Effects of immune supplementation and immune challenge on bacterial assemblages in the avian cloaca

Relationships between avian physiology and bacterial assemblages in the cloaca are poorly understood. We used molecular techniques to analyze cloacal swabs from pigeons that were subjected to two immunological manipulations: lysozyme supplementation and endotoxin challenge. From the swabs, we derived ecological indices of evenness, richness, and diversity of bacterial assemblages. Challenge led to changes in evenness that depended on supplementation. When analyzing these changes, we neutralized the effects of a possible statistical artifact by including the starting values as a covariate. Repeatability calculations suggested that swabbing reliably captured the evenness but not the richness or diversity of bacterial assemblages in the cloaca. Auswirkungen von Immunsupplementierung und experimenteller Immunantwort auf die Bakterienzusammensetzung in der Kloake Zusammenhänge zwischen der Physiologie von Vögeln und der Bakterienzusammensetzung in der Kloake sind weitgehend unbekannt. Wir haben molekulare Methoden verwendet, um Kloakenabstriche von Tauben zu untersuchen, die zwei verschiedenen Manipulationen des Immunsystems ausgesetzt waren: einer Unterstützung des Immunsystems durch Zugabe von Lysozymen und einer experimentell ausgelösten Immunreaktion. Von den Kloakenabstrichen haben wir ökologische Indices von Gleichverteilung, Artenzahl und Diversität der Bakterienzusammensetzung bestimmt. Eine Immunreaktion führte zu Veränderungen in der Gleichverteilung, die von der Unterstützung des Immunsystems mit Lysozymen abhängig war. Bei der statistischen Analyse der Veränderungen haben wir jeweils den entsprechenden Ausgangswert der Bakterienzusammensetzung als Kovariate verwendet, um einen potentiellen statistischen Artefakt auszuschließen. Berechnungen zur Reproduzierbarkeit der Ergebnisse deuten darauf hin, dass Abstriche die Gleichverteilung der Bakterienzusammensetzung verlässlich wiedergeben, nicht jedoch deren Artenzahl und Diversität.


Introduction
Ornithologists are increasingly interested in the physiological relationships between birds and the microbes that they harbor (Kohl 2012 relationships, investigators continue to develop, refine, and employ a variety of molecular genetic techniques, including denaturing gradient gel electrophoresis (DGGE), which can be useful alternatives to culture-based bacteriological techniques (van der Wielen et al. 2002;Grizard et al. 2014). In studies of domestic birds, such molecular analyses of gastrointestinal (GI) bacterial assemblages often utilize samples that are impractical or unethical to collect from wild birds, e.g., terminal samples (van der Wielen et al. 2002). Swabbing orifices, mucosa, and other surfaces of animals is a widely applicable and nonlethal alternative. Recent avian studies include data from swabs of cloacae (Klomp et al. 2008), choanae (Waite et al. 2012), and eggshells (Grizard et al. 2014). However, some swabbing procedures collect only subsets of the bacteria residing in swabbed areas (Grizard et al. 2014). The extent to which a bird's physiological state influences the samples collected via swabbing remains poorly studied. The physiological state of a bird comprises elements that naturally change over timescales both short (e.g., acute phase responses) and long (e.g., seasonal or age-related patterns in reproduction). We investigated whether acute systemic changes in physiology translate into measurable changes in bacterial assemblages in the avian cloaca. To do so, we used DGGE to analyze cloacal swabs that were collected from pigeons before and during a study of the effects of immune supplementation and immune challenge (van de Crommenacker et al. 2010;Matson et al. 2012). While these manipulations clearly affect the physiology of pigeons (van de Crommenacker et al. 2010), the impacts of the same manipulations on the microbial ecology of the cloaca are unknown. Immune supplementation was achieved by oral administration of lysozyme, a non-specific antimicrobial enzyme that is produced endogenously by pigeons (Gavilanes et al. 1982;van de Crommenacker et al. 2010). Immune challenge took the form of an injection with lipopolysaccharide (LPS), a bacterial endotoxin that triggers energetically expensive febrile responses in pigeons (Nomoto 1996;van de Crommenacker et al. 2010) and affects bacterial assemblages at the murine cecum (Sze et al. 2014). We used ecological indices of evenness, richness, and diversity (Hill 1973;Hooper et al. 2005) to compare the LPS-induced changes in bacterial assemblages between the lysozyme-supplemented and control groups (van de Crommenacker et al. 2010).

Study subjects and setup
We studied 16 homing pigeons (Columba livia; University of Groningen Animal Experimentation Committee license no. 5095). Since van de Crommenacker et al. (2010) have comprehensively described the study subjects and setup employed, we provide only key information here.
The flock was divided into lysozyme-treatment and control groups (four $, four # per group). Treatment and control birds underwent similar seven-day experimental routines. On each of the first 6 days, we orally dosed each bird either with 180 mg of lysozyme (from chicken egg white; L6876; Sigma, St. Louis, MO, USA) in 1 mL phosphate-buffered saline (PBS, P4417; Sigma) or with 1 mL of PBS only. On days four and six, birds were removed from their aviary for metabolic measurements, which meant they were isolated and fasted for 22 h. The protocols on days four/five and six/seven were identical with one exception. On day six, 4 h into the 22-h fast, all birds received an endotoxin challenge: an intraperitoneal injection of 2 mL kg -1 body mass of a 1.25 mg mL -1 solution of LPS (from Salmonella enterica serotype typhimurium; L7261; Sigma) dissolved in PBS. Upon removing each bird from its metabolic chamber on the mornings of day five (i.e., 30 h pre-challenge) and day seven (i.e., 18 h post-challenge), we collected the swabs that are the focus of the current report. At the 18 h postchallenge time point, multiple biomarkers (e.g., haptoglobin concentration, metabolic rate, etc.) indicate acute phase responses in pigeons (van de Crommenacker et al. 2010). We also collected swabs ca. 4 months before the experiment as part of a monitoring program. We used these samples to check whether groups differed by chance before the experimental treatments.

Cloacal swab collection
We sampled birds by fully inserting the rayon tip of a sterile swab (112598; LP Italiana, Milan, Italy) into the cloaca and rotating the plastic shaft back and forth for about 10 s. Immediately upon removal from the cloaca, the rayon tip and a small section of adjoining shaft (below the point that was handled) were snapped off into a sterile 1.5-mL tube. Samples were stored at -20°C until further processing.

DNA extraction, PCR, and DGGE
We extracted DNA from 48 cloacal swabs using the Fas-tDNA SPIN Kit for Soil (116560200; MP Biomedicals, Santa Ana, CA, USA) following the manufacturer's protocol.
We used an INGENYphorU system (Ingeny, Goes, Netherlands) for DGGE. We followed the protocols of Brons and van Elsas (2008), except that we used a denaturant gradient range of 45-65 % in our two gels. We scored the DGGE banding patterns using images imported into GelCompar II (Applied Maths, Sint-Martens-Latem, Belgium). One sample (#, lysozyme, post-challenge) failed to produce a visible PCR product or any DGGE bands. This individual was excluded from all analyses due to the repeated measures design.

Bacterial assemblage variables and statistical analyses
We normalized the intensities of all bands within each sample (i.e., values summed to one). Based on the number and the normalized intensities of the bands in each sample, we derived variables to describe the bacterial assemblages using Canoco (version 4.55). We focused on the four variables originally annotated by Hill (1973): total number of bands present, also known as richness (N 0 ); exponent (Shannon's index), also known as effective diversity of Shannon's index (N 1 ); 1/(Simpson's index), also known as effective diversity of the Simpson's index (N 2 ); and evenness (N 2 /N 1 ).
We used linear models (''lm'' in R version 2.15.0) followed by ANOVA (''Anova'' with type = 2 in R package ''car''). Initially all linear models of within-individual LPSinduced changes (D LPS ) in bacterial assemblage variables included sex and group (lysozyme-treated or control) and the interaction between sex and group. These linear models also included the pre-LPS-challenge values of the dependent variable under consideration as a covariate. With P [ 0.4, both sex and the interaction between sex and group were always removed. Final linear models included only group and the covariate. Using linear mixed effects models (''lme'' in R package ''nlme'') with LPS challenge status (pre or post) as a fixed effect and individual identity as a random effect, we calculated the repeatability of the bacterial assemblage variables using the control (non-lysozyme-treated) birds (Lessells and Boag 1987). With the effect of LPS challenge accounted for, a high repeatability would indicate that the corresponding bacterial assemblage variable reflects a characteristic trait of individuals regardless of challenge status.
As an aside, we wish to note a post hoc exploratory analysis in which we employed nonmetric multidimensional scaling (NMDS) to investigate experimental effects on the structure of bacterial assemblages in the cloaca (in contrast to Hill's summary variables). NMDS failed to reveal any visually compelling grouping of assemblage structure based on lysozyme treatment or LPS challenge.

Results
Around 4 months before initiating the experiment, the control and lysozyme-treated groups did not differ in any of the bacterial assemblage variables (F 1,12 \ 1.1, P [ 0.3). Following van de Crommenacker et al. (2010), we calculated descriptive statistics for the four groups in the experiment (Table 1). The experiment revealed that LPS-induced changes in evenness differed between the lysozyme-treated and control groups (Table 2). Moreover, LPS-induced changes in evenness and in effective diversity of the Simpson's index correlated negatively with prechallenge values (Table 2; Fig. 1). Additionally, we conducted auxiliary analyses on four data subsets: we investigated the effects of LPS in the control group and in the lysozyme-treated group (Table 3), and we investigated the effects of lysozyme at the pre-challenge time point and at the post-challenge time point (Table 3). These auxiliary analyses reiterated the importance of the combined treatments (i.e., effects of lysozyme at the post-challenge time point). Evenness exhibited the highest repeatability, R = 0.60 ± 0.23 (standard error; P = 0.06), meaning that this variable might be a characteristic feature of individuals. The same cannot be said of the other variables, which had exceedingly low repeatabilities (R \ 0.01).

Discussion
In ecology, evenness and richness contribute to diversity (Hooper et al. 2005). We showed that LPS-induced changes in the evenness (but not richness or diversity) of bacterial assemblages in the cloaca of pigeons differed between the lysozyme-treated and control groups. LPS challenge led to a larger decrease in evenness in the lysozyme-supplemented group than in the control group. Evenness relates to myriad ecosystem properties (Hooper et al. 2005), including stability, as illustrated by a study showing that high initial evenness makes microbial communities more resistant to environmental stress (Wittebolle et al. 2009). Our result hints at a similar relationship: the lysozyme-treated group, with its lower mean evenness, exhibited a further reduction in evenness after LPS challenge.
Lysozyme supplementation, LPS injection, and their interaction affect avian physiological function, including metabolism and inflammation (van de Crommenacker et al. 2010). However, the same manipulations in the same birds were generally less effective at eliciting differences in bacterial assemblages in the cloaca. One possible reason is that the experimental manipulations did not differentially impact the cloacal microenvironment. For example, LPS-induced increases in cloacal temperature are small (\1°C) and do not differ between lysozyme-treated and control groups (van de Crommenacker et al. 2010). Furthermore, the 22-h fast which preceded the collection of swabs and covered the entire LPS challenge experimental timeline eliminated the possibility of LPS-induced differences in food consumption, and thereby likely lessened other related differences [e.g., in defecation, since mean digesta retention times are\9 h in pigeons (Hatt 2002)]. Similar digestive states might have led to relatively homogeneous cloacal microenvironments.
At the individual level, challenge-induced changes in evenness correlated inversely with pre-challenge values of evenness. This correlation suggests a biologically germane relationship linking initial evenness and resistance to change (Wittebolle et al. 2009). However, caution is warranted since  The difference between the two regression lines suggests that lysozyme treatment affects how LPS challenge influences evenness (F 1,12 \ 5.2, P = 0.04). The data also suggest a statistical artifact known as regression to the mean. Specifically, individuals with low initial evenness generally increased after LPS challenge (6/8 below-average individuals showed increases), while individuals with high initial evenness generally decreased (6/7 aboveaverage individuals showed decreases). The vertical line indicates the overall mean evenness (0.844) individuals with low initial evenness generally increased after LPS challenge (6/8 below-average individuals showed increases) and individuals with high initial evenness generally decreased (6/7 above-average individuals showed decreases). Consequently, these results suggest a statistical artifact, regression to the mean (RTM), which ''can make natural [random] variation in repeated data look like real change'' (Barnett et al. 2005). RTM occurs when measurements further from the mean are followed by measurements closer to the mean. It should be stressed that the identification of RTM at our data analysis stage does not negate the difference between the lysozyme-treated and control groups in the current study. That said, using mean values calculated from two or more baseline swabs could help eliminate RTM in future studies (Barnett et al. 2005). Measurement error can contribute to RTM, yet the high repeatability of evenness suggests that our swab technique is capable of reliably capturing this facet of bacterial assemblages in the cloaca. However, the extremely low repeatabilities of richness and diversity raise concerns about the technique and could help explain the lack of experimental effects. The need to investigate the repeatability of bird-associated bacterial assemblages has been identified previously (Klomp et al. 2008). Such investigations should consider methodological factors (i.e., measurement error) and biological ones, such as transit within birds or transmission between birds (Klomp et al. 2008). Nevertheless, our study sets the stage for future studies of repeatability and more generally suggests that acute and systemic changes in avian physiology can interact to influence bacterial assemblages in the cloaca.