M. T. Siva-Jothy, Department of Animal & Plant Sciences, University of Sheffield, Sheffield S10 2TN, U.K. e-mail: firstname.lastname@example.org
Abstract Measurement of insect immune effector system function aimed at identifying costs has largely been stimulated by the ideas of Hamilton & Zuk (1982), who proposed that choosy females may derive some genetic benefit from selecting parasite-resistant males. Field studies of such systems assume that most variation in measured immune traits is affected strongly by genes and pay little attention subsequently to the role of nutritional status in determining the magnitude of assayed immune effector systems. In this paper the effects of nutrient deprivation on immune function are measured in the mealworm beetle (Tenebrio molitor L.) reared in otherwise ideal conditions. The results suggest that immune effector system function is down-regulated during short-term nutritional deprivation, but is rapidly up-regulated to pre-deprivation levels after animals are allowed access to food. This rapid modulation of immune function in the context of nutritional status has important implications for measuring immune function in the field, as well as the interpretation of those measures.
Interest in how insects and other organisms modulate life-history investment in immune function has grown dramatically in the last few years, largely because of the advent of theoretical frameworks for understanding how parasites mediate sexual selection (e.g. Hamilton & Zuk, 1982; Folstad & Karter, 1992). Most of the studies that have investigated the relationship between immune function and male performance under sexual selection have, necessarily, focussed on field-based systems and are, by definition, examining condition-dependent traits (Anderson, 1994). Sexual signals of male ‘quality’ are such condition-dependent traits (see Anderson, 1994); consequently a correlation between the magnitude of the sexual trait and the magnitude of a measured immune effector system can be caused by a shared dependence on, for example, nutrient availability. This potential correlation may be expressed more strongly in iteroparous organisms, because the males of such species have more opportunity to adjust investment in their sexual traits throughout life. Insects are semelparous and most quantitative and qualitative aspects of their cuticular traits related to reproduction are fixed at imaginal eclosion, so it is possible that adult male insects can, to some extent, uncouple nutrient co-dependence between secondary sexual traits and immune traits. More importantly, perhaps, is the implicit assumption in many studies of ‘immunocompetence’ that results showing reduced immune function in measured effector systems are indicative of variation in the genetic basis of the response, which theory predicts is the main fitness component driving the system (see Hamilton & Zuk, 1982; Folstad & Karter, 1992; Anderson, 1994).
Studies of vertebrates, however, have revealed that nutritional stress affects parasite resistance (e.g. Murray et al., 1998) and, somewhat counter-intuitively, that reduced nutrient intake can even promote immune function (Kyriazakis et al., 1998). In either case, variance in nutrient intake/condition can affect vertebrate immune function (for related and wider discussions see Norris & Evans, 2000). Likewise, there are several studies linking nutritional status with plasticity in several invertebrate life-history traits (e.g. Boersma & Vijverberg, 1996), as well as documented examples of the effects of several forms of stress on insect immunity (see Brey, 1994). However, there are relatively few studies that have examined the role of nutrient availability on insect immune function. Measures of insect humoral immunity often centre on the ability of the insect to melanize target immunogens. Sang & Burnet (1963) were the first to demonstrate a nutrient-mediated effect on the genetic ability to respond to immune insult via melanin production (Söderhäll et al., 1996), whereas Suwanchaichinda & Paskewitz (1998) showed that larval nutrition could significantly affect the ability of adult Anopheles gambiae to respond to synthetic immune challenge (Sephadex CM C-25 beads). As well as its effect on the humoral response, nutrient constraint has been shown to affect insect cellular responses to pathogens. Vass & Napi (1998) showed that 24 h exposure to a supplemented diet increased the survivorship of Drosophila melanogaster larvae in the face of parasitoid attack, and that that survival was due to variance in the cellular encapsulation response. Intriguingly, Hoover et al. (1998) produced results that suggest that increased nutritional intake can increase the negative effects of baculoviruses.
Most insect immune responses are believed to result from the coordination between components of the humoral immune system and the cellular immune system. The first is manifest largely as soluble proteins and peptides that respond to non-self either by neutralizing it via toxic effects, or by signalling its presence in the haemocoel. The cellular response consists of cellular reservoirs of the soluble humoral components which are released upon contact with suitable signals and cells that respond to non-self by undergoing morphological and behavioural change and encapsulating the immunogenic substrate. In short, the cells attach to the surface of the immunogen and flatten out, thereby slowly covering its surface with a layer of ‘self’ tissue, which is melanized (see Lackie et al., 1985). This process effectively externalizes the pathogen: the process of melanization is controlled by the enzyme phenoloxidase and produces cytotoxic products which may of themselves kill the pathogen, but which also provide structural integrity to the encapsulating cells.
This paper addresses two questions. First, what effect does short-term nutrient deprivation have on immune function in adult male and female Tenebrio molitor? Second, how rapidly does immune function recover from any affects of temporary nutrient deprivation? The results are presented from a controlled laboratory experiment that examined how the magnitude of a cellular (the encapsulation response) and a humoral (the amount of haemolymph phenoloxidase) immune effector system were modulated in the face of short-term nutrient constraint applied to adults that previously (as larvae) had access to food ad libitum. The aim was to determine whether short-term nutrient deprivation in the reproductive stage of an organism (otherwise exposed to ideal nutrient conditions) affected immune function. The manner in which the focal immune effector systems recovered after treatment as well as the effects of the treatments on adult condition (dry fatless mass and fat) were also observed.
Materials and methods
Eggs were collected from four cultures maintained at 26 ± 2 °C under a LD 12 : 12 h photo regime. All larvae were reared at a constant density (c. 150 larvae in 1.5 kg of powdered rat chow) with free access to water, fresh apple and rat chow throughout larval development.
Upon pupation, beetles were removed from the cultures, sexed and then weighed. Animals were allocated at random to one of the following treatment regimes immediately after imaginal eclosion.
Treatment 1 Five days starvation followed by an immune insult (insertion of nylon monofilament on day 4) and phenoloxidase assay (day 5).
Treatment 2 Five days access to food ad libitum, followed by an immune insult (day 4) and phenoloxidase assay (day 5).
Treatment 3 Five days starvation followed by 3 days access to food ad libitum followed by an immune insult (day 7) and phenoloxidase assay (day 8).
Treatment 4 Eight days access to food ad libitum, followed by an immune insult (day 7) and phenoloxidase assay (day 8).
The ‘starvation’ treatment consisted of no access to food but access to water ad libitum.
In the treatments where food was provided ad libitum, animals had access to fresh (replenished daily) apple, rat chow and water.
Haemolymph extracts were taken by perfusing the abdomen of chilled adults with 2 mL of ice-cold sodium cacodylate buffer (0.01 m Na-cacodylate, 0.005 m CaCl2). Samples were immediately frozen at − 90 °C to disrupt the haemocytes. The frozen samples were thawed to 4 °C, followed by vortexing and then centrifugation (4 °C, 2800 g, 15 min). An aliquot of 500 µL of chilled supernatant was mixed with 1 mL of 3 mm L-DOPA in chilled sodium cacodylate buffer and the reaction allowed to proceed at 30 °C in a spectrophotometer (Pharmacia Biotech Ultraspec 2000, Sweden). Readings were taken at 490 nm and analysed using Swift II software (Pharmacia Biotech). Enzyme activity was measured as the rate of substrate conversion (determined using the auto slope function in Swift II software, Sweden) during the linear phase of reaction (between 5 and 15 min after the reaction began) during Vmax (see Barnes & Siva-Jothy, 2000). The repeatability of this assay was assessed by splitting the perfused blood from single individuals into three 0.5 mL samples and then freezing them. Each sample was then treated and enzyme activity measured, as described above. There was highly significant repeatability of the measurement of enzyme activity within an individual using the method described above (rmanova, F19,59 = 166.4, P = 0.0001, repeatability = 99.4%).
All experimental animals had a ∼1 mm length of nylon monofilament (diameter: 0.128 mm) inserted into the haemocoel through a puncture in the pleural membrane between the 2nd and 3rd sternite to the right of the midline. The monofilament was retrieved by dissection 24 h after insertion, once haemolymph extraction was completed as described above. Encapsulated monofilaments were stored in 70% ethanol, and the volume of the encapsulating cell-mass (melanized and nonmelanized cells) was measured after re-hydrating the sample, dissecting the cell mass from the nylon and squashing it underneath a cover-slip bridge. The outline of the flattened cell-mass was then drawn with the aid of a camera lucida and the drawn image digitized using a CCD camera (PULNiX tm-765, California, U.S.A.). The cell area and exact dimensions of the nylon monofilament were then calculated from the captured image using Optimas 6™ software (Washington, DC, U.S.A.). The total volume (mm3) of encapsulating cells attached to a piece of monofilament is expressed per unit surface area (mm2) of monofilament.
Two measures of repeatability were assessed for this assay. First, the spatial repeatability of the response by an individual (i.e. we measured whether the same individual produced a consistent, repeatable, response to nylon) was assessed by inserting two pieces of nylon into the same individual at the same time and then harvesting both implants 24 h later. Comparison of the individual's response to both implants revealed significant repeatability (rmanova, F19,59 = 3.5, P = 0.004, single treatment repeatability = 55.7%). Second, the repeatability of the measurement of the encapsulation response was assessed by measuring the volume of encapsulating material on each implant. The sample was then stored in ethanol and re-measured, blind, after at least 48 h. There was significant repeatability of the measurement of the encapsulation response using the method described above (rmanova, F19,59 = 6.9, P = 0.0001, single treatment repeatability = 74.9%).
After haemolymph extraction and retrieval of the nylon implant, the insect's body was dried at 30 °C for 48 h and weighed to the nearest 0.0001 g (Mettler AE160 balance) to obtain the dry weight with fat. The dried animal was then placed in a cellulose extraction thimble (Whatman, 10 × 50 mm) and the end sealed with glass wool. Thimbles were exposed to Chloroform reflux overnight (∼19 h) in a Soxhlet apparatus. After reflux the animal remains were dried to constant weight (at 100 °C for 30 min) and reweighed to obtain the dry fatless mass.
All analyses were conducted using Statview 5.0 (SAS, North Carolina, U.S.A.) for Macintosh. Means are presented ± standard errors. When necessary, and possible, data were transformed to meet the assumptions of parametric tests.
There were no differences (two-way anova) in the fresh pupal weights of animals used in the experiment with regard to either experimental treatment (F3,80 = 0.012) or sex (F1,80 = 0.003): there was no interaction between these terms (F3,80 = 0.007).
Effect of treatment and sex on immune function
There were no differences (two-way anova) in the encapsulation response in animals with respect to either treatment (F3,74 = 0.43, P = 0.73), sex (F1,74 = 3.1, P = 0.08 (re-analysis with the interaction term removed: F = 3.2, P = 0.07)) or the interaction between treatment and sex (F3,74 = 0.11, P= 0.95: Fig. 1a).
However, there were significant effects of both treatment (F3,80 = 7.66, P = 0.0001) and sex (F1,80 = 17.16, P < 0.0001) on the activity of haemolymph phenoloxidase. Males expressed significantly more phenoloxidase than females (mean difference c. 30%) (Fisher's PLSD, P < 0.0001) and starved animals expressed significantly less phenoloxidase at the end of 5 days than (a) fed animals after 5 days (Fisher's PLSD, P = 0.0002), (b) starved (5 days) and then fed (3 days) animals (Fisher's PLSD, P = 0.0007) and (c) fed animals after 8 days (Fisher's PLSD, P = 0.0001: Fig. 1b).
Effect of treatment and sex on measures of condition
There was no difference in dry fatless mass between the sexes (two-way anova, F1,79 = 0.74, P = 0.39) but this trait was significantly affected by treatment (F3,79 = 6.12, P = 0.0009), with no interaction between sex and treatment (F3,79 = 0.117, P = 0.95). Starved animals had lower dry fatless masses at the end of 5 days than (a) fed animals after 5 days (Fisher's PLSD, P = 0.007), (b) animals starved (5 days) and then fed (3 days) (Fisher's PLSD, P = 0.0001) and (c) fed animals after 8 days (Fisher's PLSD, P = 0.0015: Fig. 2a).
Fat content was affected by treatment (two-way anova (sqrt transformed fat), F3,79 = 3.04, P = 0.034). Starved animals had lower fat masses at the end of 5 days than (a) fed animals after 5 days (Fisher's PLSD, P = 0.004), (b) fed animals after 8 days (Fisher's PLSD, P = 0.03). Fat was significantly affected by sex (F1,78 = 13.34, P = 0.0005) such that females had significantly more fat than males (mean difference c. 20%) (Fisher's PLSD, P = 0.0004): there was no significant interaction (F3,79 = 0.683, P = 0.565: Fig. 2b).
Short-term nutrient deprivation has a rapid effect on immune function in T. molitor. The observed changes were reversed soon after provision of food. The patterns of change in investment suggest that there may be important differences in the costs associated with maintaining specific immune effector systems.
The effect of treatment and sex on encapsulation
Neither treatment nor sex affected the level of encapsulation, despite measurable effects in other traits. There are at least two possible reasons for this. First, the period over which encapsulation was measured (24 h) may have been too long to detect differences. For example, if the slowest rate of encapsulation induced by the treatment completed the encapsulation process in 20 h, all individuals would appear to have similar encapsulation rates. However, this explanation is unlikely because studies of other insects show that the encapsulation response can progress for several days after nylon insertion (e.g. Ryder, 1999). A second explanation could be that encapsulation is such an important determinant of fitness that individuals can not afford to down-regulate this process, despite the conflicting demands of other life-history traits. This explanation suggests that efficient encapsulation of internalized non-self is a crucial life-history trait regardless of its cost. Studies of other insects suggest that there may be an energetic cost to maintaining a viable encapsulation response and that there are situations where this trait is down-regulated (e.g. Koenig & Schmidt-Hempel, 1995; Siva-Jothy et al., 1998).
There was a marginally non-significant (P = 0.07) tendency for females to produce a smaller encapsulation response than males. Sex differences in immune function have been studied in detail in other insect groups (e.g. Kurtz et al., 2000) and are probably derived from the physiological and metabolic differences between adults related to gamete production and reproductive behaviour.
The effect of treatment and sex on haemolymph phenoloxidase levels
Male T. molitor express about 30% more phenoloxidase activity in their haemolymph than females (Radhika et al., 1998). Studies of another insect have shown that females express phenoloxidase in the ovaries (Ferdig et al., 1993): it is possible that these demands result in the lower levels in the haemolymph. Another possibility is that males are exposed to higher risks of immune challenge than females (for example, because of sexually dimorphic behaviour and/or physiology). By maintaining higher levels of haemolymph phenoloxidase, males may be better able to resist such challenge (for the effects of higher haemolymph phenoloxidase on refractoriness to parasites see Nigam et al., 1997).
Starvation leads to a significant decrease in the activity of phenoloxidase in the haemolymph of both males and females, despite the presence and maintenance of relatively large fat reserves. Phenoloxidase activity increased soon after access to food. This suggests that high levels of phenoloxidase activity may be relatively energetically costly to maintain, hence the lower levels during periods of food deprivation. However, the rapid return to pre-deprivation levels suggests that this immune effector system plays a relatively important role in immune defence. The lack of an effect on encapsulation despite the measurable effect on phenoloxidase in this study may be explained by the fact that the encapsulation response was measured during the phase when cells were being rapidly recruited to the immune challenge. Changes in phenoloxidase activity may only have a measurable effect on encapsulation later in the process (when the capsule becomes melanized).
The effect of treatment and sex on fatless body mass
In both males and females there is a significant decrease in dry fatless body mass during food deprivation, but this is recouped rapidly in both sexes after provision of food. This suggests that early on during periods of food deprivation adult T. molitor maintain physiological function by utilizing reserves of protein and/or carbohydrate in addition to fat.
The effect of treatment and sex on fat
Females carry about 20% more fat than males despite the similarity in dry fatless body mass. Fat represents a relatively large proportion of body mass (c. 16% in males and c. 23% in females, see Fig. 2) and, not surprisingly, animals utilize this reserve during food deprivation. The near-significant increase in fat stores after access to food may stem from the fact that the fat-body is an important immune organ (e.g. Hetru et al., 1998) and this fact may explain why phenoloxidase activity is reduced during periods when the fat-body reserves are being utilized.
These results show that food deprivation over the time course examined in the present experiments has a measurable affect on a humoral immune effector system, but has no measurable effect on a cellular immune effector system. This suggests that phenoloxidase activity is more plastic and is modulated more rapidly than encapsulation with respect to nutrient condition. This result has important implications about the use of these assays as indicators of an individual's refractoriness to parasites (both have been shown to be phenotypically correlated with refractoriness to parasites, e.g. Gorman et al., 1996; Kraaijeveld & Godfray, 1997) and the effects of nutrient availability on other commonly used assays of immune function in ecological immunity studies of insects and (potentially) vertebrates. Food shortage is probably common in the wild (e.g. Kells et al., 1999) and is probably compounded by competition at higher population densities, when parasite transmission is more likely (Steinhaus, 1958). If parasites reduce a host's ability to assimilate food, then infection may further increase susceptibility to parasites. Such potentially large environmental effects on immune system function, and thereby host refractoriness to parasites, over such short time-scales are potentially at odds with the current tendency to interpret quantitative assays of immune effector systems in wild individuals solely in the context of heritable variation. The results presented here suggest that condition (or heritability) should be assessed or controlled for in studies of ecological immunity before individual variation in immune function is slotted into an evolutionary interpretation based on heritable parasite resistance.
This work was supported by Natural and Environmental Research Council grant # GR3/12121. Jens Rolff (G & K), Sophie Armitage and two anonymous referees made comments that greatly improved the manuscript.