Selective under-representation of transcripts encoding membrane-integrated architectures following neural tissue vitrification Authors: Dominika Wilczok, Qian Long, Zhuoan Huang, Joseph Kangas, Yewei Liu, Minyan Wang, Ferdinand Kappes Published: 2026 ABSTRACT AbstractCryopreservation is essential for long-term storage of biological tissues. Yet, surprisingly, the impact of cryopreservation on tissue transcriptomes remains poorly defined. This study used bulk RNA sequencing to examine how preservation method (snap freezing or vitrification) affects transcriptomes in mouse cerebral cortex and hippocampus. Tissues processed under vitrification condition showed selective under-representation of a small but structurally coherent group of transcripts, with the hippocampus exhibiting greater vulnerability than the cortex. UniProt annotation revealed that affected transcripts were strongly enriched for proteins with membrane-associated, secretory-pathway, and multi-pass topologies, indicating that structurally complex membrane-integrated architectures could be disproportionately sensitive to vitrification. Pathway-level analysis using iPANDA further showed that negative preservation scores in vitrified tissue clustered primarily within signal transduction and metabolic pathways, suggesting coordinated pathway-level disruption rather than global transcript loss. This work highlights molecular vulnerability during vitrification and emphasizes the need for transcript-level evaluation when optimizing cryopreservation approaches for neural system. Introduction Cryopreservation is a foundational technology in biotechnology, enabling the long-term storage of biological materials and supporting a wide range of applications in basic research, medicine, and conservation, such as routine access to viable cells for culture and transplantation, long-term preservation of tissues for molecular analysis, in vitro disease models, and genetic resource banking for regenerative medicine, biodiversity conservation, and reproductive medicine [1]. Cryopreservation relies on cooling biological specimens to cryogenic temperatures, most commonly storage in liquid nitrogen (−196 °C) or in the vapor phase above liquid nitrogen (≈−150 °C), and in some cases in ultra-low mechanical freezers (≈−80 °C). At these temperatures, molecular diffusion and biochemical reactions are greatly reduced, effectively arresting biological activity [2]. Depending on the goal of preservation and sample type, different cooling approaches are used. When the objective is to preserve molecular composition for downstream analyses such as proteomics or metabolomics, samples are commonly snap frozen by rapid immersion in liquid nitrogen. Although snap freezing renders the sample non-viable, it preserves its molecular composition. Rapid cooling arrests metabolic and enzymatic activity and limits ice crystal growth, allowing DNA, RNA, proteins, and metabolites to be retained in a state suitable for downstream biochemical analysis. On the other hand, for applications where structural and functional preservation is required, the samples are vitrified. Vitrification refers to the solidification of a solution into an amorphous glass without ice crystallization, achieved through sufficiently high concentrations of cryoprotective agents (CPAs) combined with rapid cooling that suppresses ice nucleation and growth. CPAs are chemical compounds added to biological samples to mitigate damage during freezing. CPAs function by reducing ice nucleation and growth, lowering the freezing point of intracellular and extracellular solutions, and stabilizing cellular structures such as membranes and proteins [2]. CPAs are typically categorized as penetrating agents (e.g., dimethyl sulfoxide (DMSO), glycerol, ethylene glycol, propylene glycol), which permeate cell membranes and reduce intracellular ice formation, and non-penetrating agents (e.g., sucrose, trehalose, hydroxyethyl starch), which primarily exert osmotic and extracellular protective effects. However, despite substantial progress at the level of cells and small tissues, scaling cryopreservation to whole organs has thus far proven technically prohibitive, with a few notable examples, due to challenges associated with CPA delivery [3], [4], CPA toxicity [[5], [6], [7], [8], [9], ice crystal formation [[10], [11], [12]], and safe, uniform rewarming scaling nonlinearly with biological size and complexity [13], [14]. Notably, a rat kidney was successfully vitrified, rewarmed, transplanted, and shown to restore close to normal physiological function in the recipient [15]. Similar approaches, although without transplant validation, have been reported for other organs, including rat hearts [16], sheep ovaries [17], and porcine livers [18], with preservation of structural integrity and partial functional outcomes. Importantly, functional or structural recovery at the organ level does not necessarily imply preservation of molecular integrity. Current organ cryopreservation strategies rely on vascular perfusion with CPA to achieve uniform distribution prior to cooling. It implicitly assumes that a single cryoprotectant formulation should preserve molecular integrity across diverse tissue types exposed simultaneously during loading, vitrification, unloading and rewarming. Heterogeneous tissue composition of a whole organism, metabolic state, and permeability, however, raises the clear possibility that cryoprotectants may differentially affect molecular preservation across organs and regions, even under identical perfusion conditions. Transcriptome-wide RNA sequencing (RNA-seq) provides a means to systematically interrogate these molecular consequences. RNA-seq enables highly sensitive detection of method-specific transcript representation or loss that may not be apparent from structural or functional assessments alone. Despite widespread use of RNA-seq in biomedical research, its application in deciphering cryopreservation-induced molecular bias remains limited. Although cryopreservation-induced molecular damage such as protein misfolding or DNA double strand breaks may occur in any tissue, their implications are especially consequential in the brain, where preservation of molecular integrity is essential for maintaining higher-order neural function. Consequently, brain cryopreservation represents a technically demanding test case for evaluating molecular preservation strategies. Notable success cases in brain cryopreservation include the functional revival of hippocampal slices after vitrification [19] and successful cryopreservation and rewarming of brain organoids [20]. However, state-of-the-art approaches were set by German et al. (2025), showing that basic electrophysiology can be regained in hippocampi dissected from whole vitrified mouse brains [21]. Although the whole brain showed dehydration and darkening, the hippocampal granule cells obtained from the dissected hippocampi showed regained functional electrophysiological measures like basic excitability, E/I balance (a measure of the relative strength of excitatory versus inhibitory synaptic transmission), spontaneous activity, input–output transmission, and long-term potentiation were preserved relative to untreated control slices. The whole dissected hippocampi, however, showed reduced mitochondrial respiration and impaired short-term plasticity. The work of German and colleagues (2025) suggests that although whole-brain vitrification may introduce global injury, specific regions such as the hippocampus can still retain measurable functional capacity. Yet long-term goals in cryobiology aim for preservation methods that maintain the integrity of the entire brain, not only isolated regions. Consequently, understanding the structural and physiological differences between major brain regions is essential, as these intrinsic features may reveal region-specific susceptibilities to cryoinjury. Two brain regions of interest to neuroscience given their high neuronal density and great functional relevance are the cerebral cortex and the hippocampus (a comparison is outlined in Fig. 1). While cortical organization is dominated by six distinct layers (I–VI) with granular, pyramidal, and multiform layers, the hippocampus contains compact laminae such as the dentate granule cell layer and the CA pyramidal cell layer [22], [23]. Both regions contain pyramidal neurons, interneurons, and dense excitatory glutamatergic projection systems, however, the cortex contains a wide variety of projection neurons (pyramidal and spiny stellate) and at least 15 interneuron subtypes, the hippocampus contains fewer neuronal classes but extremely dense and well-defined granule and pyramidal layers [22], [23]. The cortex, compared to a hippocampus, exhibits a denser and highly organized network of perineuronal nets, which are specialized extracellular matrix structures that wrap around certain neurons (mostly parvalbumin-positive interneurons) and regulate synaptic plasticity, ion buffering, and protection against oxidative stress [24]. Astrocytes in the cortex are highly coupled, while in the hippocampus they are only coupled in the CA1 region [25]. Both regions have a high proportion of pericytes compared to other brain regions [25]. Histological comparison of murine cerebral cortex and hippocampus. Schematically shown are major differences between the murine cortex and hippocampus. See text for details. These region-specific structural features imply that cryopreservation may not impact all neural tissues uniformly, making it necessary to examine how preservation strategies affect molecular integrity. To address these gaps, this work applies RNA sequencing to compare transcriptomic preservation following snap freezing and vitrification in neural tissue, proxied by isolated mice hippocampi and cerebral cortex. Snap freezing, which is not intended to support post-thaw viability or functional recovery but is widely used to preserve molecular content, serves here as a molecular reference condition [26]. In contrast, vitrification has been shown to enable structural and, in select cases, functional preservation of complex tissues and organs, yet its molecular consequences remain poorly characterized [13,18]. The analysis focuses on transcripts that are selectively under-represented after vitrification while remaining relatively stable under snap freezing, hypothesizing that such transcripts are selectively vulnerable to CPA exposure rather than cold-induced effects. The scope of the study is limited to RNA representation and does not address post-thaw viability, functional recovery, or organismal revival, but instead evaluates transcriptomic preservation with relevance for cryoprotectant and protocol optimization. Methods Experimental model and study participant details Three adult female mice (Mus musculus, C57BL/6 strain; 12 weeks of age) were used. The mice were obtained from GemPharmatech Co., Ltd. and housed in Room 905 of the SPF-grade animal facility at the Laboratory Animal Center of Soochow University. At the time of the experiment, the animals weighed 22–25 g and were clinically healthy, with no visible signs of disease. A trained and authorized professional euthanized the animals by cervical dislocation to minimize distress and ensure rapid cessation of vital functions. The cerebral cortex and bilateral hippocampus were collected because of their roles in memory and higher-order cognitive function and their use as reference regions in studies of neural integrity and molecular preservation. All procedures were performed at the Laboratory Animal Centre, Soochow University, in collaboration with Prof. Wang's laboratory under an agreement between Xi'an Jiaotong-Liverpool University and Soochow University. Procedures followed the institutional animal-care and use guidelines of Xi'an Jiaotong-Liverpool University and were approved under protocol ER-SRR-1088616620231127132732. Only female animals were included; consequently, sex-associated differences could not be evaluated and the findings may not generalize to male animals. Method details Artificial cerebrospinal fluid Artificial cerebrospinal fluid (ACSF) contained 125 mM NaCl, 2.5 mM KCl, 1.2 mM NaH2PO4, 12.5 mM d-glucose, 2 mM CaCl2, 2 mM MgSO4, and 26 mM NaHCO3 in double-distilled water to a final volume of 1 L. The solution was bubbled with carbogen (95% O2/5% CO2) for approximately 10 min, adjusted to approximately pH 7.3 using less than 1 mL of 12 N HCl, and stored at 4 °C until use. Tissue acquisition and allocation Immediately after euthanasia, each mouse was decapitated and the brain was removed from the skull. The intact brain was placed immediately in ice-cold ACSF continuously aerated with 95% O2/5% CO2 and maintained on wet ice. The cerebral cortex and bilateral hippocampus were dissected under a stereomicroscope into 3x3x3 millimeter cubes while minimizing damage to adjacent structures. Tissues were pooled so that each resulting sample contained material from all three mice and were then allocated to fresh, snap-frozen, or vitrified conditions. The pooled material for each condition was subsequently divided into three sequencing reactions. These sequencing reactions represent technical, not independent biological, replicates. Fresh control Fresh-control tissue was processed immediately for homogenization in TRIzol as described below. Snap freezing Samples assigned to snap freezing were transferred to microcentrifuge tubes and immersed rapidly in liquid nitrogen for 5 min. The intact tissues were then maintained in liquid nitrogen for 4 days until processing. Vitrification solution The full-strength vitrification solution (100 wt%) was prepared from ethylene glycol (28 wt%), dimethyl sulfoxide (DMSO; 22.3 wt%), sucrose (17.1 wt%), and phosphate-buffered saline (PBS; 32.6 wt%). Ethylene glycol and DMSO were combined to achieve the high permeating-CPA fraction required for vitrification while avoiding reliance on a single permeating agent at the complete target concentration [27]. Sucrose was included as a non-permeating osmotic agent to promote controlled dehydration and reduce water available for ice formation. PBS served as the buffered aqueous carrier. The mixture was heated in a microwave for a total of 3 min in 1-min intervals, with vigorous stirring between intervals to dissolve the sucrose. The stock solution was diluted with PBS to prepare 75%, 50%, and 25% working solutions. CPA loading and vitrification Hippocampal and cortical tissues were transferred to separate microcentrifuge tubes and exposed sequentially to 100 μL each of 25%, 50%, 75%, and 100% vitrification solution for 5 min at each concentration. The stepwise loading procedure was used to moderate osmotic stress while permitting CPA penetration. After exposure to the 100% solution, tubes containing tissue and vitrification solution were immersed in liquid nitrogen for 5 min and then maintained in liquid nitrogen for 4 days until rewarming. Rewarming solution The full-strength rewarming solution was prepared to a total mass of 100 g from 28 g ethylene glycol, 22.3 g DMSO, 34.2 g sucrose, and 15.5 g PBS. The solution was heated in a microwave for a total of 5 min in 1-min intervals, with vigorous stirring between intervals, until the sucrose dissolved completely. The stock was diluted with PBS to prepare 75%, 50%, and 25% rewarming solutions. The greater sucrose concentration relative to the vitrification solution was used to increase extracellular osmolality during CPA efflux and moderate rehydration-associated volume excursions [28]. Rewarming and CPA unloading Upon removal from liquid nitrogen, vitrified samples were rewarmed immediately with gentle agitation in a 37 °C water bath until the vitrification solution returned to the liquid phase. The 100% vitrification solution was aspirated and replaced with 100 μL of 75% rewarming solution. CPA unloading proceeded sequentially through the 75%, 50%, and 25% solutions, with a 5-min incubation at each concentration. Samples were then submerged in 100 μL PBS for 5 min. No delayed post-rewarming recovery interval was introduced before homogenization. TRIzol homogenization and storage After condition-specific processing, samples were homogenized in TRIzol (Thermo Fisher Scientific) using a hand-held tissue homogenizer (QIAGEN), following the TRIzol manufacturer's instructions. The resulting TRIzol homogenates were stored at −80 °C overnight before shipment on dry ice to Beijing Genomics Institute (BGI) for RNA extraction and sequencing. All experimental groups underwent the same post-homogenization handling. RNA sequencing Bulk RNA sequencing was performed by BGI using the DNBSEQ platform with paired-end 150-bp reads. Reads were aligned to the Mus musculus reference genome GCF_000001635.26_GRCm38.p6. The mean genome-alignment ratio was 99.11%, the mean gene-set alignment ratio was 75.91%, and 18,389 genes were detected. All libraries showed stable base composition and uniformly high Phred quality scores. Bioinformatic analysis Principal-component analysis and heatmaps were generated using BGI's Dr. Tom software suite. Differential-expression outputs were examined in PandaOmics (Insilico Medicine), which was used to generate volcano plots and prioritize downregulated genes using the iPANDA pathway-scoring algorithm. iPANDA integrates weighted expression changes across predefined pathways and applies noise-resistant normalization and penalization to emphasize coordinated pathway-level shift [29]. The resulting pathway-impact metrics were used to identify pathways disproportionately associated with preservation conditions and to rank affected gene sets. Gene annotation and functional metadata were obtained from UniProt. Quantification and statistical analysis The study used tissue from three animals. Tissues were pooled before allocation to the experimental conditions; consequently, the animals do not constitute three independent biological replicates within each condition. Each pooled condition was divided into three sequencing reactions, which represent technical replicates. The study was exploratory and was not designed to estimate inter-animal variability or population-level effect sizes. Transcript-abundance comparisons followed thresholds applied by the sequencing provider, BGI, typically an adjusted p value < 0.05 and an absolute log2 fold change ≥ 1 unless otherwise specified. Summary statistics, data sorting, filtering, and figure preparation were conducted in Microsoft Excel. No missing values were imputed, and pathways with insufficient gene coverage were excluded from scoring. For these analyses, transcripts with lower measured abundance than the fresh condition were described as under-represented. This term denotes lower relative abundance in the RNA-sequencing dataset and does not imply biological downregulation. Because all samples underwent the same extraction and sequencing procedures, exhibited uniformly high sequencing-quality metrics, and showed selective rather than transcriptome-wide differences, a generalized reduction in extraction efficiency is unlikely to explain the observed pattern. Nevertheless, transcript-selective differences in RNA recovery or library preparation cannot be excluded. Sample-size estimation, randomization, or blinding procedures were not performed. Key resources table REAGENT OR RESOURCESOURCEIDENTIFIERC57BL/6 female mice, 12 weeks oldGemPharmatech Co., Ltd.N/ASodium chloride (NaCl)MacklinS805275-500gPotassium chloride solution, 1 M reference material (KCl), 500 mLCoastal HongmengBW017401 (CRM)Sodium phosphate monobasic, anhydrousAdamas LifeCAS 7558-80-7d-Glucose monohydrateAdamasCAS 5996-10-1Calcium chloride solution, 0.25 M, sterileBeyotimeST365Magnesium sulfate, anhydrousAdamas LifeCAS 7487-88-9Sodium bicarbonateAdamas LifeCAS 144-55-8Hydrochloric acid, 37%AdamasCAS 7647-01-0Ethylene glycol, ≥99%AdamasCAS 107-21-1DMSO, cell-culture gradeBiosharpBL165BSucrose, molecular-biology grade, 2.5 kgMacklinS62121× PBS buffer, cell-culture gradeBiosharpBL302ATRIzol™ Reagent, 100 mLThermo Fisher Scientific15596026Software and algorithmsDr. Tom software suiteBeijingGenomics Institute2026 versionPandaOmicsInsilicoMedicine6.0MicrosoftExcelMicrosoftVersion 16.110.3 (26070318)UniProtUniProtConsortiumRelease 2026_02Reference genomeNCBI RefSeqGCF_000001635.26_GRCm38.p6OtherHand-held tissue homogenizerQIAGENTissueRuptor IIDNBSEQ sequencing platformBeijingGenomics InstitutePE150 Results To evaluate the transcriptomic consequences of different preservation strategies, the analysis first examined global gene expression patterns across fresh, snap-frozen, and vitrified samples (Fig. 2). This initial assessment aimed to determine whether preservation method and tissue type explained major sources of transcriptomic variance and to establish the overall scale and organization of preservation-induced differences before pathway-level and transcript-level investigations. Graphical abstract of methodological procedures. The tissues extracted from the brain were equally divided into 3 groups, vitrified, snap frozen and control, as indicated. Subsequently, RNA-seq data obtained from all groups were subjected to bioinformatics data analysis. Transcriptomic divergence across preservation methods and tissues For the purposes of these analyses, transcripts showing reduced abundance relative to the fresh condition are referred to as under-represented, reflecting reduced preservation or recovery rather than inferred biological downregulation. The analysis identified a substantial number of differentially expressed genes across all pairwise comparisons, with both preserved and under-represented genes observed in each condition (Fig. 3a). Principal component analysis revealed clear separation between preservation conditions at the transcriptome-wide level (Fig. 3b). The first principal component (PC1), explaining 96.20% of the total variance, captured the dominant differences among samples, while PC2 accounted for an additional 1.76% of the variance. Samples clustered tightly by treatment conditions, indicating high within-group consistency and a strong effect of preservation strategy on global gene expression profiles. To further examine the structure of transcriptomic differences observed across preservation conditions, hierarchical clustering heat maps were generated for differentially expressed genes in each pairwise comparison (Fig. 3c-f). These heat maps illustrate expression patterns across individual samples and provide additional resolution on the consistency of transcript representation within and between preservation methods. Heat maps are shown for vitrified versus fresh hippocampus (HV/HF) (Fig. 3c), snap-frozen versus fresh hippocampus (HS/HF) (Fig. 3d), vitrified versus fresh cortex (CV/CF) (Fig. 3e), as well as snap-frozen versus fresh cortex (CS/CF) (Fig. 3 f). In both hippocampus and cortex, vitrification-relative comparisons showed coherent blocks of transcript under-representation and over-representation compared to fresh tissue, whereas snap-frozen samples exhibited more modest deviations. To further characterize the direction and magnitude of transcriptomic differences between preservation methods, differential expression results were visualized using volcano plots for each pairwise comparison (Fig. 3g-j). These plots integrate effect size (log2 fold change) and statistical significance, enabling identification of transcripts exhibiting the largest changes in representation relative to the fresh condition. Comparisons involving vitrification showed a pronounced skew toward transcripts with negative log2 fold change, whereas snap freezing resulted in fewer and more symmetrically distributed changes relative to fresh tissue. The preservation method drives global transcriptomic separation and differential gene representation. (a) Number of differentially expressed genes identified across all pairwise comparisons between preservation conditions. (b) Principal component analysis of transcriptomic profiles across all samples. C: cortex, H: hippocampus, V: vitrified, F: fresh and S: snap-frozen. (c-f) Hierarchical clustering heatmaps of differentially expressed genes comparing vitrified, snap-frozen, and fresh hippocampal and cortical tissues. Gene-level expression values are shown as log2(exp + 1), scaled by row, with red indicating higher expression and blue indicating lower expression relative to the gene's mean. Dendrograms represent hierarchical clustering of both samples and genes (d) hippocampal vitrified vs hippocampal fresh. (d) hippocampal snap-frozen vs hippocampal fresh. (e) cortical vitrified vs cortical fresh. (f) cortical snap-frozen vs cortical fresh. (g-j) Differential transcript representation across preservation comparisons visualized by volcano plots. (g) Hippocampal vitrified vs hippocampal fresh. (h) Hippocampal snap-frozen vs hippocampal fresh. (i) Cortical vitrified vs cortical fresh. (j) Cortical snap-frozen vs cortical fresh. System-level and pathway-level consequences of preservation To determine whether transcript-level changes observed across preservation methods may cluster into coordinated biological pathways, differential expression profiles were further analyzed using the PandaOmics platform, a software suite used in biomedical research. PandaOmics's iPANDA score integrates gene-level fold changes with pathway topology and regulatory directionality to infer pathway-level activation states, enabling assessment of higher-order structure within the transcriptomic response beyond individual genes [29]. Subsequent analyses focused on pathways with negative iPANDA scores, with p values < 0.05 as these represent pathways predicted to be inhibited relative to fresh tissue and, in the context of this study, are most consistent with reduced transcript representation associated with preservation-related effects. Each pathway with a negative iPANDA score across both tissues and conditions was filtered. In the hippocampus, iPANDA analysis identified 302 pathways with negative scores in the snap-frozen condition and 873 pathways with negative scores following vitrification (Fig. 4). Of these, 210 pathways were shared between the two preservation methods. In the cortex, there were 130 pathways with a negative iPANDA score in the snap frozen cortex, and 896 negative iPANDA pathways in the vitrified hippocampus, of which 105 of these pathways overlapped. Number of pathways with a negative iPANDA scores across conditions. In the hippocampus the vitrified condition yielded nearly three-fold as many pathways with a negative iPANDA score as the snap frozen condition (with 210 overlapping pathways between both groups). In the cortex, the vitrified condition yielded nearly seven-fold more pathways with a negative iPANDA score than the snap frozen condition (with 105 overlapping pathways between both groups). These differences in the magnitude and overlap of negatively scored pathways prompted a deeper assessment of how these pathways cluster into functional categories within each tissue and preservation condition. Fig. 5a summarizes the number of pathways within each functional category that were shared between snap-frozen and vitrified conditions within individual tissues. Overlap was evaluated separately for hippocampus and cortex, and pathways were not required to be common across tissues. The largest numbers of shared pathways were observed in the signal transduction, metabolism, disease-related, developmental biology, and immune system categories. When this criterion was further restricted to retain only pathways shared across both preservation conditions and both tissues, the number of overlapping pathways was reduced to 13 (Fig. 5b). Impact of preservation method on cellular pathways. (a) Within-tissue pathway overlap between vitrified and snap-frozen samples (b) Pathways shared across tissues and preservation conditions. To assess whether negative iPANDA pathway scores reflected coherent gene-level behavior or mixed transcript responses within pathways overlapping in both tissues and conditions, genes contributing negatively to scoring pathways were further classified based on the consistency of their directionality across preservation comparisons. Transcripts exhibiting uniform directionality across all tissues and preservation conditions either consistently positive or consistently negative log2 fold changes were classified as consistently preserved or consistently under-represented, respectively. Transcripts showing mixed directionality, defined as a negative log2 fold change in at least two comparisons irrespective of tissue or preservation method, were classified as bidirectionally preserved. To reduce noise from minor fluctuations, only transcripts with absolute log2 fold-change values greater than 0.1 were included in this analysis. Transcripts exhibiting consistent directionality in only three out of four tissue–condition comparisons were excluded from this analysis to ensure stringent classification and to avoid ambiguity arising from partial or condition-specific effects. None of the transcripts contributing to any of the pathways in the categories of gene expression (transcription), metabolism of RNA, transport of small molecules or neuronal system categories met the selection criteria. Gene-level classification within negatively scoring pathways indicated that only a small fraction of transcripts exhibited consistent directionality across tissues and preservation methods, while the majority did not conform to a uniform preservation or loss pattern (Fig. 6). Distribution of total, consistently preserved, under-represented, and bidirectionally preserved genes within negatively scoring iPANDA pathways. Gene distribution within negatively scoring iPANDA pathways by preservation behavior. Bars represent total gene counts per functional category, partitioned into genes preserved across both methods (dark blue), consistently under-represented genes (purple), and genes with opposite directionality between tissues or preservation conditions (green). Selective transcript vulnerability to CPA exposure Rather than pursuing individual gene examples, the analysis further focused on transcript abundance patterns and preservation logic. Transcripts that remain comparable between fresh and snap-frozen tissue but are markedly under-represented following vitrification were hypothesized to represent transcripts selectively vulnerable to cryoprotectant exposure rather than cold-induced effects. To operationalize this preservation logic at the transcript level, log2 fold-change values were computed for all detected transcripts for each tissue and preservation method, including log2(CS/CF) and log2(CV/CF) for cortex, and log2(HS/HF) and log2(HV/HF) for hippocampus. Transcripts were then filtered independently within each tissue to identify those showing minimal deviation between snap-frozen and fresh samples, alongside pronounced under-representation following vitrification. Specifically, transcripts were required to exhibit near-zero-fold change under snap freezing relative to fresh tissue and a substantial negative fold change under vitrification, consistent with selective vulnerability to cryoprotectant exposure rather than cold-induced effects. Following tissue-specific filtering, overlap analysis was performed to identify transcripts meeting these criteria in both hippocampus and cortex (Fig. 7a). Functional characterization of altered transcripts. a)Transcripts selectively under-represented after vitrification but preserved after snap freezing. In the hippocampus, 190 transcripts were uniquely under-represented, whereas 104 transcripts were uniquely under-represented in the cortex. The central overlap indicates the 14 genes that exhibited vitrification-specific under-representation in both regions, suggesting shared susceptibility across tissues. b) Preservation-dependent changes across brain regions and subcellular architecture. c) Subcellular architecture and structural enrichment in snap-frozen versus vitrified samples. To assess whether transcripts that were selectively reduced after vitrification shared common structural features, we examined the annotated subcellular localization of the proteins they encode using UniProt annotations in both tissues (Fig. 7b). UniProt successfully mapped 95 of 104 cortex genes and 156 of 190 hippocampus genes. In both regions, vitrification-sensitive transcripts showed a strong and consistent enrichment for proteins routed through the secretory pathway. In the cortex, 33 of 95 genes (34.7%) encoded membrane-associated proteins and 18 (18.9%) encoded secreted or extracellular proteins, such that 53.6% of the set corresponded to surface-exposed or extracellularly trafficked protein architectures. Hippocampus-derived samples showed a very similar pattern, with 59 of 156 genes (37.8%) encoding membrane-associated proteins and 20 (12.8%) encoding secreted proteins, yielding a combined 50.6% representation of secretory-pathway targets. In contrast, classical intracellular proteins—including nuclear and cytosolic species—accounted for only 24.2% of cortex transcripts (23/95) and 19.8% of hippocampal transcripts (31/156). Transcripts encoding proteins localized to specialized structural compartments such as cilia and axonemes were also consistently overrepresented (7.4% in cortex and 8.3% in hippocampus), indicating a bias towards non-cytosolic, architecturally complex compartments. In contrast, protein (transcript) length distributions did not reveal a dominant size class and showed no evidence that extreme protein lengths drive transcript loss. In the cortex, proteins clustered primarily within a mid-length range: 62 of 95 proteins (65.3%) were 200–600 amino acids long. Hippocampus displayed the same pattern, with 95 of 156 proteins (60.9%) falling in the mid length category. Beyond subcellular localization and protein size, additional structural annotations revealed further convergence on secretory-pathway and membrane-associated architectures (Fig. 7c). Signal peptides were present in 35.8% of cortex proteins (34/95) and 25.6% of hippocampal proteins (40/156), and transmembrane segments were identified in 32.6% (31/95) and 32.1% (50/156), respectively. When considered together, more than half of the transcripts in each tissue encoded proteins requiring either signal-peptide-directed ER entry, membrane insertion, or both (54.7% in cortex; 42.9% in hippocampus). Hippocampus displayed a disproportionate enrichment of multi-pass membrane proteins, with 27 of 50 transmembrane proteins (54.0%) containing two or more membrane-spanning helices, including a prominent cluster of 7-TM architectures characteristic of GPCR-like or transporter-like topologies. Cortex had 31 total transmembrane proteins, of which 13/31 were multipass. Glycosylation patterns reinforced this theme: 47 of 156 hippocampal proteins (30.1%) carried annotated glycosylation sites—primarily N-linked—and most of these were simultaneously membrane-associated or signal-peptide–containing (70.2% and 59.6%, respectively), consistent with proteins that depend on ER/Golgi processing falling between 200 and 600 amino acids. Discussion Cryopreservation imposes multiple biophysical and biochemical stresses on biological systems, yet the transcriptome-wide consequences of different preservation strategies remain incompletely understood, particularly in structurally complex neural tissues. This study identified a selective set of transcripts that were preserved under snap freezing but under-represented following vitrification in both murine cortex and hippocampus. Although this group represents only a small fraction of the ∼17000 detected transcripts, its composition was highly non-random. Affected genes were consistently enriched for membrane-associated proteins, multi-pass transporters, secreted factors, and signaling components. The limited scale of differential expression observed here is consistent with prior transcriptomic studies of cryopreserved multicellular systems. Anjos et al. found that only 2–3% of Crassostrea angulata larval transcripts changed significantly after freeze–thawing despite clear physiological impairment, with effects concentrated in membrane biology, receptor-mediated signaling, and intracellular trafficking rather than diffuse transcriptomic disruption [30]. Similarly, Cordeiro et al. demonstrated that vitrification of bovine oocytes produced structured, pathway-specific transcriptomic shifts enriched for glycoproteins, transporters, and membrane-associated signaling molecules even when cryoprotectant exposure alone produced minimal alterations [31]. Together with our data, these findings point towards specific vulnerability affecting membrane-integrated and extracellular signaling architectures across diverse taxa, tissues, and cryopreservation strategies. Although comparable transcriptome-wide vitrification studies in mammalian brain tissue are lacking, there are studies demonstrating feasibility of reversible cryopreservation in microglia [32], small-molecule neural precursor cells [33], and brain organoids [20]. Based on previous cryoprotectant toxicity resistance research [34], genes identified as conferring to resistance were compared with this dataset (e.g., Myh9, Nrg2, Pura, Fgd2, Pim1, Opa1, Hes1, Hsbp1l1, Ywhag). No coordinated pattern emerged, and effects were generally small and inconsistent across contrasts. This suggests that the selective under-representation reported here reflects a preservation-outcome axis distinct from pathways identified through forward genetic screens in cell culture. Finally, in this dataset a subset of transcripts appeared increased after vitrification relative to snap-frozen and fresh tissue. Without direct measures of transcriptional activity, we interpret these changes cautiously. They may reflect differential RNA stability in high CPA concentrations, selective depletion of other transcripts, or method-specific biases in extraction or library preparation rather than bona fide induction. De novo transcription during CPA exposure or rewarming cannot be excluded but remains speculative. Although comprehensive transcriptomic profiling of mammalian neural tissue following vitrification remains limited, ischemic injury serves as a valuable pathophysiological proxy for understanding molecular vulnerability due to shared mechanisms of metabolic crisis and membrane disruption. In both settings, the hippocampus exhibits heightened sensitivity compared to the relatively resilient cortex; in ischemic models, this manifests as acute excitotoxic death in CA1 pyramidal neurons and abundant transcriptional dysregulation, whereas cortical populations sustain significantly milder atrophy and fewer gene expression changes [35,36]. This established biological vulnerability parallels the molecular susceptibility observed in vitrified tissue, where the hippocampus displays a more profound under-representation of transcripts than the cortex. Furthermore, the specific functional classes compromised by vitrification, predominantly membrane-associated architectures, signal transduction components, and metabolic networks, mirror the primary axes of dysregulation in ischemia, which is characterized by the breakdown of ion channel homeostasis, the disruption of calcium signaling pathways, and maladaptive metabolic shifts such as upregulated ribosome biogenesis [36,37]. Thus, while ischemia represents a biological reaction to injury and vitrification represents a biophysical stress, both conditions reveal that complex membrane-integrated systems and high-demand metabolic pathways constitute the pathogenic centers of neuronal vulnerability. Limitations Several limitations of this study should be acknowledged. First, this analysis is restricted to bulk RNA sequencing and does not distinguish between cell-type-specific effects or regional heterogeneity within tissues. Second, transcript representation was assessed after a fixed storage duration and a single vitrification protocol, limiting generalizability across cryoprotectant formulations, loading rates, or rewarming strategies. Given that transcriptional responses are among the earliest and most dynamic molecular changes triggered by cellular stress, RNA profiles offer a sensitive first layer for detecting preservation-related perturbations. Accordingly, this study focused on transcriptomics as an initial readout, with proteomic, epigenomic, and functional assessments reserved for future work. Finally, the use of pathway inference tools, while valuable for identifying higher-order structure, introduces dependence on curated pathway definitions that may not fully capture preservation-induced molecular artifacts. The present design cannot determine whether vitrification-associated transcript under-representation reflects molecular changes within the tissue or condition-dependent technical effects. Although all samples underwent the same downstream extraction and sequencing workflow and exhibited high sequencing-quality metrics, differences in RNA extraction efficiency, transcript recovery, residual CPA interference, or library preparation cannot be excluded. No exogenous RNA spike-ins or orthogonal transcript-quantification methods were included to distinguish biological loss from technical under-recovery. Accordingly, the findings are interpreted as transcriptomic associations with the complete vitrification workflow rather than direct evidence of RNA damage or degradation. Future directions Future studies should determine whether the transcriptomic signatures identified here correspond to functional preservation outcomes. This will require coupling RNA-seq with viability and physiological assays in vitrified neural tissue, including electrophysiology, mitochondrial function, and synaptic activity. Achieving functionality in hippocampal slices after vitrification has been previously demonstrated [21]. Systematic evaluation of individual cryoprotectants and defined cocktails will be essential to establish whether specific CPA chemistries can mitigate the pronounced membrane- and secretory-pathway vulnerability observed in this study. Proteomic profiling will further clarify whether transcript under-representation translates into proportional protein loss or whether post-transcriptional buffering preserves functional architecture. Finally, if vitrified neural tissue can be rendered viable, longitudinal analyses after rewarming will be critical to distinguish transient CPA-loading responses from persistent molecular injury. Together, these directions will define the mechanistic basis of vitrification-induced transcript loss and guide the design of preservation strategies with improved molecular fidelity. Despite the limitations, the findings have important implications for cryopreservation research, particularly in the context of organ-scale and whole-organism strategies that rely on vascular perfusion with a single cryoprotectant formulation. The assumption that a given vitrification protocol preserves transcriptomic integrity uniformly across tissues is not supported by the transcriptomic patterns observed here. Instead, the data show selective transcript under-representation following the vitrification workflow, with patterns varying across transcripts, tissues, and pathways. Whether this under-representation reflects biological alteration, differential transcript recovery, or a combination of both remains unresolved. For applications where downstream molecular analysis, omics-based profiling, or long-term informational preservation is a priority, these effects warrant explicit consideration. Conclusion Organ cryopreservation research is advancing rapidly, yet its effects on transcriptomic integrity remain largely unexplored. Our data show that vitrification conditions are associated with selective under-representation of transcripts encoding membrane-integrated and secretory-pathway proteins, revealing a dimension of molecular vulnerability that is not detectable through morphology or functional assays alone. Integrating molecular readouts into cryoprotectant development and protocol optimization may therefore be essential for advancing cryopreservation toward applications that demand not only survival or function but faithful preservation of biological information. Ethics declaration This study was conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. This study was approved by the institutional animal care and use committee (IACUC) guidelines of XJTLU. (Approval No. ER-SRR-1088616620231127132732) Funding This project is supported by DKU Office of Academic Services through a Signature Work Research Grant and additional funding by the Signature Fund. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbrep.2026.102782. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 Data availability data is publicly available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE325022. References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement data is publicly available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE325022.