Genetic suppression of cryoprotectant toxicity

We report here a new, unbiased forward genetic method that uses transposon-mediated mutagenesis to enable the identification of mutations that confer cryoprotectant toxicity resistance (CTR). Our method is to select for resistance to the toxic effects of M22, a much-studied whole-organ vitrification solution. We report finding and characterizing six mutants that are resistant to M22. These mutants fall into six independent biochemical pathways not previously linked to cryoprotectant toxicity (CT). The genes associated with the mutations were Gm14005, Myh9, Nrg2, Pura, Fgd2, Pim1, Opa1, Hes1, Hsbp1, and Ywhag. The mechanisms of action of the mutations remain unknown, but two of the mutants involve MYC signaling, which was previously implicated in CT. Several of the mutants may up-regulate cellular stress defense pathways. Several of the M22-resistant mutants were also resistant to dimethyl sulfoxide (Me2SO), and many of the mutants showed significantly improved survival after freezing and thawing in 10% (v/v) Me2SO. This new approach to overcoming CT has many advantages over alternative methods such as transcriptomic profiling. Our method directly identifies specific genetic loci that unequivocally affect CT. More generally, our results provide the first direct evidence that CT can be reduced in mammalian cells by specific molecular interventions. Thus, this approach introduces remarkable new opportunities for pharmacological blockade of CT.


Introduction
The limited shelf life of donated organs negatively affects the success of human organ transplantation [1,22,35]. Efficient organ cryopreservation could potentially overcome many obstacles to organ transplantation, thereby reducing costs, improving tissue matching, reducing organ wastage, enabling elective scheduling of transplantation, and facilitating tolerance induction. It could also, in combination with the generation of bioartificial organs or xenograft organ sources, enable an end to the organ shortage. There is now good evidence that the banking of large organs is possible [3,5,14,21,24,29,31,42,54,57,60], but the toxicity of the cryoprotective agents needed to preserve certain organs, such as the kidney, appears to be the major limiting factor [30].
Cryoprotectant toxicity (CT) is particularly problematic during cryopreservation by vitrification (ice-free cryopreservation) [16,53], which has been gaining in popularity in recent years [23,27,28,35,45]. Vitrification requires extremely high concentrations of cryoprotectants [17,28], and the higher the concentration, the more likely it is that CT will arise. Further, since whole organs must be cooled relatively slowly [12], this necessitates using relatively high cryoprotectant concentrations [28]. These high concentrations have high viscosity, which means slow introduction by perfusion and consequently relatively long exposure times to intrinsically toxic concentrations of cryoprotectant. Therefore, direct remedies for CT are needed for vitreous organ banking, and should have numerous practical advantages in cryobiology in general.
Here we describe an entirely new way of addressing the problem of CT. First, we created libraries of transposon-mutagenized mouse embryonic stem cells (ESC). We then selected mutant cells that could survive exposure to 9% of full-strength M22 (an organ vitrification solution; [21]) at 37°C. This approach (illustrated in detail in Fig. 1) allows the cell itself to identify pathways that are crucial for reducing CT. We demonstrate here, for the first time, the existence of genes whose modulation can confer CT resistance (CTR). We also show that the same genes that confer CTR also provide significant protection

Cells and cell culture conditions
We generated a mouse ESC line from an F1 hybrid background (C57BL/6J × 129X1/SvJ), referred to here as "C9" [7,8]. For all experiments, cells were thawed and grown for 24 h under standard culture conditions prior to harvesting for experimentation, unless otherwise stated. (We plated 100,000 cells in 5 ml ESC medium, and cultured cells at 37°C [2].)

Toxicity of M22 to C9 cells at 2°C
Cells were trypsinized, divided into equal-volume aliquots of 150,000 cells, pelleted, and supernatants drawn off. The method of M22 addition and removal was modified from Guan et al [36]. All pellets were suspended in 333 μl ESC medium at 2°C, and brought to 60% M22 by stepwise addition of 2°C 100% M22 in LM5 to yield 10, 20, 40 and finally 60% M22, at 10-min intervals. After 0, 1, 2, 3, 4 and 8 h of exposure to 60% M22 at 2°C, single samples were returned from 60% M22 to near 0% M22 using sequential 10-min intervals. The M22 concentration of each tube was adjusted to 30% by adding 600 mM mannitol solution (at 2°C) in LM5 carrier solution, and then in subsequent intervals, adjusted to 15, 7.5 and 3.75% M22, by adding 300 mM mannitol (at 2°C) in LM5 [36]. Following return to near 0% M22, each sample was pelleted and cells were resuspended in 5 ml fresh ESC medium and cultured in a T25 flask at 37°C . At 48 h after the end of each exposure period, the numbers of cells surviving (adherent to the floor of the flask) were counted.

Mutagenesis and gene identification
Briefly, mouse ESC were mutagenized via random insertion of a piggyBac (pB) transposon [7,8]. We estimate that the mutant library includes about 42,000 independent mutations (of which approximately 12,000 have been tested). Use of the pB transposon allows subsequent excision from the genome (reversal of the mutation) to verify causality. Although most insertions cause loss of function, overexpression has also been observed [8]. The mutant cell lines recovered remain heterozygous for the mutation unless treated further. Full details are available elsewhere [8].

Selection for CTR
To select mutant cell lines resistant to CT, cells from the library were plated in 15-cm plates at 6 × 10 6 cells per plate in 9% M22 and cultured at 37°C for 48 h. Normal ESC medium without M22 was reinstated for 7 days of standard culture, after which six surviving colonies were picked and transferred to fresh individual flasks for recovery and analysis. All pB insertion sites were determined by Splinkerette PCR [59]. This allowed identification of the genomic insertion sites associated with CTR.

Validation of M22-resistant mutants
Mutants initially selected for resistance to M22 (as well as C9 cells used as controls) were re-cultured and seeded at 20,000 cells per well of a 96-well plate in 0-6% M22. After 48 h at 37°C growth within each well was measured by the MTT assay [48]. The same mutants were also tested for resistance to Me 2 SO at concentrations of 0-5%. (3) Mutations putatively conferring CTR are identified. (4) Causality is verified by demonstrating that CTR is conferred by other mutations in the candidate gene and is abolished by excising the pB transposon and (5) generating genomic equivalents of that mutation. (6) Bioinformatics are used to link the identified mutation to previously-identified pathways. (7) A search is made to identify known drugs that may induce CTR. Candidate drugs are then tested on wildtype cells for efficacy in CTR. Drug effects are characterized (8) molecularly and (9) phenotypically. (10) Ultimately, mice carrying the original mutation are generated and novel improved drugs are developed using a variety of distinct methods, and (11) their organs are tested for CTR.

Cryopreservation with 10% Me 2 SO
Cells were trypsinized, pelleted, and resuspended at a concentration of 250,000 cells per 250 μl. An equal volume (250 μl) of 2X freezing medium (60% ESC medium, 20% FBS, 20% Me 2 SO) was added to each tube, all samples were mixed, promptly placed in a Mr. Frosty cellcryopreservation container (Thermo Fisher; Waltham, MA) and the container placed in a −80°C freezer for 24 h. After 24 h, samples were thawed by immersion in a 37°C water bath, resuspended in 5 ml ESC medium, pelleted, resuspended in a final volume of 500 μl ESC medium and tested for survival using trypan blue and colony formation assays [52,55].

Viability assays
To test viability using the Trypan Blue assay, 200 μl of each cell suspension in ESC medium was added to 500 μl of 4% (w/v) trypan blue premixed with 300 μl 1X DPBS, mixed thoroughly, and maintained at room temperature for 15 min, followed by scoring of blue (dead) and clear (live) cells.
To test viability using colony formation ability, 300 μl of each cell suspension in ESC medium were plated in 5 ml of ESC medium in a T25 flask and cultured for 48 h, at which time dead cells (not adhered to the flask) were removed by rinsing twice with DPBS. This assay relies on counting colonies by eye to assess viability. To assess reliability, we used two independent estimates provided (blind) by two scorers counting the same sample, and results were compared ( Fig. 2; note high value of r 2 ).

Statistics
All statistics (t-test; proportional survival) were performed in Microsoft Excel. For Trypan Blue assays, all p-values are tests of independent proportions of living cells. For colony formation assays, all p-values are one-way t-tests of the mean number of colonies observed. For MTT assays, all p-values are two-way t-tests of the mean absorbance (representing living cells) at a given concentration.

M22 toxicity in wild-type ESC
Initially we tested the toxicity of the cryoprotectant solution M22 at 60% and 2°C, in wild-type mouse ESC, to determine reasonable conditions for later mutant selection. We observed an exponential rate of loss that continued until hour 4, after which about 30% of the cells continued to survive for as long as 8 h with little additional toxicity (Fig. 3).
An effective selection of mutants requires conditions that kill nearly all cells (mutant or wild-type), selecting only resistant mutants. We therefore tried increasing the M22 concentration to 80% at 2°C, which resulted in more cell death, but also resulted in living cells that displayed exceptionally slow growth and abnormal appearance. To avoid dealing with such slow-growing cells, we asked whether exposure of mutant ECSs to lower concentrations (5% or 10%) of M22 for prolonged periods at higher temperatures would result in a smaller population of surviving cells that still displayed reasonable growth rates. We found that selection in 5% M22 permitted too many ESC to survive, while selection in 10% M22 provided no clones capable of further replication (Fig. 4A).

Identification of M22-resistant mutants
We found that exposure of normal ESC to 9% M22 at 37°C over seven days resulted in near-complete lethality (Fig. 4B). We next exposed mutated ESC to 9% M22 for 48 h and found nine rare resistant mutant clones. We were able to identify the insertion site of the pB transposable element for six of these (Table 1).

Resistance of mutants to standard cryopreservation conditions
We next asked whether mutants displaying resistance to M22 under otherwise-normal cell culture conditions would also display greater viability under standard cryopreservation conditions with Me 2 SO. Viability was measured by both the trypan blue dye exclusion assay [55] and a colony formation assay (the ability of cells to form colonies adherent to the floor of the tissue culture flask [52]).
Mutant 2.1, containing a pB insertion in Gm14005, displayed particularly high survival after freezing and thawing as measured by trypan blue exclusion ( Fig. 6A-C). In Fig. 6A, mutant M2.1 showed a 12.7-fold increase over control survival (indicated as being outside the range of all other mutants in multiple experiments). Other tested mutants also displayed more moderate but still much higher survival relative to non-mutant controls (improvements of 3-4-fold; Fig. 6B-C). The colony formation assay confirmed that the mutant M2.1/ Gm14005   Fig. 2. Independent counts of cell colonies by different scorers.

Mutations can confer CTR and cryoprotection
To our knowledge, this is the first report of mutations in mammalian cells that confer CTR and resistance to freeze-thaw injury. Although there are genes known to confer protection against freezing and thawing in free-living bacteria [47], genes that modulate similar processes in mammalian cells have not been previously described.
The protective effects of the mutations described here were found to be very dependent upon the nature of the stressor. Survival of the mutants was at best only about 20% higher after the severe treatment of prolonged M22 exposure at 37°C. Survival was about 60% higher after similar treatment with Me 2 SO. However, this translated to survival after a more typical freezing and thawing stress that was up to four times higher than for normal ESC. This greater effect on survival after freezing may occur because freezing and thawing induces injury that may include significant CT caused by freeze-concentration of the cryoprotectant [15,18,19,25]. Injury caused by freeze-concentration is expected to be easier to control than toxicity at 37°C due to the protective effect of much lower temperatures [13,31,54] and the relatively short exposure times involved in cryopreservation vs. the extreme exposure times tested at 37°C.
Although mutations that were protective at high temperatures were also protective at low temperatures, the temperature of exposure is expected to affect at least some of the mechanisms of CT [2,19,28]. Therefore, different protective mutations might emerge from different types of selection in future experiments. For example, selection experiments could be performed using shorter exposure times and higher concentrations at either 37°C or much lower temperatures.
The results presented here indicate that the detrimental effects of cryoprotectant exposure depend on normal biochemical processes and can be inhibited by mutations. In this regard, selection at 37°C will certainly favor the discovery of pathways that have detrimental effects in the presence of cryoprotectants, since any such pathways will be maximally active at this temperature. However, our observations of a biphasic response to high concentrations of M22 at low temperatures within a genetically homogeneous population (Fig. 3) is also consistent with the possible existence of an adaptive state. This adaptive state may pre-exist in a small subset of cells, or it may be induced by exposure to M22. There is precedent for an adaptive state in the observation that mammalian cells in log phase growth are considerably more resistant to  Sequences identifying genomic locations of pB insertions sites associated with resistance to M22. "Start" and "End" indicate the position of each pB insertion and the 5′ and 3' ends of the sequence used in BLAST for alignment. All sequences used in BLAST were 95 base pairs, found in BLAST searches using the Mouse Genomic Plus Transcript (Mouse G + T) database against mouse genome assembly version GRCm38.p6. Asterisks indicate intergenic insertions between the two flanking genes. For exact sequences see Table 2. freezing and thawing than are stationary phase cells [33]. Further, in bacteria [47] and yeast [44,50], freeze-thaw survival is also dependent on cell cycle stage and metabolic conditions, including starvation and other applied stresses [50]. Such a protective state would support the possibility of deliberately altering gene expression to significantly reduce CT and open up the prospect of discovering additional protective mechanisms of clinical relevance.

Identified mutants suggest multiple possible pathways modulate CTR
The specific mutated gene providing the strongest protection in our freeze-thaw assays, Gm14005, is poorly characterized. It is most likely transcribed only into long non-coding RNA (lncRNA) species (see http://www.informatics.jax.org/sequence/marker/MGI:3652191? provider=RefSeq.). lncRNAs have regulatory effects [46], so the identification of Gm14005 points to previously unknown mechanisms, both for CTR and for freezing injury.
The mechanism of action is not clear. Our colony formation assay requires adhesion of the ESC to their culture dishes, and adhesion requires myosin-related function. It is not clear why reducing or knocking out expression of a single myosin gene would improve that function, or  what other myosin-related function may be key, but interestingly, the reduced freeze-thaw survival of mammalian cells in stationary phase vs logarithmic growth phase is reversed by trypsinization [33]. Other studies have indicated that impairment of this non-muscle myosin also protects against oxidative damage [43]. However, so far we have not observed other CTR mutants with evident roles in modulating oxidative stress.
The other four identified locations of pB insertion fall between known genes. For example, the mutations in clones M2.2 and M3.1 fall between Nrg2 and Pura and between Fgd2 and Pim1, respectively. It is not obvious why modification of NRG2 or FGD2 function might reduce CT. However, the functions of PURA and PIM1 may be related to previous observations of the transcriptomic response to CT in rat liver slices exposed to two M22-like vitrification solutions [37]. One of the most interesting findings was a 40% increase in expression of the transcript for MYC [37]. MYC is a regulator of ribosome biogenesis and protein synthesis, and might have been upregulated because of its role in facilitating the replacement of proteins chemically damaged or denatured by exposure to cryoprotectant [27,28]. It is tempting to think that, in our experiments, PURA modulation may better activate MYC (http://www.uniprot.org/uniprot/P42669) and PIM1 modulation may better stabilize MYC (http://www.uniprot.org/uniprot/P06803). MYC is, of course, also associated with cell division [6], which might relate to the observation that rapidly dividing mammalian cells are protected against freezing injury [33]. In principle, however, the Myc upregulation observed in the transcriptomics study could be maladaptive rather than protective, given that protein synthesis in the presence of previously denatured proteins can lead to apoptosis or other adverse events via the unfolded protein response (endoplasmic reticulum stress) [34,41]. Furthermore, mice heterozygous for loss of MYC function display at least an 11% increase in longevity without significant difference in cancer as a cause of death, as well as an increased healthspan [38]. Improvements in these traits are generally associated with increased resistance to manifold types of stress [11,40], which indicates that MYC may inhibit stress responses. In any case, it is notable that two studies based on entirely different methods for elucidating mechanisms of CT have both pointed to a common gene, and it will be interesting to investigate the implications.
One of the two genes flanking the M4.2 mutation (Opa1) modulates the mitochondrial permeability transition pore, which plays a key role in apoptosis [51]. Possibly, the mechanism of protection is blockade of apoptosis long enough for surviving cells to repair sub-lethal injury, although apoptosis was not upregulated in either of two transcriptomics studies of CT [10,37].
The pB insertion site of the last resistant clone, M5.1, is flanked by two genes (Hsbp1 and Ywhag) that may modulate stress resistance. The former may downregulate the overall stress response [56], and the latter encodes a 14-3-3 protein [39]. 14-3-3 proteins have been shown to retain the FOXO3 protein in the cytoplasm, a state associated with reduced stress resistance [4,49].
The studies reported here provisionally establish causality between the described mutations and protection against injury, but an additional step should be taken to completely prove this causality. Our choice of the pB transposon-to mutagenize ESC will enable us to excise the transposon in future studies. If doing so restores the wild-type cell susceptibility to cryoprotectant exposure and cryopreservation, causality between the mutation and the observed protective effects will be beyond doubt. Validation of the resistance of these (or newly-generated) mutants could also be supported by generation and testing of similar mutations using CRISPR-Cas9 or other genomic alterations.

Our model probably selects for resistance to non-specific CT
The maximum tolerated concentration of M22 at 37°C was about 6% (v/v) M22, or about 4.5% (w/v) M22 cryoprotectants, which translates to about 0.65 M cryoprotectant, and 1.6% (w/v) Me 2 SO. The maximum tolerated concentration of Me 2 SO by itself was 5% (v/v), which equals about 0.71 M, or 4.5% (w/v) Me 2 SO. Therefore, at 37°C, ESC respond to the total cryoprotectant concentration roughly equally on a molar basis, and not to the concentration of Me 2 SO or any other specific component. This suggests that the effects of cryoprotectants when tested as described are not chemically specific, even though in principle, metabolism of some of the cryoprotectants might be possible, and Me 2 SO might be able to react with -SH groups [58]. Non-specific cryoprotectant toxicity is a desirable standard for comparison between different cryoprotectant formulations and is believed to be the type of toxicity that is most relevant to organ cryopreservation by vitrification [20,26].
Me 2 SO is known to be a differentiating agent when applied under the right conditions [9,32], but we do not think differentiation of ESC modulated either CTR or cryopreservation injury. First, our mutant cells behave and appear like ordinary ESC, rather than as differentiated cells, and none of the mutations we identified are expected to reverse or induce differentiation. Second, it is not obvious that differentiating ESC would make them either more or less susceptible to cell death caused by CT or cryopreservation, and it is cell death that we selected against. Finally, mutants displaying resistance to M22 also resisted Me 2 SO and cryopreservation in Me 2 SO, but M22 and Me 2 SO would presumably not have the same differentiating effects.

Identified mutant cell lines may be convertible into resistant whole mouse strains
Mutated ESC have been grown into adult mice displaying the same phenotype as the mutant ESC [7]. This suggests that CTR may also be transmissible to whole animals derived from CT-resistant ESC. The cells, tissues, and organs of such animals might serve as excellent models for further efforts to ablate CT. Furthermore, if acceptably mutated large animal ESC can be developed into adult tissues that resist cryoprotectant toxicity and cryopreservation injury, it may be possible to create unlimited supplies of replacement organs bearing such mutations for human transplantation.

Resistant mutants can be used to identify extant drugs with the same function
Our mutants' utility for normal cryopreservation lies in the possibility of using drugs to mimic the effects of these mutations. Given MYC's relationship to cancer, many extant strategies for inhibiting MYC to inhibit cancer may be repurposed to inhibit CT and cryopreservation injury. If a conventional drug implicated by a given mutant is not available, other strategies such as microRNA or CRISPR-Cas9 technology might be used to create novel drugs with the same therapeutic effects. Therefore, our novel fundamental discovery that inhibiting gene expression can inhibit CT and cryopreservation injury has implications for the cryopreservation of many cell types, tissues, and even whole organs.
4.6. The advantage of forward genetics for elucidating and countering mechanisms of CT Cryoprotectants interact with literally every molecule in the cell, and in the case of organs, with every cell type in the organ. This makes it difficult to elucidate mechanisms of CT and find remedies for it. There are now two microarray studies of the transcriptional responses of organ components to cryoprotectant exposure [10,37], which appropriately apply a global method of investigation to understanding the global influence of cryoprotectants. However, the mechanistic and therapeutic inferences that can be made from such studies are limited. Invariably, many pathways are found to be up-or down-regulated, and it can be difficult to determine whether specific pathway changes are relevant to the essence of CT or are merely secondary effects. It is also difficult to know in many cases whether observed changes in gene expression are therapeutic (compensatory) or pathologic. For example, as mentioned above, one transcriptomic study [ [3,37]] found up-regulation of MYC after CT, which was interpreted as a possible compensatory response, but in the present study, CTR was associated with a mutation that nominally would be expected to reduce rather than to increase MYC expression. A priori, it is impossible to know whether activation of MYC is protective or detrimental, but since we know our mutant is protective, we can answer the question of benefit vs harm directly, simply by determining in a follow-up study whether MYC expression is up-or down-regulated in our MYC-related mutants. In principle, a finding of a change in MYC expression in transcriptomic studies can be followed up with drug studies to see whether a drug that opposes the observed change in transcription is beneficial or detrimental, but in transcriptomic studies, there may be 100 candidate genes of uncertain relevance to so investigate [10,37], rather than just one gene of certain relevance for any given mutant discovered using forward genetics.
In summary, our results open up new opportunities for improving both freeze-thaw cellular viability and the viability of multicellular systems after cryopreservation by vitrification. By challenging cells under the extreme but experimentally convenient conditions of high temperature and long exposure time, we were able to identify protective mutations providing in some cases profound improvements in survival under conditions that are actually of interest to the cryobiologist. It will be interesting to see whether future studies uncover additional protective effects and to see whether combining CTR mutations and/or equivalent drugs can augment protection.