Siberian hamsters (Phodopus sungorus) exhibit changes in reproductive and immune function in response to seasonal variations in day length. Exposure to short days induces gonadal regression and inhibits testosterone secretion. In parallel, short days enhance immune function: increasing leukocyte numbers and attenuating cytokine and behavioral responses to infection. We examined whether photoperiodic changes in leukocyte phenotypes and sickness behaviors are dependent on concurrent photoperiodic changes in gonadal function. Male hamsters were gonadectomized or sham-gonadectomized and either exposed to short days (9 h light/day; SD) or kept in their natal long-day (15 h light/day; LD) photoperiod for 10–13 wk. Blood samples were obtained for leukocyte enumeration, and hamsters were challenged with bacterial LPS, which induced behavioral (anorexia, reductions in nest building) and somatic (weight loss) sickness responses. Among gonad-intact hamsters, exposure to SD increased total and CD62L+ lymphocytes and CD3+ T lymphocytes in blood and significantly attenuated LPS-induced sickness responses. Independent of photoperiod, castration alone increased total and CD62L+ lymphocyte and CD3+ T lymphocyte numbers and attenuated somatic and anorexic sickness responses. Among castrated hamsters, SD exposure increased lymphocyte numbers and suppressed sickness behaviors. In castrated hamsters, the magnitude of most immunological effects of SD were diminished relative to those evident in gonad-intact hamsters. The SD phenotype in several measures of immunity can be instated via elimination of gonadal hormones alone; however, photoperiodic effects on immune function persist even in castrated hamsters. Thus, photoperiod affects the immune system and neural-immune interactions underlying sickness behaviors via gonadal hormone-dependent and -independent mechanisms.
- sickness behaviors
- neural-immune interactions
seasonal changes in physiology and behavior are legion, most notably in reproductive function (8). Many mammals have evolved mechanisms that permit cessation of reproduction during intervals of the year when environmental conditions of food availability and ambient temperature are unfavorable for successful weaning of offspring (44, 47). In nonequatorial regions, changes in day length (photoperiod) predict these environmental constraints (46) and function as proximate cues for triggering anticipatory changes in reproductive physiology and behavior. In long-day breeders, decreasing photoperiods of late-summer trigger gonadal atrophy, withdrawal of gonadal hormone secretion, and cessation of sex behaviors (8). The full complement of changes in the reproductive system can be elicited in the laboratory by exposing reproductively photoperiodic animals to winter-like day lengths (<12 h of light/day) (22).
Seasonal changes in immune function also occur in wild and laboratory animals (34), and a growing body of research indicates photoperiod likewise drives these cycles. Among long-day breeding rodents, many aspects of immune function are relatively enhanced under short days; during winter, animals apparently reallocate energy away from reproduction and toward facilitating maintenance of host defense (15, 33). Siberian hamsters (Phodopus sungorus) exhibit multiple changes in immune function following adaptation to short photoperiods: numbers of circulating leukocytes, T and B cells are greater; natural killer cell cytotoxicity is facilitated; and skin inflammatory responses are enhanced (3, 18, 55). Among the most robust changes evident in the hamster immune system is the short-day attenuation of metabolic and behavioral symptoms of infection. After treatment with a bacterial mimetic (lipopolysaccharide of Escherichia coli; LPS) proinflammatory cytokine production is lower (4, 40), and the magnitude of febrile, anhedonic, anorexic, and cachexic responses to infection are reduced (4).
The mechanisms by which photoperiod effects changes in peripheral immunity and in neural-immune interactions that mediate sickness behaviors are not fully understood. In common with reproductive responses, photoperiodic changes in immunity are dependent on pineal melatonin secretion (49). Temporally, photoperiodic changes in immune function appear to parallel changes in reproductive function (41, 44, 56). In light of the immunomodulatory role of gonadal hormones (5, 29, 40), photoperiodic changes in the immune system may arise exclusively as a consequence of photoperiod-driven changes in the reproductive system. Data bearing on this issue have not yielded consistent results. In hamsters, photoperiodic changes in antibody production and skin inflammatory responses occur independent of changes in the reproductive system (18, 39); in contrast, other indices of immunity more closely track gonadal condition (spontaneous blastogenesis, leukocyte concentrations) (41, 45). Determining the extent to which seasonal changes in the immune system depend on concurrent changes in the reproductive system remains a key step toward understanding how photoperiod drives seasonal immunity, and more generally, may provide insight into how neuroendocrine activity impinges on the immune system.
This experiment tested the hypothesis that photoperiodic changes in blood leukocyte populations and in behavioral symptoms of infection are dependent on photoperiodic changes in gonadal activity. Castrated and gonad-intact hamsters were exposed to long or short photoperiods for 12 wk, and leukocytes were enumerated using flow cytometry. Hamsters were then challenged with a simulated bacterial infection to assess sickness behaviors. Photoperiodic changes in any aspects of immune function abolished by castration would be consistent with gonadal hormone-dependence; in contrast, differences preserved following castration would indicate gonadal hormone-independent effects of photoperiod on immune function.
MATERIALS AND METHODS
Animals and Housing
Male Siberian hamsters (Phodopus sungorus) were obtained from a breeding colony maintained at the University of Chicago. Hamster pups were weaned at 21 days of age and housed 2–4 per cage with same-sex siblings in polypropylene cages (28 × 17 × 12 cm) with wood-shaving beddings (Harlan Sani-Chips, Harlan, Indianapolis, IN) in a 15:9 light-dark cycle (lights off at 1800 CST) until 3–4 mo of age. Ambient temperature of the room was 20 ± 0.5°C, and relative humidity was maintained at 53 ± 2%. Food (Teklad Rodent Diet 8604, Harlan) and filtered tap water were provided ad libitum. Cotton nesting material was constantly available in the cage until aliquoted for nest-building tests after week 12. All procedures conformed to the U. S. National Research Council's Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Chicago.
At 90–110 days of age, hamsters were surgically castrated (cast, n = 42) or sham-operated (sham, n = 41) under pentobarbital sodium (Nembutal, 0.05 mg/g ip) anesthesia. Each testis was externalized via a single midline incision; testes were either ligated and removed with scalpel and forceps (cast), or returned to the scrotal cavity (sham). After surgery, hamsters received analgesia (Buprenex, 0.5 μg/g sc) twice per day for two successive days.
Four weeks after surgery (= week 0), hamsters were transferred either into short days (9 h light/day, lights on at 0900 CST, SD; SD-cast, n = 27; SD-sham, n = 25) or remained in long-days (15 h light/day, LD; LD-cast, n = 15; LD-sham, n = 16). Hamsters were weighed (± 0.1 g) on week 0 and 10. In most Siberian hamster laboratory populations, there exist individuals that do not respond to SD with gonadal regression and fur molt (43). To identify these individuals, at week 16, stage in the seasonal pelage color cycle was assessed in each hamster using an integer scale of 1 to 4 (1 = dark “summer” fur, 4 = white “winter” fur; 19) by a single trained observer who was blind to the treatment conditions. Animals from the SD photoperiod that failed to exhibit evidence of molt to a winter fur (fur score =1 on week 16) were regarded as nonresponsive to SD and were excluded from all subsequent analyses.
On week 11, blood samples (∼270 μl) for flow cytometry were drawn under light anesthesia (between 1245 and 1315, 5 h prior to lights off) from the right retro-orbital sinus using heparinized Natelson collection tubes and dispensed into vials loaded with heparin (30 units). Blood collections were performed in a room separate from the general animal colonies. Following the procedure, hamsters were separated from the colony until all collections were completed. Animals were placed into a presaturated (5% isoflurane) anesthesia chamber. The time required to obtain a blood sample, from initiation of anesthesia to filling the collection tube, was ∼1 min. Following blood collection, hamsters were given 0.5 ml sc of sterile 0.9% saline for rehydration. Blood samples were kept at room temperature and analyzed by flow cytometry within 24 h according to methods described previously (54). Briefly, specific leukocyte subtypes were measured by immunofluorescent Ab staining and analyzed using single-color flow cytometry (FACSCalibur, Becton Dickinson, San Jose, CA). Lymphocyte, neutrophil, and monocyte subpopulations were identified and gated by using forward- vs. side-scatter characteristics. T cells were identified by using CyChr-labeled anti-CD3. Neutrophils and monocytes were identified by using forward- vs. side-scatter patterns. L-selectin (CD62L)-positive cells were identified by using PE-labeled anti-CD62L. All monoclonals were directly conjugated, rat anti-mouse Abs (BD-Pharmingen, San Diego, CA). Blood samples were incubated with antibody for 20 min at room temperature, washed with PBS, and read on the FACSCalibur with 3,000 to 5,000 events being acquired from each preparation. Control samples matched for each fluorochrome and each antibody isotype were used to set negative staining criteria. Data were analyzed using CELLQuest software (Becton Dickinson).
Beginning at week 12, sickness responses to LPS were evaluated according to established methods in our laboratory (54). Briefly, body mass and food intake measurements (performed at 1730, 30 min before the onset of darkness) were obtained daily, and after 3 days of baseline measurements, shortly before lights off (1730–1800), hamsters were injected intraperitoneally with either bacterial LPS (625 μg/kg; isolated from E. coli strain 026:B6, Lot 064K4077, Sigma, St. Louis, MO) or sterile 0.9% saline (0.1 ml), in a counter-balanced design, with successive injections separated by 10–14 days. At the time of injection, each hamster was given a small piece of cotton batting (∼3 g), which was weighed before and again 3 h after presentation, to assess nesting material use. Hamsters typically shred this cotton to construct a small nest within the first few hours after the offering (54). Food intake, body mass, and nest building measures were obtained for 7 successive nights following injections.
Paired t-tests were used to determine whether hamsters exhibited significant changes in body mass between week 0 and 10. Rates of nonresponsiveness were compared between castrated and intact SD hamsters using a χ2-test, and pelage scores were compared using a Mann-Whitney U-test.
Leukocyte concentrations were compared using 2 (LD, SD) × 2 (Cast, Sham) factorial ANOVA. Because photoperiod significantly affected body mass and food intake before injection treatments, changes in body mass and food intake following LPS and Saline injections were each expressed as a percentage of individual baseline values (mean of the three daily measurements immediately preceding the injection). Values were then compared between groups using 2 (LD, SD) × 2 (Cast, Sham) × 2 (LPS, Saline) factorial ANOVA, according to the methods of Bilbo et al. (4). Data collected between the time of LPS/saline treatment and 24 h later are designated as day +1 values, those collected between 24 h and 48 h after injection are referred to as day +2 values, and so forth. Nesting material use during nightly 3-h tests was compared between groups using a Kruskal-Wallis test followed by Mann-Whitney U-tests.
Effects of injection treatment order on change in body mass, food intake, and nest building were evaluated with ANOVA. No significant effect of injection order was observed (P > 0.05 in all comparisons); therefore, injection groups were collapsed within treatment across the counterbalanced blocks for all analyses.
All statistical calculations were conducted using Statview 5.0 (SAS Institute, Cary, NC). Where permitted by significant F statistics, pairwise comparisons were conducted using Fisher's PLSD tests. Exact probabilities and test values were omitted for simplicity and clarity of the presentation of the results. To protect against Type I error from the large number of pairwise comparisons (8 treatment groups × 7 days), alpha was set to 0.01 for evaluation of pairwise comparisons of mean body mass following injection treatments. Otherwise, differences were considered statistically significant if P < 0.05.
Analyses of Effect Sizes
For several immunological and behavioral endpoints, there were significant effects of photoperiod in both intact and castrated hamsters. To compare the magnitude of photoperiodic modulation of these traits under intact and gonadectomized conditions, two descriptive measures of effect size, percentage difference, and estimated omega square, were calculated. Percentage difference simply yields an expression of the actual magnitude of the difference between LD and SD groups for a given trait. For a given trait, t, the relative increase attributable to SD treatment was calculated as the difference between SD and LD values expressed as a percentage of the LD value, that is, [SDt − LDt]/[LDt] × 100%.
Omega square (ωA2), in a single-factor design, is estimated from two variances in the treatment population, one based on the differences between the population treatment means (σTreatment2), and the other based on the variability within each population (σResidual2) as follows: ωA2 = σTreatment2/[σTreatment2 + σResidual2] (Eq. 4–1 in Ref. 28). ωA2 = 0 when treatment effects are absent and varies from 0 to 1.0 when treatment effects (in this case, photoperiod) are present. Effect magnitude, or strength, as estimated by ωA2, is a relative measure reflecting the proportional amount of the total population variance that is attributable to variation between treatment groups. ωA2 differs qualitatively from an F statistic because it is insensitive to changes in sample size (9, 30). This index is often referred to as the proportion of variation “accounted for” by the experimental manipulation, or “explained variance.” On the basis of meta-analyses of behavioral data, ωA2 values less than 0.06 are considered “small”, less than 0.15 are considered “moderate,” and ωA2 values greater than 0.15 are considered “large” (13, 14).
Somatic Responsiveness to Photoperiod Manipulations
At week 10, significant main effects of photoperiod (P < 0.0001) and gonadal condition (P < 0.0001) were evident on body mass. Castrated and gonad-intact hamsters exhibited decreases in body mass over 10 wk of exposure to SD (P < 0.0001, both comparisons; Fig. 1); castration also caused LD hamsters to lose body mass (P < 0.0001). Within each surgical condition, LD hamsters weighed more than SD hamsters at week 10 (P < 0.05, both comparisons).
The eventual development of winter pelage color was used as an independent measure to retrospectively determine whether hamsters had responded to the SD photoperiod treatment (19). The criterion for responsiveness was a fur score ≥2 following 16 wk of exposure to SD. Among gonad-intact hamsters, 16 of 25 hamsters reached criterion; among castrated hamsters, 18 of 27 hamsters reached criterion (Fig. 1). Castration affected neither the incidence of nonresponsiveness to SD (P > 0.80), nor the degree of molt to the winter fur (P > 0.90; cf. 19).
Photoperiod significantly affected total lymphocyte numbers, CD62+ lymphocytes, and CD3+ T lymphocytes (main effects; P < 0.05, all comparisons; Fig. 2). Significant main effects of gonadal condition were evident on total lymphocyte and neutrophil numbers (P < 0.005, both comparisons). Photoperiod and gonadal condition interacted to affect CD3+ T cells (P < 0.05).
Among gonad-intact hamsters, exposure to SD resulted in increases in total blood lymphocyte numbers (P < 0.0005), CD3+ T cells (P < 0.005), and CD62+ lymphocytes (P < 0.001; Fig. 2). Total monocyte and neutrophil concentrations were not affected by photoperiod (P > 0.20, all comparisons). Castrated hamsters in SD also exhibited higher concentrations of total lymphocytes (P < 0.05) and CD62+ lymphocytes (P < 0.05) relative to LD hamsters (Fig. 2). In contrast, CD3+ T cells were comparable among castrated LD and SD hamsters (P > 0.80; Fig. 2). Monocytes and neutrophils were comparable among LD and SD castrates (P > 0.20, all comparisons).
Among LD hamsters, gonadectomy resulted in significant increases in the numbers of total lymphocytes, CD62+ lymphocytes, and CD3+ T cells (P < 0.05, all comparisons). Among SD hamsters, castration did not cause further increases in the concentrations of any leukocyte subpopulation by exposure to SD (P > 0.05, all comparisons). In both LD and SD hamsters, gonadectomy significantly decreased neutrophil concentrations (Fig. 2).
Changes in food intake following injections were significantly affected by gonadal condition (P < 0.0001) and injection type (P < 0.0001), and by an interaction between photoperiod and injection type (P < 0.005; Fig. 3). Among gonad-intact hamsters, food intake was significantly decreased in hamsters injected with LPS compared with those treated with saline; this decrease persisted for 3 days following LPS injection in LD hamsters, compared with 2 days in SD hamsters (P < 0.05 vs. saline, all comparisons; Fig. 3). LPS-induced decreases in food intake were significantly greater in LD relative to SD hamsters on day +1, day +2, and day +3 (P < 0.01, all comparisons). In castrated hamsters, LPS treatments likewise suppressed food intake, but in contrast to gonad-intact hamsters, anorexia was evident in both LD-cast and SD-cast hamsters on day +1 (P < 0.005 vs. saline, both comparisons) and day +2 (P < 0.05, both comparisons) but was resolved thereafter in both groups (P > 0.05, all comparisons; Fig. 3). In common with intact hamsters, the magnitude of suppression of food intake was greater in LD relative to SD castrates, as measured on day +2 (P < 0.05; Fig. 3). Lastly, on day +1 and day +2, the magnitude of LPS-induced anorexia was significantly greater in LD-sham relative to LD-cast hamsters (P < 0.05, both comparisons).
There were significant effects of photoperiod (P < 0.005) and injection type (P < 0.0001), and an interaction between injection and photoperiod (P < 0.001), on changes in body mass following injections (Fig. 4). Across all surgical and photoperiod groups, body mass loss was significantly greater in hamsters injected with LPS compared with those treated with saline. In intact hamsters, decreases in body mass following LPS relative to saline treatment persisted for 4 days in SD hamsters, but endured for the entire 7 days posttreatment in LD (P < 0.01, all comparisons; Fig. 4). Relative to SD-sham hamsters, LD-sham hamsters sustained significantly greater body mass losses beginning on day +3 and continuing through day +7 (P < 0.01, all comparisons). LD-sham hamsters reached a significantly lower nadir (∼8% decrease from baseline) relative to SD hamsters (∼5% decrease; P < 0.01). Among castrated hamsters, decreases in body mass following LPS treatment persisted through day +7 in LD hamsters (P < 0.01, all comparisons), but were resolved after day +3 in SD hamsters (P > 0.01, all comparisons; Fig. 4). In common with intact hamsters, LD-cast hamsters sustained significantly greater decreases in body mass relative to SD-cast hamsters, from day +4 through day +7 (P < 0.01, all comparisons). In contrast to intact hamsters, both LD-cast and SD-cast hamsters reached comparable nadir values (4–5% decrease; P > 0.01).
Nesting material use.
There were significant main effects of photoperiod (P < 0.05) and injection type (P < 0.005), and an interaction between injection and photoperiod (P < 0.001), on the use of nesting material following injections (Fig. 5); there was no main effect of castration on nesting material use (P > 0.30). Across all groups, suppression of nesting material use was confined to the initial hours following LPS treatment (day 0, 0–6 h after injection). Use of nesting material was suppressed by LPS relative to saline treatment in LD-sham (P < 0.0001) and SD-sham hamsters (P < 0.05), and the magnitude of suppression was significantly greater in LD-sham relative to SD-sham hamsters (P < 0.005). Among castrated hamsters, only LD hamsters exhibited reductions in nesting material use following LPS treatment (LD-cast: P < 0.0001; SD-cast: P > 0.50). In common with intact hamsters, LPS suppression of nesting material use was significantly greater in LD relative to SD castrates (P < 0.0001). Castrated and intact hamsters exhibited comparable nest use responses to LPS in both LD (P > 0.80) and SD (P > 0.05).
Analyses of effect sizes.
Effect size assessments were performed on endpoints for which pairwise statistical tests revealed significant (P < 0.05) effects of photoperiod (Fig. 6). In intact hamsters, exposure to SD caused significant increases in total and CD62+ lymphocytes of 71% and 213%, respectively; in castrated hamsters, effects of SD were likewise significant, but were reduced in magnitude (25% and 55%, respectively). The ameliorating effect of SD on LPS-induced anorexia was most evident on day +2 after injection (cf. 54). SD exposure attenuated LPS-induced anorexia by 46% in gonad-intact hamsters, and by 50% in castrated hamsters. Mean body mass loss over the 7 days following LPS injections was 58% less in SD-intact relative to LD-intact hamsters and 52% less in SD castrates relative to LD castrates. Lastly, SD exposure attenuated LPS-induced suppression of nest building behavior by 40% in intact hamsters and by 65% in castrates. For all traits except LPS-induced inhibition of nest-building, omega square values for photoperiod were substantially (2- to 5-fold) higher in intact relative to castrated hamsters.
Exposure to SD for ∼3 mo altered several aspects of the immune system of Siberian hamsters, in agreement with previous reports (3, 4, 21, 37, 40). Absolute numbers of peripheral blood lymphocytes, L-selectin (CD62) positive lymphocytes, and CD3+ T lymphocytes were significantly increased in SD relative to LD hamsters, and SD hamsters exhibited diminished ingestive, somatic, and behavioral thermoregulatory responses to a simulated infection. Castrated hamsters housed in LD exhibited leukocyte numbers and behavioral responses that were, in most respects, comparable to those of gonad-intact SD hamsters, suggesting that withdrawal of gonadal hormone secretion is sufficient to mimic many effects of SD on immune function. However, despite the immunoenhancing effects of castration, photoperiodic differences in immune function were still evident among castrated hamsters, suggesting that photoperiod also affects the hamster immune system via mechanisms that do not require changes in gonadal hormone secretion. Together, the data indicate that changes in day length affect the immune system and neural-immune interactions underlying sickness behaviors via gonadal hormone-dependent and -independent mechanisms.
Effects of photoperiod on immunity were notably diminished, though not absent, following castration. In castrated hamsters, mean differences in leukocyte counts between LD and SD groups were smaller, the number of days of anorexia (relative to saline-injected controls) was fewer, and the interval of decreased body mass following LPS treatment was shorter. To better quantify the relative contribution of gonadal hormone-dependent vs. -independent mechanisms toward seasonal changes in immunity, two descriptive post hoc statistical approaches were applied to each dependent variable by 1) calculating the change induced by SD relative to LD and 2) calculating omega-square, an estimate of effect strength (13). The magnitude of the relative effect of SD on immunity varied as a function of gonadal condition in a trait-specific manner (Fig. 6A). SD induced a substantially greater percentage increase in lymphocyte numbers in gonad-intact relative to castrated hamsters; but the relative effects of SD on ingestive and ponderal responses to LPS were comparable among intact and castrated hamsters; and the relative effect of SD on nest-building responses to LPS was greater in castrated hamsters. Omega square values indicated that, in castrated hamsters, the magnitudes of SD effects on most dependent variables, although significant, were diminished relative to those obtained in intact hamsters (Fig. 6B). Indeed, omega-square values for the effect of SD on lymphocyte counts, and on LPS-induced anorexia and cachexia were all large (>0.15) in gonad-intact hamsters, whereas in castrated hamsters omega square values for these effects ranged from small to moderate. Taken together, the data suggest that, in the photoperiodic control of immune function, the contributions of gonadal hormone-dependent mechanisms tend to outweigh those of gonadal hormone-independent mechanisms.
Previous data from this and other species have demonstrated a similar principle; namely, that phenotypic responses to photoperiod often have both gonad-dependent and gonad-independent components. Using an analogous design, a previous study revealed gonadal hormone-dependent and -independent effects of photoperiod on body mass and thermoregulatory behavior (daily torpor) in this species (53); castration induced larger decreases in body weight in SD relative to LD hamsters, supporting the conclusion that the decline in gonadal hormones could only partially account for SD-induced decreases in body mass (53). In Syrian hamsters, photoperiodic regulation of gonadotropin secretion is also mediated by both steroid-dependent and independent mechanisms (20, 50–52).
Earlier work in this species provided somewhat conflicting evidence about the role of photoperiodic changes in gonadal hormone secretion in the control of the immune system. Decreased testosterone concentrations in hamsters housed in SD are associated with increases in lymphocytes, T cells, and CD62+ lymphocytes (3). In juvenile hamsters, however, castration abolished the enhancing effect of SD on total leukocyte counts, largely through an increase in leukocyte concentrations in LD hamsters following castration (41; cf. Fig. 2). Differences in age (juveniles vs. adults) and dependent variables (total leukocytes vs. lymphocyte subsets) may complicate direct reconciliation of these two prior experiments with the present work. Still other reports have indicated that SD enhancements in the magnitude of skin inflammatory responses occur independently of gonadal responses to photoperiod (3, 37) and persist following gonadectomy or testosterone replacement (38). Increases in the number of cells expressing CD62 cell-surface markers—adhesion molecules that participate in the migration of lymphocytes through endothelium—could contribute to enhanced skin inflammatory responses in SD (7, 12). Thus, earlier reports, suggesting gonadal hormone independence of skin inflammatory responses (39), are consistent with the persistence of photoperiodic differences in CD62+ cells following gonadectomy in the present study. Studies in other reproductively photoperiodic rodents have likewise indicated several gonadal hormone-independent effects of photoperiod on the immune system (16). The effects of photoperiod manipulations and gonadal hormones on immunity may differ categorically depending on the measure of immunocompetence, underscoring the importance of addressing neuroendocrine-immune interactions on a trait-by-trait basis (35).
This is the first report to directly test whether seasonal changes in sickness behaviors are dependent on gonadal responses to day length, although several prior studies have generated data that indirectly addressed this issue. In one report, melatonin treatments (for 1 wk), which were inadequate to initiate gonadal regression, also failed to induce SD-like changes in sickness behaviors (6). In another report, pinealectomy, which prevented gonadal regression in SD, likewise blocked the effects of SD on sickness behaviors (54). Taken together, these reports are consistent with the conjecture that photoperiodic changes in sickness behavior are dependent on concurrent gonadal responses. In a recent report, however, hamsters provided with central melatonin implants sufficient to induce complete gonadal regression failed to exhibit complete attenuation of LPS-induced anorexia or suppression of nest-building behavior (21). This latter report suggests that gonadal regression alone is not sufficient to impart the SD phenotype in immune function and thereby complements the present study's major observation that, absent the ability to exhibit changes in gonadal hormone production (4 mo. postcastration), changes in photoperiod are still sufficient to regulate the immune system.
In addition to providing evidence for gonadal hormone-independent effects of photoperiod, the present report confirms that, independent of photoperiod manipulations, withdrawal of gonadal hormones increases survival and/or proliferation of lymphocytes. Such effects may be mediated through any of several mechanisms described in other rodents. In mice, castration increases thymus size and proliferation of T lymphocytes (10, 36), and testosterone reverses these effects (11, 36) through a direct action on membrane-bound androgen receptors (2, 25). Testosterone induces apoptosis in CD4+ and CD4+CD8+ lymphocytes, in part by increasing production of the proinflammatory cytokine TNF-α (23). Testosterone also exacerbates the symptoms of simulated bacterial infection. In vivo, LPS treatments induce greater febrile responses in male relative to female mice (31, 32) and humans (1, 17), and testosterone enhances LPS-elicited TNF-α responses in ungulates (27). Effects of sex steroids on inflammatory responses may occur directly on substrates within the immune system, and/or indirectly, e.g., potentiation of anti-inflammatory HPA responses to inflammatory signals such as LPS or IL-1β (48, 49).
The gonadal hormone-independent mechanisms by which hamsters enact photoperiodic changes in the immune system are potentially quite numerous and may involve seasonal changes in pineal melatonin production and glucocorticoid secretion (3, 40, 47). Nocturnal melatonin secretion tracks seasonal changes in photoperiod (25) and is a necessary and sufficient time-of-year signal for the reproductive system (21). Pinealectomy (PINx), in addition to eliminating photoperiodic gonadal responses (21), abolishes effects of SD on blood lymphocytes and behavioral responses to LPS (49). Melatonin likely figures prominently in the transduction of day length information into the hamster immune system. The present work suggests that the efficacy of PINx in abolishing photoperiodic changes in the immune system does not solely arise from the elimination of reproductive responses to photoperiod. Gonadal hormone-independent, melatonin-dependent photoperiodic time measurement by the immune system occurs in part via an action of melatonin in the brain (in the suprachiasmatic nucleus) (20) and may also involve peripheral actions of melatonin (23). In light of a recent report that, in reproductively nonphotoperiodic rats, SD exposure increases blood leukocyte numbers and attenuates cytokine responses to LPS (41), it may be that photoperiod-driven seasonal changes in immune and reproductive function occur in parallel with one another. In reproductively photoperiodic animals, robust immunomodulatory effects of gonadal hormones on immunity may simply mask latent gonadal hormone-independent processes.
Perspectives and Significance
In the present work, hamsters housed in SD exhibited increases in blood lymphocytes and diminished acute phase symptoms of bacterial infection. Photoperiodic changes in immunity were attenuated, but clearly not abolished, by castration. The results are compatible with the hypothesis that seasonal changes in immune function do not manifest simply as a consequence of concurrent reproductive responses to photoperiod, but rather, the immune system and the reproductive system may independently gain access to pineal-mediated day length information. If the presence of gonadal hormone-independent effects of photoperiod on the immune system is a common feature among mammals, then specification of these nongonadal factors may provide novel insights into seasonal rhythms in the incidence and outcome of infectious diseases, especially in mammals whose reproductive systems are relatively unaffected by changes in day length.
This work was supported by National Institutes of Health Grant AI-67406 from the National Institute of Allergy and Infectious Diseases and an Andrew W. Mellon Minority Undergraduate Fellowship from the University of Chicago.
We thank Jarvi Wen, Jerome Galang, Tanita Mason, and Justin Wagner for technical assistance, Jean M. Tillie of the Dhabhar Laboratory for conducting flow cytometry, and Leah Pyter for providing helpful comments on the manuscript.
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. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society