Metabolic networks in a porcine model of trauma and hemorrhagic shock demonstrate different control mechanism with carbohydrate pre-feed

The animal protocol has been discussed previously [5, 6]. The experimental protocol was approved by the University of Minnesota Animal Use Committee and was conducted in accordance with established guidelines for the ethical and humane treatment of laboratory animals. A well-established model of porcine hemorrhagic shock which has been previously described [7, 8] was utilized (Instrumentation details are presented in the Additional file 1).

Sixty-four male Yorkshire pigs (Manthei Hog Farm, LLC, Elk River, MN) weighing 15–20 kg were randomized to experimental group (n = 32, pre-fed and n = 32, fasted). The sample size was dictated by power calculations for statistical significance to be observed for multiple variables (e.g. metabolites) at multiple timepoints (baseline, shock, resuscitation). Calculations were done using Statistical Sample Size Calculator PASS 2005. A power of 80 % and type 1 error of 5 % were assumed.

The pre-fed group received a bolus of Karo® Syrup at 7 cc/kg with an equal amount water 30 min prior to experimental manipulation while the fasted group received only water for 12 h prior to experimental manipulation. This dosing was based upon previous work [9]. NMR analysis of Karo® Syrup revealed its components as glucose, fructose, maltose, and sucrose. Serum samples from all animals were taken at set timepoints throughout the experiment (see Table 1). A total of 407 samples were obtained.

For a timeline of the experimental protocol, please see Fig. 1. A captive bolt device was used to create a blunt percussive injury to the right chest. Hemorrhagic shock was then induced by withdrawal of blood from the inferior vena cava until a systolic pressure in the lower 50’s was reached (typically 35 % of total blood volume). Shed blood was placed in an acid-citrate-dextrose bag for later use. A liver crush injury was induced using a Holcomb clamp technique [10], with two crush injuries created in the liver parenchyma. The animal was left in shock for a period of 45 min prior to beginning resuscitation.

Animals were resuscitated with lactated Ringer’s fluid given as 20 cc/kg intravenous (IV) boluses to maintain a systolic blood pressure greater than 80 mmHg for 1 h of limited resuscitation, then underwent full resuscitation by algorithm (see Additional file 2). Auto-transfused warmed blood was given at 10 cc/kg IV boluses for a target hemoglobin of greater than 6 g/dL and a urine output of greater than 1 cc/kg/hr. Lactated Ringer's at 20 cc/kg IV boluses or blood at 10 cc/kg/hr were given as needed to maintain blood pressure. After a resuscitation period of 20 h, animals were extubated and sent to recovery. At 24 h after the end of resuscitation, animals were sedated and serum samples were obtained. At 48 h after resuscitation, animals were sedated for endpoint sample harvesting and then euthanized with Beuthanasia D (1 ml/10 kg IV).

Harvested blood samples were immediately centrifuged for 15 min at 3000 g to extract serum and were stored at −80 °C until prepared for Nuclear Magnetic Resonance spectroscopy (NMR) analysis. Stored serum was thawed at the time of analysis and internal standard (1 mM trimethylsilylpropionic acid in D₂O) was added to 500 μL of serum. Samples were transferred into 5 mm NMR tubes (Wilmad, LabGlass, USA).

Proton NMR spectra were taken with a Bruker Avance spectrometer with autosampler and 5 mm triple resonance 1H/13C/15 N TXI CryoProbe with Z-gradient, running TopSpin v. 2.16 (Bruker, Bilerica, MA USA) at 700.13 MHz. A 1D CPMG (Carr-Purcell-Meiboom-Gill) pulse sequence was used to collect spectra for each serum sample. The $90^{\circ}$ pulse width was calibrated for each sample, and was generally 12–13 μs. The relaxation time was defined by each sample's 90° pulse width. During the relaxation period, the protein signal was suppressed. The relaxation delay was 2 s, the acquisition time was 3 s, the spectral width was 10 kHz, the total number of data points collected was 63,000, and 128 transients were collected. All spectra were collected at a temperature of 298 K. Line broadening at 0.5 Hz was applied before FFT; autophasing and auto-baseline correction were applied by TopSpin.

Chenomx software [11] was used to identify and quantify a portion of the metabolites present in each sample. Fine manual phasing and baseline corrections and the software's Reference Deconvolution algorithm were applied to each spectrum before targeted profiling of the metabolites was performed. Forty-eight metabolites were quantified in each sample, resulting in a profile containing the concentration of each identified metabolite in millimoles per liter (mM). Chemical shifts of identified metabolites were compared with those available in the Human Metabolome Database to confirm metabolite identities [12, 13].

Common metabolic interactions were investigated for the fasted and fed groups according to the following procedure. The serum samples provided the time trajectory of the metabolite concentrations at sampling points of 45 min, 3, 5, 9, 17, and 21 h after hemorrhage. This time series data was normalized and shifted by the baseline metabolite concentration before hemorrhage. For each metabolite in series, the rate of change for all animals in the group was estimated using a first difference approximation. Metabolites with sufficiently small error were considered to be ‘controlled nodes’. A regularized linear regression was performed to find the best low order predictor of the rate of change for the controlled node as a function of all 48 measured metabolite concentrations. The metabolite concentrations that were associated with non-zero coefficients in the predictor were considered to be causally affecting the controlled node and were identified as ‘controlling nodes’. The prediction was considered significant if the mean prediction error was less than 50 % than the mean magnitude of the rate of change. The details of the regularization problem are given in the Additional file 3.