astrocytes), activated microglia/macrophages or neutrophils, whereas moderate SUR1 expression was found in all these cells-types after contusion. Most circulating neutrophils were SUR1 negative; however, those from brain parenchyma were SUR1 positive. The same group also identified KIR6.2 overexpression in contusional GFAP+ glial cells (astrocytes), consistent with findings in rodents. TRPM4 expression was not assessed in this study [51].
Deconstructing regional expression of SUR1, TRPM4, and KIR6.2 in contusion core vs. penumbra, analogous to the rodent experiments described above, yielded similar results [50]. Although temporal trends were not evaluated in human tissue, core lesions demonstrated SUR1 and TRPM4 expression, colocalization and heteromer coassembly (by FRET imaging) in microvessels ( Figure 3D). KIR6.2 was not detected in these microvessels. SUR1 and TRPM4 colocalization was also demonstrated in CD68+ small round cells, i.e., microglia/macrophage; however, all three channel subunits (including KIR6.2) were expressed in these cells. Penumbral samples prominently expressed SUR1 in astrocytes, with minimal levels in microvessels or microglia/macrophages. TRPM4 and KIR6.2 levels were also increased in astrocytes, and FRET experiments demonstrated heteromers of both SUR1-TRPM4 and SUR1-KIR6.2. Neuronal expression was not evaluated in this study other than noting minimal expression of KIR6.2.

SUR1-TRPM4 Theranostic Biomarker Potential after TBI
De novo upregulation of SUR1-TRPM4 after TBI (and other CNS disorders) affords it unique potential as a theranostic biomarker. However, serum ± CSF biomarker research regarding this pathway in TBI is currently in its infancy. Pilot work demonstrated markedly elevated SUR1 levels in cerebrospinal fluid (CSF) of 28 patients after TBI vs. undetectable levels in 15 controls with normal pressure hydrocephalus ( Figure 4A) [110]. Remarkably, in some patients, ICP trajectories mirrored SUR1 expression patterns after a temporal delay ( Figure 4B). SUR1 levels were associated with radiographic edema detected on CT scans, and with the initial degree of intracranial hypertension. SUR1 trajectories (rather than absolute values) were associated with both intracranial hypertension and outcome after TBI. No patients with declining SUR1 levels between 48-72 h had any episodes of intracranial hypertension or unfavorable outcome. Unfortunately, assessment of therapeutic intensity level was not available in this study, and the mechanism of SUR1 entry into the CSF remains unknown. While it is possible that the results relating SUR1 to measures of edema reflect an epiphenomenon, there did not appear to be a correlation between SUR1 level and initial injury severity as measured by the Glasgow coma scale (GCS) score ( Figure 4C). Despite the multiple limitations of this exploratory study, including its single center and associative nature, the results are consistent with preclinical work and known pathophysiology in human studies, and warrant further exploration in larger cohorts. It is currently premature to consider SUR1 a biomarker in TBI, and further extensive functional and mechanistic studies are necessary to determine its utility. This work should include detailed phenotypic evaluation and take into consideration the multimeric expression of both SUR1-TRPM4 as well as SUR1-KIR6.2. Since SUR1 is the regulatory subunit of both these heteromeric channels (with opposite functional effects), the utility of SUR1 alone may be limited and dampen enthusiasm for clinical use in isolation. Additionally, it may be valuable to correlate SUR1-related biomarkers with several existing and promising biofluid markers of neuronal (ubiquitin C-terminal hydrolase-L1 [UCHL1], neuron specific enolase), astroglial (S100B protein, GFAP), axonal (neurofilaments, myelin-basic protein), and neurodegenerative (P-and T-tau) injury [111]. Since SUR1 heteromers with TRPM4 and KIR6.2 have been identified post-TBI in multiple cell types, correlation of biofluid levels with known markers of neuronal or astroglial injury may be informative. UCHL1 in particular may be valuable to correlate with SUR1-related biomarkers given its time course of upregulation over 24 h and its utility as an acute biomarker in severe TBI [111,112]. The discovery and validation of SUR1-pathway related biomarkers in the serum and CSF may become easier as the sensitivity and specificity of proteomic and transcriptomic technology improves. SUR1 quantification in serum/plasma samples after TBI has not yet been published, but represents another potentially important avenue for future research.
Other biomarkers related to SUR1-TRPM4 pathophysiology may also be important in TBI, but are currently understudied. In a recent Phase-2 trial of large hemispheric infarction, plasma MMP-9 levels were reduced in patients treated with intravenous GLI [91]. While largely unexplored in TBI, given the importance of MMP-9 in BBB integrity, vasogenic edema, and contusion expansion related secondary injury in this disease, MMP-9 may be a promising biomarker to investigate. A small study in 8 TBI patients demonstrated declining microdialysate MMP-9 levels over time, with a trend towards higher levels in patients with severe brain tissue hypoxia [113]. Another study in 12 TBI patients reported higher levels of pericontusional (vs. contralateral) microdialysate MMP-9 levels, but did not evaluate a relationship with cerebral edema or contusion expansion [114].

Impact of Genetic Variation in ABCC8 (Encoding SUR1) and TRPM4 in TBI
As mentioned earlier, approximately 65% of TBI outcome variability remains unexplained by existing large multivariable models that incorporate primarily non-modifiable injury characteristics identified on presentation [2,3]. It follows that the host-response is a key factor in regulating secondary injury processes such as cerebral edema and contusion expansion that significantly influence outcome after TBI. Genetic variation is increasingly recognized as an important contributor to variability in post-traumatic host response [28,[115][116][117][118][119][120][121][122][123][124].
Targeted investigations into the impact of ABCC8 and TRPM4 genetic variation on cerebral edema, intracranial hypertension, and outcome after TBI have been performed in 385-485 patients with severe TBI from a single-center [28,[122][123][124]. Genotyped ABCC8 single nucleotide polymorphisms (SNPs) were distributed across the length of the gene with good coverage ( Figure 5A) [122,123]. The majority of these were tag-SNPs (i.e., representative SNPs in high linkage disequilibrium [LD] with a group of SNPs/ haplotype). Eight regionally clustered ABCC8 SNPs (rs4148622, rs1799857, rs11024286, rs2237982, rs2283261, rs2283258, rs3819521, rs7105832) were significantly associated with measures of CT edema, ICP and/or outcome (GOS) after TBI [122,123]. Minor allele frequencies were~25-40%. Effect sizes were large and withstood the B-Y correction for multiple comparisons. These significant SNPs (and their associated 33 proxy-SNPs in LD) were all located upstream, between intron-2 and intron-15 of the 39 exon gene ( Figure 5A) [123]. The spatially clustered polymorphisms were in LD with regions of DNA encoding the SUR1 site and transmembrane domain 0 (juxtaposing the TRPM4 binding site) ( Figure 5B). This geography is in stark contrast to most SNPs reported in disorders of glucose metabolism, approximately two-thirds of which regionally cluster downstream towards introns/exons 16-39, in closer proximity with the adjacent gene KCNJ11 (encoding KIR6.2). The true implications of the identified ABCC8 SNPs in TBI remains unknown. Other than rs1799857 (synonymous variant in exon-12), the remaining significant SNPs are intronic, consistent with known SNP distributions in the human genome. Given their strategic spatial location, they may have important functional, splicing or gene expression/regulatory consequences. Further research is necessary to evaluate their impact in biological models and identify causal variants of disease.

Impact of Genetic Variation in ABCC8 (Encoding SUR1) and TRPM4 in TBI
As mentioned earlier, approximately 65% of TBI outcome variability remains unexplained by existing large multivariable models that incorporate primarily non-modifiable injury characteristics identified on presentation [2,3]. It follows that the host-response is a key factor in regulating secondary injury processes such as cerebral edema and contusion expansion that significantly influence outcome after TBI. Genetic variation is increasingly recognized as an important contributor to variability in post-traumatic host response [28,[115][116][117][118][119][120][121][122][123][124]. adjacent gene KCNJ11 (encoding KIR6.2). The true implications of the identified ABCC8 SNPs in TBI remains unknown. Other than rs1799857 (synonymous variant in exon-12), the remaining significant SNPs are intronic, consistent with known SNP distributions in the human genome. Given their strategic spatial location, they may have important functional, splicing or gene expression/regulatory consequences. Further research is necessary to evaluate their impact in biological models and identify causal variants of disease. Here, patients homozygous wild-type for TRPM4 rs8104571 have almost no proportion of their ICP measurements recorded at > 25 mmHg regardless of their ABCC8 rs2237982 genotype. This is also true for patients heterozygous for TRPM4 rs8104571 who are also wild-type for ABCC8 rs2287982 (blue). Those who are heterozygous for TRPM4 rs8104571 and are also heterozygous for ABCC8 rs2287982 SNPs (red) have an average of ~5% of their ICP measurements recorded at > 25 mmHg. However, patients who are both heterozygous for TRPM4 rs8104571 and homozygous-variant for ABCC8 rs2237982, have an average of ~60% of their ICP measurements recorded at > 25 mmHg. Similar interaction effects between TRPM4 rs8104571 and all three significant ABCC8 SNPs in intron 10 (not shown) were obtained for multiple measures of ICP (including peak, average ICP values). Adapted with permission from Jha et al., 2018.
Two TRPM4 SNPs were associated with intracranial hypertension after TBI: rs8104571 and rs150391806. Effect sizes were large (e.g., 7-10 mmHg average ICP, 20-29 mmHg peak ICP with presence of the variant SNP), and they withstood corrections for multiple comparisons [124]. These In the putative octameric structure, the four SUR1 subunits are brown and the four TRPM4 subunits are gray. Amino acid sequences highlighted in red are those encoded by regions of TRPM4 DNA (in proximity to rs8104571 and rs150391806), and those encoded by DNA regions in linkage disequilibrium with the eight significant ABCC8 SNPs in TBI. (C) Representative example of ABCC8 and TRPM4 SNP interaction effect on ICP. This graph demonstrates the direction and magnitude of the interaction effect on ICP between TRPM4 rs8104571 and ABCC8 rs2237982. Here, patients homozygous wild-type for TRPM4 rs8104571 have almost no proportion of their ICP measurements recorded at > 25 mmHg regardless of their ABCC8 rs2237982 genotype. This is also true for patients heterozygous for TRPM4 rs8104571 who are also wild-type for ABCC8 rs2287982 (blue). Those who are heterozygous for TRPM4 rs8104571 and are also heterozygous for ABCC8 rs2287982 SNPs (red) have an average of 5% of their ICP measurements recorded at > 25 mmHg. However, patients who are both heterozygous for TRPM4 rs8104571 and homozygous-variant for ABCC8 rs2237982, have an average of~60% of their ICP measurements recorded at > 25 mmHg. Similar interaction effects between TRPM4 rs8104571 and all three significant ABCC8 SNPs in intron 10 (not shown) were obtained for multiple measures of ICP (including peak, average ICP values). Adapted with permission from  Two TRPM4 SNPs were associated with intracranial hypertension after TBI: rs8104571 and rs150391806. Effect sizes were large (e.g., 7-10 mmHg average ICP, 20-29 mmHg peak ICP with presence of the variant SNP), and they withstood corrections for multiple comparisons [124]. These were both rare variants with minor allele frequencies of 1% and 0.4%, respectively. Rs150391806 in exon 24 is a missense mutation. These two SNPs also demonstrated regional clustering (downstream) between intron 20 and exon 24 of the 25 exon TRPM4 gene corresponding to regions of DNA that encode the TRPM4 channel pore and SUR1-TRPM4 binding interface ( Figure 5B). However, as with ABCC8, the true functional impact of these variants is currently unknown. Of note, rs150391806 has also been associated with cardiac conduction disorders, including progressive familial heart block and Brugada syndrome [124].
A significant interaction effect has been reported between rs8104571 and three of the significant ABCC8 SNPs, all located in intron-10 (rs2237982, rs2293261, rs11024286), that influences intracranial hypertension [124]. A representative example of this interaction is illustrated in Figure 5C, where patients heterozygous for TRPM4 rs8104571 and homozygous-variant for ABCC8 rs2237982 havẽ 60% of their ICP measurements >25 mmHg vs.~0-5% in rs2237982 homozygous-wild types or heterozygotes. Those homozygous for TRPM4 rs8104571 had almost no ICP recordings >25mmHg, regardless of their ABCC8 rs2237982 genotype. This interaction effect was consistent across multiple measures of ICP where average ICP was~30 mmHg vs.~10 mmHg and peak ICP was~60 mmHg vs. 20-30 mmHg [124]. Virtually identical findings were reported for TRPM4 rs8104571 interactions with all three of the intron-10 ABCC8 SNPs across the different measures of ICP [124].
AQP4 genetic variation may also be relevant to cerebral edema pathophysiology related to SUR1-TRPM4. The role AQP4 in post-TBI edema generation is complex and multifaceted: AQP4 on perivascular foot processes is involved in increased intracellular water movement and cellular swelling, whereas decreased AQP4 expression on subpial, subependymal, and glymphatic surfaces reduces water elimination [14]. As mentioned earlier, SUR1-TRPM4 has been reported to co-assemble with AQP4 to form hetero-multimeric complexes mediating water influx and oncotic edema; this has not specifically been studied in models of brain contusion, but has been demonstrated in astrocytes after cold-injury to the cerebellum [55]. Interactions between genetic variations in all three protein structures may thus have major implications regarding edema formation. Previous candidate-gene research has demonstrated that polymorphisms in the AQP4 gene are related to outcome after TBI [116], but has not evaluated effects on measures of cerebral edema [125]. Nonetheless, AQP4 variants and mutations (including gain of function polymorphisms) have been reported to influence water permeability of cell membranes [126]. In ICH, AQP4 variant rs1054827 has been associated with perihematomal edema formation in a small sample of 128 patients [127].
Small sample sizes in TBI (particularly severe injury) have precluded unbiased genome wide association studies (GWAS) to identify novel genetic targets or validate the above findings. Despite 5 years and 11 enrolling sites, there are <300 patients with severe TBI in TRACK-TBI with genetic material available for evaluation. Efforts to increase sample sizes for genetic evaluation in TBI and validate these findings are paramount. These initiatives are currently ongoing in a large international multi-center collaboration known as GAIN (Genetic Associations in Neurotrauma) combining cohorts from TRACK-TBI (North America) and CENTER-TBI (Europe). Nonetheless, the reported findings are provocative, promising and consistent with known SUR1-TRPM4 pathophysiology. They support the theory that ABCC8 and TRPM4 genetic variation may play a role in cerebral edema development, intracranial hypertension and outcome after TBI. The spatial clustering of significant SNPs around critical regions of DNA encoding sequences contained in the channel pore, sulfonylurea receptor, and SUR1-TRPM4 binding interface, and the interaction effect between ABCC8 and TRPM4 variants moderating ICP adds further credence to this hypothesis. Validating effects of genetic variability in this pathway may facilitate risk stratification, prognostication, and guide patient selection for future trials.
In different models, severities, and species of contusional TBI, GLI has improved BBB integrity and markedly reduced hemorrhage progression at multiple time points ranging from 45 min to 7 d [50,67,82,83]. In both mouse and rat CCI models (from two independent laboratories), extravasated blood from a leaky BBB at 24 h was approximately two fold greater in vehicle animals vs. those treated with GLI [61,83]. This difference with GLI treatment was notable by 3 h, became significant by 6 h post-injury, peaked at 12 h after which it stabilized by 24 h (Figure 3C) [61]. Reduced BBB leakiness was accompanied by reduced loss of ZO-1 and occludin in endothelial cells [83]. It should be noted that GLI-treated animals had minimal extravasated blood, and the amount quantified at 24 h was no different than that at 45 min, suggesting minimal progression of BBB breakdown with treatment [61]. Antisense oligodeoxynucleotide against SUR1 produced similar results in this study [61]. Not surprisingly, likely as a result of limited BBB breakdown, hemorrhagic progression of contusion by 24 h was reduced by~58% [50]. Antisense oligodeoxynucleotides against Abcc8/SUR1 and Trpm4/TRPM4 but not Kcjn11/Kir6.2 reduced both hemorrhage progression and hemispheric swelling vs. controls after CCI [50]. Quantification of contusion volume over 7d of GLI treatment has demonstrated consistently lower volumes (~30-40%) at all time points (3 h, 24 h, 72 h, 7 d; all p < 0.01) [82]. GLI has also been tested by the rigorous, blinded, multi-center Operation Brain Trauma Therapy (OBTT) consortium, which evaluates promising therapies in three standardized preclinical models of TBI, including CCI [131]. Most drugs tested by OBTT performed "below or well below expectations based on the published literature" [132]. Although results from OBTT are pending publication for GLI, preliminary reports support a significant benefit in contusion volume reduction [85,130]. GLI has been identified by OBTT as the second highest scoring drug in the consortium (after levetiracetam), and the only agent tested by the consortium to reduce contusion/lesion volume in CCI [130].
GLI significantly decreases cerebral edema 24 h after CCI [50,82,83]. In two of the three studies, there was a 52-66% reduction in ipsilateral hemispheric swelling [50,83]. GLI also reduces diffuse (contralateral) edema exacerbated by detrimental second insults such as hypotension [84]. Most studies evaluating effects of GLI on post-traumatic cerebral edema have used percent brain water (%BW) for quantification. While this is considered the gold standard measure, it is biased against detection of cellular edema. In order to detect increases in %BW, some degree of vasogenic edema/BBB breakdown and perfusion is required; isolated cellular swelling in the absence of this is merely redistribution of water from the interstitial to intracellular space without actually increasing total water content [15]. MRI endophenotyping of edema may be valuable to evaluate such granular effects of SUR1-TRPM4 blockade. Pilot data from a combined injury mouse model of CCI plus hypotension indicate reduction of T2 hyperintensity and volume (measuring vasogenic edema) after treatment with GLI [129]. This was noted as early as 3 h post injury in some mice, and statistically significant at 24 h. GLI effects on cellular swelling measured by mean diffusivity are currently unknown in preclinical models, but MRI based outcome measures may facilitate quantification of cytotoxic edema. Thus, although results of GLI on cerebral edema as measured by %BW are promising, this method may underestimate potential benefits of SUR1-TRPM4 blockade on cellular swelling, e.g., in astrocytes. In a model of FPI, TRPM4+ astrocytes had double the soma size of TRPM4-astrocytes after injury, as assessed by confocal and electron microscopy. While an important methodology for evaluating preclinical benefit on cellular edema and proof-of-concept, microscopy-based techniques are less clinically translatable than radiographic metrics.
Upregulation of SUR1-TRPM4 in endothelial cells as early as 3 h (peaking before 24 h) coincides with the timing of the GLI benefit on vasogenic edema, BBB breakdown, and contusion expansion after preclinical contusional TBI. Based on this pathophysiology, early treatment with GLI is likely imperative to maximize therapeutic benefit on this process. Given delayed SUR1-TRPM4 upregulation in astrocytes and microglia/macrophages relative to endothelial cells, it is possible that continued treatment with GLI may influence late edema formation and/or neuroinflammation and repair. However, this hypothesis requires further investigation, particularly given the unknown implications of concurrent SUR1-KIR6.2 expression in some cell-types.

Clinical Studies of GLI in TBI
Three small clinical trials evaluating GLI in TBI have been published ( Table 3). Two of these studies, conducted in Iran, explored effects of oral GLI in different subtypes of TBI, including contusion-TBI and diffuse axonal injury(DAI)/traumatic microhemorrhages.
Zafardoost et al., randomized 40 patients with moderate-severe DAI to 1.25 mg oral GLI every 12 h (this dose was increased to 2.5 mg if no hypoglycemia was noted at the lower dose) vs. control (no placebo) [133]. Power calculations were designed to detect an effect size of Cohen's d ≥ 0.7 with α = 0.05 and 1 − β = 0.80 in this sample size. The distribution of presenting GCS scores between treatment vs. control groups is not clearly presented (although reported to not be significantly different). Outcome was assessed at 1 week and at discharge using ordinal measures of GCS and the Glasgow outcome scale (GOS) score, all of which were significantly better in the treated group vs. control (all p ≤ 0.004). Long term functional outcomes were not evaluated.
A blinded placebo-controlled randomized trial of oral GLI in contusion-TBI was performed by Khalili et al. in 66 patients with moderate-severe TBI (GCS 5-12) [134]. Multiple outcome measures were obtained, including contusion volumes (initial, day-3, day-7), and 3-month ordinal functional outcome metrics, including GOS, modified Rankin Scale, and disability rating scale. To have 80% power at an α = 0.05, a sample size of 60 was required to detect a 5% difference in outcomes. Patients were randomized to placebo or 10 mg oral GLI for 10 days. No differences were noted in any of the functional outcome measures or absolute contusion volumes. Patients treated with GLI had significantly lower contusion expansion ratios between both baseline-day3 (p < 0.001), and baseline-day 7 (p < 0.003).
Although pharmacokinetic values were not reported in the studies by Zafardoost et al., or Khalili et al., intermittent oral GLI dosing can be problematic, with sharp supratherapeutic peaks (risking hypoglycemia) and prolonged subtherapeutic troughs [36]. Variability in stomach pH can further influence plasma levels of the drug [36,94]. A small Phase-2 multi-center randomized trial reported by Eisenberg et al., utilized an intravenous formulation of GLI, currently known as BIIB093 [135]. This drug has also been used in a published Phase-2 trial in large hemispheric infarction (LHI) and is being evaluated in an ongoing Phase-3 trial of LHI (NCT02864953) as well as a large Phase-2 contusion-TBI trial (NCT03954041) [91,135]. The intravenous formulation enables rapid and stable plasma levels, reducing the risk of hypoglycemia and maximizing therapeutic duration.
The radiographic findings by Eisenberg et al. (with BIIB093) are similar to those reported by Khalili et al.,in contusion [135]. Here, 28 patients (GCS [4][5][6][7][8][9][10][11][12][13][14] were randomized within 10 h of TBI to a 72 h infusion of placebo or BIIB093. Of these, only 14 patients had contusional TBI. Primary outcome was MRI-defined edema and or hemorrhage on a 1.5T scanner at 72 ± 12 h. GOS-extended at 90 days was a pre-specified secondary outcome. Analysis of contusions (7/group) revealed a 1036% increase in lesion volume (hemorrhage plus edema) in placebo vs. 136% in controls; this tenfold difference was not significant (p = 0.15), likely due to small sample size and contusion heterogeneity. It should also be noted that baseline contusion volumes were much higher in the treatment group of this study. Hemorrhage volumes decreased 29.6% with BIIB093 vs. an 11.6% increase with placebo; however, this was not significant. Nonetheless, one may speculate that the pathophysiologic basis for this relates to GLI enhanced RBC phagocytosis by macrophages (Simard et al., [136]). MRI measures of edema in this study included mean diffusivity (MD) as a proxy for overall water diffusivity, free water (FW) as a proxy for extracellular water, and tissue MD (MDt) as a proxy for intracellular water. These measures were compared in lesions vs. uninjured white matter (WM). Untreated lesions had increased edema by all three measures vs. uninjured WM (p < 0.02). In BIIB093 treated patients, there was no difference in any of these measures vs. uninjured WM. There was no improvement in 90-day functional outcome with BIIB093 vs. placebo in this small cohort; however, the study was not powered for this metric.  The compelling molecular and preclinical evidence supporting the benefit of GLI, particularly in contusion models, combined with encouraging clinical data and radiographic results in this subcohort despite very small studies, has generated excitement to test this therapy in a larger Phase-2 study of contusion-TBI. ASTRAL is a multicenter, international, double-blind, multidose, placebo-controlled, randomized Phase-2 clinical trial evaluating the safety and efficacy of BIIB093 in supratentorial contusion-TBI (NCT03954041, antagonizing SUR1-TRPM4 to reduce the progression of intracerebral hematoma and edema surrounding lesions). The trial is sponsored by Biogen and is actively enrolling patients. The proposed enrollment is 160 patients across several centers in North America, Europe, and Japan. Two doses will be tested (3 mg/day and 5 mg/day × 96 h) with matched placebo groups for each dose. In addition to extremely focused patient selection (contusion-TBI), the ASTRAL trial design leverages the mounting evidence that GLI reduces cerebral edema and hemorrhage progression, and has identified its primary outcome to be contusion expansion by 96 h (Figure 6). Multiple secondary outcomes will also be obtained, including acute neurological status, functional outcome, and survival. The temporal course of microvascular SUR1-TRPM4 expression and known risk-window for contusion expansion has also informed trial strategy: the treatment window for BIIB093 is thus 6.5 h. This is significantly earlier than 10 h in the aforementioned BIIB093 trial in TBI reported by Eisenberg et al. [135], as well as the aforementioned ongoing Phase 3 study of BIIB093 in LHI (NCT02864953).  Figure 6. ASTRAL clinical trial design schematic. Diagram outlining the clinical design trial for ASTRAL, a Phase-2 multicenter, international, double-blind, multidose, placebo-controlled, randomized clinical trial evaluating the safety and efficacy of BIIB093 in supratentorial contusion-TBI (NCT03954041, Antagonizing SUR1-TRPM4 to reduce the progression of intracerebral hematoma and edema surrounding lesions). Distribution of participants to high vs. low dose BIIB093 vs. matching placebo is presented; patients are randomized within 6 h of time of injury. The primary end point is documented by 96 h MRI or non-contrast CT and related to proportion of participants with contusion expansion vs. baseline. Additional secondary efficacy endpoints will also be obtained as illustrated. Time-points for key follow up assessments are also shown including 7 days after treatment, prior to hospital discharge, 30-day, 90-day, or 180-day evaluations.

Conclusions
There is substantial preclinical and clinical evidence supporting a pivotal role of the SUR1-TRPM4 pathway in cerebral edema development, contusion expansion, and outcome after TBI. Consistent findings of benefit with GLI treatment-particularly in contusion across models, injury severity, species, and drug doses-suggests robustness and is encouraging for successful clinical Figure 6. ASTRAL clinical trial design schematic. Diagram outlining the clinical design trial for ASTRAL, a Phase-2 multicenter, international, double-blind, multidose, placebo-controlled, randomized clinical trial evaluating the safety and efficacy of BIIB093 in supratentorial contusion-TBI (NCT03954041, Antagonizing SUR1-TRPM4 to reduce the progression of intracerebral hematoma and edema surrounding lesions). Distribution of participants to high vs. low dose BIIB093 vs. matching placebo is presented; patients are randomized within 6 h of time of injury. The primary end point is documented by 96 h MRI or non-contrast CT and related to proportion of participants with contusion expansion vs. baseline. Additional secondary efficacy endpoints will also be obtained as illustrated. Time-points for key follow up assessments are also shown including 7 days after treatment, prior to hospital discharge, 30-day, 90-day, or 180-day evaluations.

Conclusions
There is substantial preclinical and clinical evidence supporting a pivotal role of the SUR1-TRPM4 pathway in cerebral edema development, contusion expansion, and outcome after TBI. Consistent findings of benefit with GLI treatment-particularly in contusion across models, injury severity, species, and drug doses-suggests robustness and is encouraging for successful clinical translation. Results from ASTRAL, the ongoing Phase-2 trial of intravenous GLI (BIIB093) in brain contusion are eagerly awaited. Future research on theragnostic biomarkers and identifying/validating functionally consequential genetic variants in this pathway may eventually inform clinical risk-stratification, prognostication, identification and characterization of treatment-responders, patient selection for precision-medicine based clinical trials, and the development of novel gene-based therapies.