The use of stress-70 proteins in physiology: a re-appraisal



There are few factors more important to the mechanisms of evolution than stress. The stress response has formed as a result of natural selection, improving the capacity of organisms to withstand situations that require action. The ubiquity of the cellular stress response suggests that effective mechanisms to counteract stress emerged early in the history of life, and their commonality proves how vital such mechanisms are to operative evolution. The cellular stress response (CSR) has been identified as a characteristic of cells in all three domains of life and consists of a core 44 proteins that are structurally highly conserved and that have been termed the ‘minimal stress proteome’ (MSP). Within the MSP, the most intensely researched proteins are a family of heat-shock proteins known as HSP70. Superficially, correlations between the induction of stress and HSP70 differential expression support the use of HSP70 expression as a nonspecific biomarker of stress. However, we argue that too often authors have failed to question exactly what HSP70 differential expression signifies. Herein, we argue that HSP70 up-regulation in response to stressors has been shown to be far more complex than the commonly accepted quasi-linear relationship. In addition, in many instances, the uncertain identity and function of heat-shock proteins and heat-shock cognates has led to difficulties in interpretation of reports of inducible heat-shock proteins and constitutive heat-shock cognates. We caution against the broad application of HSP70 as a biomarker of stress in isolation and conclude that the application of HSP70 as a meaningful index of stress requires a higher degree of validation than the majority of research currently undertakes.


All organisms will at some point during their lifespan encounter environmental conditions that challenge the physiological functioning of their cells. When such an effect becomes severe enough to require countermeasures to maintain cellular homoeostasis, it may be considered stressful. Such environmental factors play an integral role in the ecology and evolution of biological systems. The term ‘stress’ has been a stalwart of the scientific community since the mid-twentieth century (Selye 1956); however, universal definitions have been hard to come by and still provide much controversy (Koolhaas et al. 2011). For the purposes of this article, stress will be defined as ‘an environmental or genetic factor that causes a change in a biological system, which is potentially damaging and which has some consequence on the organisms’ Darwinian fitness’ (adapted from Hoffmann & Parsons 1991).

Generalized responses at a cellular level are known as the ‘cellular stress response’ (CSR), which can be defined as ‘a defence reaction to a strain imposed by environmental forces on macromolecules’ (Kültz 2005). The CSR is thought to have originated in the earliest examples of life on Earth and through stabilizing selection is now characteristic of all cells (Koonin 2003; Kültz 2003). The CSR's ubiquity across all forms of life is due, in part, to the fact that stressors, despite being numerous and diverse, result in nonspecific macromolecular damage. 44 proteins with known functions in the CSR have been identified that are highly conserved in all three domains of life. These 44 proteins make up what has been termed the universal ‘minimal stress proteome’ (MSP) (Kültz 2005). Because of the generality of the damage caused by stress within cells, and the ubiquity of the 44 proteins that make up the MSP, such proteins have been advocated as good biomarkers of stress.

Undoubtedly, the most famous and widely studied group of proteins involved in the CSR are known collectively as the heat-shock proteins (HSPs) (Feder 1999; Feder & Hofmann 1999). Heat-shock proteins are a large subset of proteins termed molecular chaperones. Their name (chaperone) is taken from their function in the prevention of inappropriate protein aggregations; they also play other vital roles in folding and unfolding, assembly and disassembly, transport and degradation of misfolded/aggregated proteins (Benarroch 2011). Such tasks are imperative under normal cellular conditions; however, during periods of stress when the potential for proteins and other macromolecules to function irregularly is amplified, the need for molecular chaperones is also increased. It is important to restate that nonspecific macromolecular damage activates the CSR. In the case of heat-shock regulation, this is controlled by the ‘cellular thermometer’ model first outlined by Craig & Gross (1991). These authors have proposed that under nonstressful conditions, heat-shock factor 1 (HSF1) monomers are associated with molecular chaperones including members of the stress-70 family (HSP70s and HSC70s). However, under stressful conditions, as misfolded protein levels increase in the cytosol, the chaperone molecules dissociate from HSF1 freeing it to move into the nucleus and bind to heat-shock element (HSE) promoter regions to promote HSP70/HSC70 gene transcription (Craig & Gross 1991; Tomanek & Somero 2002). Therefore, HSP70 transcription is not activated directly by stress but simply an increase in macromolecular damage.

The first genes coding for HSPs were discovered in Drosophila busckii in 1962, seen as chromosomal puffing (as a result of RNA synthesis) after exposure to increased temperatures (Ritossa 1962). Subsequent studies have found an ever-increasing suite of stressors induce a heat-shock response (HSR) (Feder & Hofmann 1999). As a result, the HSR is now considered to be a fundamental component of an organism's CSR. The best studied family of heat-shock proteins are known as the stress-70 family and are named in relation to their molecular weight (70 kDa). This family comprises of three main isoforms (Sørensen 2010):

  1. Solely constitutive: present during normal cell functioning, amongst other roles, they carry out folding of nascent polypeptides under normal cellular conditions (known as heat-shock cognates, e.g. HSC70) (Boutet et al. 2003).
  2. Solely inducible: up-regulated in cells in response to stressful stimuli (Ravaux et al. 2007).
  3. Constitutive and inducible: expressed during normal cell functioning and also up-regulated in response to stressful stimuli (Callahan et al. 2002).

The HSP70 group shows unique versatility, demonstrating the ability to assist in a large number of protein-folding processes (Richter et al. 2010), from de novo polypeptide folding to the translocation of proteins across membranes (Benarroch 2011). The ability to perform such a large array of tasks arises from the fact that unlike many other HSPs that fully or partly enclose their substrates, the HSP70 family member binds via the C-terminal substrate-binding domain (SBD) with only a short segment of the target polypeptide (Lewis et al. 1999). This allows HSP70 to bind to a variety of substrates of differing sizes and configurations. Such flexibility reflects in the frequency of up-regulation and concentration during stress situations, HSP70 being the most commonly up-regulated of all HSPs in most organisms in response to stress (Lewis et al. 1999).

HSP70 as a stress biomarker

Heat-shock proteins, and in particular the HSP70 family, have been suggested by many in the past two decades as a good universal biomarker for stress, and superficially, it is easy to see why. Effective stress biomarkers need to meet certain criteria, they must be: quantifiable, universal within the study group, sublethal, and reliable for interpretation (Ingeborg Werner 1999; Lewis et al. 1999). HSP70 family proteins are ubiquitous and highly conserved in nearly all organisms, making the genes encoding them easy to isolate and identify within a genome. They are responsive to a large variety of stresses; in fact, the majority of known/tested stresses will at some magnitude induce a HSR (Lindquist 1986; Mukhopadhyay et al. 2003; Sørensen 2010). Induction of HSP70 within organisms is much more sensitive to stress than other more traditional markers such as growth inhibition, lethal dose measurements or fertility (Sørensen et al. 2003). HSP70 has unique versatility in terms of its activity range in comparison with all other stress proteins (Mayer 2010) and will therefore, in most cases, be the easiest to detect.

Since the discovery of HSP70 and the identification of its role in the cellular stress response, a plethora of research has used HSP70 differential expression as a marker of stress. During this period, technological advances have drastically altered our understanding of the HSP70 family. As an example, we can now probe with great detail how individual proteins within the heat-shock collective alter their structural conformation to successfully perform tasks within the cell (Mayer 2010). Work on certain model species (Drosophila sp., Arabidopsis sp. or Danio rerio, etc.) is continually pushing forward our understanding of the relationship between stress-70 family expression and organismal stress (see Calabria et al. (2012) as an example). In doing so, such research is also determining previously unidentified limitations in the use of HSP70 as a stress biomarker. Nevertheless, a large portion of the current HSP70 stress biomarker research is carried out on nonmodel organisms; in many of these cases, the way stress-70 differential expression is viewed has not kept a pace and remains rather simplistic. Our now in-depth understanding of the complexities of HSP70 expression in key model taxa (examples to follow) appears, in a large number of other cases, to have been overlooked in studies using HSP70 as a marker of stress. This may be in the form of incorrect assumptions of what the differential expression of a particular stress-70 family member represents, or inappropriate experimental design. The following sections will highlight the disparity between our current understanding of stress-70 family expression at the cutting edge of the science and their improper use as stress biomarkers.

Examples of common usage

It has now become common practice to expose organisms to a potential stressor under laboratory conditions and subsequently use the transcriptional up-regulation or protein level increase of a specific member of the HSP70 family as evidence of stress and the activation of the CSR. Many authors also compare the relative levels of up-regulation to infer the degree of stress experienced by the organism. For example, an organism showing a 20-fold induction of a particular member of the HSP70 family may be thought to be experiencing a higher degree of stress than if only a four-fold induction was seen. In 2000, Tomanek and Somero conducted experiments on two congeneric marine snails (Genus Tegula) that occupy different intertidal zones to determine whether the vertical limits of their distributions were set by thermal stress (Tomanek & Somero 2000). Accordingly, they exposed individuals from both species to acute heat-shock treatments and plotted relative expression levels of a number of HSPs including two HSP70 proteins over a 50-hour recovery period. Both species showed evidence of an up-regulation in different HSPs during the postexposure recovery periods. This up-regulation was suggested as evidence of heat stress. The low intertidal/subtidal species synthesized a higher level of HSP70 over the recovery period than the mid-intertidal species. It was concluded that the low intertidal/subtidal species experienced a higher degree of thermal stress and incurred greater thermal damage than the midintertidal species (Tomanek & Somero 2000). In another example, Hauton et al. (2009) conducted sea-water acidification experiments on the marine amphipod Gammarus locusta choosing to use HSP70 as a marker of pH stress caused by disruption of an organisms’ internal acid/base balance (Hauton et al. 2009). An HSP70 gene sequence was found and tested to determine whether it was an inducible form. After acute heat-shock treatment, the HSP70 showed a 2000-fold increase in expression, proving that it can be stress induced, at least in response to heat exposure. In the main experiment, however, no significant change in expression of the HSP70 gene was resolved in samples taken 14 and 28 days after initial exposure. From the data, it was concluded that prolonged exposure to a reduced pH environment did not induce a sustained acid/base imbalance (Hauton et al. 2009). HSP70 has been used in these ways or similar for many years with very few questioning the validity of such methods and analysis. We now contend, however, that because the up-regulation of HSP70, or lack of, occurs in response to so many varying factors, to make even simplistic assumptions may be subject to many complications. In such cases, we question what exactly HSP up-regulation signifies and whether experiments designed in this way can elucidate those answers.

Multistressor: laboratory versus field measurements

The majority of HSP70 stress biomarker studies have investigated the link between a single stressor and levels of HSP70 as controlled laboratory exposures. The results from such studies are commonly used to explain field observations or scenarios. The natural environment, however, is characterized by having many parameters that affect organisms simultaneously: in fact, in the field, multiple stressors are the rule rather than the exception. A vast array of stressors have been shown to provoke HSP70 induction (Fig. 1a). It is known that HSP70 levels can display a ‘toxic cocktail’ style phenomenon where a number of minor stressors occurring simultaneously produce higher inducible HSP70 levels than the sum of each stress individually (Hamer et al. 2004). Lockwood & Somero (2011) recently compared gene expression profiles for salinity stress and heat stress and found that 45 genes changed expression significantly in response to both stressors. Interestingly, the most strongly down-regulated heat stress ‘genes in common’ were the most strongly up-regulated in response to salinity stress. This differential expression profile between the two stressors may be due to the antagonistic effects of salinity and temperature on cross-membrane ion transportation (Lockwood & Somero 2011). This is a clear example of the differential effects of certain stressors on the expression of stress-inducible genes. The above examples highlight that the usefulness of HSP70 as a stress biomarker in field studies or research involving multiple stressors is significantly curtailed.

Figure 1.

An illustration showing the variety of factors that can induce or alter the expression of members of the stress-70 family, as well as experimental and methodological factors that can affect the way HSP70 up-regulation is observed during analysis. (a) Highlights some examples of the vast array of factors that can induce HSP70 expression. These fall into three categories: (i) environmental factors; (ii) external nonenvironmental factor; and (iii) internal nonenvironmental factors. (b) Gives examples of methodological/experimental approaches that may alter the observed HSP70 expression during subsequent analysis. (c) Highlights how the current state of the test organism and its life history may alter the way it expresses HSP70. The diagram draws attention to the interconnected nature of all the above factors, which belies the simple interpretation of HSP70 expression profiles regularly found in HSP70 stress biomarker research. HSP70 up-regulation must be analysed and interpreted with thorough consideration of this complexity.

Within-organism variation

HSP70 expression has also been found to be tissue specific in some species. Rabergh et al. (2000) found a variation in HSR amongst certain tissues of Danio rerio (zebrafish) (Rabergh et al. 2000). On exposure to heat stress, it was found that HSP70 levels were higher in the fish's gonad under all conditions than the liver and gills (Rabergh et al. 2000). It is apparent that certain stresses are going to have greater effects on certain tissues, that is, heat stress is likely to affect, and therefore induce a greater HSR, in tissues that are more heat sensitive such as fish gonads. In such cases, different organs from a single organism might give very different HSP70 expression profiles that without prior knowledge can alter the analysis of perceived stress (Fig. 1b). An understanding of these potential interorgan variations is particularly important when working with smaller organisms where whole-organism homogenisations may be employed. In such cases, stress signals from small but significant tissues may be masked by larger trivial signals (Pyza et al. 1997; Rabergh et al. 2000).

Stress-70 family proteins function in all major subcellular compartments of the cell. As such, distinct isoforms can be differentiated and split into mitochondrial members, endoplasmic reticulum (ER) luminal member, chloroplastic members or cytosolic members. Each isoform may have very similar but distinct roles and obviously differ in terms of subcellular spatial distribution (Sung et al. 2001). Cytosolic HSP70 members, for example, may play an important role in HSR regulation through the negative repression of heat-shock factor (HSF)–mediated transcription (Shi et al. 1998). Evidence is mounting that suggests that many species’ genomes contain a large number of HSP70 genes: the human genome is thought to contain at least 11 HSP70 genes (Tavaria et al. 1996); the yeast stress-70 family has 14 known members (Rassow et al. 1997), and recently, it was discovered that the genome for the Pacific oyster, Crassotrea gigas, contains 88 HSP70 genes (Zhang et al. 2012). In contrast, Escherichia coli has only two stress-70 family members (Bardwell & Craig 1984; Seaton & Vickery 1994). Sung et al. (2001) compiled an expression profile analysis of 12 HSP70 genes in an Arabidopsis plant, and Li et al. (1999) brought together a similar expression analysis of 10 HSP70 members in spinach (Li et al. 1999): both studies found similar results. In response to elevated temperatures, all stress-70 family members examined showed coordinated up-regulation. Conversely, subsequent repression profiles varied amongst isoforms with some members showing rapid repression after 60 min, whilst others still showed maximal expression after 90 min. This temporal profile of expression itself raises significant problems when considering experimental and sampling designs (as discussed below). In response to lowered temperatures, both studies found no coordination in stress-70 member expression profiles; some members showed no induction, while other members showed similar expression to elevated temperature. Sung et al. (2001) found induction by low temperature was limited to cytosolic and mitochondrial members of the stress-70 family, and induction and repression timescales varied individually. Sung et al. (2001) also found differential expression of stress-70 family members in different organs: all members were expressed in the root, with higher levels shown in one mitochondrial and one cytosolic member. In contrast, two chloroplastic members showed higher levels in leaf tissue. These findings highlight that HSP70 expression varies not only between organs, but also at a subcellular level. Variation occurs not only in magnitude but also temporally in terms of induction and repression. This is of high importance in the application of stress-70 members as stress biomarkers. A large portion of the research that uses HSP70 as a stress marker makes no reference to which of the potentially numerous members of that species’ stress-70 family was analysed and whether that particular member can be thought of as characteristic of the species’ stress response. Analysing different members of the stress-70 family may lead to vastly different and erroneous interpretations of the level of stress an organism is being exposed to. To produce reliable results, all members of the stress-70 family should be characterized and analysed to determine which, if any, members accurately portray the study organisms’ level of stress. This process is undoubtedly highly costly and time-consuming for all those not working on one of the few very well characterized model organisms.

Experimental design

The timing of sampling during or poststress can have a large bearing on the way stressor impact is observed. The sampling target, be it mRNA or protein, can also have a significant bearing on perceived stress levels (Fig. 1b/c). HSP70 is involved in repair and restoration, not in direct cell death prevention. When an acute stressor is nearing lethal severity, cellular mechanisms are primarily concerned with the avoidance of stress-induced cell death. At this point, the cost of mounting an HSR outweighs the benefits, and all available energy is directed towards cell survival. Only when imminent death is prevented are the mechanisms of repair and restoration set in motion: DiDomenico et al. (1982) found the up-regulation of an HSP70 was delayed by around an hour after a 39 °C heat shock in comparison with a 36.5 °C heat shock. Sampling during or immediately after a near-lethal acute stressor exposure is unlikely to yield an HSP70 up-regulation signature despite the fact that the cells in question are likely to be severely stressed. Buckley et al. (2006) compared mRNA expression and protein production of an HSP70 in the goby Gillichthys mirabilis over timescales during and poststress exposure; they found that an increase in mRNA levels correlated with an increase in protein production in all cases. More interestingly, however, they found that mRNA versus protein production varied in relative timing, magnitude and according to the organ sampled. For example, a 5-fold increase in HSP70 mRNA in gill tissue peaking at 180 min into heat shock was followed by a 12-fold increase in protein levels peaking at 350 min into heat shock. By contrast, in muscle tissue, an 18-fold increase in mRNA peaking at 220 min into heat shock was followed by a three-fold increase in protein levels peaking at 300 min (Buckley et al. 2006). These data suggest that using mRNA levels to predict subsequent protein levels may be erroneous, and thus, using mRNA expression alone may not properly describe the condition of the organism. The timing of mRNA or protein measurement during or after stress exposure is also of significance as the peak magnitude of fold increase varied between mRNA and protein and also between tissues sampled. With this in mind, we caution against single ‘snapshot’ time point measurements of either mRNA or protein levels, as this method will likely misrepresent organismal stress levels. This snapshot style measurement is prevalent in many HSP70 stress biomarker studies (Hauton et al. 2009, for example).

It is common practice in acute stress studies to allow a recovery period before preserving samples for HSP70 analysis (Cottin et al. 2010): this is usually a minimum of 30 min after acute stressor exposure, followed by regular sampling from a few hours to several days after exposure. In the early 1980s, when HSP70 research was still in its infancy, DiDomenico et al. (1982) showed that the severity of a heat stress treatment affected the time taken to express HSP70 and subsequently recover to basal levels. Results from Tomanek & Somero (2000) clearly show that a stress-induced HSP70 is up-regulated substantially during the first 5 h of recovery and reaches a peak after 15 h. In contrast, an HSP77 was also analysed and showed no significant up-regulation until 12 h after exposure. Once again this is an example of the differential behaviour of members of the stress-70 family. Had Tomanek & Somero (2000) only sampled one stress-70 member, as many HSP70 studies do, their evaluations of the study organisms’ stress would have been very different (Fig. 1c). Hauton et al. (2009) only sampled one HSP70 and did so only at 14 and 28 days after initial exposure, when no significant HSP70 up-regulation was found, it was concluded that no sustained acid/base imbalance stress was occurring. In fact it seems more likely that at some point, probably before the 14-day sampling, HSP70 was up-regulated in response to the pH change, particularly as the experimental organisms were subjected to an instantaneous pH change with no acclimation period at the start of the experiment. It may have been that this acute pH change invoked an HSP70 response, which was later replaced by more cost-effective stress-specific mechanisms. As highlighted above, different members of the stress-70 family will react to stressors over different timescales, so sampling only one HSP70 member is highly erroneous. The effective use of HSP70 as a stress biomarker requires the use of numerous stress-70 members alongside a rigorous postexposure time series sampling process at the very least: a procedure very few HSP70 biomarker studies conform to.

Nonenvironmental HSP70 variation

Studies have found that many nonenvironmental cues can induce and affect the way HSP70 is expressed (Fig. 1a). Genetic stressors, such as mutation (Zhao et al. 2002) and inbreeding (Kristensen et al. 2002), have been shown to influence levels of HSP70 expression. Kristensen et al. (2002) found that inbred larvae of Drosophila buzzatii expressed higher levels of an HSP70 in response to all but the very highest of temperatures within the larvae's thermal limits compared with outbred larvae. Inbreeding depression and the subsequent expression of deleterious alleles may lead to an increase in misfolded proteins at a cellular level, which in turn would up-regulate HSP70 (Kristensen et al. 2002). HSP70 levels are also dependent on the fitness of the organism, regardless of any stressors acting on the organism at a given time. The costs associated with HSP70 up-regulation are essentially concerned with development, fecundity and energy (Sørensen et al. 2003). The trade-off balance between the costs and benefits of HSP70 up-regulation may be dependent on the relative fitness of the individual. In such cases, the costs of mounting a generalized stress response may outweigh the benefits, and other forms of response may be more appropriate. This has been shown in populations that are exposed to frequent or continuous stress, which have been observed to produce lower levels of certain HSP70 proteins than populations that only experience stress infrequently (Lansing et al. 2000); effective stress responses are potentially achieved in such populations by other more cost-effective and stressor-specific means (Kohler et al. 2000).

Another study conducted by Neargarder et al. (2003) found that the frequency of phosphoglucose isomerase (PGI) alleles varied with latitude and altitude in populations of the willow beetle Chrysomela aeneicollis. It was also found that organisms with alternative PGI genotypes differed in their expression of inducible HSP70. This result suggests that genetic particularities that may not obviously be linked to HSP70 or the heat-shock response such as PGI allele variation can modulate the expression of HSP70 (Neargarder et al. 2003). Intracellular parameters have also been shown to affect the way HSP70 is expressed. Heat shock is known to increase intracellular cytosolic-free calcium concentration (Ca2+). An increase in free Ca2+ in the cytosol enhances the production of inducible HSP70 forms by facilitating the binding of HSF and HSE (Ding et al. 1996). Therefore, intracellular free calcium levels may affect the level of HSP70 production in response to stressors.

Each of the aforementioned factors can profoundly alter the way in which HSP70 is up-regulated within organisms. Up-regulation signals from nonenvironmental cues may be incorrectly interpreted as an environmental stress signal. Such nonenvironmental parameters are inherently hard to quantify, can vary significantly amongst individuals even within a single population and may distort any stress-related HSP70 induction. HSP70 has many well-documented roles within the cell. Research has also found that extracellular HSP70 (eHSP70) can act as a highly effective pro-inflammatory response (Basu et al. 2000; Srivastava 2002). Such a response is of particular relevance to this discussion because eHSP70, and therefore also intracellular HSP70, has been shown to be up-regulated in response to perceived danger. Fleshner et al. (2004) found that exposure to an exclusively psychological stressor triggered both intracellular and extracellular HSP72 up-regulation. In the study, adult male rats were exposed to a cat, but without any physical contact, thus a purely psychological stressor (Fleshner et al. 2004).

HSP70 expression has been shown to vary according to the life stage and sex of the organism in question. Garbuz et al. (2008) demonstrated that HSP70 expression levels under nonstressful conditions were twice as high in the larvae of soldier flies than in related adults. This pattern may be due to a reduced need for thermotolerance in adults of the species because increased mobility allows them to escape potential stressors with greater ease (Garbuz et al. 2008). Constitutive expression of stress-70 family members has also been shown to vary considerably according to life cycle in Arabidopsis plants (Sung et al. 2001). Sex has also been shown as a source of variation in HSP70 expression within species; Sørensen et al. (2001) found that HSP70 expression varied between males and females in two Drosophila buzzatii populations, and this may be a result of differing sensitivities to stress between the sexes, or the differential cost of HSP70 expression on reproduction. Further studies have demonstrated differential HSP70 expression according to development stage and sex (Feder et al. 1996; Pyza et al. 1997), the latter study showing opposite HSP70 responses to increasing stressor intensity between 1-day-old and 5-day-old pupae in Musca domestica. Madeira et al. (2012) found high intraspecific variability in HSP70 expression in several coastal and estuarine marine species, citing a host of potential influencing factors, including sex, age, nutritional status, all of which alter the relative expression of HSP70 by individuals when exposed to stressors. These results have implications for the selection of study organism's life stage and sex, as well as subsequent organism handling and the environment within which the experiment is conducted, all of which may have profound effects on HSP70 expression.

Constitutive versus inducible HSP70

Members of the stress-70 family have been split into two functionally distinct groups: those that are constitutively expressed (HSC70) and those that are stress inducible (HSP70). However, as of yet, no clear evidence has been presented that allows any consistent interspecies differentiation between the two groups, either functionally or structurally. Some research has suggested that genes encoding HSP70 lack introns as opposed to those encoding HSC70, which possess them (Gunther & Walter 1994; Boutet et al. 2003). It was suggested that the lack of introns in the stress-inducible form allowed for a more rapid response to stress because RNA splicing was not necessary. However, recently, it has become clear that the presence or absence of introns cannot be used to differentiate between HSP70 and HSC70. Qin et al. (2003) amongst others found introns within genes encoding stress-inducible HSP70 (Qin et al. 2003). Leignel et al. (2007) examined other characteristic regions of genes encoding HSC/HSP70 in hydrothermal vent and coastal crabs and found no structural characteristics that could differentiate between inducible and constitutive forms.

From a functional perspective, differentiation has also been problematic. Research conducted on silver sea bream found that, although structurally distinct in this case, HSP/HSC70 both showed stress-associated induction (Deane & Woo 2005). A number of studies on molluscs have found high levels of both HSP70 and HSC70 in apparently unstressed individuals (Buckley et al. 2001). Franzellitti & Fabbri (2005) found that gene sequences in a Mediterranean mussel (Mytilus galloprovincialis) for HSP70 and HSC70 were more closely related to other bivalve HSP/HSC70s, respectively, than to each other. This interspecific gene homology was suggested to indicate that HSC/HSP70 could be paralogs: performing slightly different but highly related roles within the cell (Franzellitti & Fabbri 2005). Nowhere is the functional confusion of HSP/HSC70 better underlined than in the Antarctic (highlighted in Clark & Peck 2009). In the past decade, many studies have found that Antarctic fish lack the ability to up-regulate some HSP70s in response to external stresses (Hofmann et al. 2000). To further complicate matters, a number of studies found several of the fish studied showed permanent expression of an inducible HSP70 (Clark et al. 2007; Clark & Peck 2009), which appeared to assume the same role as the constitutively expressed HSC70. What would normally be considered an inducible HSP70 was in this case noninducible, but was permanently (i.e. constitutively) expressed. This phenomenon is by no means restricted to the Antarctic; Gehring & Wehner (1995) demonstrated that two species of Saharan ants, known to be able to survive body temperatures above 50 °C, both accumulate high levels of two inducible HSP70 (i.e. express constitutively) isoforms regardless of whether they are experiencing temperatures outside their optimal range or not. Similar patterns were found in nine Saharan lizard species (Ulmasov et al. 1992). Another more recent study was conducted on four Stratiomyid larvae (Soldier fly) occurring in the vicinity of hot volcanic springs (Garbuz et al. 2008). Once again, each of the four larval species showed high constitutive expression of a number of HSP70 isoforms including inducible forms. This pattern of chronic HSP70 expression is certainly not the norm (continuous or super expression of HSP70 has been suggested to be harmful (Krebs & Feder 1997); however, as the above examples reveal, chronic HSP70 expression may be adaptive in thermally ‘extreme’ environments such as the Sahara or Antarctica. Certainly, in the above examples, the benefits of increased thermo-tolerance acquired from high constitutive HSP70 levels must outweigh the deleterious effects of high cellular HSP70 levels (Garbuz et al. 2008). In contrast to high constitutive HSP70 expression in ‘extreme’ but stable environments, species occupying highly variable environments have also been shown to constitutively express high levels of HSP70. Dong et al. (2008) showed that limpets occupying high intertidal zones (more variable) had higher constitutive expression of a reported HSP70 than limpets occupying mid- or low-intertidal zones. One of the high intertidal species (Lottia scabra) also showed no induction of this particular HSP70 in response to elevated temperatures (Dong et al. 2008). It was suggested that this may be an adaptive ‘preparative defence’ strategy to cope with highly variable and unpredictable environments such as the high intertidal zone. The examples from stable ‘extreme’ and high variable/unpredictable environments given above have important implications for the use of HSP70 as a stress biomarker. For example, contrary to conventional interpretations, high levels of HSP70 may have no relation to any recent stress exposure or the stress status of the organism and just simply reflect an adaptive strategy to the environment under study. Clearly, the structure and function of individual genes within the stress-70 family varies interspecifically and between habitats, and at present, it would seem erroneous to extrapolate characteristics of one species' stress-70 family to others. Also, the differing strategies employed by species from different environments may lead to incorrect assumptions of the condition of the organism. These matters undoubtedly complicate the interpretation of stress-70 family proteins as stress biomarkers. Strategies such as ‘preparative defence’ and others mentioned in this article may be ecologically important in allowing species to occupy certain habitats. However, some inferred strategies seem convoluted (see Clark & Peck 2009) and may have been developed by authors to explain stress-70 expression in the absence of appropriate knowledge of the stress-70 members being studied. With better understanding of the stress-70 family, and their roles, more simplistic and intuitive biological theories may be established.


At its most fundamental level, the up-regulation of HSP70 signifies the presence of proteins whose native, functional conformations have been altered by stress. When using HSP70 as a stress biomarker, we make assumptions that this shift in conformation is the result of stress and that the level of HSP70 expression at any time point is related to the level of partially unfolded proteins and thus the level of stress. This article highlights the vast array of factors that affect the level and timescale at which HSP70 is up-regulated in response to diverse stressors, and the way the experimental design and analysis can alter the way stress is perceived. Many of these factors break our simplistic assumptions. Without any prior understanding of how each of these factors alters the way HSP70 is expressed individually, and when combined, environmental stress levels may be misinterpreted. In many cases, uncertainty remains over the differential role of members of the stress-70 family. What is clear is the function and structure of specific members of the stress-70 family varies between species and regions exhibiting different potential stressor profiles and magnitudes. Without using a number of members of the stress-70 family, both inducible and constitutive, and profiling their expression levels over an appropriate timescale, stress signals can be easily missed or misinterpreted.

This article highlights the degree of validation required before HSP70s can provide real insight into an organisms' stress levels. The most significant advances in the field are being made as a result of rigorous validation, particularly in areas working on model organisms such as Drosophila sp. A large portion of current stress biomarker research is conducted on nonmodel organisms, and HSP70 stress biomarker validation in these cases will be much more difficult and time-consuming. We suggest that future research may benefit from eliciting stress-specific responses rather than generalized ones. Stress- or taxa-specific biomarkers are unlikely to be subject to such in-depth validation as is required for HSP70. A stress-specific approach to the search for stress biomarkers may yield more insightful progress in our understanding of stress physiology.

The article was written by J.P.M. with input from S.T. and C.H. The ideas and opinions discussed were jointly conceived by all three authors.