Study cases were selected based on autopsy reports and cause of death (Table 1). In total, 23 control cases [7 coronary artery disease cases, 5 intoxications (drugs and medication or ethanol), 5 suffocations (include hanging), 1 chronic drug abuse subject with acute pneumonia, 1 chronic alcohol use subject, 1 case with multiple rib fractures, and 1 case with ruptured aneurysm of thoracic aorta] and 17 hypothermia cases (including 10 cases of hypothermia as a contributory cause of death) were included. The diagnosis of hypothermia was based on multiple criteria: documented cold exposure (including environmental circumstances, clothing, signs of paradoxical undressing), autopsy findings suggestive of cold exposure and hypothermia (such as external signs and the presence of Wischnewsky’s spots in the gastric mucosa), toxicological and biochemical investigations (including urine catecholamine levels, concentration of vitreous glucose, lactate and β-hydroxybutyrate in blood), and the exclusion of other potential causes of death [1, 15]. The bodies were stored at 4 °C until autopsy. Kidney tissue samples were collected during autopsy, immediately frozen at -80 °C and stored until extraction of RNA and protein. Permission to use cadaveric tissue samples and cause-of-death investigation data was granted by Finnish Institute for Health and Welfare (Dnro THL/2542/5.05.00/2021). The Northern Ostrobothnia Hospital Research Ethics Committee approved the original hypothermia study plan (31/2007). Tissue samples were collected for research purposes in medicolegal autopsies performed in the Forensic Medicine Unit, Finnish Institute for Health and Welfare, Oulu, Finland.

Blood and tissue FFA and triglyceride (TG) concentrations were measured by enzymatic chemical reaction (LabAssay™ Non-Esterified Fatty Acid and LabAssay™ Triglyceride, FUJIFILM, Osaka, Japan). Serum was obtained from blood collected from autopsies by centrifugation at 1000 × g for 10 min. To analyze the contents of FFA and TG in renal tissue, cortical portions of the kidney were homogenized in 0.5% NP-40 in phosphate-buffered saline, and the centrifuged supernatant was used as a sample.

Oil Red O (ORO) staining was performed on frozen sections to visualize lipid droplets in renal tubular cells. Tissue specimens were initially fixed in 10% solution of formaldehyde, then embedded in an OCT compound (SAKURA, Tokyo, Japan) and frozen using hexane coolant. Kidney samples were immersed in 10% solution of formaldehyde overnight, followed by immersion in 30% sucrose. The tissues were then in embedded in an OCT compound (SAKURA), frozen using FlashFREEZE system (Milestone, Valbrembo, Italy), and sectioned at a thickness of 5 μm using a Leica CM3050 S cryostat (Leica, Deer Park, IL, USA). Lipid droplets in renal tubular cells were analyzed by ORO application. ORO-positive cells were graded on a three-tier scale at high magnification, as described by Preuss. et al. [2].: Negative (no lipid droplets), Low (slight presence of lipid droplets in random renal tubules per high-power field), and High (moderate to severe lipid droplet presence, with more than 50% of renal tubules showing positive lipid droplets per high-power field). Formalin-fixed, paraffin-embedded tissue samples were prepared in accordance with a standard protocol. For immunohistochemical analysis, 3-µm-thick paraffin sections were sequentially treated before application with the primary antibody as follows: deparaffinization, rehydration, endogenous peroxidase quenching, and antigen retrieval with 0.1 M sodium citrate buffer in a microwave. Sections were then incubated with a rabbit polyclonal FATP1 primary antibody (Novus Biologicals, Littleton, CO, USA), followed by detection with a horseradish peroxidase (HRP)-conjugated secondary rabbit IgG antibody (Vector Labs, Burlingame, CA, USA). FATP1 expression in renal tubular cells was graded by a similar grading system of immunohistochemistry staining with HSP70, as reported by Preuss et al. [2]. We graded as follows in 20 high-power fields magnification: 0, no staining; 1+, mild but clear cytoplasmic positive staining; 2+, moderate cytoplasmic positive staining; and 3+, strong and complete cytoplasmic positive staining.

Human renal proximal tubular epithelial cells (HK2) were purchased from ATCC (Manassas, VA, USA). These cells were cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% (vol/vol) fetal bovine serum and penicillin-streptomycin, in a humidified atmosphere containing 5% CO2. For low-temperature stimulation, cells were seeded into culture plates containing culture medium and subsequently treated with growth medium containing 0.1 mM palmitic acid (PA) and/or a FATP1 inhibitor (FATP1-in-1; TargetMol, Wellesley Hills, MA, USA). Cells were then incubated at either 37 °C (Cont) or 28 °C (Hypo) for 15 h. BODIPY staining was performed to assess FFA uptake under low-temperature conditions. After formalin fixation, cells were stained with BODIPY (Thermo Fisher Scientific, Waltham, MA, USA), observed under a microscope, and quantified using a fluorescence plate reader.

Total RNA was isolated from cultured HK2 cells and kidney samples with an miRNeasy Mini Kit (QIAGEN, Hilden, Germany) using an automated QIAcube sample preparation instrument (QIAGEN) according to the manufacturer’s protocol. One-step real-time RT-PCR with SYBR Green I detection was used. RT-PCR was performed using the Rotor-Gene Q system (QIAGEN) for human SLC27A1, PPARA, CD36, and ACTB mRNA as an internal control. Each threshold cycle was obtained, and the double-delta threshold cycle method was used to calculate expression values. Primer sequences were as follows for human SLC27A1, 5ʹ-CGTGCTAGTGATGGATGAGC-3ʹ and 5ʹ-GCCTCGTCTTCTGGATCTTG-3ʹ; PPARA, 5ʹ-CTGGAAGCTTTGGCTTTACG-3ʹ and 5ʹ-CAATGCTCCACTGGGAGACT-3ʹ; CD36, 5ʹ-GGCTGCAGGTCAACCTATTG-3ʹ and 5ʹ-GCAACAAACATCACCACACC-3ʹ; and ACTB, 5ʹ-GGCATCCTCACCCTGAAGTA-3ʹ and 5ʹ-GGGGTGTTGAAGGTCTCAAA-3ʹ. RT-PCR was carried out as follows: after the reverse-transcription reaction (50 °C, 30 min), amplification was performed with one cycle of denaturation (95 °C, 15 min) followed by 40 cycles of three-stage PCR (95 °C, 30 s; 60 °C, 30 s; and 72 °C, 1 min). Each cycle threshold (CT) value was obtained and the 2− ΔΔCT method was used to calculate relative expression values.

We extracted protein from cultured HK2 cells and the cortical portion of kidneys with RIPA buffer. Kidney samples were homogenized using a bead crusher, and the supernatants were isolated after centrifugation of 1000 ×g for 5 min. Protein samples were then separated in polyacrylamide-sodium dodecyl sulfate gels and electro-transferred to nitrocellulose membranes. After blocking using 5% skim milk, the membranes were probed with an anti-FATP1 rabbit monoclonal antibody (Mybiosouce, San Diego, CA, USA) and anti β-actin mouse monoclonal antibody (BD Biosciences, Franklin Lakes, NJ, USA). The membranes were then incubated with the respective HRP-conjugated anti-rabbit or anti-mouse IgG secondary antibodies (R&D Systems, Minneapolis, MN, USA). Antibody binding was visualized using the SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific). β-actin was used as an internal control to normalize the level of target protein. Quantification was performed by densitometry analysis using ImageJ (National Institutes of Health, Bethesda, Maryland, USA).

We used GraphPad Prism (GraphPad Software, San Diego, CA, USA) for all statistical analyses. Differences in experimental values were analyzed via the Mann–Whitney U test or one-way ANOVA followed by the Tukey–Kramer method. To assess correlations, Spearman’s rho test was applied. A P-value < 0.05 was considered statistically significant for evaluating differences and correlations in the experimental data.