The concept of cellular “fight-or-flight” reaction to stress

Michael S. Goligorsky


As animals respond to environmental stress with a set of default reactions described as the “fight-or-flight” response, so do epithelial and endothelial cells when they are confronting stressors in their microenvironment. This review will summarize a growing body of data suggesting the existence of a set of stereotypical cellular reactions to stress, provide some examples of diseases that are characterized by excessive flight reactions, describe the cellular mechanisms whereby the fight-or-flight reaction is accomplished, as well as cellular mechanisms triggering either fight or flight. It is proposed that cell-matrix adhesion is a sensitive indicator of the severity of stress. This indicator is interfaced with several default programs for cellular survival or death, thus dictating the fate of the cell. Some diagnostic and therapeutic applications of the concept, presently used and potentially useful, are outlined. The essential feature of this concept is its ability to categorize cellular events in terms of either type of default reaction, predict the details of each, and potentially exploit them clinically.

  • cell adhesion
  • migration
  • stress
  • preconditioning
  • endothelium
  • epithelium

 I have been trying to think of the earth as a kind of organism, but it is no go.... Then, satisfactorily for that moment, it came to me: it is most like a single cell....

 When we sense lipopolysaccharide, we are likely to turn on every defense at our disposal; we will bomb, defoliate, blockade, seal off, and destroy all the tissues in the area. All of this seems unnecessary, panic-driven.... The self-disintegration of the whole animal that follows a systemic injection can be interpreted as a well-intentioned but lethal error. The mechanism is itself quite a good one, when used with precision and restraint, admirably designed for coping with intrusion by a single bacterium: the hemocyte would be attracted to the site, extrude the coagulable protein, the microorganism would be entrapped and immobilized, and the thing would be finished. It is when confronted by the overwhelming signal of free molecules of endotoxin, evoking memories of vibrios in great numbers, that the limulus flies into panic, launches all his defenses at once, and destroys himself. Lewis Thomas, The Lives of a Cell

animals, when confronting a stressful situation, behave in a rather predictable way, which had been described in the classic work by Cannon (8) as a default set of reactions directed toward a “fight-or-flight” end point. Neurohormonal regulation of this behavioral response has been linked to the activation of neurons in the hypothalamus and brain stem, leading to the increased input to the sympathetic preganglionic neurons regulating cardiac and adrenal medullary functions (42). Selye (76) extended this view to disease stress and the hypothalamo-pituitary-adrenal cascade, demonstrating the duality of adaptive reactions. He wrote that “the first effect of a stressor acting upon the body is to produce a nonspecific stimulus,” which was identified as the humoral messenger, corticotropin-releasing factor. This, in turn, increases release of adrenocorticotropic hormone, followed by the production of corticosteroids and autonomic nervous system-induced release of catecholamines (76). The outcome of this cascade of reactions is both beneficial (modulation of energy metabolism, hemodynamics, immune and inflammatory reactions) and deleterious (exemplified by stress ulcers). The concept outlines the workings of fight-or-flight responses in the whole organism.

Recent years have provided significant, albeit scattered, evidence that epithelial, endothelial, and many other cells in the body respond to a variety of stressors in a reasonably standardized fashion, which allows them to combat the offending stimulus or escape it. Following I shall summarize the plethora of data obtained in diverse cells and propose a unifying view that the cellular response to stress falls into the category of fight-or-flight reactions. By attempting to conceptually unify these reactions and understand the mechanisms driving them, it may become possible to regulate the activity of each and the eventual outcome of the pathophysiological processes in which these reactions participate.

How should we define fight-or-flight reactions in cells? Fight reactions represent a set of cellular actions directed to attenuate the offending stimulus or to temporally and spatially limit its destructive actions. Examples of this type of reaction are production of natural antibiotics like defensins, secretion of IgE, induction of stress-response genes, production of glycolytic and antioxidant enzymes, and recruitment of polymorphonuclear leukocytes or macrophages, etc. The fight reaction has been evolutionarily selected for as an essential survival element in diverse species, from plants to animals. One of the particularly striking defensive elements is the ability to synthesize cationic antimicrobial 12–50 amino acid-long peptides, with >500 such peptides identified to date (32).

Flight reactions use cell motility to escape the sphere of influence exerted by noxious stimuli. These reactions are probably vestiges of unicellular organisms searching for the optimal environment, which became evolutionarily selected for and conserved in multicellular organisms. In metazoa, such reactions were preserved for the “purposes” of 1) exclusion of genetically damaged cells from the rest of the population and 2) migration toward the most favored microenvironment. Typical examples of these reactions are desquamation or exfoliation of cells, forming natural barriers to escape from the influence of a toxic agent or suboptimal microenvironmental conditions; but flight can also take place in cells escaping a hypoxic microenvironment, resulting in the metastatic spread of tumor cells. In fact, it has been proposed that survival of malignant cells and development of resistance to chemotherapeutic agents have inherent similarities with the SOS reaction of unicellular organisms, like bacteria (41).

On the basis of these definitions, it is obvious that the fight strategy most probably represents the modus operandi for the majority of cells; usually, this combat is successful, clinical manifestations of it are mute, and thus physicians are never called on. Simultaneously, a low level of flight is constantly occurring as a means for cell renewal. However, when the intensity of the insult transcends a threshold, default responses result in overt pathophysiological sequelae of fight-or-flight, clinical manifestations ensue, and medical attention is sought. These basic considerations are schematically outlined in Fig. 1. Several typical clinical situations, by and large related to the flight reaction, follow.

Fig. 1.

Schematic representation of cellular “fight-or-flight” reaction and its clinical manifestations. A cell is viewed as a vessel with 2 separate compartments, each responsible for the execution of fight-or-flight responses. These compartments are interconnected, and the intensity of cross-talk between them depends on the severity of stress.


Exfoliation of the Epithelium in the Respiratory Tract

This process has been described in epithelia all along the respiratory tree. In an experimental model of respiratory viral infections in neonates, 6-day-old rats infected with rat-adapted influenza virus exhibited mild erosions, inflammatory, exudative, and exfoliative changes of the nasal epithelia (16). In humans, a characteristic feature of bronchial asthma is represented by epithelial shedding with potential obstruction of the airways, a process that is mediated by platelet-activating factor chemoattraction and activation of eosinophils (95). Ciliated bronchial epithelium exposed to elevated levels of transforming growth factor-β1 (10–50 pM) showed cell loss through exfoliation and inhibition of proliferation without signs of stimulated differentiation, assessed as ciliogenesis (2). Nonciliated bronchiolar epithelial cells are exquisitely sensitive to naphthalene. In mice, naphthalene exposure results in an early permeabilization, blebbing, and exfoliation of terminal bronchiolar epithelial cells into airways (91).

Gastric and Duodenal Ulcers

The flight reaction is a prerequisite for ulcerogenesis in the gastrointestinal tract. For instance, Helicobacter pylori has been shown to colonize the regenerating epithelium, resulting in, among other sequelae, massive cell exfoliation and denudation of lamina propria (63). A similar result occurs when exfoliation is triggered by primary abnormalities in the vasculature. Injury to the gastric microvasculature or severe vasoconstriction induced by endothelin-1 can cause almost complete exfoliation of the interpit cells and apoptosis of superficial cells of gastric mucosa, with the eventual formation of ulcers (84). In a clinically important model, exposing gastric mucosa to aspirin results in exfoliation of surface epithelium and deep mucosal necrosis, which is preceded by microvascular injury consisting of rupture of capillary walls, necrosis of the endothelium, deposition of fibrin, and platelet adhesion (87). Similar findings were reported after intragastric administration of ethanol in healthy volunteers (88). Exfoliation of surface epithelial cells and a drop in gastric mucosal blood flow have also been observed after application of bile acid (taurocholate, pH 1.2), and both phenomena were reduced by intra-arterial infusion of sodium bicarbonate in experimental dogs (85). Pepsin induces changes in glycoproteins of gastric mucus, degrades the ultrastructure of the mucus layer, and leads to exfoliation of the gastric epithelium (44). Although ischemia-reperfusion-induced gastric mucosal damage is characterized by exfoliation of surface epithelial cells and increased vascular permeability, both are reduced by pretreatment with epidermal growth factor (40). Interestingly, activation of heat shock proteins (Hsp) 90, 70, and 60 induces resistance of gastric mucosa against ethanol-induced exfoliation of epithelial cells (35), as will be discussed later.

Endothelial Damage and Denudation

Endothelial denudation is a frequent companion of diverse cardiovascular diseases. Patients with diabetes mellitus, glomerulonephritis, or preexisting cardiovascular diseases, including hypertension, display various degrees of spontaneous exfoliation of endothelial cells, as detected with scanning electron microscopy of sampled microvessels (80). In experimental rabbits, oral or intravenous administration of oxidized cholesterol, but not cholesterol free of oxidation products, led to rapid unltrastructural alterations in vascular endothelial cells: vacuolization, subendothelial edema, and subsequent endothelial denudation (5). In a model of decompression disorder, exposure of healthy salmon to gas-supersaturated water resulted in vascular lesions characterized by intravascular formation of gas bubbles, vasodilation of dermal blood vessels, and endothelial denudation (83). Glomerular capillary endothelial injury induced by infusion with cationized ferritin results, within a few hours, in the detachment of endothelial cells and podocytes from the glomerular basement membrane, leading to glomerular obsolescence within 7 days (48).

Exfoliation of Tubular Epithelial Cells in Renal Ischemia

Renal tubular epithelia respond to stressors with a reversal of the polarity of integrin receptors from the predominantly basolateral location to the apical cell membrane (25) and loss of anchorage to the basement membrane and cell desquamation. The expression of functionally competent integrin receptors on the apical cell membrane may lead to promiscuous interactions, e.g., the adhesion of desquamated cells to the cells remaining in situ, thus initiating the process of tubular obstruction; conglomeration of the desquamated cells via integrin receptors further aggravates tubular obstruction (30). Importantly, these integrin-based interactions can be blocked by synthetic Arg-Gly-Asp (RGD) peptides (59,60). Gly-Arg-Gly-Asp-Ser-Pro peptide injected into the renal artery on restoration of blood flow to the kidney prevented the elevation of proximal tubular hydrostatic pressure, which was uniformly and characteristically detected in animals with renal ischemia receiving a vehicle or an inactive peptide. In vivo studies of RGD peptides in ischemic acute renal failure in rats demonstrated attenuation of acute renal failure and accelerated recovery of renal function. Using linear RGD peptide labeled with 99mTc, we have shown that this probe was retained in ischemic kidneys (60), a finding consistent with the increased availability of unoccupied RGD-recognizing integrins. To visualize RGD binding sites at the cellular level, we next performed their mapping by using fluorescent derivatives of two RGD peptides, a cBt-RGD peptide and a linear rhodamine G-RGD peptide (74). Application of both fluorescent probes, despite structural differences outside the RGD sequence, brought about similar results. These probes revealed RGD binding sites along ischemic tubules, on the luminal surface of in situ epithelium, and on the plasma membrane of desquamated epithelial cells. This distribution of RGD binding sites is consistent with the model of a direct action of systemically injected RGD peptides on the exposed unoccupied epithelial integrins. Engagement of these previously unoccupied integrins deters them from interacting with desquamated cells, thus resulting in prevention of tubular obstruction. In addition, RGD binding sites were detected on the endothelial cells lining blood vessels, suggesting a role of RGD-recognizing integrins in extravasation of inflammatory cells. These studies not only confirm a sizable body of observations on the diagnostic significance of desquamated tubular epithelial cells (see below) but also provide mechanistic insights into its pathophysiology and targets for prevention.

Peritoneal Membrane Dysfunction

The efficacy and selectivity of dialysis across the peritoneal membrane can be compromised when the membrane loses the integrity of the mesothelium. This complication of long-term peritoneal dialysis, characterized by exfoliation of mesothelial cells, is not infrequently observed in patients, especially those overusing high-glucose-containing dialysate prescriptions (94).

Corneal Exfoliation

Corneal epithelium normally moves centripetally to undergo eventual shedding from the central apex due to the shearing from the upper lid. This process is distorted in patients with keratoconjunctivitis sicca, neurotrophic keratitis, or in those using contact lenses (51). Red eye syndrome in patients with chronic renal failure has been attributed to the deposition of calcificates, leading to erosion of the corneal epithelium (47).


Mobilization of mature germ cells represents another example of the physiological exfoliation in seminiferous tubules. Aging is accompanied by premature exfoliation of spematids and spermatocytes, and, eventually, maturation arrest reaches spermatogonia, with the subsequent involution and sclerosis of seminiferous tubules (65). Under pathological conditions, increased desquamation of immature germ cells is frequently accompanied by asthenozoospermia and teratozoospermia (81).

Urinary Tract Infections

Uropathogenic strains of Escherichia coliexpress type 1 pili, which attach the organism to the uroepithelial membrane glycoproteins and uroplakins, resulting in a rapid apoptosis of host cells and their exfoliation (57), with concurrent inflammation and deranged detrusor function.

Staphylococcal Scalded-Skin Syndrome (Epidermal Exfoliation)

Staphylococcus aureus produces epidermolytic toxins that cause this ominous blistering skin disorder predominantly affecting children. The hallmark of the disease is a lesion to the zona granulosa of the epidermis, producing at times massive epidermal loss (reviewed in Ref. 49).

Exfoliation of Enterocytes

Salmonella typhimurium interacts with the follicle-associated epithelium of the ileum, which rapidly engulfs it (minutes), resulting in exfoliation (30 min-2 h) of enterocytes containing bacteria, thus providing a clearance ofSalmonella (24). Similarly, Clostridium difficile toxins B and A target human colonic enterocytes, causing patchy damage to the integrity of the epithelial layer with exfoliation of superficial cells but not crypt epithelium (73). It has been noted that Bcl-2-knockout mice display defects in the integrity of the small intestine characterized by accelerated exfoliation of epithelial cells and decreased numbers of mitotic progenitor cells (43).

Exfoliative Cheilitis

Up to 30% of autoimmune deficiency syndrome patients develop this complication, predominantly on the lower lip, which is due toCandida albicans in more than half of cases (72). Scale formation and exfoliation predispose these immunocompromised patients to generalized infections.

In conclusion, although there are some obvious cell type-dependent and offender type-dependent variations in the repertoire of cellular fight reactions (these are by and large outside the focus of this review), the flight reaction is a ubiquitous process characterized by various degrees of cell detachment from its surrounding microenvironment. The actual mechanisms of this reaction may, however, differ from cell to cell, as discussed below.


One of the enzymes almost uniformly activated by stress is represented by the c-Jun NH2-terminal kinases (JNKs) encoded by three genes and is also known as the stress-activated protein kinases (SAPK). These enzymes are activated by MAP kinase kinases (MKK 4 and MKK 7) and phosphorylation of specific threonine and tyrosine residues and inactivated by Ser/Thr, Tyr, and possibly dual-specificity Ser/Thr/Tyr MAP kinase phosphatases (reviewed in Ref.37), as illustrated in Fig.2. It has been recognized, however, that there are variations in the programs unveiled by cells subjected to different stressors. JNK or SAPK is rapidly activated by stressors such as oxidative insult or cytokines (i.e., tumor necrosis factor-α) via a process that is independent of caspases and results in activation of nuclear factor (NF)-κB and antiapoptotic signaling. This reaction belongs to the category of fight responses. However, this process is contrasted by a slower, Fas ligand-induced activation of SAPK, which requires cleavage by activated caspase 3 and cellular redistribution of MAP kinase kinase/extracellular signal-regulated kinase (Erk) [MEK1 kinase (MEKKI)], thus prestaging apoptotic cell death (14). The latter process could be inhibited, at least in Jurkat T cells, by signals activating Erk (e.g., phorbol esters and MEK1) that, acting as response modifiers, rescue these cells from Fas-induced apoptosis and thus switch the reaction to the fight mode (93).

Fig. 2.

General schema of mitogen-activated protein (MAP) kinase signaling pathways induced by stress. Diverse cytokines and stressors elicit activation of MAP kinase (MAPK) kinases via several intermediaries, like βγ-subunits of G proteins, Ras- and Rho-GTPases, c-Abl tyrosine kinase, and PI 3-kinase (PI-3-K); elevation of cytosolic calcium, in turn acting through calmodulin kinase and protein kinase PYK-2; and changes in redox potential of intracellular glutathione. There is at least partial specificity of MAPK kinases, such that MKK3, MKK6, and MKK4 activate p38, whereas MKK4 and MKK7 are specific for c-Jun NH2-terminal kinase (JNK) activation. JNK and p38 further activate (by phosphorylation or translocation) several transcription factors that enable this system to participate in inflammatory responses (p38 and, in some cases, JNK pathway) and pro- and antiapoptotic cell responses.

Offenders that act predominantly through disruption of endoplasmic reticulum (ER) function (typical ER stressors are thapsigargin, which depletes luminal calcium, tunicamycin, which blocks protein glycation, and dithiothreitol, which interferes with formation of disulfide bonds) achieve activation of SAPK by oligomerization and phosphorylation of IRE1p (45, 79, 90). A recently cloned gene, PERK, encodes a protein in the ER, which contains a luminal IRE1-like domain and a cytoplasmic domain strongly resembling the eukaryotic initiation factor-2α (eIF-2α) kinases (33). ER stress activates PERK kinase, which in turn serine phosphorylates eIF-2α, leading to the inhibition of mRNA translation.

In the process referred to as anoikis, namely, apoptotic cell death induced by the loss of cell-matrix adhesion, JNKs are activated in a caspase-dependent fashion (22, 56). One of the substrates for the Asp-Glu-Val-Asp motif-specific caspase is represented by MEKK1. The flight reaction, manifesting in the loss of cell-matrix adhesion, results in a cleavage of MEKK1, its activation, and stimulation of apoptosis by a cleavage product (9). It is important to emphasize that resistance to anoikis can be induced by activating integrin signaling through pp125FAK (23), inhibiting caspase activity (21), or expressing a cleavage-resistant mutant of MEKK1 (9). Caspases can also cleave pp125FAK in detached cells (see below).

Stressors leading to DNA damage unveil three major programs: DNA repair mechanisms, a halt of cell cycle progression to prevent cells with damaged DNA from replicating, and activation of programmed cell death. The actual mechanisms executing each of these functions are beyond the scope of this review. Suffice it to mention that the ataxia telangiectasia mutant (ATM) family of kinases is the major regulator of cell responses, especially those related to the regulation of cell cycle checkpoints (reviewed in Ref. 92). Activation of p53 proceeds in an ATM-dependent but also ATM-independent (i.e., ultraviolet light) manner. Another ATM-dependent protein tyrosine kinase affected by DNA damage is c-Abl, which regulates tyrosine phosphorylation of the catalytic subunit of RNA polymerase II. It is important to emphasize that the activity of c-Abl is also regulated by cell-matrix adhesion: it is active in attached cells but inactivates when cells detach from the matrix (53). These data raise the possibility that integrin-c-Abl link may serve important functions in determining cell fate when cell-matrix adhesion becomes compromised.

Acquisition of resistance to stressors, referred to as “preconditioning,” is emerging as an important phenomenon with therapeutic potential. Earlier work by Selye (76) considered exogenous and endogenous conditioning factors to be responsible for individual variations in the level of the local adaptation syndrome induced by “qualitatively different agents of equal toxicity” or “even the same degree of stress, induced by the same agent, producing different lesions in different individuals.” It has been demonstrated that various Hsps can increase the threshold for injury. A few specific examples are presented. Resistance to ischemic injury has been linked to Hsp10, 27, 60, 70, 72, 90, and crystallin; resistance to oxidant stress with hydrogen peroxide is afforded by Hsp27, Hsp70, and crystalline, whereas resistance to ultraviolet irradiation depends on the induction of Hsp27 and 70 (reviewed in Ref. 18). It has been demonstrated that ischemic preconditioning requires activation of the Ras-Raf-Mek-Erk cascade and is dependent on the ability of cells to generate nitric oxide and induce Hsp70, Hsp72, and Bcl-2 (53). However, precise molecular mechanisms responsible for Hsp-induced preconditioning await elucidation. Similarly, studies into mechanisms of developing drug resistance, a phenomenon of tumor cell preconditioning by specific chemotherapeutic agents, are needed (89).


What kind of message(s) forces a cell to reprogram itself for the flight reaction? Recent years have brought about some intriguing discoveries. On the one hand, cells that have sustained and accumulated a critical level of DNA damage are destined to collapse their focal contacts with the surrounding matrix proteins and undergo apoptotic death (reviewed in Ref. 92). Stress-induced changes in cell-matrix adhesion initiate the flight reaction, which may, eventually, result in anoikis (22, 56). Here, cell adhesion to its matrix serves as a sensor of the severity of stress, triggering cell exfoliation and suicide. Both reactions appear to be “altruistic” in the sense that they protect the organism from the accumulation of genetically compromised cells.

Stress-induced changes in cell-matrix adhesion occur via several routes that invariably lead to the disassembly of the focal adhesion complexes (matrix protein-integrin receptor-recruited adaptor proteins of the focal adhesion-cytoskeleton axis). The site along this axis, where initial disassembly occurs and the cellular mechanism employed in the process may have profound significance, is illustrated below. It has been demonstrated that the calcium-dependent protease II, calpain, is localized in adhesion plaques in epithelial cells and fibroblasts (4). One of the substrates for calpain, talin, a component of adhesion plaques, is colocalized with the protease. It has been proposed that elevation of cytosolic calcium concentration activates calpain, which in turn leads to the irreversible proteolytic cleavage of talin and reversible disassembly of focal adhesion sites.

Another scenario is initiated by matrix metalloproteases (MMP), which digest matrix proteins and trigger outside-in signaling. MMP-2 digestion of laminin 5 generates a γ2-chain fragment, which promotes cell migration much more than the native matrix protein (26). Similarly, thrombin-cleaved osteopontin possesses promigratory properties (78). A rapid disassembly of focal adhesions in smooth muscle cells has been shown to be initiated by collagenase-degraded collagen fragments (10). The process is driven by fragment-induced activation of α2-integrins and calpain I, resulting in proteolytic cleavage of p125FAK, paxillin, and talin, which leads to the loss of focal adhesions and cell rounding. This process can be prevented by calpain I inhibitors, but inhibitors of caspase-3, another enzyme with a potential to cleave components of focal adhesions (see below), appear to be ineffective. Notably, hypoxia has been shown to activate calpain I degradation of NF-κB inhibitor IκBα with the activation of NF-κB (98). In contrast to this scenario, endothelial cell detachment, caused by the withdrawal of growth factors, is initiated by activated caspase-3, which cleaves p125FAK to generate a COOH-terminal fragment exhibiting structural and functional similarity with the FAK-related nonkinase FRNK, a competitive inhibitor of assembly of focal adhesions (52). Hence, under these circumstances, the same process of cell detachment is driven by the proapoptotic proteolytic cascade.

ADAMs, a growing family of extracellular matrix proteins, which can also exist on the cell surface, express both disintegrin and MMP domains. A member of this family, Kuzbanian, has been shown to cleave a ligand from the plasma membrane, Delta, which then interacts with a cell surface receptor, Notch, critical in determining cell fate (68).

Yet another proapoptotic mechanism is initiated at the same boundary but utilizing proteolytic cleavage of extracellular matrix proteins. Angiostatin, an NH2-terminal fragment of plasminogen released by MMP-12, is a potent antiangiogenic factor that inhibits endothelial cell proliferation and induces apoptosis (13). Two other products of proteolytic cleavage of collagen XVIII and XV, endostatin and restin, respectively, exhibit antiproliferative and proapoptotic properties (15, 62,71).

Activation of protein tyrosine phosphatases (PTPs) provides yet another route for disruption of focal adhesions. Well-established substrates of these phosphatases are present in focal adhesions and include p125FAK, p130CAS, and paxillin (1). One of these PTPs, transmembrane LAR, has been shown to localize at focal adhesions, another receptor-like PTP-α regulates Src activity and cell-matrix adhesion, whereas a bacterial PTP, Yersinia YopH, has been shown to dephosphorylate p130CAS and p125FAK and destabilize cell-matrix adhesion (6). A ubiquitously expressed PTP containing a Src homology-2 domain, SHP-2, is stimulated by platelet-derived growth factor to dephosphorylate p125FAK, stimulating cell migration (69). It is important to bear in mind that oxidant stress and changes in redox state are believed to inhibit PTPs, all of which express a vulnerable cysteine residue that is essential for enzyme activity (19). An important addition to this family was the recognition of a tumor suppressor bifunctional phosphatase, PTEN, which induces FAK and Shc dephosphorylation, thus inhibiting cell proliferation and migration, as well as phosphatidylinositol 3,4,5-trisphosphate, thereby suppressing protein kinase B (AKT) and promoting apoptosis (86). It has recently been demonstrated that the main target for PTPs is regulation of RhoA activity: PTPs prevent guanine nucleotide exchange factor activation of RhoA and thus depress tensile forces and suppress formation of focal adhesions (reviewed in Ref. 75).

An important role, which is not yet fully understood, in disassembly of focal adhesions and acquisition by the cell of a flight phenotype belongs to nitric oxide (NO). Inducible NO synthase (NOS) is a universal companion of diverse stress conditions, such as hypoxia and cytokines (77). Induction of this enzyme results in a protracted generation of large amounts of NO (64), which is accompanied by the loss of focal adhesions, cell rounding, and detachment (29). This process is mediated by a rapid cycle of phosphorylation-dephosphorylation of p125FAK and paxillin, a phenomenon compatible with our electrophysiological demonstration of increased turnover of focal adhesions, which was termed “podokinesis” (61). Although p125FAK plays an important role in assembly of focal adhesions and their turnover, inhibition of p125FAK signaling, either in cells obtained from mice with targeted deletion of the gene (37) or in cells loaded with a fusion protein containing p125FAK-targeting domain, leading to the displacement of endogenous p125FAK (27), produced a phenotype that displayed enhanced focal adhesions and decreased migration rate. This was precisely the phenotype of endothelial cells pretreated with a NOS inhibitor nitro-l-arginine methyl ester, to mimic endothelial dysfunction: they exhibited reduced immunolocalization of p125FAK to focal contacts and suppressed motility.

Another important observation is related to the role of NO in tyrosine phosphorylation of focal adhesion components. In vitro studies of a low-molecular-weight phosphotyrosine protein phosphatase, previously known as acid phosphatase, demonstrated that NO-generating compounds inhibit the activity of the enzyme (11). Tyrosine phosphorylation of focal adhesion-associated proteins like p125FAK or paxillin has been shown to be dependent on the integrity of the actin cytoskeleton. Disruption of the actin cytoskeleton by cytochalasin D causes a profound and selective inhibition of p125FAK and paxillin tyrosine phosphorylation (82, 97). Tyrosine phosphorylation of these proteins has been suggested to stabilize interactions between focal adhesions and the cytoskeleton (7,96), whereas their destabilization is attributed to decreased tyrosine phosphorylation of focal adhesion components (54). In this context, the phenotype of human umbilical vein endothelial cells subjected to increased concentrations of NO is that of a cell with reduced tyrosine phosphorylation of p125FAK (after the initial brief elevation of tyrosine phosphorylation), increased turnover of focal adhesions, and the propensity for migration. These observations are consistent with the proposed role of NO in destabilization of contacts between focal adhesions and the cytoskeleton, thus releasing the cytoskeleton from focal adhesion complexes.

In human umbilical vein endothelial cells and Swiss 3T3 cells treated with NO-generating compounds, we observed areas containing elements of focal adhesion complexes bound to the extracellular matrix devoid of cytoskeletal structures, cell ghosts. The phenomenon of “ripping” of integrin receptors with attached focal adhesion proteins, but not the cytoskeleton, from a trailing edge of fibroblasts has been proposed as an important mechanism for releasing the trailing edge of the cell in the process of locomotion (35, 48). Similarly, Nakamura et al. (58) have observed a “peeling-off” phenomenon in endothelial and other cell types subjected to the stimulated neutrophils or the neutrophil-derived oxidant NH2Cl. Specifically, the inference to be made on the basis of data presented above ascribes to NO the ability to facilitate the disassembly of focal adhesion-stress fiber complexes. Although a low-output NO synthesis should be sufficient to accelerate cell migration via this action at the leading and trailing edges, the high-output NO release, as occurs when inducible NOS is upregulated, may eventually lead to cell detachment. Therefore, we hypothesize that the function of endogenous NO production is to seek the optimal set point on the bell-shaped curve, correlating tightness of adhesion with rate of migration (Fig.3). In this situation, the ability of a cell to modulate NO generation serves a role of a tuning fork that adjusts cell-matrix adhesion to the specific environment, thus determining the need for fight or flight. Is this a physiological or a pathophysiological effect of NO? Probably, two extreme situations are relevant to the observed effect of NO on focal adhesions. The first is exemplified by the states of profound inhibition of NOS, when endothelial cells may attain a nonmotile phenotype, similar to that observed in p125FAK-deficient cells (27, 37), thus explaining the reduced rate of wound healing and angiogenesis. The second is related to the conditions of NO overproduction, which may result in endothelial retraction and denudation of the vascular wall, as characteristically seen in endotoxemia (55).

Fig. 3.

Cell-matrix adhesion as an indicator of cell fate. Loss of cell adhesion reflects the severity of cell injury and, in turn, predisposes cell to anoikis. Diverse mechanisms promoting cell detachment are presented in blue. This detached phenotype is accompanied by the defective DNA repair. On the other hand, growth factors, protein kinase B (AKT), and phosphorylation of focal adhesion kinase (pp125FAK) prevent cell detachment. Mechanisms supporting cell-matrix adhesion are presented in green. Adhered phenotype is characterized by competent mechanisms of DNA repair. Migratory phenotype denotes a transitional state of cells expressing the “flight” reaction. This graphic emphasizes the property of such a flight reaction to either succumb to injury or combat it (apoptosis and survival, respectively). NO, nitric oxide; PTPs, protein tyrosine phosphatases; GF, growth factor; RTK, receptor tyrosine kinase; GPCR, G protein-coupled receptor; FRNK, truncated form of pp125FAK; PTEN, tumor suppressor protein; ILK, integrin-linked kinase; MMPs, matrix metalloproteases.

Loosening and loss of cell-matrix adhesion appear to represent a critical step in cell fate: would the cell succumb to anoikis, become vagrant with chances of survival remaining unanchored, or be capable of reassembling cell-matrix contacts elsewhere? Ingber's laboratory (12, 38, 67) has provided compelling evidence that cell-matrix adhesive interactions and developing tensile forces are a prerequisite for cell survival. Nonetheless, some cells, like circulating metastatic cells, escape the death pathway and acquire anchorage independence. Furthermore, there is growing evidence for the existence of a circulating pool of endothelial cells or their precursors (reviewed in Ref. 70); it is assumed that these cells can attach to the basement membrane at areas of denudation and participate in angiogenesis.


Although fight responses have been used for clinical diagnostic purposes for many decades (eosinophilia, lymphopenia), the diagnostic value of flight responses is only beginning to emerge. These responses can be exploited for the purposes of establishing the very fact of cellular stress and even identifying the nature of an offender without actually examining an offended tissue in situ. Flight responses provide excellent examples of such distanced diagnostics. For instance, rejection of human renal allografts is accompanied by the desquamation of renal tubular epithelial cells from different nephron segments and their appearance in the urine, together with lymphocytes. It has been demonstrated, using the Papanicolaou staining of cytocentrifuge preparations of urine samples, that detachment of collecting duct-specific epithelial cells at numbers exceeding 20 cells/10 high-power fields is a more sensitive and specific criterion of acute rejection than lymphocyturia at its optimal diagnostic level of 13 cells/10 high-power fields (17). This procedure resulted in establishment of correct diagnosis in 64% of rejection episodes later diagnosed and confirmed clinically. In contrast, cyclosporine nephrotoxicity results in the desquamation of epithelial cells originating predominantly in the proximal convoluted tubule, which are found in the urine in excess of collecting duct cells. Furthermore, the finding that renal allograft rejection is accompanied by urothelial exfoliation in 54% of cases raises the possibility of using these cells as diagnostic markers (44). In addition, urine cytology can be utilized as a means of monitoring the efficacy of therapy in acute rejection. Rejections responsive to solumedrol are accompanied by a gradual decrease in the number of cells in the urine, in contrast to those with no response to therapy. On the other hand, OKT3 therapy, in cases that were later categorized as successful, leads to the increased excretion of necrotic cells, debris, lymphocytes, and collecting duct epithelia, whereas rejections unresponsive to OKT3 administration show no interval changes in urine cytology.

In fact, biological fluids represent a rich source for diagnostics. Testicular carcinoma in situ can be diagnosed noninvasively by examining seminal cytospin smears with in situ hybridization assay with a probe for chromosome 1 detecting >2.6% hyperploid, exfoliated germ cells (0.2% in control) (28). Cytology of peritoneal lavage has been used to diagnose exfoliated malignant cells in patients with colorectal cancer or ovarian carcinoma (20, 34). Cytological examination of breast secretions has been advocated for diagnosing breast cancer, duct papilloma, cystic fibrosis, and inflammatory lesions (3). Moreover, discovery of exfoliated cancer cells in pleural lavage in patients with adenocarcinoma (before any manipulations on the lung) has been found to be associated with a four times lower survival rate compared with those with negative cytology (64).


The data presented above provide a reasonable foundation for a hypothesis on the existence of fight-or-flight reaction at the cellular level, demonstrate the broad applicability of this concept, both for understanding cell stress responses and their diagnostic monitoring, and unify an array of pathophysiological observations made in different organ systems. The canvas displayed, however, has a multitude of white or only lightly sketched areas. To complete the painting, the following problems need to be addressed in the future.

1) It is expected, due to the inherent heterogeneity of epithelia and endothelia, that fight-or-flight responses should exhibit certain specificity, depending on the organ or the stressor. This field needs to be examined in much greater detail.

2) It is becoming clear that fight-or-flight responses described above in fact represent different gradations of fight-and-flight reaction, whereby, depending on the biological context and local modifiers, one or another component of the reaction predominates.

3) Hence, if both reactions are not mutually exclusive, what are the criteria that cells use in making the decision as to fight or flight? Furthermore, is it possible that such a decision can be reversed and an alternative route taken, or once a cell enters each program does it become committed to its completion?

4) It is necessary to find detailed mechanisms for different routes of cell-matrix and cell-cell detachment and construct maps of such routes, their respective checkpoints, and sites where cells could exit each route.

5) The cross-talk relationship between integrin and growth factor receptor occupancy appears to play a substantial role in rescuing cells from anoikis. Therefore, the details of this process await elucidation.

Answers to these and many other questions should benefit vexing biological problems related to the pathophysiology of ischemic and toxic injuries, metastatic processes, and inflammatory and metabolic diseases, to name a few. In addition, diagnostic studies of cells examined in a flight mode should benefit from the deeper understanding of processes that led to it.

In conclusion, this review draws together a number of observations, which support a general idea that cell-matrix adhesion is a sensitive indicator of the severity of stress. This indicator is interfaced with several default programs for cellular survival or death, thus dictating the fate of the cell. The remaining big problem is: How can we tune this indicator up or down, not perturb the basic wiring, and gain access to the decision making as to fight or flight? And there is another problem. By tinkering with these vital processes, how can we avoid a potential selection and propagation of damaged cells; or in other words, what is the safety diapason for pharmacological regulation of fight-or-flight processes?


I am obliged to Dr. D. Mynarcik (Dept. of Medicine, SUNY Stony Brook) for a critical reading of the manuscript and to all former and present postdoctoral fellows for contributions.


  • Studies performed in my laboratory, referred to in this review, were supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-45695, DK-52783, and DK-45462.

  • Address for reprint requests and other correspondence: M. Goligorsky, Dept. of Medicine, SUNY Stony Brook, Stony Brook, NY 11794-8152 (E-mail: mgoligorsky{at}


View Abstract