Selye defined stress as the nonspecific response of the body to any demand. Stressors elicit both pituitary-adrenocortical and sympathoadrenomedullary responses. One can test Selye’s concept by comparing magnitudes of responses at different stress intensities and assuming that the magnitudes vary with stress intensity, with the prediction that, at different stress intensities, ratios of increments neuroendocrine responses should be the same. We measured arterial plasma ACTH, norepinephrine, and epinephrine in conscious rats after hemorrhage, intravenous insulin, subctaneous formaldehyde solution, cold, or immobilization. Relative to ACTH increments, cold evoked large norepinephrine responses, insulin large epinephrine responses, and hemorrhage small norepinephrine and epinephrine responses, whereas immobilization elicited large increases in levels of all three compounds. The ACTH response to 25% hemorrhage exceeded five times that to 10%, and the epinephrine response to 25% hemorrhage was two times that to 10%. The ACTH response to 4% formaldehyde solution was two times that to 1%, and the epinephrine response to 4% formaldehyde solution exceeded four times that to 1%. These results are inconsistent with Selye’s doctrine of nonspecificity and the existence of a unitary “stress syndrome,” and they are more consistent with the concept that each stressor has its own central neurochemical and peripheral neuroendocrine “signature.”
hans selye in 1936 (31) reported that exposure to any of several noxious agents produced the same syndrome: adrenal enlargement, gastrointestinal ulceration, and thymicolymphatic involution. From this pathological triad, he developed a theory of stress that attained wide popularity and aroused intense research interest but also incited controversy that persists to this day (32). He defined stress as the nonspecific response of the body to any demand (33), emphasizing that the same pathological triad would result from exposure to any stressor. In this report, we call this the doctrine of nonspecificity.
Selye focused on the hypothalamic-pituitary-adrenocortical (HPA) system as a key effector of the stress response. The HPA activation attending stress was thought to be part of a constellation of stereotypical neural and endocrine responses. Administration of ACTH can elicit all three components of the pathological triad, and some have even defined stress empirically as that which increases plasma ACTH concentrations (8). Selye did not claim that ACTH, or high circulating levels of adrenocortical steroids, can elicit the pathological triad.
Numerous studies, however, have demonstrated differential neuroendocrine responses during exposure to different stressors. For instance, glucopenia evokes selective adrenomedullary and pituitary-adrenocortical activation (10, 22, 23); orthostasis, hyperthermia, and cold exposure evoke selective sympathoneural activation (16, 27, 37); water deprivation evokes selective vasopressinergic activation (29); and manipulations of dietary salt intake produce selective effects on renin-angiotensin-aldosterone system activity (17). Moreover, during exposure to different stressors, sympathetic responses exhibit pronounced heterogeneity among different organs (6, 36); response patterns also depend on the duration and intensity of the stressor (24, 30) and vary with the experience of the individual (13).
This patterning does not of itself refute the doctrine of nonspecificity, because the different patterns could result from superimposition of different stressor-specific homeostatic responses on the same nonspecific stress response. According to Selye’s theory, stress refers only to the nonspecific component revealed after subtraction of the specific components from the total response. As demonstrated mathematically in thediscussion, it is not possible to test Selye’s theory by comparing effector responses to single intensities of different stressors.
Acceptance of certain assumptions, however, renders the doctrine of nonspecificity amenable to experimental testing. This testing was the main purpose of the present experiments. We assessed plasma ACTH and catecholamine responses to different intensities of various stressors, all of which are known to evoke increases in circulating ACTH levels. By comparing ratios of differences in responses to low- and high-intensity stressors above a threshold level of stressor intensity, we examined the hypothesis that all stressors evoke the same nonspecific response pattern, a hypothesis that has survived more than half a century without undergoing a direct test.
All experiments described in the present study were approved by the National Institute of Neurological Disorders and Stroke Animal Care and Use Committee.
Male Sprague-Dawley rats (Taconic Farms, Germantown, NY), weighing 280–370 g, were maintained in an animal housing room at 22 ± 2°C and 40% humidity, with lights on from 0600 to 1800. The rats were fed a normal laboratory diet. Tap water was provided ad libitum. At least 4 days for acclimatization in the housing room elapsed before the start of the experiment.
Conscious animals with an indwelling arterial catheter were exposed to one of the following stressors: insulin (Ins); hemorrhage (Hem); immobilization (Immo); formaldehyde solution (Form) injected subcutaneously to produce pain and tissue damage; cold exposure; physiological saline (Sal) injected subcutaneously after handling in preparation for the injection; and painless intravenous Sal (without handling) as a control.
Preparation of animals.
Twenty-four hours before the acute experiments, each rat was anesthetized with pentobarbital sodium (40 mg/kg ip). Polyethylene catheters (PE-50) filled with heparinized 0.9% Sal (50 IU/ml) were inserted into a femoral artery and, in some experiments, into a femoral vein. The catheters were sutured in place and tunneled subcutaneously to the nape, where they were attached to a metal spring. The cannulas were flushed with heparinized 0.9% Sal (100 IU/ml, 0.2 ml) at the end of light cycles (1700–1800).
All acute experiments began between 0800 and 0900 on the day after operation. Each rat was assigned randomly to one of the experimental treatments described below.
Blood samples (0.4 ml each) were collected into tubes containing heparin (5 ml, 1,000 IU/ml) for catecholamine determinations or tubes containing EDTA for ACTH determinations. Sal was added to the blood cells, and the same volume as had been drawn was injected into the animal immediately after each blood sample was collected. Blood samples for catecholamine determinations were centrifuged immediately at 13,000 rpm for 90 s to separate the plasma. Blood samples for ACTH determinations were centrifuged at 2,900 rpm for 20 min at 4°C to separate the plasma. Plasma samples were stored at −70°C and analyzed within 3 wk.
The neurochemical results for the various stressors were compared with those after intravenous injection of Sal as a control, as described below.
Handling and subcutaneous Sal.
Even brief, gentle handling of conscious rats produces marked increases in plasma levels of catecholamines (18). Subcutaneous injection would also be expected to evoke a catecholaminergic response. Some animals underwent subcutaneous injection of physiological Sal, after handling to position the animal for the injection, as described below.
After the baseline collection, the animals received 0.3 (n = 7; 313 ± 9.4 g final body wt), 1.0 (n = 5), or 3.0 (n = 7) IU/kg of insulin (Eli Lilly, Indianapolis, IN) or physiological Sal intravenously (0.1 ml solution/100 g body wt). Arterial blood samples were obtained at 15, 45, 75, and 105 min after insulin administration.
Formaldehyde solution-induced pain and tissue damage.
After baseline samples were collected as described above, the animals received 1% formaldehyde solution (Baker, Phillipsburg, NJ;n = 6), 4% formaldehyde solution (n = 7; 344 ± 11.3 g final body wt), or Sal (n = 5) subcutaneously into the right leg (0.1 ml/100 g body wt). Arterial blood samples were obtained at 15, 45, 75, 105, and 135 min after formaldehyde solution administration.
Nonhypotensive and hypotensive hemorrhage.
The experiment began with collection of a baseline arterial blood sample. The animals were bled to either 10 (n = 6) or 25% (n = 7) of estimated blood volume (EBV), with EBV calculated from the equation, EBV = 0.06 × body weight (g) ± 0.77 (21). Arterial blood samples were obtained 15, 45, 105, and 135 min after hemorrhage.
The experiment began with collection of a baseline arterial blood sample at room temperature (24 ± 2°C) outside the cold chamber. The unshaved animal was then carefully transferred (<30 s) from its home cage (placed next to cold chamber) into the cold chamber [chamber temperature 4 (n = 9) or −3°C (n = 7)] and kept in the chamber for 3 h. Arterial blood samples were collected during exposure to cold at 15, 45, 105, and 165 min. After 3 h, the cooling system in the cold chamber was switched off, the doors of the cold chamber were opened, and room temperature was attained inside the chamber after ≤10 min. Arterial blood samples were obtained 15, 45, and 105 min after cold exposure. To assess effects of the transfer itself, three rats in a control group were exposed to room temperature for 1 h in the chamber.
The experiment began with collection of a baseline arterial blood sample. Each animal then underwent 2 h of Immo, which consisted of taping the rat’s limbs to a metal frame with hypoallergic tape (18,19). Arterial blood samples samples were obtained at 15, 30, 45, 75, 105, and 120 min during Immo.
Plasma concentrations of norepinephrine (NE) and epinephrine (Epi) were assayed using reverse-phase liquid chromatography with electrochemical detection after partial purification by adsorption on alumina (5,12, 14). Catechol concentrations in each sample were corrected for recovery of the internal standard, dihydroxybenzylamine. The limits of detection for NE and Epi were 2–4 pg (30 fmol) per volume injected.
Plasma levels of ACTH-(1—39) were measured in duplicate without prior extraction, using a commercially available RIA kit (ICN Biomedicals) (26). The intra-assay coefficient of variation was <5%.
Results are presented as means ± SE.P < 0.05 defined statistical significance.
Plasma levels of ACTH, NE, and Epi were assayed before, during, and after exposure to the stressor. Changes in levels of ACTH, Epi, and NE were analyzed either by one- or two-way ANOVA for repeated measures. An area under the curve (AUC) was calculated based on concentration × time as a measure of the magnitude of the response. Thus each animal provided one data point, the AUC, for each dependent variable. Total AUCs for plasma catecholamines and plasma ACTH were calculated by rectangular integration. Net AUCs from the baseline were calculated as the difference between total AUCs and basal AUCs. For statistical purposes, the length of testing for all stressors was limited to the duration of Immo (2 h). Differences among stressors or neuroendocrine measures were also assessed by ANOVA as appropriate, as were responses to different stressor intensities. Independent-meanst-tests were used for examining differences in responses to stressors at two intensities.
At their highest intensities, all the stressors increased levels of ACTH, NE, and Epi significantly compared with levels after intravenous Sal (Figs.1-5). Although Immo evoked large increases in plasma levels of ACTH, NE, and Epi, other stressors induced disproportionately large NE or Epi responses compared with ACTH responses. Thus cold evoked large NE responses, whereas low-dose Ins evoked large Epi responses. There was no effect of the transfer of rats into the cold chamber on plasma NE, Epi, or ACTH (data not shown). The plasma catecholamine and ACTH results in the rats receiving intravenous Sal postoperatively were similar to those previously reported in conscious, unrestrained rats (18).
The 4% Form concentration elicited about a twofold larger plasma ACTH response and about a fourfold larger plasma Epi response than did the 1% concentration, the differences for both variables being statistically significant (P < 0.05 for ACTH, P < 0.01 for Epi).
Hem evoked small NE and Epi responses relative to ACTH responses. The 25% Hem elicited about a fivefold larger ACTH response than did the 10% Hem (P < 0.01). There were no differences in plasma NE or Epi responses to 10 and 25% Hem.
The present results, based on several experiments involving various intensities of different stressors and multiple peripheral and central neurochemical-dependent measures, demonstrate the marked heterogeneity of neuroendocrine responses. Ins, regardless of dose, markedly stimulated adrenomedullary secretion, as reflected by plasma Epi levels, with much less intense sympathoneural activation, as reflected by plasma NE levels; cold exposure produced much larger proportionate increments in plasma NE levels than in plasma Epi or ACTH levels; Hem produced large ACTH responses but relatively small NE and Epi responses; and Immo increased plasma levels of all three compounds. In addition, although ACTH and Epi responses to Ins were highly correlated, ACTH and Epi responses to other stressors were more weakly correlated or were not related at all, despite significant increases in ACTH levels and therefore HPA activation in response to all the stressors.
Heterogeneity does not disprove the doctrine of nonspecificity.
As noted above, this heterogeneity does not of itself support or refute Selye’s stress theory. The stress response was thought to occur in three stages, the “general adaptation syndrome” (GAS), consisting of “alarm,” “resistance,” and “exhaustion.” Selye (33) wrote that the GAS denoted the whole response, of which the alarm, resistance, and exhaustion stages were merely successive phases. In other words, although during the different stages of the GAS the intensity of the response might vary, the neural and endocrine pattern would be the same, regardless of the stage. The stressors used by Selye and his students (33) included acute manipulations, several of which are used in the present study, hemorrhage, cold, insulin, and immobilization.
To explain the heterogeneity of neuroendocrine responses to stressors, Selye hypothesized the existence of specific reactions. Only after subtraction of these from consideration could one identify the core, nonspecific, shared element, the stress, and its stereotypical consequence, the stress response.
The doctrine of nonspecificity cannot be disproved.
No one appears to have considered formally the possibility or impossibility of disproving the doctrine of nonspecificity.
In a well-known review, Munck et al. (25) argued persuasively against Selye’s concept of “diseases of adaptation,” pointing out correctly that glucocorticoids produce potent anti-inflammatory effects, in contrast with Selye’s view that collagen diseases, allergy, and rheumatic diseases result from an abnormal or excessive GAS. Munck also correctly noted that, after the 1960s, the concept of diseases of adaptation was largely ignored. The review did not consider the validity of Selye’s doctrine of nonspecificity.
It can be shown mathematically that, without simplifying assumptions, the doctrine of nonspecificity cannot be disproved.
Consider responses of effector systems,A andB, e.g., ACTH and Epi, to the different stressors, x andy. The overall magnitude of the response of effector system A to stress x(Ax ) is the sum of the nonspecific component (x ⋅ a n) and the specific component (x ⋅ ax ), where x is a particular intensity of the stressor anda n anda x are constants relating the intensity of the stress to the nonspecific and specific components of the response. (Most stimulus-response curves approximate a linear relationship between the magnitude of the response and the logarithm of the stimulus intensity.) The nonspecific and specific components of the response of effector systemB can be represented similarly. Thus, for stressors x andy and responsesA andB Equation 1The doctrine of nonspecificity predicts that the ratioa n/b nfor each stressor is the same, i.e., that Equation 2Note that the four parts of Eq. 1 contain eight variables. From a comparison of the responses to a single intensity of two (or any number) of different stressors, it is theoretically impossible to test the prediction in Eq.2 .
Concepts about the neuroendocrinology of stress have continued Selye’s line of reasoning, proposing that above a threshold stressor intensity, a “stress syndrome” occurs (3, 15). If the magnitude of the specific component were directly related to the intensity of the stressor and if the nonspecific component were expressed after a threshold of stressor intensity, this would not render the above equations solvable, as follows.
One can obtain data for different intensities of the same stressor. Ifx is the low-intensity value of stressor x andX the high-intensity value, then applying Eq. 1 Equation 3and For the other stressor, Y, the constant for the specific response isay . If there were a threshold stressor intensity, t, for eliciting the stress response (nonspecific component), and ifX >x > t, then for effector systemsA andB Equation 4
If one assumes that the termsX ⋅ ax ,x ⋅ ax ,X ⋅ bx , andx ⋅ bx , corresponding to the specific components, either reach a maximal value or are negligible at stressor intensities eliciting the stress syndrome, thenX ⋅ ax =x ⋅ ax .
From these equations, it can be shown that Equation 5and analogously for stressor y The doctrine of nonspecificity predicts that the neuroendocrine pattern corresponding to the nonspecific component is the same for all stressors and is therefore the same for stressorsx andy, i.e., for all stressors Inspection of the four parts of Eq. 4 , however, demonstrates that if one does not accept either of the above assumptions, then even measuring responses of different effector systems to different intensities of different stressors could not solve the equations and therefore could not test the prediction of the doctrine of nonspecificity.
Assumptions rendering the doctrine of nonspecificity testable.
Inspection of Eq. 4 indicates that if the constantsax ,bx ,ay , andby were all equal to zero (or were very small with respect toa n andb n), then one could test the doctrine of nonspecificity. Assuming the constants for the specific components (ax ,bx ,ay , andby ) were small relative to the constants for the nonspecific component (a n andb n) would be tantamount to assuming that above the threshold intensity, t, the specific components would contribute negligibly to the overall responses of the effector systems. In this situation, regardless of the stressor, the ratio of the intensity-related increment in the value for effector system A to the intensity-related increment in the value for effector systemB would be a constant, namely,a n/b n.
Another assumption would also enable testing of the doctrine of nonspecificity. This assumption is that the magnitudes of both the specific and nonspecific components vary directly with the intensity of the stressor over the whole range of stressor intensities, i.e., that there is no ceiling for the specific component and no threshold for the nonspecific component. The reasoning is as follows. If one considers ratios of values for the dependent variables at different stressor intensities Equation 6 Equation6 cannot be solved; however, if t = 0, thenEq. 6 is reduced to Similarly, for stressor Y Analogously, for dependent measure B leading to the testable predictions,BX /Bx =AX /Ax andBY /By =AY /Ay .
A test of the doctrine of nonspecificity.
Even with simplifying assumptions, appropriate testing of the doctrine of nonspecificity would require a very complex study, involving multiple stressors at different intensities and multiple simultaneously assessed dependent measures. Although an enormous literature describes effects of graded intensities of stressors on neuroendocrine-dependent measures, none would apply to the issue of the validity of the doctrine of nonspecificity.
As demonstrated mathematically above, the doctrine of nonspecificity could be tested, given one of two assumptions about the existence of a threshold stressor intensity for the nonspecific response or the magnitude of the specific response above the threshold stressor intensity. In the present study, for some stressors, the data obtained were inconsistent with both assumptions and therefore irrelevant as tests of the doctrine of nonspecificity. For instance, there appeared to be a threshold for enhanced ACTH responses to Ins; yet Epi responses continued to increase substantially above the threshold. This would be reasonable, since one might expect an intensity-related increase in the need for a specific, hormonal adrenergic response as a glucose counterregulatory mechanism; however, the Ins results could not be used to test the doctrine of nonspecificity, because the obtained data did not fit either assumption.
The results about Immo also could not be applied, because the intensity of the stressor was not varied quantitatively. The results about cold could not be applied, because the specific noradrenergic component appeared to predominate regardless of stressor intensity. This would be consistent with a specific, noradrenergic neuronal response to conserve heat and expend energy; however, the responses of ACTH levels were so small, the nonspecific component could have contributed only negligibly to the overall response.
The data about ACTH and Epi responses to Hem and Form did seem to fit the assumptions required to test the doctrine of nonspecificity. Figure6 depicts the summary data in forms related to the theoretical predictions. As shown in Fig. 6 A), for plasma Epi, the ratio of the response for the more severe Hem to the less severe Hem was smaller than the ratio of the response for the more severe Form to the less severe Form. The doctrine of nonspecificity would predict that the difference between Hem and Form would also obtain for plasma ACTH; in fact, however, for plasma ACTH, the ratio of the response for the more severe Hem to the less severe Hem was much larger than the ratio for the more severe Form to the less severe Form. As shown in Fig. 6 B), the increment in plasma Epi levels between the two intensities of Form was larger than the increment in plasma ACTH levels; yet the increment in plasma Epi levels between the two intensities of Hem was smaller than the increment in plasma ACTH levels. Thus, by both tests, the doctrine of nonspecificity failed to predict the experimental results for Hem and Form.
With regard to the doctrine of nonspecificity, we therefore reach the following conclusions. Without simplifying assumptions, which may or may not be acceptable for particular stressors, the doctrine of nonspecificity is impossible to disprove and is therefore of little scientific value. Yet with acceptable simplifying assumptions, the doctrine of nonspecificity fails to predict the obtained experimental results. Given the simplifying assumptions, then, the data are inconsistent with Selye’s stress theory and refute the existence of a unitary stress syndrome.
An alternative proposal: primitive specificity of stress responses.
The present results are more consistent with an alternative proposal: that each stressor has a neurochemical “signature,” with quantitatively if not qualitatively distinct central and peripheral mechanisms. These neurochemical changes would occur not in isolation but in concert with physiological, behavioral, and even experiential changes. In evolutionary terms, natural selection would have favored this patterning of stress responses by enhancing the protection and propagation of genes (4). We call this “primitive specificity.”
In line with the notion of stressor-specific alterations in central neurotransmission, Lachuer et al. (20) reported differential early time course activation of brain stem catecholaminergic groups in response to various stressors. Ceccatelli and Orazzo (2) reported increased expression of enkephalin mRNA and neurotensin mRNA in the paraventricular nucleus after ether exposure and after immobilization but not after swimming or exposure to cold; none of the stressors increased corticotropin-releasing hormone (CRH) mRNA expression. Romero et al. (28) reported stressor-specific responses of oxytocin, CRH, and vasopressin accumulation after colchicine blockade of axonal transport. Gaillet et al. (7) noted that the involvement of noradrenergic ascending pathways in stress-induced HPA activation depends on the type of stressor.
Possible limitations of present experiments.
Assessing the extent of experienced “distress” in laboratory animals is at best difficult. If one accepts that emotional distress increases plasma levels of Epi and ACTH, then prior experience with the preparation has indicated little if any distress in conscious, unrestrained animals 1 day after insertion of an indwelling arterial cannula. Studies over the past 5 years have involved preparations for stress testing and blood and microdialysate sampling beginning 24 h after surgical placement of catheters and a microdialysis probe (1). After this period of recovery, food and water intake and plasma levels of catecholamines, corticosterone, and ACTH averaged about the same to those in animals studied 48 h after surgery (34). In humans, arterial plasma concentrations of Epi and ACTH return postoperatively to preoperative values within 24 h of surgery, without further changes during the subsequent 2 days (35).
Hypothermia induced by pentobarbital sodium anesthesia and heat loss during catheter implantation probably did not affect responses of levels of NE, Epi, or ACTH during cold exposure beginning 22–24 h after the surgery. Body temperature was maintained by using a heating lamp throughout the surgery, and there was no evidence that by the time of the acute study the animals were hypothermic (36.8 ± 0.2°C). Moreover, arterial plasma levels of ACTH and catecholamines at 1 day after catheter implantation under halothane anesthesia (11) averaged about the same as at 1 day after catheter implantation under pentobarbital sodium anesthesia in the present study (60.8 ± 7.6 vs. 49.5.8 ± 5.9 pg/ml).
Future studies in this area should focus on further examination of the notion of stressor-specific patterns of central neurotransmitter release and regional neuronal activation and on altered responsiveness of organisms with specific genetic changes. Meanwhile, we hope that the present findings spur attempts to elaborate concepts that incorporate the specific and nonspecific elements of stress responses and that yield testable hypotheses.
The present findings emphasize the stressor specificity of neuroendocrine response patterns. Although Selye defined stress as the nonspecific response of the body to any demand, the present analysis has demonstrated the untestability of this idea, unless one accepts certain assumptions; yet given these assumptions, the doctrine of nonspecificity did not predict the present empirical results. According to an alternative “homeostatic” theory of stress (9), survival advantages afforded by multiple, shared effectors have led to the evolution of primitively specific generators for patterned neuroendocrine responses to different stressors. Studies relating neurochemical alterations in specific brain pathways with simultaneously monitored dependent neuroendocrine, physiological, and behavioral variables should enable identification of the components and elucidate the regulation of these patterns. Moreover, studies of appropriate animal models and human pedigrees should elucidate the genetic bases of these patterns and provide insights about the mechanisms underlying the predisposition to develop clinical stress-related disorders.
Address for reprint requests: D. S. Goldstein, National Institute of Neurological Disorders and Stroke, National Institutes of Health, 9000 Rockville Pike, Bldg. 10, Rm. 6N252, Bethesda, MD 20892–1424.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 1998 the American Physiological Society