Chinese hamster ovary (CHO) cells stably expressing human aquaporin-1 (AQP1) or mouse aquaporin-4 (AQP4) and Madin-Darby canine kidney cells (MDCK) cells stably expressing AQP4 were maintained at 37°C in a humidified atmosphere with 5% CO₂-95% air in a growth medium consisting of Ham's F-12 (Wako, Japan) for CHO and D-MEM (Wako) for MDCK fortified with 10% FBS (fetal bovine serum) (Wako), 1% penicillin/streptomycin (Life Technologies) and 0.5 mg/mL G418 antibiotics (Nacalai tesque) in 10-cm culture dishes. Stable CHO cell clones with or without AQP4 were then established for reference, see the previous paper [21]. Mixtures of stable CHO cells expressing either AQP4-IRES-EGFP or IRES-EGFP were cultured in a dish at a density of 1×10⁵ cells/dish at a ratio of 1∶0, 3∶1, 1∶1, 1∶3 or 0∶1 for a few days. CHO cells were seeded onto 60-mm dishes at a density of 1×10⁵ cells/dish and were transfected with ET_AR-IRES-EGFP gene using Lipofectamine and plus reagents (Life Technologies), according to the manufacturer's instructions.

Primary cultured astrocytes derived form either WT- or AQP4-null mice (Acc. no. CDB0758K: http://www.cdb.riken.jp/arg/mutant%20mice%20list.html) were prepared as described previously [22]. This study was carried out in strict accordance with the recommendations in the guide for the care and use of laboratory animals of the MEXT of Japan. The protcol was approved by the Animal Care Committee of Keio University School of Medicine (Permit Number: 080007). All surgery were made under the protcol, it was performed to minimize suffering. The 5′- and 3′-homology arms were obtained from the BAC clone RP23-189N2 (BACPAC Resources).

Transgenic mouse embryonic stem (ES) cell lines with Tet-off system were cultured on the feeder dishes with conventional medium for undifferentiated-cell state, supplemented with 15% FCS, recombinant Leukemia inhibitory factor (LIF), 2-mercaptoethanol and so on as describe elsewhere. Doxycycline (Dox) and LIF (1 µg/ml) were added into the culture medium for the suppression of transgene and maintenance of undifferentiated cell state. For differentiation and transgene induction, cells were cultured in α-MEM medium without Dox and LIF. These differentiated cells were collected post 3day. Spontaneously differentiated ES cell were also harvested as a typical differentiated ES cells [18], [19].

XhoI and SacII sites were added to rat endothelin receptor A (ET_AR) cDNA [23] using PCR with the primers 5′-CTCGAGAAGATGGGTGTCCTTTGCTTTCTG-3′ and 5′-CCGCGGTTAGTTCATGCTGTCCTTGTGGC-3′. cDNA encoding the mouse AQP4 M1 isoform was connected in frame with the initiation codon of an internal ribosomal entry site (IRES-AQP4), and a NotI site was added at the 3′ end using 2-step PCR with the following primer sets: 5′-ACCGGACTCAGATCTCGAGCTCAAGCTTCG-3′ and 5′-CTCGCTGCAGCTCCGTCACTCATGGCCATATTATCATCGTG-3′; and 5′-CACGATGATAATATGGCCATGAGTGACGGAGCTGCAGCGAG-3′ and 5′-GCGGCCGCCTATACGGAAGACAATACCTC-3′. The PCR products were inserted into pGEM-T vector (Promega) to confirm their sequences. Then, the IRES-AQP4 cDNA was excised with HindIII and NotI and was inserted between the HindIII and NotI sites of a pIRES2-EGFP vector (Clonetech) to generate the pIRES-AQP4 vector. The pIRES-AQP4 vector was digested with XhoI and SacII, and the ET_AR cDNA was inserted to produce the pET_AR-IRES-AQP4 vector.

Cultured cells were trypsinized, collected, and suspended in culture medium containing 10% FBS and cryoprotectant agent at a density of 1×10⁶ cells/mL for cryopreservation. In the cryopreservation, cryoprotectant agent, dimethylsurfoxide (DMSO) was used for final 10% concentration except for experiments regarding to a dose dependent effects of DMSO. The Bicell freezer box (Nihon Freezer) was provided at cooling rate of −1°C/min in a −80°C freezer. At cooling rates of −10, −30 and −50°C/min in mechanical freezers (Planer) were verified the monitoring of vapor phase and sample on the each cooling conditions. At the cooling rate of −120°C/min in a liquid nitrogen tank, frozen samples were directly immersing to liquid nitrogen. This condition was verified the monitoring of the temperature in the freeze preservation medium. All samples were frozen at each cooling rates of starting at room temperature, and finally stored in vapor phase nitrogen for a minimum 1 hour. Each experiment was repeated more than 3 times.

The numbers of AQP4-expressing cells before and after freezing were analyzed using a FACScalibur flow cytometer (Becton Dickinson) equipped with an argon laser emission of 488 nm, described in Miyazaki et al. [21]. AQP4-expressing cells were detected by staining with a monoclonal antibody against the extracellular domains of mouse AQP4 (1∶5) followed by the anti-mouse IgG conjugated with phycoerythrin (PE)(1∶100). A primary gate based on forward and side light scatters (FSC and SSC, respectively) was set to exclude dead cells or debris. The background level was estimated by omitting the primary antibody.

RT-PCR was performed to identify transcripts encoding mammalian AQPs in undifferentiated and differentiated mouse ES cells mouse. PCR-amplification was performed using specific primers for mouse AQP4: 5′-CTGGAGCCAGCATGAATCCAG -3′ and 5′-TTCTTCTCTTCTCCACGGTCA -3′ [24]. The experiments were carried out with a number of cycles that precedes saturation. PCR products (10 µl) were separated by electrophoresis on a 2.0% agarose gel and visualized after ethidium bromide staining under UV radiation. The expected PCR product size was 310 base pairs.

Transfected pET_AR-IRES-AQP4 cells with or without 3 cycles of freezing were cultured in 96-well thin-glass bottom dishes (1×10⁴ cells/well). The cells were loaded with 10 µM of the Ca²⁺ sensitive fluorescence indicator OregonGreen 488 BAPTA-1/AM (Life Technologies) at 37°C for 15 min in DMEM/F12 (1∶1) buffer (Life Technologies). Bolus injections of Oregon Green 488/AM were performed as described elsewhere [25]. The imaging of Oregon Green 488 fluorescence in cells was performed using a FV1000 confocal microscope equipped with a 60×/1.2 NA water-immersion objective (Olympus). Fluorescence data were analyzed using custom Fluoview software (Olympus). Ca²⁺ imaging was mediated using 10 µM of ET1 (Peptide Inc.) added 60 seconds later.

Statistical analysis was performed using Microsoft Excel with the Statcel2 add-on (OMS). Data are presented as mean±S.D. The Student t-test was used for paired data of each of the groups compared (Statistical significance was defined as *P<0.05, **P<0.01, ***P<0.001).

In this study, we evaluated possible roles of AQP in cryopreservation using mammalian cells, and its application in selecting a specific type of cells. Table 1 shows cell survival rates after ultra-quick freezing. Both, Chinese Hamster Ovary (CHO) and Madin-Darby canine kidney (MDCK) cells were affected by low survival rates of only 2.4±1.4% and 0.7±0.8%, respectively under cooling rate of −120°C/min (ultra-quick freezing) in 10% dimethyl sulfoxide (DMSO) solution. Interestingly, survival rates dramatically increased for cells stably expressing AQP4: 60.5±16.7% for AQP4-CHO cells and 37.2%±8.7% for AQP4-MDCK cells (Table 1). Cryoprotective effect was not limited to AQP4 since CHO cells expressing AQP1 also revealed high survival rate after ultra-quick freezing (Figure S1A). Surprisingly, survival rate of AQP4-CHO cells was still high even the concentration of DMSO is decreased from 10% to 1%. It should be noted that decreased concentrations of DMSO from 10% to 1% are associated with the decrease of osmolality from 1,600 mOsm to 432 mOsm (Figure S1B), which leads to decrease in osmotic stress and cellular toxicity [26], [27]. We further evaluated whether endogenously expressed AQP4 is sufficient to exert a cryoprotective effect, using primary cultured astrocytes derived from either wild-type or AQP4-null mice. Astrocytes from AQP4-null mice exhibited limited cell survival after ultra-quick freezing (2.5±2.3%), while the survival of wild-type astrocytes that express AQP4 endogenously, was significantly higher (54.1±6.4%) (Table 1).

In order to understand the mechanisms behind the effects of AQP on cell survival after ultra-quick freezing, we next examined cell survival rate at different cooling rate. No significant difference in the cell viability was seen at freezing rates of −1, −30, or −50°C/min between the control CHO cells (dotted line) and those expressing AQP4 (AQP4-CHO cells) (solid line). However, at −120°C/min, a significant difference in survival rate was observed: 2.4±1.4% for CHO cells, and 60.5±16.7% for AQP4-CHO cells, indicating that the expression of AQP resulted in freezing tolerance at high cooling rate (Figure 1A).

A colony forming assay revealed an even clearer difference: significantly better survival rates were observed for AQP4-CHO cells at freezing rates of −30, −50, and −120°C/min. However, survival of CHO cells without AQP4 was not deteriorated by increasing freezing rates (Figure 1B).

To evaluate the cell membrane damages after ultra-quick freezing (−120°C/min), the cells were stained with phycoerythrin (PE)-conjugated Annexin V and 7-amino-actinomycin D (7-AAD) and then analyzed using flow-cytometry. The control CHO cells shifted to late apoptosis or cell death (both Annexin V and 7-AAD-positive) as freezing rate was increased, whereas AQP4-CHO cells remained viable for all tested freezing rates (Figure 1C). These results strongly suggest that the expression of AQP reduces membrane damage caused by freezing/thawing, thereby inhibiting cell death. Scanning electron microscope (SEM) imaging further confirmed the rescue of the cells from membrane damage after ultra-quick freezing by expression of AQP4. No difference in morphological characteristics (smooth and spherical cells with a slight bubbling effect) was observed between CHO cells and AQP4-CHO cells before freezing (Figure 1D, left panels). After ultra-quick freezing, CHO cells appeared to be burst, with flattering and invaginations caused by cell membrane damage, whereas AQP4-CHO cells exhibited spherically smooth surfaces, similar to cells before freezing (Figure 1D, right panels).

Having identified that AQP expression can acquire cryo-resistant feature of the cells, we hypothesize that cells expressing AQP can be selected by ultra-quick freezing. CHO cells (open circle) and AQP4-CHO cells (filled circle) were mixed in ratio of 1∶0, 3∶1, 1∶1, 1∶3, or 0∶1 and then cultured. The population of mixed-cell cultures before and after ultra-quick freezing was analyzed using flow-cytometry by staining with anti-AQP4 antibody followed by a secondary antibody conjugated with phycoerythrin (PE) (Figure. 2, upper panels). A clear shift in the PE intensity towards higher values occurred, regardless of the ratios of the mixture, indicating that only the AQP4-CHO cells were retained after ultra-quick freezing/thawing (Figure 2, lower panels). These data indicate that outstanding selection of cells can be achieved by ultra-quick freezing based on the AQP expression profile.

Bioinformatics approach revealed that AQP4 expression is gradually increased during ES cell differentiation into neural ectoderm lineage triggered with Achaete-scute homolog1 (Ascl1) induction in Tet-off system and differentiated cells were confirmed by morphological changes of the cells (Figure 3A). To evaluate freeze tolerance along with AQP4 expression, we examined the viability of Ascl1 induced cells after ultra-quick freezing. The expression of AQP4 mRNA was detected in the differentiated but not in the undifferentiated ES cells (Figure 3B). Then we examined cell viability of both undifferentiated and differentiated ES cells after ultra-quick freezing at different concentration of DMSO. Cell viability was 1.0±1.4% or 6.8±2.1% for the undifferentiated cells and 25.5±2.1% or 38.0±12.6% for the differentiated cells in media containing 3% or 10% DMSO medium, respectively (Figure 3C). These results strongly suggest that differentiated ES cells, if they express AQP4, can be highly selected since undifferentiated ES cells are more efficiently excluded by ultra-quick freezing with medium containing 3% DMSO comparing to 10% DMSO.

We next examined if this method of selection can be applied to a transient expression system [28]. CHO cells were transiently transfected with a plasmid containing AQP4 and EGFP cDNAs connecting with internal ribosome entry site (IRES) (Figure S3A). EGFP was used to monitor the transfection efficiency. The transfection efficiency was significantly increased by ultra-quick freezing (61.3±7.3%), compared to cells that were not frozen (13.7±10.3%), 2 days after transfection (Figure S3B). This effect was transient, since the number of GFP positive cells gradually decreased and disappeared after 5 days post thawing (Figure S3C, blue and red line). However, we found that the population of GFP positive cells remained high even 5 days post thawing when we performed more than three cycles of freezing/thawing, suggesting that stably transfected cell line can be mimicked by this strategy (Figure S3C, green and purple line). We noticed that the order of insertion of cDNA in the IRES construct is also important [29] since flow-cytometry analyses revealed different transfection efficiency assessed with first gene (AQP4) or second gene (EGFP), 73.0±4.7% and 52.8±5.0%, respectively after 3^rd ultra-quick freezing/thawing (Figure S3D). IRES has been widely used to translate multiple gene products from a transcript and little known the efficiency of IRES-dependent second gene expression relative to first gene expression. In addition, the latter part is sometimes truncated when the construct is integrated into the genomic DNA of host cells. Our results showed the reduced expression of the second gene in pAQP4-IRES-EGFP (Figure S3D). It is, therefore, important to insert the gene of interest as the first gene and AQP as the second gene in the IRES construct for this purpose.

To examine if this method can be applied for a functional assay of the gene of interest, a plasmid containing ET_AR (Endothelin Receptor A)-IRES-AQP4 construct was transiently transfected into CHO cells (Figure 4A). We performed functional assay for ET_AR in the AQP4-positive cells concentrated after multiple ultra-quick freezing/thawing or cultured post transfection without freezing. We confirmed that the cell survival rate significantly increased by transfecting the ET_AR-IRES-AQP4 construct and stably retained high values after the third freezing/thawing and during three weeks after transfection (Figure 4B). A flow-cytometry analysis further confirmed that most of the cells were AQP4-positive (Figure 4C). We next performed Ca²⁺ imaging to assess the function of ET_AR. The [Ca²⁺]i increases, evoked by stimulation with endothelin-1 (ET-1), were almost negligible in the non-freezing cells. In contrast, the [Ca²⁺]i increases were clearly observed in treated cells (91%) even three weeks later (Figure 4D and Video S1). These results strongly suggest that multiple cycles of ultra-quick freezing/thawing episodes not only increase the transfection efficiency of the genes co-transfected with AQP, but also allow the cells to stably express them.