Looking at the cell's passive response in real time has enabled us to observe previously unseen shape changes. Immediately upon osmotic shock, water efflux causes a reduction of the cell and cytoplasmic radii but little change in the lengths. We find that the rate of volume reduction is ~10% per second and that during this initial shrinking the inner membrane is not detached from the cell wall. This phase of volume reduction lasts for ~5 seconds and the E.coli
Aquaporin Z protein is not required for fast water efflux. Immediately after shrinking, we observe an increase in the cell radius and attribute this to diffusion of sucrose into the periplasm of the cell. As the total cell radius increases, cytoplasmic cell length decreases such that the inner membrane detaches from the cell wall mostly at the cell poles. The final result is a typical plasmolyzed cell as seen in electron microscopy experiments 
. As with the fast phase of shrinking, AqpZ played no significant role in the rate of volume reduction on longer time scales. The only difference we found between cells with and without AqpZ was the value of Vmin
upon hyperosmotic shock for the cells grown to late stationary phase. We find that AqpZ+
cells shrink less when compared to the AqpZ−
strain. It is currently not clear why this might be and further experiments are needed to fully address the role of AqpZ, if any, in the passive flux of water through the E. coli
In rich media, volume adaptation upon hyperosmotic shock shows characteristic phases that depend on the magnitude of the shock. On the fast time scales, lasting ~20 minutes, recovery is likely due to the import of osmolytes by transporters that are constitutively expressed during normal growth. Larger shocks require an even larger accumulation of solutes into the cytoplasm. This happens during a slower phase of recovery and requires protein synthesis. We hypothesize that the typical level of transporters that are present during normal growth are capable of regulating the turgor pressure during small changes to the external osmolarity, shocks of magnitude less than ~0.6 Osmol/kg, that may normally be encountered in the environment. Larger perturbations, which are encountered far more infrequently in the wild, require the synthesis and action of secondary recovery pathways that are only used in extreme circumstances.
In addition, protein synthesis-driven mechanism seems to switch the fast recovery phase off while at the same time inducing the slower phase. This is most readily seen in , where the volume plateaus in time after ~20 minutes without chloramphenicol, but keeps increasing for another ~15 minutes when protein synthesis is inhibited. Previous research suggests that the increase in cytoplasmic potassium concentration during the fast recovery phase triggers the synthesis of potassium efflux channels. Efflux of potassium is then observed to accompany trehalose synthesis such that the cell exchanges one osmoprotectant for another 
. It is reasonable to predict, therefore, that the slower phase of volume recovery we observe corresponds to the export of potassium and the accumulation of trehalose.
A number of differences exist between the experiments referenced above and our experimental protocol, which must also be considered. First, trehalose accumulation is observed on the order of 30 min after shock, while the slow recovery we observe lasts several hours 
. Second, the growth conditions and shocking media, including media osmolarity and potassium concentrations, varied between the different protocols. Third, a Kdp- strain was used in that work, which at the osmolality and potassium concentrations they probed was indistinguishable from the wild type. In LB media, this is not the case and we expect protein-synthesis to be responsible for the production of nascent transporters, such as the Kdp potassium pump 
. Furthermore, the secondary transporter ProP is both constitutively expressed and expressed under conditions of osmotic stress as a result of increased internal potassium concentrations 
. Induction of the proU
gene is fast and has been observed within ~4 min after osmotic shock 
. Future experiments using sucrose and LB media that monitor the cellular concentrations of potassium and trehalose at the single cell level will be needed to further connect these aspects of the recovery and address possible redundancies among the different recovery pathways.
For shocks in the range of 0.50 to 1.45 Osmol/kg the slower recovery phase enables the cells to reach their initial volume, but it is followed by a long pause. Unshocked cells and cells shocked with a small amount of sucrose continue to grow during this same length of time in our experimental chambers indicating that the pause at the end of the slow rate recovery is unlikely due to nutrient limitation. In liquid culture at room temperature and with no shaking, these cells will eventually grow to high density, usually within a day. Therefore at present we do not understand the nature of this pause. It is possible that growth resumes at a later point and that only very few cells continue to grow after larger shocks. Future experiments that will monitor the cell volume changes on even longer time scales can help us understand the observed pause.
For very large sucrose shocks, above 2.5 Osmol/kg, no recovery is observed. These large shocks produce a free-energy barrier to osmoprotectant accumulation during the initial recovery rate. We estimate this maximum energy, ΔGmax
, for potassium pumping using the known external, ~8 mM, and internal, ~250 mM, concentrations of potassium for cells grown to exponential phase in LB 
. A hyperosmotic shock of ~2.5 Osmol/kg decreases the cytoplasmic volume by about a factor of two, so that the concentration of potassium in non-recovering, shocked cells is estimated to be ~500 mM. To include the electrostatic energy-contribution for transport of potassium across the inner membrane, we assume the membrane voltage is on the order of 150 mV 
. Using these values, we calculate a maximum energy of ΔGmax
~25 kJ mol−1
, of the same order of magnitude as the maximum energy output of light-powered proteorhodopsin and other cellular pumping systems 
Lastly, we comment that most of the previous research on osmoregulation in E. coli
used the addition of sodium chloride to produce an external hyperomostic shock 
. We chose sucrose because E.coli
membrane permeability to it is understood. A previous report found drastic differences in potassium uptake depending on the shocking agent 
. These results are unsurprising. Charged solutes influence the overall energetic state of the cell, including ΔGmax
for a particular pump in a given condition. Consequently the sequence of events during the initial stage of osmorecovery can be altered. Methods to monitor E.coli
cells membrane voltage on a single cell level and in real time with high temporal resolution have been recently reported 
. Simultaneous measurements of cell volume changes and changes in cell energetic state promise to bring additional answers on cells' overall stress response.
High-resolution measurements of cell volume changes upon hyperosmotic shock have given us new insight into the dynamics of cytoplasmic water efflux and cell volume recovery. In a similar manner we hope to extend our measurements to fully understand and answer open questions regarding the sequence of events involved in osmoregulation and other stress responses.