Fatigue during heavy exercise is generally explained in terms of a dysfunction within the muscle itself, including excessive accumulation of metabolites and/or depletion of energy substrates (10). The central nervous system has also been implicated in exercise fatigue resulting in diminished neural drive to the muscle owing to impaired firing frequency and motor unit recruitment, and/or lack of motivation (7).
Infection and other illnesses are also often accompanied by fatigue, malaise, and reduced performance, but the mechanisms are understudied and largely unexplained (12,17,27). Studies have found reductions in isometric strength and run time to fatigue on a treadmill in men infected with sandfly fever virus (6,11,12). Common markers of peripheral fatigue could not explain fatigue in these studies, and therefore it was suggested that central nervous system mechanisms might be involved. Interest in the mechanisms of fatigue during illness has increased with the recent characterization of a syndrome of chronic debilitating fatigue (chronic fatigue syndrome), which is believed to involve immune system dysfunction (27). The presence of elevated cytokines in conjunction with chronic low grade infection is among some of the proposed mediators of this condition, but specific factors involved in fatigue have not been determined (17,21,27).
Recent research involving possible mediators of fatigue during infection has focused on several cytokines released from immune cells during infection. In particular, a set of similar substances known as Interferon-alpha, -beta, and -gamma (IFNα/β/γ) has attracted much interest. These substances are glycoproteins capable of inhibiting viral replication and regulating the immune response to pathogens (29). They are thought to be responsible for the fatigue experienced during influenza (30,31,33) and perhaps in chronic fatigue syndrome (20,27), but the experimental evidence is extremely limited. Thus far, the evidence is limited to the observations that cancer patients receiving IFN therapy report excessive fatigue and malaise (1,30,33) and that rodents decrease voluntary activity in response to IFN administration (31).
There have been no studies of the influence of IFN on fatigue during motivated treadmill exercise. Clearly, a distinction needs to be made between fatigue that might be better defined as malaise versus fatigue that occurs during motivated treadmill exercise. Fatigue during treadmill running in rodents is typically defined as the inability/unwillingness to maintain a prescribed power output (e.g., treadmill speed) that may involve both central nervous system as well as peripheral mechanisms (7). This study was designed to determine the effects of IFN-α/β on exercise time to fatigue using a standard treadmill exercise protocol in mice (36). In this study, endogenous IFN-α/β was induced by the injection of a synthetic polynucleotide (RNA), poly I:C (polyriboinosinic acid: polyribocytidylic acid), a known inducer of IFN-α/β (9). A specific role for IFN-α/β would be indicated if pI:C administration caused early fatigue during treadmill running and if this effect was blocked by immunoneutralization of endogenously produced IFN by the injection of a specific anti-IFN-α/β antibody.
Mice. Pathogen-free male CD-1 mice (Charles River) 6-8 wk of age were used in this study. All mice were housed in conventional cages with filter bonnet tops and were maintained in a facility that observed a 12-h light/dark cycle at 22°C, with 50% RH. Animals were given food (Purina Chow) and water ad libitum.
Reagents. The medium used in all assays was RPMI-1640 (GIBCO, Grand Island, NY) containing penicillin (100 U·mL−1), streptomycin (100 μgAmL1, glutamine (20 mM), and 10% fetal bovine serum (Environmental Diagnostics, Burlington, NC). pI:C, gamma irradiated, was purchased from Sigma (St Louis, MO). Vesicular stomatis virus (VSV) used for the IFN bioassay was generated from stock virus. Rabbit polyclonal antimouse IFN-α/β IgG was purchased from Lee BioMolecular Laboratories, Inc. (San Diego, CA). This antibody has less than 0.02% cross-reactivity against heterologous interferon species and is highly specific for IFN-α/β.
Exercise protocol. This protocol was approved by this University's Institutional Animal Care and Use Committee and conforms to the policy statement of the American College of Sports Medicine on research with experimental animals. Mice underwent treadmill acclimation for 4 d before the actual experiments. This was intended to adapt mice to investigator handling and to allow for the identification of poor runners. Acclimation consisted of 15 min of treadmill running beginning at 17 m·min−1 and gradually increasing the speed every 5 min up to 20.7 m·min−1. Exercise consisted of treadmill running to fatigue at 18.7-24 m·min−1, 5% grade, 12 or 24 h after injection of pI:C, or saline. Fatigue was defined as the point at which mice continually refused to run despite being given mild physical prodding continuously for 3 min. Electrical shock was not used as a negative reinforcement, as this adds undue extraneous stress that is typically not associated with volitional exercise to fatigue.
Treatments. The first experiment (EXP 1) was designed to examine the effects of pI:C on plasma IFN-α/β and forced exercise time to fatigue. Mice (N = 10/group) were randomly assigned to one of two groups, pI:C or saline (CON). Vials of the two treatments were prepared and coded. The investigator, blind to the treatments, injected i.p. 0.2 mL of either pI:C (5 mg·kg−1) or an equal volume of saline into the two treatment groups. Mice then underwent the exercise protocol either 12 or 24 h after the injection. These time points were chosen because it had not been determined at which time, in relation to IFN production, fatigue may be elicited. Thirty minutes before exercise, 70 μL of blood was collected in a microhematocrit tube from the retro-orbital sinus of each mouse for the analysis of IFN titers. Each animal was removed from the treadmill at the point of fatigue, anesthetized in a small bell jar containing halothane, and exsanguinated via cardiac puncture. Blood was also collected at this time point for the analysis of IFN titer.
The second experiment (EXP 2) was designed to examine the effect of anti-IFN-α/β antibody on run time to fatigue following pI:C administration. Mice (N = 24) were randomly assigned to one of three groups: control (CON), pI:C, or pI:C + rabbit anti-IFN-α/β Ab (pI:C + Ab). The two pI:C groups received 0.2-mL injections of pI:C (5 mg·kg−1), and the control group was injected with an equal volume of nonimmune IgG. The pI:C + Ab group also received the antibody to IFN-α/β (1000 neutralizing U·dose−1) intraperitoneally at times 0, 4, and 8 h after receiving a pI:C injection for a total of 3000 neutralizing U. In this investigation the time at which exercise began was 24 h posttreatment injection. Thirty minutes before exercise time (24 h), 70 μL of blood was drawn from each mouse by retro-oribital puncture using a microhematocrit tube. A second investigator, unaware of the treatments, removed each animal from the treadmill at the point of fatigue. All animals were anesthetized in a small bell jar containing halothane and exsanguinated via cardiac puncture.
VSV titration. L929 cells, from a murine fibroblast cell line, were seeded in 96-well culture plates at 2.0 × 104 cells/well and incubated for 24 h. Serial 10-fold dilutions of virus (10 to 10−6) were made in RPMI containing 2% penicillin/streptomycin. A 0.25-mL suspension of virus was added to each well. The plates were allowed to incubate at 37°C in 5% CO2 for 48 h or until significant cytopathic effect could be visually detected. The neutral red assay was used to quantify percent viability.
Interferon bioassay. Blood taken from each group of mice 30 min before exercise via retro-orbital bleeding was pooled (pI:C, control, or pI:C + rabbit antimouse IFN-α/β Ab) in a collection tube and spun in a centrifuge at 2,000 rpm for 20 min at 4°C. Serum was aliquoted and stored in cryo-vials at −70°C until use. Blood samples taken postexercise via cardiac puncture underwent the same processing but were stored individually.
L929 cells were seeded in a 96-well microtiter plate at a concentration of 4 × 104 cells per well (0.2 mL). Plates were incubated overnight at 37°C in 5% CO2 to allow cell adherence. Ten-fold dilutions of mouse serum ranging from 10−1 to 10−5 were tested. The medium was removed by aspiration from the plate, and the diluted serum samples (0.2 mL) were added to the L929 cells in duplicate. The cells were incubated overnight at 37°C in 5% CO2. Along with the serum sample, a known IFN standard (3.2 × 106 IU/mL, NIH) was run. The fluid was then removed, and 0.1 mL of an optimal dilution of VSV was added. After 1 h at 37°C the nonabsorbed virus was removed, 200 μL of 2% CRPMI was added to each well, and the plates were incubated for 48 h at 37°C in 5% CO2. Cytopathic effect (CPE) was determined by a neutral red dye uptake assay.
Neutral red uptake assay. The neutral red assay was used to quantify VSV viral titrations and interferon levels in serum samples infected with VSV. IFN titers were expressed as the reciprocal of the dilution that caused a 30% reduction in viral-induced CPE. Following incubation of the VSV-infected L929 cells, the monolayer was washed twice with medium and stained with 0.006% neutral red in medium for 1 h. The dye was then removed via aspiration, the monolayers were washed, and the residual dye incorporated in cells was extracted with 0.2 mL of Sorensen's citrate buffer (pH 4.2) containing 50% ethanol. The plate was then placed on the rotary shaker for 2 min. The optical density was read at 530 nm in a Dynatech MR 600 microplate reader (Dynatech, Alexandria, VA). CPE was evaluated by the use of a viability index, which is the ratio of dye uptake of virus-infected cells to dye uptake of uninfected cells: Equation 
Statistics. Treadmill run times to fatigue and serum IFN levels were analyzed using Student's t-tests (P < 0.05) when there were only two treatment groups, and a one-way ANOVA (P < 0.05) followed by independent t-tests when there were three treatment groups (CON, pI:C, pI:C + Ab) in the individual experiments. Interferon data from combined experiments were not distributed normally, so the nonparametric Spearman's correlation was used to analyze the association between serum IFN-α/β and run time to fatigue.
Run Time to Fatigue
Mean run times to fatigue for each of the trials varied. This was expected, as it is not unusual to have large variability in run times to fatigue in rodents (32,36). Within each trial, however, consistency in running the mice should have allowed us to detect true differences in treatment groups within an experiment.
Experiment 1. Male CD-1 mice injected i.p. with 0.2 mL of pI:C (5 mg·kg−1) or an equal volume of saline were run to fatigue 12 or 24 h after treatment. No significant difference in run time to fatigue on a motorized treadmill was found between groups at 12 h postinjection (P = 0.70) (Table 1). However, at the 24-h postinjection time point, a significantly shorter run time to fatigue was found in the pI:C group versus CON (P = 0.035).
Experiment 2. Run times for mice treated with pI:C were again, shorter than those in CON (P = 0.06) (Table 1). However, the effect on run times was partially negated in animals receiving pI:C combined with anti-IFN-α/β antibodies (pI:C + Ab).
Experiment 1. pI:C significantly increased IFN-α/β titers in all cases; however, there was variability in the magnitude of the interferon induction in the different experiments (Fig. 1). For example, the largest differences in IFN titer between CON and pI:C were found in the 24-h trial. Differences in titer were not nearly as great in the 12-h experiment, which might explain the lack of effect of pI:C on fatigue in this trial, and actually supports the hypothesis that interferon is related to the fatigue. IFN titer in the pooled preexercise blood (retro-orbital bleed) was similar to the postexercise mean values in each group (data not shown).
Experiment 2. Significant differences were found in plasma IFN titers between CON, pI:C, and pI:C + Ab in this experiment. The relatively large increase in IFN with pI:C was attenuated by the administration of rabbit anti-IFN-α/β Ab (Fig. 2). The larger control group IFN titer in this experiment compared with the control groups in the first experiment might be explained by the administration of IgG versus saline, which may have elicited a slight inflammatory response. Shorter fatigue times in this experiment compared with the first also coincide with generally higher IFN titers.
A significant negative correlation was found between run time to fatigue and the serum IFN titers across all experiments in animals in which IFN was induced (r = −0.81, P < 0.0001) (Fig. 3). A log-linear scale is used because of the nonnormal distribution of the IFN titers. Control animals not treated with pI:C had very low or nondetectable IFN activities.
The purpose of this study was to determine if increased IFN induction following immune system activation with pI:C may play a role in fatigue during forced treadmill exercise. Induction of IFN-α/β was accomplished by the use of pI:C, a potent IFN-α/β inducer in rodents (9,13,15,25). pI:C induces primarily IFN-α/β from macrophages but may also induce IFN-γ production (16,25).
No significant difference in the run time to fatigue was found between groups (pI:C vs CON) when exercise was given 12 h post-pI:C injection in experiment 1. However, at this time point, the increase in IFN titers was relatively small (Fig. 1). At 24 h postinjection, IFN titers were much greater (Fig. 1), and run time to fatigue was shorter in the pI:C group (Table 1). The evidence in support of an association between IFN-α/β and fatigue was strengthened in experiment 2 by the fact that administration of the specific antibody to IFN-α/β abrogated the effects of pI:C on fatigue. Furthermore, a significant correlation existed between run time to fatigue and plasma IFN titers for data from all experiments in which these variables were measured (r = −0.81, P = 0.0001). It seems that the relationship between plasma IFN and run time to fatigue is strongest when IFN titers reach approximately 900-1000 U·mL−1. Below this apparent threshold, the effects of IFN on exercise fatigue are less dramatic. This is consistent with our findings in the 12-h postinjection trial in which lower levels of IFN were detected. Incidentally, viral infections, which are commonly associated with fatigue, generally produce levels in excess of 1000 U·mL−1 (15).
It should also be noted that the antibody treatment did not completely abrogate the induction of IFN by pI:C, and although the run time did increase with antibody treatment, it was still somewhat shorter than in the control group. It is certainly possible that pI:C may have induced other cytokines or may have had other actions that negatively effect exercise fatigue. For example, transforming growth factor (TGF-α) has been linked to a decrease in voluntary wheel running and delayed initiation of grooming after swimming (4), IL-1 has been shown to decrease reaction to novel stimuli and to promote sleep (8,19), and TNF-α, along with IL-1, may suppress feeding behavior (24,26).
Other studies have attempted to measure lethargy or malaise as a result of immune stimulation, but no studies have looked at the specific effects of IFN as a promoter of fatigue during prolonged exercise. This line of research is relevant as exercise has been shown to increase IFN levels in the blood (34). Chao et al. (4) studied the reduction in voluntary wheel running (defined as general fatigue) and the delayed initiation of grooming behavior (defined as postexertional fatigue) after swimming in response to heat-killed bacterium (Corynebacterium parvum). C. parvum was effective in reducing running wheel behavior by day 6 after inoculation to 50% of baseline running activity, and the effect persisted for 8 d. Also, animals treated with C. parvum required approximately 6 min before they initiated grooming behavior, whereas grooming behavior normally begins immediately after exercise in control animals. Chao et al. reported threefold increases in transforming growth factor (TGF-β) in the C. parvum inoculated group compared with the saline control group (4). However, there is no other evidence in the literature to suggest that TGF-β can cause fatigue. Although not measured in the study by Chao et al., C. parvum is also known to induce IFN-α/β. Kirchner et al. (18) found small elevations in IFN levels (20 IU) in the serum of mice injected with small amounts of C. parvum (350 μg), but Chao et al. (4) used 1400 μg of C. parvum, which might induce higher serum levels of IFN-α/β. Therefore, it is possible that IFN may have played a role in decreased voluntary wheel running in this study.
Segall and Crnic (31) reported depressed horizontal activity, headpokes into a food chamber, and food intake in DBA/2 J mice both 10 and 24 h after the injection of a single 1600 U·g−1 dose of murine IFN. They measured activity in a Plexiglas box equipped with photobeams sensitive to activity (activity counts). Most of the changes in horizontal activity occurred within the first 10 h. Similar results were found with repeated doses of IFN-α (30 μg), which were able to modify open field activity of Swiss-Webster mice in a 16-square Plexiglas chamber (35). These relatively quick responses to IFN injection, in comparison to our finding that it took 24 h to affect fatigue during treadmill running, may be explained by the fact that peak IFN titers were not achieved until approximately 24 h following induction by pI:C.
It is unlikely that generalized fatigue during voluntary activity in these models involving the induction of IFN can be explained by traditional peripheral markers of muscular fatigue, as these animals could choose to rest at any time. In our model of motivated treadmill running to fatigue, it is likely that both peripheral and central factors played a role in fatigue (7). This study was not designed to investigate the potential mechanisms of early fatigue with elevated IFN, and there are no previous studies in this area. In fact, there is very little information regarding the mechanisms of immunologically mediated fatigue in general. The exact mechanisms behind the reduced exercise tolerance that occur in subjects with acute viral or bacterial infections (6,11,12), or in subjects with chronic fatigue syndrome (22,23), are not known. However, studies have failed to show significant morphological or enzymatic changes in muscle tissue (6,12), and depleted fuel stores, lack of available metabolic substrates, or lactic acidosis do not seem to be the limiting factors (2,17).
There is some evidence that the CNS may be involved in exercise fatigue resulting from the induction of IFN, although pharmacological studies of this cytokine suggest that very little IFN penetrates the blood-brain barrier. Only 0-4% of the reported serum levels of IFN injected into monkeys was found in the cerebrospinal fluid (14). In contrast, evidence for the interaction of plasma IFN and the brain can be seen in cancer patients receiving IFN therapy. Intravenous IFN therapy is associated with changes in EEG patterns despite little or no detection of IFN in the cerebrospinal fluid (1,30,33). Additionally, both clinical and experimental trials have shown that IFN is capable of producing fatigue, sleepiness, and other neuropsychiatric symptoms (3,28).
Other ways in which IFN may result in increased fatigue is through effects on opiate receptors or through increased lipolysis and serum-free fatty acid (FFA) levels. It is known that IFN is capable of binding to opiate receptors in the brain that induce changes in locomotion (3). Memon et al. (28) has shown that inoculation of mice with IFN (5-50 μg) increases lypolysis and serum FFA in a dose-dependent manner. FFA compete with tryptophan for binding sites on albumin, displace tryptophan, and lead to an increase in free plasma tryptophan (5). Increases in free plasma tryptophan are associated with elevated 5-HT (serotonin) in the brain, which is involved in centrally mediated fatigue (7).
In conclusion, pI:C administered to mice 24 h before exercise reduced run time to fatigue on a motorized treadmill during what would be the peak activity time for mice. Since the administration of anti-IFN-α/β antibody attenuated the effect of pI:C on run time to fatigue, and a significant correlation was found between plasma IFN titers and run time to fatigue, it is likely that IFN plays at least a partial role in fatigue under these conditions of immune system stimulation by pI:C. The specific mechanisms of an effect of IFN on run time to fatigue in this model and other models of immunologically mediated fatigue remain to be determined.
1. Adams, F., J. R. Quesada, and J. U. Gutterman. Neuropsychiatric manifestations of human leukocyte interferon therapy in patients with cancer. JAMA
2. Barnes, P. R. J., D. J. Taylor, G. J. Kemp, and G. K. Radda. Skeletal muscle bioenergetics in the chronic fatigue syndrome. J. Neurol. Neurosurg. Psychiatry
3. Blalock, J. E., and E. M. Smith. Human leukocyte interferon (Hu IFN-alpha): potent endorphin-like opioid activity. Biochem. Biophys. Res. Commun.
4. Chao, C. C., M. Delahunt, S. Hu, K. Close, and P. K. Peterson. Immunologically mediated fatigue: a murine model. Clin. Immunol. Immunopathol.
5. Curzon, G., J. Friedel, and P. J. Knott. The effect of fatty acids on the binding of tryptophan to plasma proteins. Nature
6. Daniels, W. L., J. A. Vogel, D. S. Sharp, G. Friman, J. E. Wright, W. R. Beisel, and J. J. Knapik. Effects of virus infection on physical performance in man. Mil. Med.
7. Davis, J. M., and S. P. Bailey. Possible mechanisms of central nervous system fatigue during exercise. Med. Sci. Sports Exerc.
8. Dunn, A. J., M. Antoon, and Y. Chapman. Reduction of exploratory behavior by intraperitoneal injection of interleukin-1 involves brain corticotropin-releasing factor. Brain Res. Bull.
9. Field, A. K., A. A. Tytell, and G. P. Lampson. Double stranded polynucleotides as interferon inducers. J. Gen. Physiol.
10. Fitts, R. H., and J. M. Metzger. Mechanisms of muscular fatigue. In: Principles of Exercise Biochemistry,
J. R. Poortmans (Ed.). Basel: S. Karger, 1993, pp. 248-268.
11. Friman, G. Effect of acute infectious disease on isometric muscle strength. Scand. J. Clin. Lab. Invest.
12. Friman, G., J. E. Wright, N. G. Ilback, W. R. Beisel, J. D. White, D. S. Sharp, E. L. Stephen, W. L. Daniels, and J. A. Vogel. Does fever or myalgia indicate reduced physical performance capacity in viral infection? Acta Med. Scand.
13. Galabru, J., N. Robert, C. Buffet-Janvresse, Y. Riviere, and G. Hovanessian. Continuous production of interferon in normal mice: effect of anti-interferon globulin, sex, age, strain, and environment on the levels of 2-5A synthestase and p67K kinase. J. Gen. Virol.
14. Habif, D. V., R. Lipton, and K. Cantell. Interferon crosses blood-cerebrospinal fluid barrier in monkeys. Proc. Soc. Exp. Biol. Med.
15. Hermans, H., A. Billiau, and P. Desomer. Interferon in experimental viral infections in mice: confrontation between tissue interferon levels resulting from virus infection and from exogenous interferon therapy. Infect. Immun.
16. Jullien, P., J. deMaeyer-Guignard, and E. E. A. deMaeyer. Interferon synthesis in X-irradiated animals. V. Origin of mouse serum interferon induced by polyI:C and encephalomyocarditis virus. Infect. Immun.
17. Kent-Braun, J. A., K. R. Sharma, M. W. Weiner, B. Massie, and R. C. Miller. Central basis of muscle fatigue in chronic fatigue syndrome. Neurology
18. Kirchner, H., M. T. Scott, and K. Munk. Protection of mice against viral infection by Corynebacterium parvum and Bordetella pertussis. J. Gen. Virol.
19. Krueger, J. M., J. Walter, C. A. Dinarello, S. M. Wolff, and L. Chedid. Sleep-promoting effects of endogenous pyrogen (interleukin-1). Am. J. Physiol.
20. Lloyd, A. Muscle versus brain: chronic fatigue syndrome. Med. J. Aust.
21. Lloyd, A., I. Hickie, A. Brockman, J. Dwyer, and D. Wakefield. Cytokine levels in serum and cerebrospinal fluid in patients with chronic fatigue syndrome. J. Infect. Dis.
22. Lloyd, A. R., S. C. Gandevia, and J. P. Hales. Muscle performance, voluntary activation, twitch properties and perceived effort in normal subjects and patients with the chronic fatigue syndrome. Brain
23. Lloyd, A. R., J. P. Hales, and S. C. Gandevia. Muscle strength, endurance and recovery in the post-infection fatigue syndrome. J. Neurol. Neurosurg. Psychiatry
24. Mahoney, S. M., and M. J. Tisdale. Induction of weight loss and metabolic alterations by human recombinant tumor necrosis factor. Br. J. Cancer
25. Martinez, D., R. J. Lynch, J. B. Meeker, and A. K. Field. Macrophage dependence of polyriboinosinic acid-polyrobocytidilic acid-induced resistance to herpes simplex vius infection in mice. Infect. Immun.
26. McCarthy, D. O., M. J. Kluger, and A. J. Vander. Suppression of food intake during infection: is interleukin-1 involved? Am. J. Clin. Nutr.
27. McKenzie, R., and S. E. Straus. Chronic fatigue syndrome. Adv. Intern. Med.
28. Memon, R. A., K. R. Feingold, A. H. Moser, W. Doerrler, and C. Grunfeld. In vivo effects of IFN-alpha and IFN-gamma on lipolysis and ketogenesis. Endocrinology
29. Pestka, S., J. A. Langer, K. C. Zoon, and C. E. Samuel. Interferons and their actions. Annu. Rev. Biochem.
30. Rohatimer, A. Z. S., P. F. Prior, A. C. Burton, A. T. Smith, F. R. Balkwill, and T. A. Lister. Central nervous system toxicity of interferon. Br. J. Cancer
31. Segall, M. A., and L. S. Crnic. An animal model for the behavioral effects of interferon. Behav. Neurosci.
32. Slentz, C. A., J. M. Davis, D. L. Settles, R. R. Pate, and S. J. Settles. Glucose feedings and exercise in rats: glycogen use, hormone responses, and performance. J. Appl. Physiol.
33. Smedley, H., Katrak, M., Sikora, K., and Wheeler, T. Neurological effect of recombinant human interferon. Br. Med. J. (Clin. Res. Ed.)
34. Viti, A., M. Muscettola, L. Paulesu, V. Bocci, and A. Almi. Effect of exercise on plasma interferon levels. J. Appl. Physiol.
35. Weinberger, S. B., G. Schulteis, A. G. Fernando, C. Bakhit, and J. L. Martinez. Decreased locomotor activity produced by repeated, but not single, administration of murine-recombinant interferon-gamma in mice. Life Sci.
36. Woods, J. A., J. M. Davis, E. P. Mayer, A. Ghaffar, and R. R. Pate. Exercise increases inflammatory macrophage antitumor cytotoxicity. J. Appl. Physiol.