We constructed the computational model using the SimBiology Model Builder Interface (SimBiology version 6.1, Mathworks Inc., Natick, MA). A representation of our model is presented in Figure 1. Our model has four discrete compartments, each of which was considered well-mixed: extracellular space, cytoplasm with plasma membrane, nucleus with nuclear membrane, endoplasmic reticulum with endoplasmic reticulum membrane, and mitochondria. The model was built in a modular fashion, and contains 7 integrated sub-models with their regulatory components:
1. Calcium sub-model, including NMDA receptor (NMDAR), voltage gated Ca2+ channel (VGCC), Ca2+-permeable AMPA receptor (CP-AMPAR), IP3-Ca channel, SERCA channel, Ryr channel, and Na-Ca exchanger (NCX)
2. Src kinase sub-model with regulatory components
3. Oxygen-ATP sub-model
4. Ca2+/CaM cytoplasmic signaling cascade sub-model with regulatory components
5. Ca2+/CaM nuclear signaling sub-model with regulatory components
6. PKC-Raf-MEK-ERK pathway
7. CREB-Creb binding protein (CBP) regulation of Bax transcription
The complete model was composed of 62 species, 134 reaction parameters, and 63 individual reactions. Reactions were modeled as either activation enzymatic reactions, where the enzyme E activates the molecule A at rate kact to yield Aactive:
+ E (1)
or inactivation reactions, when Aactive is inactivated at rate kinact:
Aactive A (2)
or binding/unbinding reactions, when A binds to molecule B to yield AB at rate kbind, and unbinds at rate kunbind:
A + B AB (3)
AB A + B (4)
or translocation/transport reactions, where A translocates from the cytoplasm into the nucleus at rate kin and translocates out at rate kout:
Ordinary differential equations (ODEs), starting quantities, and fluxes for reactions are listed in the Supplementary material. The initial species concentrations were derived or estimated from other published models, our lab’s data, and from conversion of label-free quantification signal intensity from localization data on PepTracker to protein counts using the Perseus ‘Proteomics Microruler’ plug-in.
Simulation of hypoxia, normoxia, and pharmacological intervention
We modeled conditions of normoxia (Nx) and hypoxia (Hx) by modifying the extracellular concentration of O2. Per Tuckerman et al., 2004 in normoxia, the intracellular O2. concentration in tissues is 10,000-40,000 ppm, in moderate hypoxia, [O2] = 1,000-10,000 ppm, and in severe hypoxia [O2] = 0-1000 ppm . Assuming a solvent density of 1g/mol, these values correspond to 312.5-1250 mM, 31.25-312.5 mM, and 0-31.25 mM respectively. We used “Events” in the SimBiology model builder interface to set discrete transitions between Nx, Hx, and intervention with the Src-inhibitor PP2 (Hx + PP2). We set the hypoxic event trigger as time ≥ 120 s, with the event function lowering the extracellular O2 concentration to 3.25 µM. In the second event, we maintained Hx, but introduced PP2 at a concentration of 1 µM. For model input during Nx, we dosed extracellular glutamate at the amount defined by the parameter “stim”, for 12 pulses, with a 10 s inter-pulse interval. During Hx and Hx + PP2, we dosed extracellular glutamate at the amount defined by “hyperstim” delivering 48 pulses with a 5 s inter-pulse interval.
Global sensitivity analysis (GSA) and multiparametric global sensitivity analysis (MPGSA)
We used the Global Sensitivity Analysis app to perform GSA on our model and compute Sobol indices[24,25]. We set the number of samples to 1000, and the output time to match the duration of our simulation: 0-360 s at a resolution of 1 s. The “interp1q” Interpolation method was used. Following computation of Sobol indices, the same simulations were reused for MPGSA with specific cut-offs for observables. Significance level was set at 0.05.
The experiments and the reporting data are performed and published in accordance to the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. Fifteen newborn Yorkshire piglets (postnatal day 2 to 3; Willow Glenn Farm, Strasburg, PA) were studied. Animals were similar at baseline clinical characteristics. Animals were housed in standard cages and were exposed to 12-hour light/ 12-hour dark cycle. Access to standard newborn formula and water was provided during study ad libitum. The experimental protocol was approved by the Institutional Animal Care and Use Committee of Drexel University (IACUC protocol #200491). All methods were performed in accordance with the relevant guidelines and regulations. Appropriate use of drugs in standardized dosage was followed during experiments to reduce or eliminate pain, according to the NIH Guidelines for the Care and Use of Laboratory Animals. Anesthesia was induced with 4% Isoflurane and maintained with 0.8-1% Isoflurane. The piglets were intubated within 10 min of induction of anesthesia and were placed on a pressure ventilator using a mixture of 75% nitrous oxide and 25% oxygen. Fluids and medication were administered using a peripheral intravenous catheter. Fentanyl (25μg/kg) and pancuronium bromide (0.1 mg/kg) were administered through the intravenous route.
Following baseline ventilation with normal blood gases for a period of 1 hour, animals were randomly assigned to either a normoxic (Nx), hypoxic (Hx) or hypoxic treatment groups (Hx + PP2) by means of block randomization. Hypoxia was induced by a gradual decrease of the FiO2 to 0.06 and maintenance at this level for 1 hour.
Following 1 hour of hypoxia, the cerebral cortex was harvested. Anesthesia and analgesia were maintained throughout hypoxia and brain harvesting through rapid midsagittal craniotomy. The parenchyma was flash-frozen in liquid nitrogen and stored at -80 °C for further biochemical analysis. The piglets were maintained at 38-39°C (normothermic). No animal died during the application of hypoxia. For the animals in treatment group, Src-inhibitor PP2 (1 mg/kg, IV) was administered a half-hour before Hx.
Isolation of Cerebral Cortical Neuronal Nuclei
The isolation procedure for the cerebral cortical nuclei was based on the method of Giuffrida et al. . 1 gram of brain tissue was homogenized in 15 volumes of a medium containing 0.32 M sucrose, 10 mM Tris–HCl and 1 mM MgCl2 (pH 6.8). The resulting homogenate was then filtered through a nylon bolting mesh (size 110 lm) and subjected to centrifugation at 850 g for 10 min. To increases the yield of large neuronal nuclei, the homogenate was passed through a discontinuous gradient with a final sucrose concentration of 2.1 M, followed by another centrifugation step at 70,000 g for 1 hour. The nuclear pellet was then collected and re-homogenized. Purity of neuronal nuclei was ensured by phase contrast microscopy. Protein content was determined according to Lowry et al. 
Isolation of the Cytosolic Fraction
One gram of cerebral cortical tissue was homogenized by a Dounce homogenizer (seven strokes with pestle clearance 0.15 mm and seven strokes with pestle clearance 0.07 mm) in 30 ml of fresh isolation medium (0.32 M sucrose, 1 mM EDTA, 20 mM Tris–HCl buffer, pH 7.1). The homogenate was then subjected to centrifugation at 1500 g, followed by a second centrifugation step at 15000 g for 10 minutes. The supernatant was finally centrifuged at 100000 g for an hour to obtain the cytosolic fraction and the protein concentration was determined.
Western Blot Analysis
Total protein extraction was performed in cell lysis buffer (0.5% NP-40, 0.5% SDS, 1.5 mM pH 7.4 Tris-HCL, 15 mM NaCl). Following electrophoresis (20 μg protein/lane) and polyvinylidene difluoride (PVDF) membrane transfer, primary antibody incubation was performed against Calcium Calmodulin Kinase dependent Kinase 2 [Sigma rabbit Anti-CAMKK2, Catalog# HPA017389, 1:500 dilution] overnight at 4°C on a rocking platform. The membranes were treated with goat-anti-rabbit HRP-conjugated secondary antibodies (Invitrogen Thermo Fisher, Waltham, Massachusetts, USA). The target protein bands were determined using the reagents visualized provided in the ECL+ plus kit (GE Healthcare, Piscataway, NJ, USA). The immunoreactive band intensities were analyzed by imaging densitometry (GS 700 Imaging Densitometer, Bio-Rad) using ImageJ (National Institutes of Health, Bethesda, MD, USA). The protein expression is presented as optical density (OD) per mm2.
Determination of ATP levels
Energy failure caused by hypoxia in cerebral cortex tissue was confirmed by determining the levels of ATP based on an assay previously described in detail [31,32]. Briefly, frozen cortical tissue was powdered under liquid nitrogen in 6% w/v perchloric acid, thawed on ice and then centrifuged at 2000 g for 15 minutes at 4°C. The supernatant was neutralized using 2.23 M Potassium Carbonate/0.5 M triethanolamine/50 mM EDTA buffer (pH 7.3) and then subjected to centrifugation at 2000 g for 15 minutes at 4°C. 300 μl of the supernatant was added to 1 ml of buffer (50 mM triethanolamine, 5 mM MgCl2, 2 mM EDTA, 2 mM glucose, pH 7.6) and 20 μl NADP (10mg/ml in 50 mM triethanolamine (TRA)-HCl buffer). Samples were incubated for 8 minutes with Glucose-6-phosphate dehydrogenase and read, following which Hexokinase was added and absorbance measured until steady state was reached. ATP concentration was calculated from the increase in absorbance at 340 nm. 20 µl each of ADP
Nuclear Ca2+ Concentration
Neuronal nuclei (150 µg) were incubated in 300 μl medium composed of 50 mM Tris buffer (pH 7.4), 1 μM 45Ca2+, and 1 mM ATP. The Ca2+ influx assay was carried out at 37 °C for 2 minutes. Samples were then filtered on a glass fiber filter and washed three times in 20 mM Tris (pH 7.2), 100 mM potassium chloride buffer. A Rackbeta scintillation counter (Pharmacia, Gaithersburg, MD, USA) was used to measure radioactivity.
Statistical analysis for experimental measurements was performed using Graphpad Prism v. 9. We used Ordinary one-way ANOVA with Tukey post-hoc multiple comparisons test. A P-value of less than 0.05 was considered statistically significant.