doi:10.1093/braincomms/fcaa078 BRAIN COMMUNICATIONS 2020: Page 1 of 12 | 1 BRAIN AIN COMMUNICATIONS Cold stress protein RBM3 responds to hypothermia and is associated with good stroke outcome Paulo Ávila-Gómez,1* Alba Vieites-Prado,1* Antonio Dopico-López,1 Saima Bashir,2 Héctor Fernández-Susavila,1 Carme Gubern,2 Marı́a Pérez-Mato,1,3 Clara Correa-Paz,1 Ramón Iglesias-Rey,1 Tomás Sobrino,1 Alejandro Bustamante,4 Sven Wellmann,5 Joan Montaner,4 Joaquı́n Serena,2 José Castillo,1 Pablo Hervella1 and Francisco Campos1 *These authors contributed equally to this work. RNA-binding motif protein 3 is a molecular marker of hypothermia that has proved neuroprotective in neurodegenerative disease models. However, its relationship to the well-recognized therapeutic effect of hypothermia in ischaemic stroke had not been studied. In this work, the expression of RNA-binding motif protein 3 was investigated in ischaemic animal models subjected to sys- temic and focal brain hypothermia, specifically the effects of RNA-binding motif protein 3 silencing and overexpression on ischae- mic lesions. Moreover, the association of RNA-binding motif protein 3 levels with body temperature and clinical outcome was evaluated in two independent cohorts of acute ischaemic stroke patients (n ¼ 215); these levels were also determined in a third co- hort of 31 patients derived from the phase III EuroHYP-1 trial of therapeutic cooling in ischaemic stroke. The preclinical data con- firmed the increase of brain RNA-binding motif protein 3 levels in ischaemic animals subjected to systemic and focal hypothermia; this increase was selectively higher in the cooled hemisphere of animals undergoing focal brain hypothermia, thus confirming the direct effect of hypothermia on RNA-binding motif protein 3 expression, while RNA-binding motif protein 3 up-regulation in is- chaemic brain regions led to functional recovery. Clinically, patients with body temperature <37.5 C in the first two cohorts had higher RNA-binding motif protein 3 values at 24 h and good outcome at 3 months post-ischaemic stroke, while RNA-binding motif protein 3 levels in the cooled third cohort tended to exceed those in placebo-treated patients. These results make RNA-binding motif protein 3 a molecular marker associated with the effect of hypothermia in ischaemic stroke and suggest its potential applica- tion as a promising protective target. 1 Clinical Neurosciences Research Laboratory, Health Research Institute of Santiago de Compostela (IDIS), Hospital Clı́nico Universitario, Santiago de Compostela 15706, Spain 2 Stroke Unit, Department of Neurology, Hospital Universitari Dr. Josep Trueta de Girona, IDIBGI, Girona 17007, Spain 3 Neuroscience and Cerebrovascular Research Laboratory, Department of Neurology and Stroke Center, La Paz University Hospital, Hospital La Paz Institute for Health Research (IdiPAZ), Autonomous University of Madrid, 28046, Madrid, Spain 4 Neurovascular Research Laboratory, Vall d’Hebron Institute of Research (VHIR), Universitat Autònoma de Barcelona, Barcelona 08035, Spain 5 Department of Neonatology, University Children’s Hospital Regensburg (KUNO), University of Regensburg, 93049 Regensburg, Germany Correspondence to: Dr. Pablo Hervella, Health Research Institute of Santiago de Compostela (IDIS), Clinical University Hospital (CHUS), Travesı́a da Choupana, s/n, Santiago de Compostela E15706 A Coru~ na, Spain E-mail: pablo.hervella.lorenzo@sergas.es Received February 17, 2020. Revised April 22, 2020. Accepted May 4, 2020. Advance Access publication June 4, 2020 C The Author(s) (2020). Published by Oxford University Press on behalf of the Guarantors of Brain. V This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com 2 | BRAIN COMMUNICATIONS 2020: Page 2 of 12 P. Ávila-Gómez et al. Correspondence may also be addressed to: Dr. Francisco Campos E-mail: francisco.campos.perez@sergas.es Keywords: hypothermia; neuroprotection; RBM3; stroke; translational study Abbreviations: AAV ¼ adeno-associated virus; CIRP ¼ cold-inducible RNA-binding protein; CSPs ¼ cold shock proteins; FC ¼ focal control; FH ¼ focal hypothermia; GFP ¼ green fluorescent protein; IS ¼ ischaemic stroke; RBM3 ¼ RNA-binding motif pro- tein 3; SC ¼ Systemic normothermic control groups; SH ¼ systemic hypothermia; WB ¼ western blot Graphical Abstract Introduction RMB3 is a glycine-rich protein (17 kDa) that promotes global protein synthesis at 32 C by accelerating ribosome Hypothermia is one of the most effective neuroprotective assembly, stabilizing mRNA and decreasing microRNA treatments in preclinical models of ischaemic stroke (IS) expression (Dresios et al., 2005). In perinatal asphyxia (Castillo et al., 1998; van der Worp et al., 2007; models, RBM3 mediates rescue from apoptotic neuronal Campos et al., 2012; Vieites-Prado et al., 2016). death during therapeutic cooling (Wellmann et al., 2010). However, its clinical translation remains hampered by Up-regulated RBM3 expression is associated with hiber- side effects such as shivering, hypotension, arrhythmia nation: it helps restore brain activity in awakening ani- and increased risk of pneumonia that usually require sed- mals and protects cells against cold damage (Peretti ation or anaesthesia (Darwazeh and Yan, 2013; et al., 2015; Wood, 2015). In mouse models of Geurts et al., 2017). We recently showed that non-inva- Alzheimer’s disease, RBM3 mediates protective cooling sive focal hypothermia applied to the cerebral ischaemic effects by reducing synaptic loss (Knott, 2015; Zhu et al., region decreases the side effects of cooling stress in 2016a). More recently, RBM3 has been found to stimu- awake animals while retaining benefits similar to those of late neuronal differentiation and inhibit hypoxic-ischae- systemic hypothermia; unfortunately, clinical translation mia-induced apoptosis in the two main areas of persistent remains a daunting challenge since the thickness of adult neurogenesis, the subventricular and subgranular human skull effectively insulates the brain, thus requiring zones (Zhu et al., 2019). However, although cooling is a the use of prolonged skin-damaging cold to achieve targe- well-recognized therapy in cerebral ischaemia, a role for tintracerebral temperatures (Vieites-Prado et al., 2016). RBM3 has yet to be sought in this regard. We therefore Despite theseclinical limitations, the physiology under- evaluated the effect of systemic and focal hypothermia on pinning the therapeutic effects of hypothermia provides cerebral RBM3 expression in ischaemic animals and ana- key protective targets for reducing ischaemic lesions lysed the relationship between changes in blood RBM3 (Han et al., 2012). At the same time as hypothermia levels and body temperature in IS patients. down-regulates global protein synthesis and cell metabol- ism, it up-regulates cold shock proteins (CSPs). The two main CSPs described in mammals are cold-inducible RNA-binding protein (CIRP) and RNA-binding motif Materials and methods protein 3 (RBM3). While CIRP is detrimental in enhanc- ing the inflammatory response (Zhou et al., 2014), inter- Animal handling est in RBM3 has significantly increased due to its critical All experimental protocols were approved by the local role in the protective effect of hypothermia (Zhou et al., Animal Care Committee according to European Union 2014; Knott, 2015; Zhu et al., 2016a). rules (86/609/CEE, 2003/65/CE and 2010/63/EU). The RBM3 in ischaemic stroke BRAIN COMMUNICATIONS 2020: Page 3 of 12 | 3 manuscript conforms to Animal Research: Reporting of 2). Animals were randomly assigned to the experimental In Vivo Experiments (ARRIVE) guidelines. Male 280– groups (n ¼ 3) using the GraphPad software randomiza- 330 g Sprague-Dawley rats were housed individually, in tion tool (www.graphpad.com). stable environmental conditions (23 C, 40% relative hu- midity, 12-h light-dark cycle), with free access to food Brain RBM3 silencing and enhancing and water. Younger 130–150 g rats were used for prema- ture viral injections. in ischaemic animals We evaluated the impact on cerebral ischaemia of silenced Cerebral ischaemia or enhanced RBM3 expression using two different con- structs of adeno-associated virus (AAV) (Vector Biosystems, Transient focal ischaemia (45 min) was induced by intra- PA, USA) and their respective empty vectors: AAV2- luminal occlusion of the middle cerebral artery as CamKII-rat-Rbm3-CamKII-eGFP and AAV2-GFP-U6-rat- previously described (Vieites-Prado et al., 2016; Rbm3-shRNA (GenBank access number for RBM3: Fernandez-Susavila et al., 2017). Cerebral blood flow was NM_053696). All animals weighing 130–150 g were monitored during surgery and evaluated by MRI during injected four times with 2 ml virus in the somatosensory occlusion, before hypothermic/normothermic treatment, to cortex. Stereotaxic injection of 0.2 ml/min was performed confirm uniformity of the initial ischaemic lesion in all using 30 Ga cemented Hamilton syringes (Hamilton, NV, animals (detailed protocol in Supplementary material). USA) under microinjector control (Stoelting Co., Wood Dale, IL, USA). The needle was left in place for 10 add- MRI itional minutes to avoid reflux. Animals were divided into MRI was performed using a 9.4 T system (Bruker four experimental groups: (i) RBM3 overexpression; (ii) BioSpin, Billerica, MA, USA) with 440 mT/m gradients RBM3 overexpressing vector without the insert; (iii) RBM3 combining a linear birdcage resonator (7 cm diameter) for silencing; and (iv) RBM3 silencing vector without the signal transmission with a 2  2 surface coil array for sig- shRNA sequence. T2*-weighted MRI was performed after nal detection (details in Supplementary material). stereotaxic surgery to confirm correct injection positioning or to discard needle-induced haemorrhage. Infected animals were allowed to recover from surgery and express the vec- Inducing systemic and focal tors for 4 weeks before the ischaemic injury was induced. hypothermia Animals were randomly assigned to the experimental Systemic hypothermia was induced to a rectal and brain groups (n ¼ 8) using the GraphPad software randomization temperature of 32 C for 4 h under 2% sevoflurane anaes- tool (www.graphpad.com). thesia animal models using a rectal thermostat-controlled pad (Neos Biotec), as previously described (Vieites-Prado RBM3 levels in IS patients et al., 2016). Animals were then returned to their home RBM3 blood levels were measured in two independent IS cages for spontaneous rewarming. Normothermic systemic cohorts: one from the Clinical Hospital of Santiago de controls were also maintained for 4 h at 37 C (rectal and Compostela, and the other from Dr. Josep Trueta brain temperature). We found differences lower than 2 C University Hospital in Girona, following approval by the between rectal and brain temperature, measured either respective ethics committees. The study was carried out in via non-invasive magnetic resonance thermometry or by accordance with the Declaration of Helsinki of the World means of invasive temperature probes (Vieites-Prado Medical Association, fulfilling the criteria of the STROBE et al., 2016). Based on these results, we selected rectal Statement (Guidelines for reporting observational studies). temperature as a reliable monitoring approach that Application of the inclusion and exclusion criteria reflected brain temperature throughout the experimental (Supplementary Tables 3 and 4) yielded a total of 215 eli- procedures. gible patients (113 from Santiago and 102 from Girona). Brain focal hypothermia was induced for 24 h via a RBM3 levels were determined at admission and at 24 h metallic spiral tube implanted over the temporal skull after stroke onset. Body temperature was measured at ad- and connected to a thermal bath, allowing free move- mission and for the first 24 h. Hyperthermia was defined ment. The tube was also implanted for 24 h in normo- as an axillary temperature 37.5 C and normothermia as thermic focal controls, after which the tube was removed <37.5 C (details in Supplementary material). and the animals returned to their home cages. This protocol selectively cools the targeted hemisphere to 32 C by the spiral device without affecting body temperature RBM3 levels in cooled IS patients (Vieites-Prado et al., 2016). Brain RBM3 expression was Blood samples for RBM3 analysis were obtained from measured by quantitative polymerase chain reaction the phase III EuroHYP-1 trial of therapeutic cooling in IS (qPCR) and western blot (WB) 3 h after cooling (details patients (van der Worp et al., 2014). The trial random- in Supplementary material; Supplementary Tables 1 and ized patients to hypothermic treatment plus standard care 4 | BRAIN COMMUNICATIONS 2020: Page 4 of 12 P. Ávila-Gómez et al. or standard care alone (controls). In the hypothermic variables identified in the bivariate analysis and the arm, cooling was started within 6 h of symptom onset results shown as odds ratios (ORs) with 95% confidence and within 90 min of starting thrombolysis (or within intervals (95% CI). 90 min of hospital admission in those not treated with thrombolysis) using 20 ml/kg refrigerated normal saline Data availability (4 C) IV over 30–60 min or a pre-specified surface cool- ing method. Cooling was maintained at 34–35 C for 24 h The authors confirm that the data supporting the findings using a surface or endovascular technique. Thereafter, of this study are available within the article and its patients were passively rewarmed at a rate of Supplementary material. 0.2 6 0.1 C/h until rectal or bladder temperature reached 36.0 C, after which the cooling device was disconnected. All patients received treatment for acute IS and secondary Results prevention according to published guidelines. Pre-defined per-protocol analyses were performed in all patients Temporal profile of brain RBM3 whose body temperature was maintained at 35.0 C for expression in hypothermia at least 6 h during active cooling. We selected 31 patients We began by measuring brain RBM3 expression at differ- (14 treated and 17 controls) for the study based on ent time-points in healthy animals subjected to systemic blood sample availability. Sample collection for biomark- hypothermia or normothermia þ anaesthesia. After cool- er measurement was pre-specified at three time-points: ing for 4 h, all animals were sacrificed at 0, 3, 6, 12 or basal (pre-cooling baseline); 26 h (2 h after rewarming) 24 h. Rbm3 mRNA expression in normothermic condi- and 72 h after stroke onset. tions after 4 h showed an increase over baseline at 3 h. The increase was greater in cooled animals. The same RBM3 analysis in human blood increasing profile was observed in RBM3 protein levels samples (Fig. 1A, Supplementary Tables 5 and 6). Based on these findings, RBM3 expression was determined 3 h post-cool- Serum RBM3 levels were measured using commercial ing in subsequent study groups. Post-focal hypothermic ELISA kits following manufacturer instructions (Biotez quantification showed selective increases in Rbm3 mRNA RBM3 ELISA, Berlin, Germany). and RBM3 protein levels in the cooled brain region of healthy animals versus controls. Focal normothermic con- Statistical analysis ditions (24 h with the cooling device without hypothermia All preclinical data were analysed in GraphPad Prism in awake conditions) did not significantly affect Rbm3 5.01 as the mean and standard error of the mean. One- mRNA or RBM3 protein expression (Fig. 1B, or two-way ANOVA followed by post hoc Bonferroni Supplementary Tables 5 and 6). All animals which passed evaluation was used to determine significant difference the surgical inclusion criteria completed the hypothermic between multiple groups, with P set at <0.05. experimental procedures. Investigators responsible for the assessment of outcomes (infarct size, behaviour and molecular biology analyses) RBM3 expression in cooled were blinded to treatment groups (coded animals, samples and MRI images). Investigators performing surgeries ischaemic animals could not be blinded to the hypothermic treatment (sys- Having established the effect of temperature in healthy temic and focal) due to the differences in the experimen- animals, Rbm3 mRNA and protein levels were then tal procedure. Sample size was established based on quantified in ischaemic animals undergoing systemic or our previous study where we have already evaluated focal hypothermia and their respective normothermic con- the effect of systemic and focal hypothermia in the same trols. MRI data determined using apparent diffusion coef- experimental ischaemic animal model (Vieites-Prado ficient maps confirmed baseline lesion volumes between et al., 2016). 35% and 45% of the ipsilateral hemisphere in each ani- Clinical data were analysed blind in SPSS 21.0 (IBM, mal (Fig. 2A), before cooling treatment. USA) as percentages for categorical variables and as the In ischaemic animals undergoing systemic normother- mean 6 standard deviation (SD) or median and range mia, Rbm3 mRNA and RBM3 protein expression tended (25th to 75th percentiles) for continuous variables de- to increase in both hemispheres compared with healthy pending on their adjustment to normality, assessed by controls. This increase was significantly enhanced by sys- Kolmogorov–Smirnov and Lilliefors tests; correlations temic hypothermia in both hemispheres (Fig. 2B, were performed using Spearman coefficients. Differences Supplementary Tables 3 and 4). In ischaemic animals between patients according to outcome group were deter- undergoing focal hypothermia, Rbm3 mRNA expression mined by ANOVA and chi-square tests. Logistic regres- increased in the ipsilateral hemisphere ischaemic region sion analysis was adjusted by the clinically significant where the cooling spiral was located. RBM3 protein RBM3 in ischaemic stroke BRAIN COMMUNICATIONS 2020: Page 5 of 12 | 5 Figure 1 Temporal profile of RBM3 expression after hypothermia. (A) Above: Temporal brain Rbm3 mRNA levels after 4 h of SH (target temperature 32 C) or normothermia (target temperature 37 C). Bottom: Temporal brain RBM3 protein expression after 4 h of SH. (B) Above: Brain Rbm3 mRNA expression in healthy animals under FC conditions for 24 h (FC-24h, target temperature 37 C) or FH for 24 h (FH-24h, target temperature 32 C) followed by 3 h of recovery. Bottom: Brain RBM3 protein expression in healthy animals subjected to FC-24h or FH-24h. Rbm3 mRNA levels were relativized to b-actin then normalized to untreated controls. RBM3 protein expression was relativized to b-actin then normalized to control animal expression. WB bands corresponding to b-actin and RBM3 (37 and 17 kDa, respectively) are shown below. Data are shown as mean 6 SEM. **P < 0.01 or ***P < 0.001; using one-way or two-way ANOVA followed by a post hoc Bonferroni test (n ¼ 3/group). FC ¼ focal hypothermia control group; FH ¼ focal hypothermia group; SC ¼ Systemic normothermia control groups; SH ¼ systemic hypothermia group. Full-size WB gels are available in Supplementary material. analysis confirmed the focal hypothermia-induced increase infection followed by confirmatory MRI. AAV injection in RBM3, although it was also observed in the contralat- selectively infected cortical neurons determined by co-lo- eral region (Fig. 2C, Supplementary Tables 5 and 6). calization with NeuN (neuronal marker) and green fluor- escent protein (GFP) as reporter gene (Fig. 3A and B). Analysis of brain RBM3 silencing No GFP co-localization was observed with markers of astrocytes [glial fibrillary acidic protein (GFAP)] or and enhancing in ischaemic damage microglia [ionized calcium-binding adaptor molecule 1 To evaluate the protective effect of RBM3 on infarct vol- (Iba1)]. One month after virus injection, animals under- ume reduction, we induced artificial expression (up-regu- went cerebral ischaemia to evaluate the effect of RBM3 lation and down-regulation) in the ipsilateral ischaemic overexpression or silencing on ischaemic damage. cortical region (before inducing ischaemia) using AAV Following the same protocol used for inducing cerebral 6 | BRAIN COMMUNICATIONS 2020: Page 6 of 12 P. Ávila-Gómez et al. Figure 2 Hypothermia reduces ischaemic lesion volume and enhances RBM3 expression. (A) MRI assessments of ischaemic injury. Apparent diffusion coefficient maps were recorded before treatment (during surgical ischaemia induction) and T2 images were acquired after 24 h. Since animals subjected to systemic normothermia or SH were sacrificed 3 h after treatment (7 h after hypothermia induction), no T2 images at 24 h are available for these groups. (B) Above: Brain Rbm3 mRNA levels in ischaemic animals subjected to systemic normothermia (IþSC-4h, target temperature 37 C) or SH (IþSH-4h, target temperature 32 C) followed by 3 h of recovery. Below: RBM3 protein expression in animals subjected to IþSC-4h or IþSH-4h, followed by 3 h of recovery. (C) Above: Brain RBM3 mRNA levels in ischaemic animals subjected to focal normothermia (IþFC-24h, target temperature 37 C) or FH (IþFH-24h, target temperature 32 C) followed by 3 h of recovery; Below: Brain RBM3 protein expression in ischaemic animals subjected to IþFC-24h or IþFH-24h, followed by 3 h of recovery. Rbm3 mRNA levels were relativized to b-actin then normalized to the expression in untreated controls; RBM3 protein expression was relativized to b-actin then normalized to the expression in controls. WB bands corresponding to b-actin and RBM3 (37 and 17 kDa, respectively) are shown below. Data are shown as mean 6 SEM. **P < 0.01 or ***P < 0.001; using one-way or two-way ANOVA followed by a post hoc Bonferroni test (n ¼ 3/group). FC ¼ focal hypothermia control group; FH ¼ focal hypothermia group; SC ¼ systemic normothermia control groups; SH ¼ systemic hypothermia group. Full-size WB gels are available in Supplementary material. RBM3 in ischaemic stroke BRAIN COMMUNICATIONS 2020: Page 7 of 12 | 7 independent association between maximum temperature at 24 h <37.5 C and RBM3 levels at 24 h 300 pg/ml (obtained from the ROC curve), with an adjusted OR of 15.83 (95% CI: 6.71–37.39; P < 0.0001) (Table 1). We observed a clear relationship between RBM3 levels at 24 h and functional outcome at 3 months measured on the modified Rankin Scale (mRS) (Fig. 5C). In agreement with the trend observed with temperature, mean DRBM3 was positive in patients with good outcome versus nega- tive in those with poor outcome (Fig. 5D). In addition, logistic regression analysis revealed that an DRBM3 value 10 (obtained from the ROC curve) was an independent marker of good functional outcome at 3 months (Table 2), after adjusting for the independent variables associated with poor functional outcome at that time- point (Supplementary Table 10). Figure 3 Histological analysis of RBM3 in brain tissue. Blood RBM3 levels in cooled IS (A) Cortical AAV infection 1 month before ischaemia induction. patients (B) GFP immunofluorescence with neuronal (NeuN), astrocyte (GFAP) and microglial (Iba1) markers. White arrows represent the In the 31 patients from the EuroHYP-1 trial (van der co-localization observed only for GFP and NeuN. Worp et al., 2014), we found no significant differences in RBM3 levels between those treated with hypothermia (n ¼ 14) or placebo (n ¼ 17), although rewarming levels tended to be higher in those that received active treatment ischaemia, all animals were scanned by MRI to confirm (P ¼ 0.131) (Fig. 6A, left). Results were similar (P ¼ 0.153) uniformity of basal infarct volume (Fig. 4A). Analysis of when including only those patients meeting the pre-defined infarct volume 24 h post-ischaemia showed no significant criteria for per-protocol analyses (Fig. 6A, right). We impact on ischaemic damage by local changes in RBM3 observed the same trend (P ¼ 0.278) towards higher expression (Fig. 4B), although rotarod testing revealed RBM3 levels in cooled patients when comparing DRBM3 functional recovery in animals with up-regulated protein values between baseline and rewarming (2 h after complet- expression (Fig. 4C and D). ing the 24-h treatment, i.e. at 26 h) with those in placebo patients (Fig. 6B, left). Results were again similar Association between blood RBM3 (P ¼ 0.601) in patients meeting the pre-defined criteria for per-protocol analyses (Fig. 6B, right). No significant correl- levels, temperature and outcome in ation was found between RBM3 levels and body tempera- IS patients ture except for a negative correlation between RBM3 after IS patient characteristics in both cohorts (Supplementary rewarming (26 h) and temperature at 24 h (R ¼ 0.414, Table 7) showed a negative correlation between basal P ¼ 0.029) (Fig. 6C, left). Separate group analysis revealed RBM3 levels and body temperature at admission similar results in the hypothermic group (R ¼ 0.565, (Spearman coefficient 0.150, P ¼ 0.028) and at 24 h P ¼ 0.44 for RBM3 after rewarming and body temperature (Spearman coefficient 0.662, P < 0.0001). Individual at 24 h) (Fig. 6C, right). No such correlation at these or correlation data (Supplementary Table 8) showed that other time-points was found in the placebo group. normothermic (<37.5 C) patients had higher RBM3 val- ues at 24 h than hyperthermic (37.5 C) patients (Fig. 5A). Interestingly, mean variation between the Discussion RBM3 levels determined at admission and 24 h after RBM3 is an evolutionarily old protein whose relationship stroke onset (DRBM3) was positive in normothermic to cold stress remained undescribed until 1997 (Danno patients versus negative in hyperthermic patients et al., 1997). Given its novelty and presumably complex (Fig. 5B). role in the cell, its functions are still poorly understood. Having observed several variables related to tempera- It has been reported to be involved in ribosome assembly, ture (Supplementary Table 9), we evaluated ORs for the limitation of the unfolded protein response, interaction association between RBM3 and maximum temperature at with microRNA regulators and cell-cycle regulation 24 h <37.5 C by performing a logistic regression analysis (Dresios et al., 2005; Matsuda et al., 2011; Pilotte et al., using the variables identified as clinically and statistically 2011; Wong et al., 2016; Zhu et al., 2016b). significant in the model. The analysis revealed an All knowledge to date concerning RBM3 and 8 | BRAIN COMMUNICATIONS 2020: Page 8 of 12 P. Ávila-Gómez et al. Figure 4 Effect of RBM3 expression on neurological ischaemic damage. (A) MRI assessments of ischaemic injury and ischaemic lesion volume in AAV-infected animals. Apparent diffusion coefficient maps were recorded before treatment (during surgical ischaemia induction) and T2 images were acquired after 24 h. White arrows show the needle track after virus injection. (B) Representation of the ischaemic injury determined in the four groups studied. Functional deficit in AAV-infected animals determined by grip (C) and rotarod tests (D) respect to the basal conditions (before ischaemia) and 48 h after ischaemia. RBM3þ: animals infected with AAV2-CamKII-rat-Rbm3-CamKII-eGFP used to enhance RBM3 expression; RBM3þ/0: control group infected with the same empty vector; RBM3: animals infected with AAV2-GFP-U6-rat- Rbm3-shRNA used to silence RBM3 expression; RBM3/0: control group infected with the same empty vector. Data are shown as mean 6 SEM (n ¼ 8/group). hypothermia has been obtained from preclinical in vitro magnetic resonance thermometry, we showed that focal and in vivo studies. Clinical relevance had not previously hypothermia exerted a local cooling effect in the ischae- been investigated. Our study is the first to demonstrate mic region where the cooling device was located, whereas an association between RBM3 and hypothermia not only systemic hypothermia lowered whole body and brain in cerebrally lesioned adult rats but also in IS patients. temperature. The present study complements these earlier Our group had already established that both hypother- findings by showing that the decreases in brain tempera- mia protocols used in this study, focal and systemic ture recorded by magnetic resonance thermometry during hypothermia, reduce ischaemic lesion volume in animal focal and systemic hypothermia are associated with models (Vieites-Prado et al., 2016). Focal hypothermia increased levels of RBM3. requires maintaining for 24 h to achieve the same protec- Selective RBM3 expression in the ipsilateral hemisphere tion as that afforded by 4 h of systemic cooling. But where the focal hypothermia device is located appears a direct while focal hypothermia can be used in awake and freely effect of temperature on cerebral tissue. Diffusion no doubt moving animals, systemic hypothermia must be performed accounts for the slight increase in RBM3 protein levels observed under anaesthesia to avoid hypothermic stress. Using in the contralateral hemisphere (Figs 1B and 2C). RBM3 in ischaemic stroke BRAIN COMMUNICATIONS 2020: Page 9 of 12 | 9 Figure 5 RBM3 is associated with a good outcome and is temperature-dependent. (A) RBM3 levels at 24 h and (B) DRBM3 in normothermic (<37.5 C) and hyperthermic (37.5 C) IS patients. (C) RBM3 levels at 24 h and (D) DRBM3 in IS patients with poor and good clinical outcome at 3 months. DRBM3 was defined as % variation of RBM3 at 24 h respect to the basal levels. Table 1 Multivariate analysis OR not adjusted 95% CI P OR adjusted 95% CI P Center (cat) 1.32 0.76–2.31 0.325 1.22 0.43–3.50 0.708 Glycaemia 0.99 0.99–0.99 0.002 0.99 0.99–1.00 0.177 Leukocytes 0.82 0.74–0.90 <0.0001 0.79 0.69–0.93 0.004 Fibrinogen 0.99 0.99–0.99 <0.0001 0.99 0.99–1.00 0.070 Systemic fibrinolysis (cat) 2.86 1.53–5.36 0.001 2.81 0.91–8.74 0.074 Endovascular treatment (cat) 0.40 0.17–0.97 0.042 0.49 0.12–1.97 0.317 Previous Rankin 0.76 0.58–0.99 0.040 0.94 0.65–1.37 0.764 NIHSS on admission 0.91 0.87–0.96 <0.0001 0.92 0.86–0.98 0.014 RBM3 at 24 h 300 pg/ml 16.11 7.81–33.26 <0.0001 15.83 6.71–37.39 <0.0001 Dependent variable: maximum temperature in the first 24 h <37.5˚C, not adjusted and adjusted by clinically relevant variables found in our study. Cat ¼ categorized; OR ¼ odds ratio. Additionally, the increased RBM3 observed during sys- did not increase in non-anaesthetized normothermic focal temic normothermia in both healthy and ischaemic ani- control groups. mals could be caused by the sevoflurane used during the The viral infection strategy used in this study to evalu- experimental procedure: sevoflurane mediates cell signal- ate the effect of enhanced or blocked RBM3 expression ling pathways related to AKT, a protein impacting on cerebral ischaemic damage has been tested in other RBM3 activity and expression (Matchett et al., 2009; models. Thus, cerebral overexpression of RBM3 mediated Wang et al., 2010). Supporting evidence is that RBM3 by lentiviral infection restored structural synaptic 10 | BRAIN COMMUNICATIONS 2020: Page 10 of 12 P. Ávila-Gómez et al. Figure 6 RBM3 levels in patients treated with hypothermia versus placebo. (A) RBM3 levels across the pre-specified collection time- points: basal (before cooling), 26 h (2 h after starting rewarming) and 72 h after stroke onset. Left panel includes all 31 patients and the right panel the SH group only, comprising patients fulfilling pre-defined criteria for per-protocol analyses (n ¼ 7). (B) Boxplots represent the median (interquartile range) of DRBM3 values (%) between baseline and rewarming time-points in the SH and placebo arms (left, all cases; right, only patients fulfilling pre-defined criteria for per-protocol analyses). (C) Correlation between RBM3 levels at rewarming and body temperature at 24 h in all patients (left) and those treated with SH (right). plasticity and reversed neuronal deficit in an animal significantly impacted ischaemic infarct volume, possibly model of Alzheimer’s disease (Peretti et al., 2015). In the due to the wide ischaemic lesion compared with the present study, the viral infection we implanted in the cor- reduced population of infected neurons. However, a trend tical region was successful (based on GFP gene reporter in functional recovery was observed in animals with up- data) in enhancing and inhibiting local RBM3 expression. regulated RBM3. Although viral tissue infection was optimized in order to The analysis of RBM3 levels in two independent maximize GFP expression with minimal tissue damage cohorts of IS patients showed an association between caused by the needle injection, neither approach increasing DRBM3 and good outcome at 3 months, RBM3 in ischaemic stroke BRAIN COMMUNICATIONS 2020: Page 11 of 12 | 11 Table 2 Multivariate analysis OR not adjusted 95% CI P OR adjusted 95% CI P Center (cat) 0.45 0.26–0.78 0.005 0.84 0.25–1.98 0.682 Age 0.97 0.94–0.99 0.030 1.00 0.96–1.05 0.955 Temperature in the first 24 h <37.5˚C (cat) 3.86 2.15–6.92 <0.0001 0.61 0.22–1.70 0.341 Cardioembolic versus non-cardioembolic 3.69 1.88–7.24 <0.0001 2.05 0.75–5.58 0.162 Previous Rankin 0.33 0.22–0.51 <0.0001 0.37 0.22–0.64 <0.0001 NIHSS on admission 0.84 0.79–0.89 <0.0001 0.87 0.81–0.95 0.001 DRBM3 10% 18.87 9.45–37.68 <0.0001 26.52 9.32–75.44 <0.0001 Dependent variable: Clinical outcome at 3 months, not adjusted and adjusted by clinically relevant variables found in our study. Cat ¼ categorized; OR ¼ odds ratio. reflecting a potentially protective role on neuronal tissue. 00033). EuroHYP-1 received funding from the European The fact that the association with good prognosis was Union’s Seventh Framework Programme (FP7/2007–2013) stronger than with temperature itself suggests the involve- under grant agreement n 278709. European Union pro- ment of RBM3 in other neuroprotective mechanisms in- gramme FEDER and the European Regional Development dependent of body temperature. High RBM3 levels have Fund – ERDF. been clinically associated with prolonged overall survival in intestinal-type gastric cancer (Ye et al., 2017), good prognosis in invasive breast cancer (Kang et al., 2018) Competing interests and colon cancer (Jang et al., 2017), and improved re- sponse and survival in metastatic colorectal cancer The authors report no competing interests. (Siesing et al., 2017), none of which benefits appear tem- perature-related. The blood samples from the IS patients that underwent References hypothermic treatment in the EuroHYP-1 trial confirmed Campos F, Pérez-Mato M, Agulla J, Blanco M, Barral D, Almeida Á, the association between RBM3 and cooling. Although the et al. Glutamate excitoxicity is the key molecular mechanism which results were not statistically significant, this trend is influenced by body temperature during the acute phase of brain observed in just 17 patients can be considered remark- stroke. PLoS One 2012; 7: e44191. able. Although a larger sample size might achieve signifi- Castillo J, Dávalos A, Marrugat J, Noya M. Timing for fever-related brain damage in acute ischemic stroke. Stroke 1998; 29: 2455–60. cance, the pool of eligible patients will remain limited Danno S, Nishiyama H, Higashitsuji H, Yokoi H, Xue J-H, Itoh K, until hypothermic treatment becomes established clinical et al. Increased transcript level of RBM3, a member of the glycine- practice. EuroHYP-1 was a pioneering phase III trial but rich RNA-binding protein family, in human cells in response to cold access to multiple samples was limited by the low num- stress. Biochem Biophys Res Commun 1997; 236: 804–7. ber of patients enrolled, incompatibility between hypo- Darwazeh R, Yan Y. Mild hypothermia as a treatment for central ner- thermic procedures and blood sampling, and the inability vous system injuries: positive or negative effects. Neural Regen Res 2013; 8: 2677–86. of many patients to complete the cooling protocol. Dresios J, Aschrafi A, Owens GC, Vanderklish PW, Edelman GM, In conclusion, these preclinical and clinical results iden- Mauro VP. Cold stress-induced protein Rbm3 binds 60S ribosomal tify RBM3 as a molecular marker associated with the subunits, alters microRNA levels, and enhances global protein syn- protective effect of hypothermia in IS and suggest its po- thesis. Proc Natl Acad Sci U S A 2005; 102: 1865–70. tential application as a promising protective target. Fernandez-Susavila H, Iglesias-Rey R, Dopico-Lopez A, Perez-Mato M, Sobrino T, Castillo J, et al. Inclusion criteria update for the rat intraluminal ischaemic model for preclinical studies. Dis Model Mech 2017; 10: 1433–8. Supplementary material Geurts M, Petersson J, Brizzi M, Olsson-Hau S, Luijckx G-J, Algra A, et al. COOLIST (Cooling for Ischemic Stroke Trial): a multicenter, Supplementary material is available at Brain open, randomized, phase II. Clinical Trial 2017; 48: 219–21. Communications online. Han HS, Park J, Kim JH, Suk K. Molecular and cellular pathways as a target of therapeutic hypothermia: pharmacological aspect. Curr Neuropharmacol 2012; 10: 80–7. Funding Jang HH, Lee HN, Kim SY, Hong S, Lee WS. Expression of RNA- binding motif protein 3 (RBM3) and cold-inducible RNA-binding This study was partially supported by the Spanish protein (CIRP) is associated with improved clinical outcome in Ministry of Economy and Competitiveness (SAF2017- patients with colon cancer. Anticancer Res 2017; 37: 1779–85. Kang SH, Cho J, Jeong H, Kwon SY. High RNA-binding motif protein 84267-R), Xunta de Galicia (GRC2014/027, 3 expression is associated with improved clinical outcomes in inva- IN607A2018/3), Instituto de Salud Carlos III (RD16/ sive breast cancer. J Breast Cancer 2018; 21: 288–96. 0019 and PI17/00540; Miguel Servet contracts CPII17/ Knott G. Neurodegeneration: Cold shock protects the brain. Nature 00027 and CPII19/00020; Sara Borrell contract CD19/ 2015; 518: 177–8. 12 | BRAIN COMMUNICATIONS 2020: Page 12 of 12 P. Ávila-Gómez et al. Matchett GA, Allard MW, Martin RD, Zhang JH. Neuroprotective ef- ischemia and reperfusion injury via PI3K/Akt pathway. Brain Res fect of volatile anesthetic agents: molecular mechanisms. Neurol Res 2010; 1357: 142–51. 2009; 31: 128–34. Wellmann S, Truss M, Bruder E, Tornillo L, Zelmer A, Seeger K, et al. Matsuda A, Ogawa M, Yanai H, Naka D, Goto A, Ao T, et al. The RNA-binding protein RBM3 is required for cell proliferation Generation of mice deficient in RNA-binding motif protein 3 and protects against serum deprivation-induced cell death. Pediatr (RBM3) and characterization of its role in innate immune responses Res 2010; 67: 35–41. and cell growth. Biochem Biophys Res Commun 2011; 411: 7–13. Wong JJ, Au AY, Gao D, Pinello N, Kwok CT, Thoeng A, et al. Peretti D, Bastide A, Radford H, Verity N, Molloy C, Martin MG, RBM3 regulates temperature sensitive miR-142-5p and miR-143 et al. RBM3 mediates structural plasticity and protective effects of (thermomiRs), which target immune genes and control fever. cooling in neurodegeneration. Nature 2015; 518: 236–9. Nucleic Acids Res 2016; 44: 2888–97. Pilotte J, Dupont-Versteegden EE, Vanderklish PW. Widespread regu- Wood H. Neurodegenerative disease: RBM3, a protein upregulated lation of miRNA biogenesis at the Dicer step by the cold-inducible during hibernation, provides new insights into neurodegeneration. RNA-binding protein, RBM3. PLoS One 2011; 6: e28446. Nat Rev Neurol 2015; 11: 124. Siesing C, Sorbye H, Dragomir A, Pfeiffer P, Qvortrup C, Ponten F, Ye FP, Jin PS, Cai XN, Cai PP, An HM. High RNA-binding motif pro- et al. High RBM3 expression is associated with an improved sur- tein 3 (RBM3) expression is independently associated with pro- vival and oxaliplatin response in patients with metastatic colorectal longed overall survival in intestinal-type gastric cancer. Med Sci cancer. PLoS One 2017; 12: e0182512. Monit 2017; 23: 6033–41. van der Worp HB, Macleod MR, Bath PM, Demotes J, Durand- Zhou M, Yang WL, Ji Y, Qiang X, Wang P. Cold-inducible RNA- Zaleski I, Gebhardt B, et al. EuroHYP-1: European multicenter, binding protein mediates neuroinflammation in cerebral ischemia. randomized, phase III clinical trial of therapeutic hypothermia plus Biochim Biophys Acta 2014; 1840: 2253–61. best medical treatment vs. best medical treatment alone for acute is- Zhu X, Buhrer C, Wellmann S. Cold-inducible proteins CIRP and chemic stroke. Int J Stroke 2014; 9: 642–5. RBM3, a unique couple with activities far beyond the cold. Cell Mol van der Worp HB, Sena ES, Donnan GA, Howells DW, Macleod MR. Life Sci 2016a; 73: 3839–59. Hypothermia in animal models of acute ischaemic stroke: a system- Zhu X, Yan J, Bregere C, Zelmer A, Goerne T, Kapfhammer JP, et al. atic review and meta-analysis. Brain 2007; 130: 3063–74. RBM3 promotes neurogenesis in a niche-dependent manner via Vieites-Prado A, Iglesias-Rey R, Fernandez-Susavila H, da Silva-Candal IMP2-IGF2 signaling pathway after hypoxic-ischemic brain injury. A, Rodriguez-Castro E, Grohn OH, et al. Protective effects and mag- Nat Commun 2019; 10: 3983. netic resonance imaging temperature mapping of systemic and focal Zhu X, Zelmer A, Kapfhammer JP, Wellmann S. Cold-inducible hypothermia in cerebral ischemia. Stroke 2016; 47: 2386–96. RBM3 inhibits PERK phosphorylation through cooperation with Wang JK, Yu LN, Zhang FJ, Yang MJ, Yu J, Yan M, et al. NF90 to protect cells from endoplasmic reticulum stress. FASEB J Postconditioning with sevoflurane protects against focal cerebral 2016b; 30: 624–34.