Tissue alkalosis in cold-ischemia time

The control of pre-analytical-factors in human biospecimens collected for health research is currently required. Only two previous reports using post-mortem brain samples have tried to address the impact of cold-ischemia on tissue pH. Here we report pH variations according to time (third-order polynomial model) in mice for liver, kidney and lung samples. Tissue alkalosis in cold-ischemia time may be an underlying mechanism of gene expression changes. Therefore, tissue-pH regulation after organ removal may minimize biological stress in human tissue samples.

Supplementary Table S1. A pH significant increase according to time (third-order polynomial model) was found in mice for liver, kidney and lung (Table 1). Fisher's least significant difference remained unchanged at time-zero ( Fig. 1). Following the ischemic insult, time dependence was always higher in lung, followed by kidney and liver, with a pH range of 0.64, 0.4 and 0.36, respectively. Although RIN's statistical power did fit a second-order time evolution ( Supplementary Fig. S5), in biological terms, this result does not modify its clinical use. In this study, the adjusted RIN was greater than 8.90, with the minimum acceptance threshold being RIN = 7. On the other hand, organ histoarchitecture (data not shown) and RNA absorbance ratios remained unchanged over the time for each organ ( Table 2). Supplementary data is available at Scientific Reports' website (Supplementary Table S1 and Figs S1-S6).

Discussion
Two phenomena must be highlighted for a better understanding of our work: on the one hand, chronic pH profile after brain ischemia evidences the duration and presumably the severity of ischemic injury. After the onset of stroke symptoms, intracellular pH has been found to increase up to 0.45 pH units in regions of cerebral infarction. Brain tissue alkalosis cannot be attributed to abnormal blood gases, because post-ischemic blood gases were found to be within normal physiological ranges. Whether alkalosis is secondary to ischemic cell damage, or it may  Table 2). There is no previous evidence of tissue pH measurement in tissues submitted to ischemia, except for the brain. Therefore, this study is novel in the way it looks at tissue pH following cold ischemia insult, and because it explores this phenomenon in three different organs. As outlined in the work of Chopp et al., and according to our data, we would hypothesize that tissue pH measurement varies substantially, regardless of blood pH 18 . Unfortunately, in the experimental design performed herein we were unable to measure blood pH, because blood flow was immediately stopped upon mice euthanasia.
On the other hand, acidic duodenal pH has been reported to alter gene expression in the pancreas of a cystic fibrosis mouse. When the duodenal pH was corrected, either genetically (breeding CFTR-null with gastrin-null mice) or pharmacologically (use of the proton pump inhibitor omeprazole), expression levels of genes measured by quantitative RT-PCR were significantly normalized 19 .
Our work could be considered a hypothesis generator study. Was the blood acidosis detected in a closed corporeal entity in hypoxia a consequence for the compensation of the tissue alkalosis produced in the first instance in the hypoxic tissues? This hypothesis should be further explored since high-throughput molecular assays are playing an ever-growing role in research and into the understanding of the pathobiology of human and animal diseases.
In conclusion, this work found significant differences in pH values between organs at the same cold-ischemia time, and in the same organ at different times in an experimental animal model, although no differences were seen in the RNA quality assessed by its integrity number or absorbance ratios. Our results suggest that pH is high in tissues undergoing ischemia, a finding that has already been reported in previous studies, and the regulation of tissue pH after organ removal may minimize biological stress in human tissue samples. The results from this study suggest that although the RIN is a powerful tool to inspect the ribosomal profile to infer RNA quality from fresh and frozen tissues (and to compare samples RIN values given the same organism/tissue/extraction method), it is not sufficient to predict the integrity of mRNA transcripts and describe the real biological conditions. Therefore, the role of pH as an underlying mechanism of gene expression changes after an ischemic insult warrants further investigation.

Materials and Methods
Fifty-two mice (Mus musculus) C57BL/6-SPF from the Animal Facility at the CIPE were used. The study was approved by the Ethics Committee in the Use of Animals (CEUA) at Antonio Prudente Foundation (N°049/12). All experiments were performed in accordance with relevant guidelines and regulations. Animals were divided into 5 groups according to the time of tissue cryopreservation. All mice were pathology free (data not shown). The 26 male and 26 female mice were carefully selected in order to have a similar weight (24.89 ± 5.21 gr) and time of life for each gender (104 ± 36 days, Supplementary Fig. S1). The procedure involving the removal of organs was carried out under laminar flow and aseptic conditions through surgical incision by the median line. The material used for the surgery was submerged -between procedures-for 2 minutes in 0.3% (v/v) Endozyme AW plus Apa solution (Ruhof Corporation Biomed Division, NY, USA). The timers were activated for the recording of the different times of tissue cold ischemia exactly upon heart removal when blood flow stopped (0, 15, 30, 45 and 60 minutes). Both lungs, both kidneys and two equal portions of the liver were removed and weighed ( Supplementary Fig. S2). Once the organs were weighed, they were divided into two parts, the largest for cryopreservation and a smaller portion (10-30-60 mg of lung, kidney and liver, respectively) for the determination of tissue pH. The organ removed was left out at room temperature until the required time elapsed. Then, pH was measured and the tissue cryopreserved.
Histopathology. The anatomopathological study was carried out using cryopreserved material. It was placed in Tissue-Tek ® O.C.T. Compound (O.C.T. optimal cutting temperature medium, Sakura ® Finetek, Torrance, CA, USA). Three μm sections were cut in a −28 °C cryostat. Excess of Tissue-Tek ® O.C.T. compound was separated and tissue was saved in the corresponding cryotube for subsequent macromolecules extraction. The histoarchitecture was studied using standard hematoxylin-eosin staining, to assess the morphological changes induced by cold-ischemia.
260/280 ratio. The ratio of absorbance at 260 nm and 280 nm (A260/280) was used to assess RNA purity.
A ratio of 2.0 or greater is accepted as pure RNA (Supplementary Fig. S3A). A lower ratio may imply the presence of protein, phenol or other contaminants strongly absorbing at or near 280 nm 20   ionic strength have shown that the change in the A260/280 ratio is primarily due to a decrease in the absorbance at 280 nm, when the ionic strength or pH is increased. This effect was achieved irrespective of the method of purification. However, the elution method can interfere with the A260/280 ratio. Thus, RNA diluted into distilled-DEPC-treated-water has usually a lower ratio compared with Tris-EDTA pH 8.0.   13 . This user-independent RIN is calculated using a software algorithm that contemplates the entire electrophoretic trace of the RNA, including presence or absence of degradation products. The Bioanalyzer software also generates an electropherogram and gel-like image of the 18S and 28S RNA ribosomal ratio. RIN numbers greater than 7 correlate with successful downstream microarray experiments, as manufacture indication ( Supplementary Fig. S4).
pH. The pH was measured at room temperature in technical duplicates for both lungs and kidneys, and for the anterior and posterior portions of the liver (Supplementary Fig. S6). The tissue was manually homogeneized without dilution using a micro pistill 16 . Thus, each animal had twelve different measurements of pH, four for each organ (Dual Channel pH-Ion-meter Accumet Excel-XL25, Thermo Fisher Scientific, CA, USA).
Statistical Analysis. Statistics were performed using InfoStat software version 2015 21 together with R version 3.2.3 packages 22, 23 for model estimation. Mice age comparison by gender was carried out using Wilcoxon rank sum test with continuity correction, since normality assumption between these two groups was not achieved ( Supplementary Fig. S1). In order to consider the complete experimental design and technical constraints, independent linear mixed-effects models were estimated. Briefly, the fixed effects were the global mean contribution (µ) of the corresponding dependent output variable; the i-th organ indicator (α i for lung, liver or kidney); the j-th linear (β j = 0, 15, 30, 45, 60 min), quadratic (γ j = β j 2 ) or third (δ j = β j 3 ) time degree; the double time by organ interactions terms (α i × β j , α i × γ j , α i × δ j ); the k-th gender indicator (θ k for male or female), and the l-th organ side indicator (ρ l for right or left). The random effects were the m-th animal a m ~ N(0, Iσ a 2 ), i.e. normally distributed with zero mean and identity the variance matrix (I) times, the σ a 2 parameter; the n-th technical replicate within animal effect b n(m) ~ N(0, Iσ b 2 ), the o-th sample extraction day c o ~ N(0, Iσ c 2 ), the p-th surgery day d p ~ N(0, Iσ d 2 ), and the random error function g (ε)ijklmnop which was modeled as independent errors, i. e., g(ε) ijklmnop = I(ε) ijklmnop = ε ijklmnop ~ N(0, Iσ ε 2 ) or dependent errors, with the lack of heterocedasticity being corrected with a constant variance function grouped by animals, formally, g(ε) ijklmnop = varIdent (~1|Animal) ijklmnop . Using the former annotations, the different experimental data measurements (y ijklmnop ) were modeled starting from the maximal model of equation (1): Model selection was carried out using backward elimination over log-likelihood test results, Akaike information criterion (AIC) and Bayes information criterion (BIC) under the parsimony criterion. The corresponding model assumptions were verified using residual vs. predicted values and normal quantile-quantile plots. When required, outlier values were removed and models were estimated again. For the final model assessment, marginal hypothesis test was used with type-III sum of squares for fixed effects. A posteriori analyses included Fisher's least significant difference (LSD) using Bonferroni adjusted p-values at each time period or main effect. All statistical tests, unless otherwise specified, were two-sided with a cut-off at 0.05 for statistical significance. Equations (2)(3)(4)(5)(6)(7)(8)(9) and Supplementary Table S1 show the final adjusted model for pH, RIN, RNA A260/280 (RNA 260/280) and A260/230 (RNA 260/230), RNA concentration (RNA cc) and relative organ weight (ROW) = (organ weight)/ (animal weight), lung or kidney relative side weight (RLW or RKW) = (organ side weight)/(animal weight).