Production of heat-shock proteins is induced when a living cell is exposed to a rise in temperature. The heat-shock response of protein DnaK synthesis in E.coli for temperature shifts T ! T 1 DT and T ! T 2 DT is measured as a function of the initial temperature T . We observe a reversed heat shock at low T. The magnitude of the shock increases when one increases the distance to the temperature T 0 ഠ 23 ± C, thereby mimicking the nonmonotonous stability of proteins at low temperature. This suggests that stability related to hot as well as cold unfolding of proteins is directly implemented in the biological control of protein folding.PACS numbers: 87.14. Ee, Chaperones direct protein folding in the living cell by binding to unfolded or misfolded proteins. The expression level of many of these catalysts of protein folding changes in response to environmental changes. In particular, when any living cell is exposed to a temperature shock the production of these evolutionary conserved proteins is transiently increased [1]. The heat-shock (HS) response in E.coli involves about 40 widely dispersed genes and is mediated through the s 32 protein [2,3]. The s 32 binds to RNA polymerase (RNAp), where it displaces the s 70 subunit and thereby changes RNAp's affinity to a number of promoters. This induces production of the heat-shock proteins. If the gene for s 32 is removed from the E.coli genome, the HS is suppressed [2,4] and also the cell cannot grow above 20 ± C.The HS is fast. In some cases it can be detected by a changed synthesis rate of, e.g., the chaperone protein DnaK, about a minute after the temperature shift. Given that the DnaK protein in itself takes about 45 seconds to synthesize, the observed fast change in DnaK production must be very close to the physical mechanism that triggers the response. We will argue for a mechanism that does not demand an additional synthesis of s 32 and thus postulate that a changed synthesis of s 32 only plays a role in the latter stages of the HS. To quantify the physical mechanism we measure the dependence of HS with initial temperature and find that the magnitude of the shock is inversely proportional to the folding stability of a typical globular protein.This paper measures the expression of protein DnaK. Steady state levels can be found in [5,6]; they vary from approximately 4000 at T 13.5 to approximately 6000 at 37 ± C. DnaK is a chaperone and it has a high affinity for hydrophobic residues [7]; as these signal a possible misfold, s 32 controls the expression of DnaK by binding to the RNAp. One expects at most a few hundred s 32 in the cell, a number which is dynamical adjustable because of the short in vivo half-life of s 32 (0.7 min at 42 ± C and 15 min at 22 ± C [8,9]). The lifetime s 32 is known to increase transiently under the HS.The measurement was on an E.coli K12 strain grown on an A 1 B medium with a 3 H labeled amino acid as described in [10]. After the temperature shift we extracted samples of the culture at subsequent times. Each sample was exposed to radioact...