Response of human cells to desiccation: comparison with hyperosmotic stress response


Corresponding author A. Tunnacliffe: Institute of Biotechnology University of Cambridge, Tennis Court Road, Cambridge CB2 1QT, UK. Email:


Increasing interest in anhydrobiosis (‘life without water’) has prompted the use of mammalian cells as a model in which candidate adaptations suspected of conferring desiccation tolerance can be tested. Despite this, there is no information on whether mammalian cells are able to sense and respond to desiccation. We have therefore examined the effect of desiccation on stress signalling pathways and on genes which are proposed to be expressed in response to water loss through osmotic stress. Depending on the severity of the drying regime, human cells survived for at least 24 h. Both SAPK/JNK and p38 mitogen-activated protein kinases (MAPKs) were activated within 30 min by desiccation as well as by all osmotica tested, and therefore MAPK pathways probably play an important role in both responses. Gene induction profiles differed under the two stress conditions, however: quantitative polymerase chain reaction (PCR) experiments showed that AR, BGT-1 and SMIT, which encode proteins governing organic osmolyte accumulation, were induced by hypersalinity but not by desiccation. This was surprising, since these genes have been proposed to be regulated by ionic strength and cell volume, both of which should be significantly affected in drying cells. Further investigation demonstrated that AR, BGT-1 and SMIT expression was dependent on the nature of the osmolyte. This suggests that their regulation involves factors other than intracellular ionic strength and cell volume changes, consistent with the lack of induction by desiccation. Our results show for the first time that human cells react rapidly to desiccation by MAPK activation, and that the response partially overlaps with that to hyperosmotic stress.

Anhydrobiosis is the state of suspended animation which certain organisms enter in response to desiccation (Crowe et al. 1992; Clegg, 2001); such organisms are found among a range of taxa, including bacteria, yeasts, plants and invertebrates (Keilin, 1959). Survival of desiccation and the dry state is one of the most intriguing phenomena in nature, but is far from fully understood (Tunnacliffe & Lapinski, 2003). The most well-developed hypotheses attempting to explain anhydrobiosis invoke the non-reducing disaccharides, trehalose and sucrose, one or other of which is present in high concentrations during desiccation of a number of anhydrobiotic organisms. However, non-reducing disaccharides are not an absolute requirement for anhydrobiosis, being absent from bdelloid rotifers, for example (Lapinski & Tunnacliffe, 2003). Other molecules are also likely to contribute significantly to desiccation tolerance, e.g. small heat shock proteins in brine shrimp cysts (Liang et al. 1997), extracellular polysaccharides and water stress proteins in the cyanobacterium Nostoc commune (Potts, 1994, 1999; Helm et al. 2000), or amphiphilic molecules in plant embryos (Hoekstra et al. 2001). Late embryogenesis-abundant (LEA) proteins, accumulated during water deficit, might also play a role in drought and desiccation tolerance of plants (Ingram & Bartels, 1996), anhydrobiotic nematodes (Solomon et al. 2000; Browne et al. 2002; Goyal et al. 2003) and bacteria (Battista et al. 2001). However, as argued previously (Tunnacliffe & Lapinski, 2003), there is little proof in living systems of the importance of any of these groups of molecules in anhydrobiosis, e.g. by mutagenesis or functional knockdown of the relevant genes.

Another strategy for defining the key adaptations for anhydrobiosis would be to use the candidate molecules identified in anhydrobiotic organisms to attempt to confer desiccation tolerance on an otherwise sensitive cell or organism. In this way, a minimum number of adaptations required for anhydrobiosis could be identified, at least for a particular cell type. We and others have begun to take this approach with mammalian cell cultures, initially by testing the role of trehalose. These studies have proved to be controversial (Guo et al. 2000; García de Castro et al. 2000; García de Castro & Tunnacliffe, 2000), but one interpretation is that, although intracellular trehalose can improve freeze tolerance (Eroglu et al. 2000) and osmotolerance (García de Castro & Tunnacliffe, 2000) in replication-competent mammalian cells, it does not confer desiccation tolerance, although membrane integrity is improved (Chen et al. 2001). This represents a major step forward in our understanding of anhydrobiosis, since it seems clear that intracellular trehalose is not sufficient for desiccation tolerance, as previously proposed (Crowe & Crowe, 1992).

One outcome of this work has been the realization that essentially nothing is known about the response of mammalian cells to desiccation. This is perhaps not surprising, since much less severe dehydration is lethal to the organism as a whole; individual cells are not expected to withstand almost complete water loss in the way that anhydrobiotic organisms do. However, not only are mammalian cells useful for our attempts to understand anhydrobiosis through definition of an anhydrobiotic gene set, but the ability to dry cells in a viable form would have many applications in medicine and research, e.g. for tissue engineering, biosensors or cell banks (Bloom et al. 2001; Tunnacliffe et al. 2001). Furthermore, how desiccation affects the well-defined stress signalling and response pathways in mammalian cells could provide insight into analogous mechanisms in less well-characterized anhydrobiotic organisms. Finally, differences between the responses to desiccation in non-desiccation-tolerant cells and anhydrobiotic organisms should highlight important physiological adaptations associated with desiccation tolerance. Given the value of mammalian cells as a model, therefore, it is important to understand the degree to which they sense and respond to impending desiccation.

Although there is currently no information on the desiccation response, there is a large literature on the partial dehydration of mammalian cells due to osmotic upshift (‘hyperosmolarity’ or ‘hypertonicity’). Despite some fundamental differences between desiccation and hypertonicity, they both lead to efflux of intracellular water and consequent concentration of inorganic salts, crowding of macromolecules, and potential damage to macromolecular structure and function, as well as mechanical damage due to cell shrinkage and distortion (Potts, 1994). These similarities suggest that both types of dehydration stress might induce similar responses in terms of signal transduction and gene regulation. Therefore, an initial examination of the desiccation stress response in human tissue culture cells should include signalling pathways and genes which are activated in the hyperosmotic response, such as kinase cascades and genes governing compatible solute accumulation, respectively.

Well-conserved and -characterized stress signal transduction routes involve mitogen-activated protein kinases (MAPKs) which modulate gene expression and cell metabolism during growth and development, apoptosis and environmental stress responses (Kültz, 1998; Kültz & Burg, 1998; Chang & Karin, 2001; Cowan & Storey, 2003). Phylogenetic, structural and functional analysis has identified three main MAPK families: extracellular signal-regulated kinases (ERKs), c-Jun N-terminal kinases (SAPK/JNKs) and the p38 group of kinases (Kültz, 1998). It is well established that MAPKs, particularly from the SAPK/JNK and p38 families, can be activated by hypertonicity and elicit a variety of cellular responses in different organisms and cell cultures (Galcheva-Gargova et al. 1994; Han et al. 1994; Cohen, 1997; Kültz & Burg, 1998; Sheikh-Hamad et al. 1998; Häussinger & Schliess, 1999; Zhu, 2002; Bildin et al. 2003), and it is possible, therefore, that they are similarly activated by desiccation.

Cellular accumulation of organic osmolytes (‘compatible solutes’) occurs in virtually all hypertonically stressed organisms examined (Yancey et al. 1982; Burg et al. 1997). Compatible osmolytes help counteract raised extracellular osmolarity and reduce intracellular ionic strength to normal levels; because they generally do not perturb macromolecular structure, they can accumulate in large amounts (Burg et al. 1997; Brigotti et al. 2003). In mammalian cells, too, compatible solutes such as sorbitol, myo-inositol and glycine betaine are accumulated in response to hypertonicity. Sorbitol is synthesized from glucose by aldose reductase; inositol and betaine are taken up from medium by relevant transporters. The genes encoding these enzymes and transporters, i.e. AR (aldose reductase; Smardo et al. 1992), BGT-1 (betaine–γ-amino-n-butyric acid (GABA) transporter; Uchida et al. 1993), and SMIT (sodium-dependent myo-inositol transporter; Yamauchi et al. 1993), are regulated by cell tonicity and have been proposed to respond specifically to intracellular ionic strength and cell volume changes (Burg et al. 1997; Neuhofer et al. 2002a).

We have therefore compared the effect of desiccation and hyperosmolarity, imposed by various osmotically active agents, on some known MAPKs and tonicity-responsive genes in human cells. Our results show that both SAPK/JNK and p38 MAPKs are activated by desiccation, as well as by all osmotica tested, indicating that desiccation is efficiently sensed by, and elicits a response in, human cells. In contrast, several genes which are purportedly up-regulated in response to elevated osmolarity were shown, in fact, to respond in an osmolyte-specific manner and therefore are unlikely to be induced during desiccation.


Cell culture

Human embryonic kidney cell line T-REx 293 (Invitrogen, Paisley, UK) was chosen since it can be used with an inducible expression system for subsequent experiments involving candidate desiccation protection genes; it is derived from HEK 293, whose response to osmotic upshift has been studied previously (e.g. Ferraris et al. 2002). Cells were routinely maintained for up to 50 passages at 37°C in an atmosphere of 5% CO2 and 95% air in 75 cm2 tissue culture flasks (Nunc, Roskilde, Denmark) containing 25 ml of high glucose (4500 mg l−1) Dulbecco's modified Eagle's medium (DMEM; Sigma, Poole, UK) supplemented with 10% fetal bovine serum (FBS; TCS Biosciences Ltd, Botolph Claydon, Buckingham, UK), 2 mml-glutamine, 100 units ml−1 penicillin, 100 μg ml−1 streptomycin (all Sigma) and 5 μg ml−1 blasticidin (Invitrogen). For real-time PCR samples, cells were grown either in the same way or in 1 ml of medium in multidishes (1.9 cm2 per well, flat bottom; Nunc). Cells for Western blot analysis were grown in 3 ml medium in multidishes (9.6 cm2 per well, flat bottom; Nunc). All treatments were performed on the same number of cells in each experiment.

Hypertonic shock

To determine suitable NaCl concentration and application method to cells for gene quantification, cells were grown to near confluence. The medium was then replaced with fresh medium (as a control), medium containing 100 or 150 mm NaCl (one-step shock), or medium with progressive NaCl increase to 150 mm (multistep increase of NaCl at 1.5 h intervals to 10, 20, 40, 60, 80, 100, 120 and finally 150 mm), and the cultures were incubated at 37°C for a further 18 h. For the time course of NaCl shock, near confluent cells were treated with three-step 150 mm NaCl (increasing NaCl at 2 h interval to 50, 100 and finally 150 mm) and the cultures were incubated at 37°C for the indicated times after the final increase of NaCl. Based on the NaCl shock results, the other salts, sugars and sugar alcohols were applied at similar final osmolarity in the same fashion of three-step shock: 50, 100 and 150 mm for KCl, NaNO3 and sodium acetate (NaOAc); 100, 200 and 300 mm for trehalose, sucrose, sorbitol and mannitol. All hypertonic samples for Western blot analysis were one-step shock for the indicated times using the respective final concentration of the above three-step treatments. The resulting final osmolarities (mosmol l−1) of the above solutions in medium, as measured with a 13/13DR-Autocal freezing-point osmometer (Roebling, Berlin, Germany), were: medium, 294; 150 mm NaCl, 561; 150 mm KCl, 561; 150 mm NaNO3, 573; 150 mm NaOAc, 581; 300 mm trehalose, 615; 300 mm sucrose, 632; 300 mm sorbitol, 599; 300 mm mannitol, 608 (Sigma) or 598 (Fisher).


Near confluent cells in multidishes were either washed with PBS or DMEM as indicated or not washed. After liquid was carefully but completely removed by aspiration, the air-vent multidishes were placed in humidity chambers and dried for the indicated times. The humidity in the chambers was maintained by saturated BaCl2 (90% relative humidity; RH), saturated CuSO4 (98% RH) or sterile water (100% RH) in a Stewart air-tight polypropylene box (in mm: 162 × 176 × 100; w × d × h) at room temperature or 37°C as indicated. RH was confirmed at room temperature using an Oakton Thermohygrometer (Cole-Parmer Instrument Co, Chicago, IL, USA).

Cell viability test

Viable cells were determined using CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA) according to the manufacturer's instructions. Cells (100 μl well−1) were passaged into 96-well plates (Nunc) and incubated at 37°C until near confluence (about 2 days). The cells were dried at 37°C as above in a humidity chamber for specific times and rehydrated for 2 h in 100 μl of medium at 37°C. After the addition of 20 μl of MTS tetrazolium reagent, the plate was incubated at 37°C and 5% CO2 for 3 h. Absorbance was measured at 490 nm using a 96-well plate reader and corrected by subtraction of background absorbance (same amount of medium without cells).

Western blot analysis

Anti-SAPK/JNK (rabbit); anti-SAPK/JNK, phospho-specific (Thr183,Tyr185), human (rabbit); anti-p38 MAPK human (rabbit); and anti-p38 MAPK, phospho-specific (Thr180,Tyr182), human (rabbit) antibodies were from Calbiochem (La Jolla, CA, USA). PhosphoPlus SAPK/JNK (Thr183/Tyr185) antibody kit, containing both phospho- and non-phospho-SAPK/JNK antibodies, was purchased from Cell Signalling Technology Inc. (Beverly, MA, USA).

Protein samples were prepared by direct lysis of cells in the 9.6 cm2 wells with equal numbers of starting cells. After complete removal of medium by aspiration or at the time points of desiccation, 100 μl of SDS sample buffer was added directly to the cell monolayer and cells were immediately scraped and transferred to a cold microfuge tube on ice. The cell lysates were then sonicated, boiled for 5 min and centrifuged at 4°C for 15 min at 15 000 g. Equal volumes of the supernatants were used for immunoblotting.

Proteins were separated on 10% SDS-PAGE gel (10 × 10 cm) and transferred to nitrocellulose membrane (0.45 μm, pore size; Bio-Rad, Hercules, CA, USA) using a Trans-Blot SD Semi-Dry Electrophoretic Transfer Cell (Bio-Rad). The membrane was washed with Tris-buffered saline (TBS; pH 7.6) and blocked with 5% skimmed milk in TBS/T (0.1% Tween-20 in TBS) at room temperature for 1 h. The membrane was washed with TBS/T (3 × 5 min) and incubated overnight at 4°C with the above primary antibody at 1: 1000 dilution with 5% bovine serum albumin (BSA) in TBS/T. After washing with TBS/T (3 × 5 min), the membrane was incubated at room temperature for 1 h with horseradish peroxidase-conjugated secondary antibody (1: 1000 dilution with 1% skimmed milk in TBS/T). After washing the membrane with TBS/T (3 × 5 min), antibody binding was detected using either the ECL Western Blotting Analysis System (Amersham Biosciences UK Ltd, UK) or Phototope-HRP Western Detection Kit (Cell Signalling Technology Inc.). After further washing with TBS/T (4 × 5 min), antibody was removed from the membrane by incubation in stripping buffer (63 mm Tris, 2% SDS and 0.7% 2-mercaptoethanol) at 50°C for 30 min. After washing again with TBS/T (10 × 5 min), the membrane was reprobed as above for another antibody. Equal loading on SDS-PAGE was demonstrated by staining gels with Coomassie Brilliant Blue after transfer, and by staining Western blots with Ponceau S.

Relative quantification of gene expression

RNA was prepared using Cells-to-cDNA II kit (Ambion Europe Ltd, Huntingdon, UK) following the instruction manual. After complete removal of medium by aspiration from the 1.9 cm2 wells or at the time points of desiccation, 100 μl of ice-cold Cell Lysis II Buffer was directly added to the cell layer. The cells were immediately scraped and pipetted quickly to a cold 1.5 ml microfuge tube in ice. The lysates were processed according to instructions until just before the reverse transcription step, i.e. 75°C for 10 min, DNase I digestion at 37°C for 15 min and 75°C for 5 min. The RNA preparations were stored at –20°C for a short period. Reverse transcription was also performed using the kit according to the instruction with random decamers provided and 5 μl of cell lysate (RNA). Resultant cDNA was diluted five times with water and stored at –20°C.

Relative quantification of cDNA by real-time PCR was performed using a Rotor-Gene Real-Time Cycler 2000 (Corbett Research, Sydney, Australia) and QuantiTect SYBR Green PCR kit (Qiagen, Hilden, Germany). The critical threshold values were used to calculate the relative amounts of cDNA according to the delta–delta method (Pfaffl, 2001). β-actin and GAPDH (glyceraldehyde 3-phosphate dehydrogenase) were used as reference gene transcripts. Target genes were AR, BGT-1, SMIT and HSP70.1. Whenever appropriate the primers used spanned exon–exon junctions to avoid potential genomic DNA contamination (although RNA preparations were digested with DNase I). Primers were chosen so that the size of resulting amplicons was 50–150 bp (see Table 1). Specific amplification of transcripts was verified by gel electrophoresis, where only one DNA fragment was observed, and by melt curve analysis of the real-time PCR products, where a single peak was seen.

Table 1.  Primers used for real-time PCR

Gene name

size (bp)

  1. F: forward primer; R: reverse primer.

β-ActinF: CCTGGCACCCAGCACAAT144Xu et al. (2000)
GAPDHF: TGCACCACCAACTGCTTAGC 87Vandesompele et al. (2002)
ARF: ATCGCAGCCAAGCACAATAA105Ferraris et al. (2002)
BGT-1F: TGTTCAGCTCCTTCACTTCTGA 68Ferraris et al. (2002)


Cell survival during desiccation

The viability of T-REx 293 cells dried at 98% RH and 37°C for various times was assessed through the effect of intracellular dehydrogenases on a MTS tetrazolium compound. Relative cell survival using this assay was expressed as the percentage of absorbance of dehydrated cells relative to that of fully hydrated control cells. As shown in Fig. 1, about 50–60% of control activity remained after drying for 12 h, which decreased to about 30% after drying for 24 h. This relative activity fell to about 4% after drying for 48 h; staining with Trypan Blue gave similar results (data not shown). Live cells could be grown readily from populations dried for at least 12 h at 98% RH, and some cells could still be grown after up to 24 h. Incubation at 90% RH was a more severe stress, with few live cells remaining after 8–10 h drying.

Figure 1.

Survival of T-REx 293 cells at 98% relative humidity
T-REx 293 cells in 96-well plates (8 wells for each treatment) were dried at 98% RH for the indicated times and rehydrated for 2 h at 37°C prior to assay using MTS tetrazolium compound (Promega CellTiter 96 AQueous One Solution Cell Proliferation Assay).

Activation of MAPKs by desiccation

Using antibodies able to recognize either the activated, phosphorylated forms of MAPKs specifically, or both phosphorylated and unphosphorylated forms of the proteins, Western blots were performed on protein extracts of cells dried for various times. Phosphorylation of both SAPK/JNK (p46 and p54 isoforms) and p38 MAPKs was induced by incubation at 98% RH, and lasted up to 24 h (Fig. 2A), by which time point the majority of cells had died. Similar results were obtained whether the incubation was performed at room temperature or at 37°C, either with or without a PBS wash prior to drying. Phosphorylation was rapid, and was apparent as early as 30 min after transfer to the desiccation chamber (Fig. 2B).

Figure 2.

Time course of SAPK/JNK and p38 MAPK phosphorylation by desiccation
After complete removal of medium, T-REx 293 cells in multiwell plates were dried at 98% RH at room temperature for the indicated times and lysed directly in multiwell plates. Similar phosphorylation was also obtained by desiccation at 37°C but not by temperature shift alone (in the medium, from 37°C to room temperature for 4 h).

For comparison, T-REx 293 cells were subjected to osmotic upshift by one-step addition of 150 mm NaCl, and protein extracts were again analysed by Western blotting with MAPK antibodies. Figure 3A shows that phosphorylation of both SAPK/JNK and p38 was induced within 30 min, although the phosphorylation of SAPK/JNK became weaker after 4 h treatment. Other osmotica, including other sodium salts, KCl and non-ionic osmolytes such as trehalose, sucrose and sorbitol, produced similar degrees of phosphorylation of SAPK/JNK and p38 at comparable osmolarities (1 h time point; Fig. 3B). Although a degree of phosphorylation of p38 MAPK was noted in the control in this experiment (other authors have made similar observations, e.g. De Smaele et al. 2001), phosphorylation was increased after osmotic upshift or desiccation (Figs 2 and 3). It was also noticeable that KCl treatment resulted in lower levels of extractable protein, and consequently Western blot signals were reduced compared to other lanes.

Figure 3.

Time course of effect of NaCl (A), and effect of different salts, sugars and sugar alcohols (B), on SAPK/JNK and p38 MAPK phosphorylation
A, T-REx 293 cells were treated with 150 mm NaCl (one-step addition) at 37°C for the indicated times and directly lysed in multiwell plates. B, T-REx 293 cells were treated with approximately equal osmolarities of the indicated osmolytes at 37°C for 1 h and lysed directly in multiwell plates.

In summary, it seems clear that impending desiccation is detected very rapidly by human cells, on a comparable time scale to detection of osmotic stress, and that cells respond by activating MAPK signalling pathways. Since activation of these signalling cascades results in gene induction in response to other stresses, we next examined mRNA levels of several genes which are implicated in the response to hypertonicity, and which might therefore be expected to respond to desiccation.

Tonicity-responsive genes are not induced by desiccation

The activities of three genes (AR, BGT-1 and SMIT) which are regulated by hypertonicity were examined in drying cells by quantitative PCR. The heat shock protein gene HSP70.1, which has been reported to be hypertonically induced by some authors (Shim et al. 2002; Cai et al. 2004), but not by others (Woo et al. 2002), was also included. Strikingly, none of the tonicity-responsive genes was induced under any of the drying conditions attempted (90%, 98% and 100% RH) at either room temperature or 37°C. A typical time course of relative transcript levels after incubation at 98% RH is shown in Fig. 4. There was some variability in relative transcript quantification with time, but this was within 0.5- to 2.0-fold of the control and therefore not considered significant. The slight difference in profile obtained when referenced to either β-actin or GAPDH mRNA levels is probably due to variable expression of the reference genes themselves, rather than the target genes.

Figure 4.

Time course of the effect of desiccation on tonicity-responsive gene expression
T-REx 293 cells were exposed to 98% RH at 37°C. Relative gene quantification was normalized to β-actin (A) and GAPDH (B). Similar results were also obtained when cells were exposed to 90% or 100% RH at either 37°C or room temperature.

To ensure that these genes behave appropriately in the T-REx 293 line, cells were subjected to osmotic upshift. Since the rate of increase of osmotic stress, as well as final osmolyte concentration, can play a role in the expression of tonicity-responsive genes (Leroy et al. 2000; Cai et al. 2002, 2004), transcripts were quantified in cells undergoing both acute and progressive increases in NaCl concentration. For AR, SMIT and BGT-1, expression levels were increased under all conditions tested – with the possible exception of AR after exposure to 100 mm NaCl – although multistep progressive increase of NaCl concentration to 150 mm was the most effective (Fig. 5). HSP70.1 showed only marginal induction (∼2-fold increase) by either one-step 100 mm or 150 mm multistep NaCl shock and no induction by one-step 150 mm shock (Fig. 5). Clearly, AR, SMIT and BGT-1 respond to hypertonicity in T-REx 293 cells and it is appropriate to expect that, if they are induced by elevated intracellular ionic strength and cell volume changes, desiccation would also up-regulate their expression.

Figure 5.

Effect of NaCl concentration and increase rate on tonicity-responsive gene expression
T-REx 293 cells were treated as indicated: 100 mm and 150 mm, 18 h at 37°C after one-step addition of NaCl; 150 mm*, 18 h at 37°C after 8-step increase at 1.5 h intervals to final 150 mm NaCl in medium. Relative gene quantification was normalized to β-actin; it was similar when referenced to GAPDH.

Factors which might account for a lack of increase in expression of these genes in T-REx 293 cells undergoing desiccation include (a) whether sufficient time is available for effective gene induction, and (b) the nature of the agent governing gene induction; these are investigated below.

Time course of gene induction by NaCl hypertonicity

Gene expression levels were followed for 24 h after cells were treated with the three-step 150 mm NaCl protocol described above. In response to hypersaline stress, SMIT mRNA levels were already increased after 4 h, and AR and BGT-1 by 12 h, but all reached maximum levels at 16–20 h (Fig. 6), in general agreement with previous work (e.g. Galvez et al. 2003). With β-actin as reference, BGT-1 mRNA increased 60- to 100-fold, SMIT 15- to 20-fold and AR 5- to 10-fold; the same treatment had no significant effect over the same time period on HSP70.1. Thus, although AR, SMIT and BGT-1 are maximally expressed 16–20 h after the osmotic stress was imposed, significant expression took place at earlier time points, suggesting that sufficient time is available during desiccation for gene induction.

Figure 6.

Induction time course of tonicity-responsive genes by 150 mm NaCl
T-REx 293 cells were treated for the indicated times at 37°C after 3-step increase of NaCl, at 2 h intervals, to a final 150 mm in medium. Relative gene quantification was normalized to β-actin.

Effect of different salts and non-ionizing osmolytes on tonicity-responsive genes

Gene induction by hypersalinity could be due to hyperosmotic and/or ionic stress. To differentiate between these, the effect of different salts, sugars and sugar alcohols on gene induction was determined (Fig. 7). Three different sodium salts (NaCl, NaNO3 and sodium acetate), plus KCl, were used as ionizing osmotica, and trehalose, sucrose, sorbitol and mannitol were used as non-ionizing osmolytes; all were used at comparable osmolarities and shock rates. All Na+ treatments up-regulated AR, SMIT and BGT-1, but KCl had no effect, suggesting that these genes respond primarily to an increase in sodium ion concentration, rather than hypertonicity or increased ionic strength per se. Interestingly, mannitol also had a significant effect on these genes but other non-ionizing osmolytes used, namely trehalose, sucrose and sorbitol, were ineffective. None of the hypertonic treatments induced HSP70.1. In summary, gene induction of AR, SMIT and BGT-1 seems to be dependent on the nature of the osmolyte, being induced by Na+ and mannitol, but not by K+, Cl or the non-ionizing osmolytes, trehalose, sucrose or sorbitol. Therefore, the regulation of these genes seems to involve factors other than overall intracellular ionic strength and cell volume changes. This is consistent with the lack of induction during desiccation, although we cannot exclude the possibility that some solutes have additional effects which counteract the influence of ionic strength and cell volume changes.

Figure 7.

Effect of different salts and non-ionizing osmolytes on tonicity-responsive gene expression
T-REx 293 cells were treated for 18 h at 37°C after 3-step increase, at 2 h intervals, of monovalent salts to final 150 mm and non-ionizing osmolytes to final 300 mm in medium. Relative gene quantification was normalized to β-actin; it was similar when referenced to GAPDH.


Although both SAPK/JNK and p38 MAPK pathways have been implicated in a number of stress responses, including that to hyperosmotic stress (Galcheva-Gargova et al. 1994; Han et al. 1994; Sheikh-Hamad et al. 1998; Häussinger & Schliess, 1999; Bildin et al. 2003; Uhlik et al. 2003), there is no information on the response to desiccation in any animal cell. We have shown for the first time that water loss through drying can be sensed and responded to by increased phosphorylation of both SAPK/JNK and p38 kinases in human cells. In plants, drought stress, which is arguably a similar stress vector to desiccation, is also known to induce MAPK activation (Jonak et al. 1996; Xiong & Yang, 2003). We consider it likely therefore that fully desiccation-tolerant organisms will also use MAPK pathways to initiate defence strategies against desiccation damage and a search for downstream targets for these kinases in anhydrobiotes should improve our understanding of the phenomenon. A similar strategy has recently been adopted in the study of freeze tolerance in vertebrates (Cowan & Storey, 2003).

Interestingly, although the p38 MAPK pathway is crucial for cell survival under various types of stress including hyperosmotic shock (Uhlik et al. 2003), sustained SAPK/JNK activation may induce cell death in response to environmental stress and tumour necrosis factor (TNF; Kyriakis, 2001; Franzoso et al. 2003). This is consistent with the data of Figs 2 and 3: under hypertonic stress which cells were able to tolerate, SAPK/JNK phosphorylation became weaker after 4 h (Fig. 3), but the ultimately lethal effect of desiccation was associated with longer term SAPK/JNK activation (Fig. 2). The real situation may be much more complicated, however, since both SAPK/JNK (Lamb et al. 2003) and p38 MAPK (Deacon et al. 2003) are capable of signalling cell survival as well as apoptosis under different conditions. Nevertheless, it is conceivable that manipulation of the stress response through induction of pro-survival pathways (e.g. via nuclear factor (NF)-κB) and suppression of pro-apoptotic pathways (e.g. via SAPK/JNK; Bennett et al. 2001; De Smaele et al. 2001; Kyriakis, 2001; Tang et al. 2001; Franzoso et al. 2003) might lead to increased desiccation tolerance of sensitive mammalian cells.

Since both desiccation and hypertonicity cause cell dehydration and may have similar consequences, it is instructive to compare the respective stress responses. There are no studies in the literature which directly compare these stresses in animal or lower eukaryotic cells, but it was recently shown that drought- and salinity-stressed plants exhibit some overlap in gene expression patterns, as well as clear differences (Liu & Baird, 2003). In the present study, we showed that three human genes induced by NaCl hypertonicity (Figs 5–7), i.e. AR, BGT-1 and SMIT, were not induced by desiccation (Fig. 4). Since these genes are reported to be induced by elevated intracellular ionic strength as well as by additional factors such as cell volume due to osmotic efflux of water (e.g. Neuhofer et al. 2002a), and since ion concentrations and cell volume in drying cells should be similarly affected due to evaporation of water, this was a surprising result. One reason for this might be that there is insufficient response time during desiccation before cell metabolism is compromised, since all three genes are only maximally expressed after 16–20 h under hypertonic conditions. However, this argument is unconvincing, at least for cells exposed to 98% RH, when there was also no gene expression although many cells survived for between 12 h and 24 h (Fig. 1). Moreover, significant gene induction was observed under hypertonic conditions well before the time of maximal expression. Finally, we have recently obtained data showing that other genes (e.g. the transcription factor genes EGR1, EGR3 and SNAI1) are up-regulated by desiccation (Z. Huang & A. Tunnacliffe, manuscript in preparation), which demonstrates that the transcriptional machinery is able to respond. Therefore, if AR, BGT-1 and SMIT are truly regulated by intracellular ionic strength and cell volume, some increase in expression should have been observed in drying cells.

A more likely explanation for the lack of expression of tonicity-responsive genes in drying cells relates to the nature of the stress signal responsible for up-regulation of these genes. Three different Na+ salts (NaCl, sodium acetate and NaNO3) were able to induce expression of AR, BGT-1 and SMIT, but KCl was unable to do so at comparable osmolarity. It has been reported previously that, whereas NaCl hypertonicity clearly increased the mRNA abundance of these genes, KCl hypertonicity did so only modestly, despite a marked increase in intracellular ionic strength (Neuhofer et al. 2002b). Taken together with our results, this could indicate that hypertonicity alone does not seem to be important, since extracellular trehalose, sucrose and sorbitol at comparable osmolarities were all ineffective at inducing gene expression, although MAPK phosphorylation was increased on exposure to these osmolytes, indicating an effect of the stress. Intriguingly, however, we were also able to show that AR, BGT-1 and SMIT were induced by hypertonicity imposed by the sugar alcohol, mannitol. Differential effects of hypertonic levels of mannitol and sorbitol on AR induction have also been reported in rat cardiomyocytes, although these two epimers with identical osmotic properties elicited the same degree of cell shrinkage (Galvez et al. 2003). Mannitol has also been reported to induce AR in human retinal pigment epithelial cells (Stevens et al. 1993), but both sorbitol and mannitol up-regulate AR in L-929 cells (Libioulle et al. 1996). These discrepancies may be due to the different types of cells used, as argued elsewhere (Galvez et al. 2003), but it seems that the regulation of these genes is more complicated than is first apparent; this was also indirectly supported by the current data from drying cells. What can be concluded, however, is that water loss alone, whether through osmosis or evaporation, is not sufficient to up-regulate the AR, SMIT and BGT-1 genes in T-REx 293 cells.

Despite some similarities, there are fundamental differences between desiccation and hypertonicity (Potts, 1994). For instance, the initial response of cells under sudden hypertonic stress is a rapid uptake of inorganic salts, accompanied by osmotic influx of water; cell volume is adjusted through regulatory volume increase (RVI) (Waldegger & Lang, 1998; O'Neill, 1999; Bildin et al. 2003) to mitigate mechanical stress; and organic osmolytes are accumulated by synthesis or uptake from the medium (Burg et al. 1997). None of these actions is possible for cells undergoing desiccation; nevertheless, anhydrobiotic organisms are able to respond to desiccation by production of, for example, non-reducing disaccharides and hydrophilic proteins. How desiccation is sensed in order to effect these changes is completely unknown, but clues might be derived from mammalian cell models. For example, hyperosmotic shock activates the p38 kinase pathway through a scaffold protein (OSM) which binds actin, the GTPase Rac and the upstream kinases MKK3 (MAPKK) and MEKK3 (MAPKKK); osmotic stress promotes changes in the plasma membrane and underlying cytoskeleton which recruit the scaffold complex to actin structures and lead to p38 activation (Uhlik et al. 2003). Similar mechanisms might also sense desiccation in anhydrobiotic organisms and bring about the induction of genes governing the synthesis of disaccharides and hydrophilic proteins.



This work was supported by grant no. 8/C17391 from the Biotechnology and Biological Sciences Research Council, UK. A. T. is the Anglian Water Fellow in Biotechnology of Pembroke College, Cambridge.