Conspecific coprophagy stimulates normal development in a germ-free model invertebrate

Insects

Periplaneta americana nymphs, adults and ootheca were obtained from a live collection maintained in the Insectary at the Ohio State University Biological Sciences Greenhouse (Columbus, Ohio).

Ootheca treatment

Ootheca were manually detached from gravid P. americana females and surface sterilized by three rounds of cleaning using a detergent scrub (1% Alconox detergent), dilute bleach (0.08% sodium hypochlorite), and then an enzymatic lysis buffer (lysozyme 10 mg/ml, EDTA, Tris-HCl and Triton X-100), each step followed by three rinses with sterile MilliQ water (MQW) to remove antiseptic solution and induce intermittent hypo-osmotic shock on surface-clinging bacteria (Schwinghamer, 1980; Salema et al., 1982). An aliquot of each third MQW rinse was reserved to monitor effectiveness of previous cleanings to reduce bacterial load. Aseptic ootheca were incubated individually in sterile microcentrifuge tubes capped with sterile cotton at 31 °C for approximately 30 days until hatching, which yielded up to 16 GF nymphs per oothecum. No decreased oothecum viability due to washing treatments was observed.

Rearing germ-free P. americana

Cohorts of three to four GF first instar nymphs were aseptically-transferred to sterile rearing chambers of approximately 16 cm³, stocked with gamma-irradiated (aseptic) rat chow and aseptic water, which was renewed weekly. Rearing chambers were passively ventilated with air filtered through a 0.22 µm membrane.

Rearing conventionalized P. americana

To introduce bacteria native to P. americana, cohorts of three to four GF first instars were exposed to frass taken from a lab-maintained colony of non-sterile or “wild-type” (WT) P. americana in lieu of food for 3 days. Subsequently, conventionalized (Conv) first instar nymphs were housed under the same conditions as GF insects except for that sterility was not maintained. As each generation of P. americana typically acquires their gut microbial community through conspecific and nestmate coprophagy (Nalepa, Bignell & Bandi, 2001; Bell, Roth & Nalepa, 2007), the conventionalization approach used reflects the normal route of gut microbe acquisition.

Rearing wild-type P. americana

One-day old first instar insects from 10 ootheca were deposited in an aquarium containing 10 adult male cockroaches from a non-sterile mixed generation colony maintained in the lab and provided with gamma-irradiated rat chow and access to MQW ad libitum; nymphs from this colony were designated “WT”. WT hatchlings were free to interact with adult cockroaches and their frass to facilitate coprophagy and subsequent acquisition of normal gut microbiota. As cannibalism of deceased nestmates is also a putative mechanism for gut microbiota acquisition, late-stage nymphs from the non-sterile colony were sacrificed and deposited in the WT colony. WT insects did not experience spatial constraints or undergo the oothecum sterilization procedure as compared to GF and Conv insects and were exposed to unfiltered air.

Quality control

Quality control measures were employed throughout experiments to ensure maintenance of GF status and to confirm colonization of Conv individuals. Oothecum rinse water, collected between each antiseptic treatment, was plated in triplicate on Luria-Bertani agar (LBA) plates (incubated at 31 °C for up to 1 week) to detect cultivable contaminants flushed from ootheca, and to provide confidence that ootheca are aseptic as they enter incubation. Ootheca shedding no contaminants at final rinses were considered to be aseptic and used for subsequent experiments. At the time of hatching and installation into habitats, one first instar nymph from each oothecum, or the oothecum itself, was sacrificed and homogenized in sterile 1% PBS and plated on LBA in triplicate to confirm aseptic status. At the termination of insect growth, and just prior to insect dissection, frass was collected from rearing chambers, suspended in 1% PBS, and plated to confirm that habitats remained GF throughout the duration of the experiment. Bi-monthly sampling of GF-reared individuals for contamination was performed by homogenizing individuals at various instars and plating homogenate on LBA. During sterile treatment method development, diagnostic PCR with universal 16S primers 27F (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1391R (5′-GACGGGCGGTGTGTRCA-3′) was performed on DNA extracted from homogenates prepared from GF insects and their frass to detect non-cultivable contaminants of the gut or habitat. Additionally, microscopic examination of 4′,6-diamidino-2-phenylindole (DAPI)-stained frass, gut contents, and gut thin-sections were performed to further evaluate the effectiveness of the GF, and conventionalization protocols.

Instar duration measurement

Germ-free and Conv nymphs raised in cohorts of three to four insects were monitored individually for molting activity, and dates of instar transitions of individuals within each cohort were recorded. Aquarium rearing of WT insects in cohorts of dozens of individuals prevented tracking of individual insects and corresponding molt date and developmental times within the WT treatment. Insects dissected at subsequent instars for gut morphological measurements resulted in decreasing sample sizes at later instars, and non-parametric statistical methods were used when making comparisons across instars. Targeting early instar insects constrains the experiment to a relatively short timeframe, as rearing to adulthood is prohibitively long (6–12 months for WT insects), with GF P. americana progression to adulthood uncertain.

Morphological measurements

Nymphs were collected as they molted to third, fourth, and fifth instars, and duration (days) of these instars were recorded. Previous work has shown that the cumulative effects of bacterial colonization, or lack thereof, in the host could be observed in fourth and fifth instars and thus all experimental measures in this study were taken from fifth instar individuals (Bracke, Cruden & Markovetz, 1978; Gijzen & Barugahare, 1992). Nymphs at designated life stages were dissected approximately 5 days after molting, at peak of feeding within instar (Valles, Strong & Koehler, 1996), to minimize variability in gut morphology associated with instar transitions. Eight morphological metrics were collected in millimeters (mm), unless otherwise noted: body length, body width, body mass (grams, g), whole gut length, foregut length, midgut length, hindgut length, and gut mass (g). FIJI image analysis package (ImageJ) was used to perform measurements of the full length of dissected guts (esophagus to anus) and their corresponding carcasses from digital images of these tissues taken immediately following dissection. Results of statistical tests performed on comparisons of morphological measurements are reported in Table S1. Additionally, gut compartments and bodies were traced and lengths measured, with measurements calibrated to a scale in each photo. Mass measurements were also collected for whole GF, Conv, and WT individuals prior to dissection and of their dissected digestive tracts using a microbalance. Additionally, qualitative observations of gut texture, color, opacity, and segmentation were collected from dissected fifth instar GF, Conv, and WT individuals.

Phenotype analysis

All statistics were performed in R using vegan and FSA packages (Ogle, Wheeler & Dinno, 2018; R Development Core Team, 2013; Oksanen et al., 2015). Morphological measurements at instar rather than calendar age yielded differences in sample size, as insects aged out of instar classes at different rates; non-parametric statistics were utilized to accommodate sample size variation. Instar duration, and morphological measurements were analyzed for statistical differences using a Mann–Whitney Test for pairwise-comparisons. Kruskal–Wallis and Dunn Tests were performed for multiple comparisons, with Bonferroni correction-adjusted p-values. Principal component analysis (PCA) was performed to simultaneously examine multiple morphological variables across treatments. The function “Varpart” within the vegan R package was used to partition variance among morphological variables and the PCA was constrained to the three variables representing 93% of the variation. Subsequently, multi-response permutation procedure (MRPP) was used to assess the significance of the observed differences between treatment centroids.

Germ-free insects exhibited reduced width and prolonged instar duration

Germ-free (n = 37) P. americana individuals remained in each instar an average of two days longer than Conv (n = 34) individuals (Fig. 1A), which resulted in longer times between molts in GF insects and prolonged developmental periods (Fig. 1B) being observed. While Conv insects molted to fifth instar after an average of 32.3 days, GF insects required an average of 38.9 days to reach the same life stage. While stadium duration of WT insects was not obtained for every instar because of the complexity of tracking individual insects in a large cohort, average age at fifth instar was 30.2 days (Fig. S1). GF insects also exhibited the lowest average body width (3.84 mm) when compared to WT (4.18 mm) and Conv (4.08 mm) insects (Fig. 1C). Body lengths of GF and Conv individuals were not significantly different at third, fourth, or fifth instars (Fig. 1D), and body mass did not differ between treatments (Fig. S2).

Conventionalization of germ-free P. americana recovers normal hindgut tissue morphology and development

Distinct visual contrasts between fifth instar WT (n = 23) (Fig. 2A; Figs. S3A–S3C) and GF (n = 32) (Fig. 2B; Figs. S3D–S3F) hindgut tissues were consistently evident. WT P. americana hindguts are characterized by numerous lateral folds along the length of the hindgut that are visible as an undulating gut margin with lateral creases (Fig. 2D). All inspected WT hindguts were opaque, turgid in texture, and generally filled with mustard-yellow hindgut contents (Fig. 2A; Figs. S3A–S3C). Conv hindguts (Fig. 2C; Figs. S3G–S3I) were similar to those from WT insects in that they were turgid in texture and more opaque than GF hindguts (Fig. 2B; Figs. S3D–S3F), and Conv hindguts exhibited frequent lateral folds and coloration characteristic of WT insects. Further, dissection of Conv revealed mustard-yellow contents, similar to those observed in WT digestive tracts, and this appears to be a mix of digestate and bacterial biomass (note the fluorescence of material in gut lumen in Fig. 3) lodged in hindgut folds (Fig. 3). DAPI stained hindgut tissue thin sections reveal a dense mat of bacterial biomass adjacent to the host gut epithelial lining and planktonic bacterial growth deep within the lumen of WT and Conv insects (Fig. 3). In contrast, GF hindguts were uniformly flaccid, translucent, and displayed less frequent and less pronounced lateral folding (Fig. 2B; Figs. S2D–S2F). Additionally, gut contents were comprised of partially digested diet and no evidence of bacteria were observed (Fig. 3). No amplicon was observed after diagnostic PCR conducted on GF frass DNA extract (Fig. S4) using bacteria-specific primers. Despite thorough cleaning of fat bodies from guts at the time of dissection, the fat body specific endosymbiont Blattabacterium was detected in diagnostic PCR reactions of gut tissue DNA extracts from GF insects, negating the utility of bacteria-specific primers for contamination detection in gut homogenates by diagnostic PCR.

Average length of the complete fore-to-hindgut tissues was shorter in GF insects (17.13 mm) than in both Conv (18.96 mm) and WT (21.24 mm) insects (Fig. 4A; Dunn Test: GF-Conv p = 0.0244, Conv-WT p = 0.0028, GF-WT p = 0.0054). When fore- (Fig. S5) and mid-gut (Fig. 4B) sections were examined separately, average lengths did not differ significantly between any treatments at any instar, except for between GF (3.30 mm) and either WT (4.83 mm) or Conv (3.48 mm) third instar midguts (Dunn Test: WT-GF p = 0.0058, Conv-GF p = 0.0455). Conversely, hindgut length was significantly reduced in GF insects at third, fourth, and fifth instars relative to either Conv or WT insects (Fig. 4C). Finally, significant differences in gut mass were only detected at fourth instar between GF and WT insects (Fig. S6), with respective masses of 5.2 and 6.7 mg (Dunn Test: WT-GF p < 0.0070), but variability was high within treatments.

Wild-type, conventionalized, and germ-free insects bear morphological signatures that define a gradient across treatments

Of eight morphological variables examined in this study whole gut length, midgut length, and hindgut length explained 91.4% of variance across treatments (Fig. S7). PCA was performed to examine the relationship among treatments at each instar in reference to these three response variables. Ordination was conducted on the first two principal components, representing 84.5% and 6.9% of variance explained, respectively (Fig. 5). Treatment centroids were significantly different (MRPP: p < 0.001), with the Conv centroid being intermediate to that of GF and WT centroid. All vector loadings associated with morphological variables were significant, with strong linearity (p < 0.001, whole gut length r² = 0.99, midgut length r² = 0.91, hindgut length r² = 0.89). Loading vectors for hindgut length and whole gut length demonstrate a gradient of low values in proximity to the GF centroid to high values toward the WT centroid. The vector for midgut length appears to follow a gradient associated with within-instar variation, as midgut length varies more within instar than across treatments. A second PCA was performed on data from all instars and treatments, simultaneously, with 93.1% and 1.9% of variance explained by the first two axes, yielding greater linearity of loading vectors but reduced separation of centroids (Fig. S8). With multiple instars represented within each treatment, vector loadings previously associated with within-instar variation (i.e., midgut length) realign to describe an across-treatment gradient, implying that variation in midgut length is better explained by treatment than instar or within-instar effects.