Insect’s intestinal organ for symbiont sorting

Significance

In general, animals have a mouth for feeding, an anus for defecation, and a gut connecting them for digestion and absorption. However, we discovered that the stinkbug’s gut is functionally disconnected in the middle by a previously unrecognized organ for symbiont sorting, which blocks food fluid and nonsymbiotic bacteria but selectively allows passing of a specific bacterial symbiont. Though very tiny and inconspicuous, the organ governs the configuration and specificity of stinkbug gut symbiosis, wherein the posterior gut region is devoid of food flow, populated by a specific bacterial symbiont, and transformed into an isolated organ for symbiosis. Mutant analyses showed that the symbiont’s flagellar motility is needed for passing the host organ, highlighting intricate host–symbiont interactions underpinning the symbiont sorting process.

Abstract

Symbiosis has significantly contributed to organismal adaptation and diversification. For establishment and maintenance of such host–symbiont associations, host organisms must have evolved mechanisms for selective incorporation, accommodation, and maintenance of their specific microbial partners. Here we report the discovery of a previously unrecognized type of animal organ for symbiont sorting. In the bean bug Riptortus pedestris, the posterior midgut is morphologically differentiated for harboring specific symbiotic bacteria of a beneficial nature. The sorting organ lies in the middle of the intestine as a constricted region, which partitions the midgut into an anterior nonsymbiotic region and a posterior symbiotic region. Oral administration of GFP-labeled Burkholderia symbionts to nymphal stinkbugs showed that the symbionts pass through the constricted region and colonize the posterior midgut. However, administration of food colorings revealed that food fluid enters neither the constricted region nor the posterior midgut, indicating selective symbiont passage at the constricted region and functional isolation of the posterior midgut for symbiosis. Coadministration of the GFP-labeled symbiont and red fluorescent protein-labeled Escherichia coli unveiled selective passage of the symbiont and blockage of E. coli at the constricted region, demonstrating the organ’s ability to discriminate the specific bacterial symbiont from nonsymbiotic bacteria. Transposon mutagenesis and screening revealed that symbiont mutants in flagella-related genes fail to pass through the constricted region, highlighting that both host’s control and symbiont’s motility are involved in the sorting process. The blocking of food flow at the constricted region is conserved among diverse stinkbug groups, suggesting the evolutionary origin of the intestinal organ in their common ancestor.

Diverse organisms are obligatorily associated with microbial symbionts, which significantly contribute to their adaptation and survival (1–3). In such symbiotic associations, the host organisms often develop specialized cells, tissues, or organs for harboring their specific microbial partners [for example, root nodules in the legume–Rhizobium symbiosis (4, 5), symbiotic light organs in the squid–Vibrio symbiosis (6, 7), and bacteriocytes in the aphid–Buchnera symbiosis (8, 9)].

These microbial symbionts are either acquired by newborn hosts from the environment every generation as in the legume–Rhizobium and the squid–Vibrio symbioses or transmitted vertically through host generations as in the aphid–Buchnera symbiosis (10). In the environmentally acquired symbiotic associations, it is essential for the host organisms to recognize and incorporate specific symbiotic bacteria while excluding a myriad of nonsymbiotic environmental microbes (6, 11). In the vertically transmitted symbiotic associations, it is important for the host organisms to selectively transmit their own symbiotic bacteria while excluding parasitic/cheating microbial contaminants (12–14). Hence, it is expected that the host organisms must have evolved some mechanisms for selective incorporation, accommodation, and maintenance of their specific microbial partners. Those controlling mechanisms are of general importance for understanding symbiosis (6, 10).

Stinkbugs, belonging to the insect order Hemiptera, consist of over 40,000 described species in the world (15). The majority of the stinkbugs suck plant sap or tissues, and some of them are notorious as devastating agricultural pests (16). These plant-sucking stinkbugs possess a specialized symbiotic organ in their alimentary tract: A posterior region of the midgut is morphologically differentiated with a number of sacs or tubular outgrowths, called crypts or ceca, whose inner cavity hosts symbiotic bacteria (17–21). Usually, a single bacterial species dominates in the midgut crypts, and elimination of the symbiont causes retarded growth and increased mortality of the host, which indicates the specific and beneficial nature of the stinkbug gut symbiosis (20–31). The initial symbiont infection is established by nymphal feeding, which may be either via vertical transmission from symbiont-containing maternal secretion supplied upon oviposition (19–21) or via environmental acquisition from ambient microbiota (21–23). What mechanisms underlie the selective establishment of a specific bacterial symbiont in the midgut symbiotic organ despite the oral inoculum contaminated by nonsymbiotic microbes has remained largely an enigma, although recent studies have started to shed light on the symbiotic mechanisms underlying the environmental acquisition of specific Burkholderia symbionts in the bean bug Riptortus pedestris (Hemiptera: Alydidae) (22, 32). Antimicrobial substances produced by the midgut epithelia (33, 34) and some symbiont factors, such as stress-responsive polyester accumulation, cell wall synthesis, and purine biosynthesis (35–37) might be involved in the selective infection of the Burkholderia symbiont to the midgut crypts.

Here we address this important symbiotic issue by the discovery of a previously unrecognized intestinal organ in the stinkbugs. Though very tiny and inconspicuous, the organ governs the configuration and specificity of the stinkbug gut symbiosis. Lying in the middle of the midgut, the organ blocks food flow and nonsymbiotic bacteria but selectively allows passing of specific symbiotic bacteria, whereby the stinkbug’s intestine is functionally partitioned into the anterior region specialized for digestion and absorption and the posterior region dedicated to symbiosis. The blocking of food flow by the organ is conserved across diverse stinkbug families, suggesting the possibility that the organ evolved in their common ancestor and has played substantial roles in their symbiont-mediated adaptation and diversification.

Results and Discussion

Identification of Constricted Region in Stinkbug Midgut.

As in diverse other stinkbugs (17–21), the midgut of the bean bug R. pedestris consists of the following morphologically distinct regions: the voluminous midgut first section (M1), the tubular midgut second section (M2), the ovoid midgut third section (M3), and the midgut fourth section (M4) with numerous crypts densely populated by a specific betaproteobacterial symbiont of the genus Burkholderia, which is orally acquired by nymphal insects from the environment every generation (22, 23, 32) (Fig. 1A). A swollen region adorally connected to the M4 is without crypts and called M4 bulb (M4B) (18, 34, 38) (Fig. 1A). Although biological roles of each section are not exactly known, it is conjectured that the M1 serves for transient food storage and digestion, the M2 and the M3 perform food digestion and absorption, and the M4 is specialized for harboring the symbiotic bacteria. Infection with a GFP-labeled strain of the Burkholderia symbiont (38) revealed symbiont localization to the M4B and the M4 (Fig. 1B). Between the nonsymbiotic M3 and the symbiotic M4B a particularly narrow intestinal region was present, which we designated as “constricted region” in this study (Fig. 1C). Probably because of being very tiny and inconspicuous structurally, the constricted region has attracted little attention in previous anatomical studies on the alimentary tract of diverse stinkbugs (17–21). However, we discovered that the constricted region plays a pivotal role in the stinkbug gut symbiosis, as detailed below.

Fig. 1.

Midgut organization and the constricted region of R. pedestris. (A) Alimentary tract dissected from a third instar nymph. (B) Symbiont localization in the M4B and M4 regions of the dissected alimentary tract visualized using a GFP-labeled Burkholderia symbiont strain. Green and blue signals indicate symbiont cells and host’s nuclear DNA, respectively. (C) Enlarged image of the constricted region. Note that the symbiont signal is preferentially detected in M4B and not in M3. (D) Sectioned image of the constricted region of a second instar nymph, which was stained with periodic acid–Schiff reagent (red) and hematoxylin (purple). (Inset) An enlarged image of the cavity of the constricted region. Note that the narrow lumen of the constricted region is filled with periodic acid–Schiff-positive material, indicating the presence of polysaccharide-rich mucous matrix. CR, constricted region; H, hindgut; M1, midgut first section; M2, midgut second section; M3, midgut third section; M4, midgut fourth section with crypts; M4B, M4 bulb. Arrows indicate the constricted region.

Burkholderia Symbiont Passing Through Constricted Region.

Histological inspection revealed that, although very narrow, a lumen was present in the constricted region, connecting the inner cavities of the M3 and the M4B (Fig. 1D). Our previous work showed that when newly molted second instar nymphs were fed with GFP-labeled Burkholderia the symbiont cells aggregated at the entrance of the constricted region around 6 h after inoculation, subsequently migrated into the narrow lumen of the constricted region, and finally reached the M4B and the M4 (38). These observations suggested that passing of the constricted region comprises a rate-limiting step, or a bottleneck, for symbiont colonization to the midgut symbiotic regions M4B and M4.

Oral Administration of Food Colorings Unveiled Restricted Food Flow at Constricted Region.

When third instar nymphs that had been inoculated with the Burkholderia symbiont at the second instar were fed with water supplemented with Congo red, the coloring stained the M1, the M2, and the M3 but, strikingly, never appeared in the M4B and the M4 (Fig. 2 A and B). This phenomenon was found not only in third instar nymphs but also throughout all developmental stages including first, second, third, fourth, and fifth instars (Fig. 2 C–N). Furthermore, this phenomenon was consistently observed not only with Congo red but also with other food colorings of different color (Fig. 2 O–T). Meanwhile, the colorings appeared in the hindgut (Fig. 2 A and O–T) and were excreted with feces (Fig. 2U). These results strongly suggested that (i) food fluid ingested by R. pedestris enters the M1, the M2, and the M3 but cannot pass the constricted region; (ii) the fluid is absorbed in the anterior regions of the midgut, excreted through Malpighian tubules into the hindgut, and discarded with feces; (iii) as a result, the M4B and the M4 do not contribute to the food flow; and (iv) therefore, the M4B and the M4 are, although structurally connected to adoral and aboral regions of the intestine, functionally isolated and specialized for harboring the Burkholderia symbiont.

Fig. 2.

Dissected alimentary tracts of R. pedestris fed with water supplemented with food colorings. (A–N) Congo red. (O) New coccine. (P) Brilliant blue FCF. (Q) Gardenia yellow. (R) Purple sweet potato color. (S) Gardenia yellow and Gardenia blue. (T) Kaoliang color. The developmental stages of R. pedestris are at the first instar (C–D), the second instar (E–J), the third instar (A and B and O–T), the fourth instar (K and L) and the fifth instar (M and N). Abbreviations are as in Fig. 1. (U) Third instar nymphs of R. pedestris fed with soybean seeds and water supplemented with New coccine. Arrowheads indicate feces of the insects, in which the red coloring is excreted.

Coadministration of GFP-Labeled Burkholderia Symbiont and Food Coloring Revealed Selective Symbiont Passing at Constricted Region.

When newly molted second instar nymphs were fed with a mixture of Congo red and the GFP-labeled Burkholderia cells, both the coloring and the bacteria were found in the M1, the M2, and the M3. However, whereas the coloring did not enter the constricted region (Fig. 3A), the Burkholderia cells appeared in the narrow inner cavity of the constricted region (Fig. 3 B and C). These results indicated that the constricted region is involved in selective passing of the Burkholderia symbiont while restricting passing of the food fluid.

Fig. 3.

Symbiont sorting at the constricted region of R. pedestris. (A–C) Coinoculation of Congo red and a GFP-labeled Burkholderia symbiont strain. Second instar nymphs of R. pedestris were fed with water containing the coloring and the bacteria, and their alimentary tracts were dissected and observed 6 h after inoculation. (A) Differential interference contrast microscopic (DIC) image, wherein Congo red signal is observed in the M3 only. (B) Fluorescence microscopic (FM) image, wherein green Burkholderia symbiont signals are found in the constricted region. (C) Merged image. (D–F) Coinoculation of the GFP-labeled Burkholderia symbiont strain and an RFP-labeled E. coli strain. Second instar nymphs of R. pedestris were fed with water containing the bacterial mixture, and their alimentary tracts were dissected and observed 6 h after inoculation. Green and red signals indicate the Burkholderia symbiont cells and the E. coli cells, respectively. (D) FM image. (E) Laser scanning microscopic (LSM) image. (F) Enlarged LSM image of the constricted region, wherein red E. coli cells are indicated by arrowheads. Abbreviations are as in Fig. 1.

Coadministration of GFP-Labeled Burkholderia Symbiont and Red Fluorescent Protein-Labeled Escherichia coli Uncovered Selective Symbiont Sorting at Constricted Region.

When newly molted second instar nymphs were fed with a mixture of the GFP-labeled Burkholderia cells and red fluorescent protein (RFP)-labeled E. coli cells (10⁴ cells per μL each), both the Burkholderia cells and the E. coli cells were found in the M1, the M2, and the M3 (Fig. 3 D and E). However, whereas the Burkholderia cells soon appeared in the constricted region, the M4B and the M4, the E. coli cells were not observed in these posterior midgut regions (infected insects/inoculated insects = 0/5) (Fig. 3 E and F). In addition to E. coli, orally administrated Pseudomonas putida and Bacillus subtilis did not colonize the M4B and the M4 as determined by plating and counting cfu of dissected midgut regions (infected insects/inoculated insects = 0/5, respectively). These results corroborated that the constricted region is involved in selective passing of the Burkholderia symbiont and suggested that the selection mechanism can discriminate the Burkholderia symbiont from nonsymbiotic bacteria.

Constricted Region As a Host Organ for Selective Symbiont Sorting.

In this way, we identified a previously unknown type of intestinal organ, the constricted region, whose function is to ensure selective passing of the Burkholderia symbiont to the posterior symbiotic midgut. Considering that R. pedestris orally acquires the Burkholderia symbiont from the microbe-rich environment every generation (22, 23), this organ must play a central role for establishment and maintenance of the specific Riptortus–Burkholderia symbiotic association. The M4 crypts are efficiently colonized by the Burkholderia symbiont during the second and third instars, whereas the colonization efficiency drops significantly from the fourth instar and afterward, indicating that there exists a permissive time window for infection (39). Whether the constricted region is involved in the establishment of the infection time window is of interest and deserves future studies.

Fine Structure of Constricted Region.

We observed the constricted regions dissected from second instar nymphs of R. pedestris using light microscopy (Figs. 1D and 4A) and transmission electron microscopy (Fig. 4 B–E). The lumen of the M3 was filled with the symbiont cells, and the inner surface of the M3 was covered with a well-developed layer of microvilli (Fig. 4 C and D). The inner surface of the constricted region was also covered with a dense layer of microvilli, which occupied most of the narrow inner space (Fig. 4 B, C, and E). Only a very thin canal was left at the center of the constricted region, which was one or a few micrometers in width and filled with a slightly electron-dense material (Fig. 4 B, C, and E). Periodic acid–Schiff staining of semiultrathin sections of the constricted region identified the material in the canal as polysaccharide-rich, presumably a mucous matrix (Fig. 1D). These observations suggested that (i) the inner cavity of the constricted region is nearly filled with the highly developed microvilli layer, (ii) the very thin canal left at the center of the constricted region is filled with the mucous matrix, (iii) the microvilli and the mucous matrix plug the inner cavity of the constricted region, thereby hindering the passing of food colorings and E. coli cells; (iv) however, the Burkholderia symbiont can manage to pass through the constricted region despite these obstacles, (v) and, therefore, although speculative, it seems likely that the Burkholderia symbiont somehow interacts with the microvilli and/or the mucous matrix for passing through the constricted region. Because no ciliary layer was observed on the inner surface of the constricted region, host-driven movement of the symbiont cells seemed unlikely. Hence, we suspected that motility of the Burkholderia symbiont may play a role in passing through the constricted region.

Fig. 4.

Fine structure of the constricted region in second instar nymphs of R. pedestris. (A) Semiultrathin section stained with toluidine blue and observed by light microscopy. (B–E) Ultrathin sections observed by transmission electron microscopy. (B) Longitudinal section of the constricted region, whose inner surface is lined with a dense layer of microvilli, leaving a very narrow canal at the center. (C) Posterior end region of the M3 in connection to the constricted region. The inner surface of the M3 and of the constricted region is lined with a dense layer of microvilli. The inner cavity of the M3 harbors numerous Burkholderia symbiont cells, whereas the inner canal of the constricted region is full of a mucous matrix. (D) Enlarged image of the M3 cavity. (E) Enlarged image of the constricted region. *, inner canal of constricted region; CR, constricted region; ML, microvilli layer; S, Burkholderia symbiont cell.

Transposon Mutagenesis and Screening for Motility-Deficient Mutants of Burkholderia Symbiont.

We performed a random transposon-insertion mutagenesis of the Burkholderia symbiont using the miniTn5 system (40), and resultant transconjugants were screened for motility on 0.4% semisolid nutrition agar. Of 6,212 transconjugants subjected to the assay, 7 were identified as motility-deficient. Furthermore, an additional 4,252 transconjugants were subjected to screening for biofilm formation, because previous studies had reported that biofilm-deficient bacterial mutants also tend to be motility-deficient (41, 42). Of 25 biofilm-deficient mutants obtained, 7 were confirmed as motility-deficient by the semisolid agar assay. Southern hybridization analysis detected a single Tn5 insertion in 13 of the 14 motility mutants. In this way, we obtained 13 Tn5-inserted motility-deficient mutants of the Burkholderia symbiont (Fig. 5A). Of these, 9 mutants without visible motility in liquid medium under light microscopy were designated as no-motility mutants NM1-NM9, whereas the other 4 mutants that were motile in liquid medium but not in semisolid agar were designated as altered-motility mutants AM1–AM4 (Table S1). Most of the mutants contained a Tn5 insertion associated with bacterial flagella-related genes: a flagellin gene (fliC) in NM1, an MS-ring gene (fliF) in NM2-NM6, a motor apparatus gene (fliM) in NM7, a flagellar export apparatus gene (fliR) in NM8, a rod regulation gene (flhA) in NM9, a regulation gene for hook length (fliK) in AM1, and a regulation gene for flagellar formation (flhF) in AM2 (Fig. 5 B and C). In addition, a chemotaxis-related gene, cheA, was disrupted by the transposon in the mutant AM3, whereas a putative gene of unknown function contained a Tn5 insertion in the mutant AM4 (Fig. 5B). All these genes were single-copy in the Burkholderia symbiont genome (43).

Fig. 5.

Motility-deficient mutants of the Burkholderia symbiont obtained by transposon mutagenesis and screening. (A) Growth of no-motility mutants (NM1–NM9) and altered-motility mutants (AM1–AM4) on a semisolid YG agar plate. Note the limited diffusion of the mutant colonies in comparison with the wild-type symbiont colony. (B) Transposon insertion sites for the motility-deficient mutants. Note that the majority of the disrupted genes are flagella-related. (C) Structural presentation of the flagella-related genes disrupted in the motility-deficient mutants.

Table S1.

Motility mutants of the Burkholderia symbiont

Symbiont strain	Disrupted gene*		Function of disrupted gene†	Flagella formation rate‡	Colonization rate to M4§
Original wild-type strain
RPE75	—		—	100% (10/10)	100% (9/9)
Tn-inserted no-motility mutants
NM1	fliC		Flagellin	0% (0/17)	0% (0/10)
NM2	fliF		MS-ring	0% (0/17)	0% (0/10)
NM3	fliF		MS-ring	0% (0/21)	0% (0/10)
NM4	fliF		MS-ring	0% (0/19)	0% (0/8)
NM5	fliF		MS-ring	0% (0/17)	0% (0/10)
NM6	fliF		MS-ring	0% (0/19)	40.0% (4/10)
NM7	fliM		Motor switch	40.0% (8/20)	0% (0/10)
NM8	fliR		Flagellar export apparatus	0% (0/18)	0% (0/8)
NM9	flhA		Rod	0% (0/15)	0% (0/10)
Tn-inserted altered-motility mutants
AM1	fliK		Regulation of hook length	78.6% (11/14)	100% (9/9)
AM2	flhF		Regulation of flagellation	31.3% (5/16)	100% (12/12)
AM3	cheA		Chemotaxis	84.6% (11/13)	100% (11/11)
AM4	putative		Unknown	92.3% (12/13)	50.0% (4/8)
fliC disrupted no-motility mutant (NM1) transformed with a blank plasmid pBBR122
NM1pBBR122	fliC		Flagellin	0% (0/17)	0% (0/11)
fliC disrupted no-motility mutant (NM1) complemented with a fliC-containing pBBR122
NM1fliC	fliC		Flagellin	92.3% (12/13)	90.0% (9/10)
Gene-deleted no-motility mutant generated by homologous recombination
ΔfliF	fliF		MS-ring	0% (0/26)	0% (0/12)

^*For Tn-inserted mutants, the insertion sites are shown in Fig. 5B. For ΔfliF mutant, fliF gene is completely deleted by homologous recombination.

^†Corresponding flagellar structures are illustrated in Fig. 5C.

^‡Number of flagellated cells per total number of cells inspected by electron microscopy (also see Fig. 6).

^§Number of infected insects per total number of insects inspected by diagnostic PCR.

Flagella formation of the 13 motility-deficient mutants was determined by electron microscopy (Fig. 6). Whereas the wild-type strain RPE75 exhibited multiple polar flagella (Fig. 6N), most of the no-motility mutants had no or few flagella (Fig. 6 A–I and Table S1). Some cells of the mutant NM7 (fliM⁻) formed morphologically intact flagella (Fig. 6G), but the mutant showed no motility, probably because of its disrupted motor apparatus. All of the altered-motility mutants AM1–AM4 were equipped with flagella (Fig. 6 J–M). The mutant AM2 (flhF⁻) formed a single flagellum on a lateral side of the cell (Fig. 6K) as reported in flhF mutants of Pseudomonas aeruginosa (44) and Vibrio alginolyticus (45).

Fig. 6.

Transmission electron microscopy of negatively stained bacterial cells of the motility-deficient mutants of the Burkholderia symbiont. (A) NM1 (fliC⁻) without flagella. (B) NM2 (fliF⁻) without flagella. (C) NM3 (fliF⁻) without flagella. (D) NM4 (fliF⁻) without flagella. (E) NM5 (fliF⁻) without flagella. (F) NM6 (fliF⁻) without flagella. (G) NM7 (fliM⁻) with flagella, though the majority of the bacterial cells are without flagella. (H) NM8 (fliF⁻) without flagella. (I) NM9 (flhA⁻) without flagella. (J) AM1 (fliK⁻) with flagella. (K) AM2 (flhF⁻) with an abnormal lateral flagellum. (L) AM3 (cheA⁻) with flagella. (M) AM4 (no homology gene⁻) with flagella. (N) RPE75 (wild type) with flagella. Bacterial cells at a midexponential growth phase are shown. Arrowheads indicate flagella. For details, also see Table S1. (Scale bars, 0.5 µm.)

Flagellar Motility Is Important for Passing Through Constricted Region.

Colonization ability of the 13 motility-deficient mutants to the symbiotic midgut regions M4B and M4 was investigated by oral administration of the cultured bacteria to second instar nymphs of R. pedestris. When the posterior midgut regions were dissected from the inoculated insects at the third instar and subjected to diagnostic PCR, all 4 altered-motility mutants exhibited infection rates ranging from 50 to 100%, whereas 8 of 9 no-motility mutants did not infect the symbiotic midgut regions (Table S1). The exceptional no-motility mutant NM6, which carries a Tn5 insertion near the 5′ end of fliF (Fig. 5B), exhibited a partial infection rate of 40%, which might be, although speculative, due to leaky or conditional expression of the flagella-related gene from a promoter encoded within the Tn5 cassette. The essentiality of fliF for flagella formation, motility, and colonization to the symbiotic midgut regions was confirmed by inspection of a fliF-deleted mutant, ΔfliF, generated by homologous recombination (Fig. 7 H–J and Table S1). These results strongly suggested that the flagellar motility of the Burkholderia mutants is important for passing through the constricted region and infection to the symbiotic midgut regions M4B and M4.

Fig. 7.

Colonization ability of the motility-deficient mutants of the Burkholderia symbiont to the symbiotic midgut regions of R. pedestris. (A) Normal motility of the wild-type symbiont RPE75 on semisolid agar. (B) No motility of the fliC⁻ mutant NM1^pBBR122 on semisolid agar. (C) Restored motility of the complemented fliC⁻ mutant NM1^fliC on semisolid agar. (D) RPE75 cell with normal flagella (arrowheads). (E) NM1^pBBR122 cell without flagella. (F) NM1^fliC cell with restored flagella (arrowhead). (G) Colonization ability of the wild-type symbiont RPE75, the fliC⁻ mutant NM1^pBBR122, and the complemented fliC⁻ mutant NM1^fliC to the midgut symbiotic regions of R. pedestris. (H) Normal motility of the wild-type symbiont RPE75 on semisolid agar. (I) No motility of the fliF⁻ mutant ΔfliF on semisolid agar. (J) ΔfliF cell without flagella. (K) Colonization ability of the fliK⁻ altered-motility mutant AM1, the flhF⁻ altered-motility mutant AM2, the fliF⁻ no-motility mutant NM2, and the fliF⁻ gene deletion mutant ΔfliF to the midgut symbiotic regions of R. pedestris. Symbiont colonization was evaluated 48 h after oral administration to second instar nymphs. Means and SDs of cfu per dissected tissue (n = 3 or 4) are shown. Asterisks indicate statistically significant differences between the strains (Kruskal–Wallis test; P < 0.05).

To quantitatively evaluate the infection process of the motility-deficient mutants, second instar nymphs of R. pedestris were orally administrated with the cultured bacteria and dissected 48 h after inoculation. The dissected midgut sections (M1, M2, M3, M4B, and M4) and hindgut were homogenized and spread on nutrition agar plates for counting cfu. The wild-type strain RPE75 was motile, flagellated, and detected in all of the midgut sections, being the most abundant in the M4 section at around 10⁵ cfu per tissue (Fig. 7 A, D, and G). By contrast, the fliC⁻ no-motility mutant NM1^pBBR122 (NM1 transformed with the empty plasmid pBBR122) was not motile, without flagella, and not detected in the M4B and M4 sections at all, although a high level of infection was detected in the M3 section at around 10⁶ cfu per tissue (Fig. 7 B, E, and G). The genetically complemented fliC⁻ mutant NM1^fliC (NM1 transformed with a pBBR122 plasmid containing intact fliC gene) restored flagella formation, motility, and colonization ability to the M4B and M4 sections (Fig. 7 C, F, and G). Furthermore, the altered-motility mutants AM1 (fliK⁻) and AM2 (flhF⁻) were detected in the M4B and M4 sections, whereas the no-motility mutants NM2 and ΔfliF (both fliF⁻) were scarcely detected in the symbiotic midgut regions (Fig. 7K). These results further corroborated the idea that the flagellar motility of the Burkholderia symbiont is important for passing through the constricted region and infection to the symbiotic midgut regions M4B and M4.

Possible Mechanisms of Selective Symbiont Sorting at Constricted Region.

It should be noted that, although E. coli, P. putida, and B. subtilis are motile with functional flagella, they were blocked at the sorting organ and unable to gain entry into the symbiotic midgut regions, which indicate that bacterial flagellar motility is indeed necessary but not sufficient for passing through the constricted region in R. pedestris. What mechanisms are involved in the selective passing of the Burkholderia symbiont through the constricted region is currently unknown and deserves future studies. It is conceivable, although speculative, that the Burkholderia symbiont may be capable of penetrating the constricted region by, for example, excreting specific enzymes that degrade the mucous matrix plugging the inner cavity of the constricted region. Alternatively, the constricted region may be producing some antimicrobials, such as antimicrobial peptides and lysozymes (33, 34), to which nonsymbiotic bacteria are sensitive but the Burkholderia symbiont is resistant. We have previously identified several bacterial factors needed for stable symbiont colonization in the M4 crypts (35–37), which might also be involved in the symbiont sorting process. Host’s control mechanisms over symbiont’s population and localization have been demonstrated or suggested in aphids, weevils, and other insect–microbe symbiotic associations (46–48). Transcriptomic analyses of the constricted region will provide further insights into the molecular mechanisms operating at the symbiont sorting machinery. Taken together, intricate host–symbiont interactions at the constricted region, where both the symbiont motility and the host selection play important roles, must be involved in the selective colonization and establishment of the Burkholderia symbiont in the midgut symbiotic organ of R. pedestris, as has been established for the colonization of the symbiotic light organ in the squid–Vibrio symbiosis (6, 7).

General Relevance of Constricted Region to Gut Symbiosis in Diverse Stinkbugs.

The structural differentiation of the midgut into the morphologically distinct regions M1, M2, M3, M4B, and M4 (though M4B is not obvious in some species) has been observed across diverse plant-sucking heteropteran bugs (17–20, 49). In an attempt to gain insight into the role of the constricted region in general, adult insects of the following stinkbug species, which represent different stinkbug families and harbor specific symbiotic bacteria within the M4 crypts, were subjected to feeding of water supplemented with Congo red and inspection of their alimentary tract: in addition to R. pedestris (the family Alydidae), Cletus punctiger and Acanthocoris sordidus (Coreidae), Togo hemipterus and Paromius exiguus (Rhyparochromidae), Dolycoris baccarum (Pentatomidae), Poecilocoris lewisi (Scutelleridae), Megymenum gracilicorne (Dinidridae), Adomerus triguttulus (Cydnidae), Elasmucha putoni (Acanthosomatidae), and Megacopta punctatissima (Plataspidae). In all of the stinkbug species examined, strikingly, the coloring stained the M1, the M2, and the M3 but never appeared in the M4B and the M4 (Fig. 8), suggesting that the sorting role of the constricted region is not restricted to R. pedestris but is commonly found across the diverse stinkbug groups. The stinkbugs of the superfamilies Coreoidea and Lygaeoidea, including R. pedestris, C. punctiger, A. sordidus, T. hemipterus, and P. exiguus, are associated with betaproteobacterial gut symbionts of the genus Burkholderia (23, 32), which are orally acquired by nymphal insects mainly from the environment every generation (22, 23, 50). In these species, the constricted region must play a pivotal role in selectively picking up the Burkholderia symbiont from the diverse environmental microbiota. On the other hand, the stinkbugs of the superfamily Pentatomoidea, including D. baccarum, P. lewisi, M. gracilicorne, A. triguttulus, E. putoni, and M. punctatissima, are associated with gammaproteobacterial gut symbionts of the family Enterobacteriaceae (24, 25, 30, 51–53), which are vertically transmitted through host generations by newborn’s feeding on either symbiont-containing excrements smeared on the egg surface (21, 24, 26) or symbiont-encasing capsules deposited near the eggs (25, 27, 54). In these species, although speculative, the constricted region may play a role in excluding microbial contaminants for ensuring stable vertical transmission of the specific bacterial symbiont. Also, it should be noted that the constricted region functionally partitions the stinkbug’s alimentary tract into the anterior region for digestion and absorption and the posterior region specialized for symbiosis. In A. sordidus and M. punctatissima, strikingly, their constricted region was reduced to a thread-like connective tissue, whereby the anterior region and the posterior region of their alimentary tract were structurally completely disconnected (Fig. 8 D and T). Such disconnected alimentary tracts have been described from some stinkbug species representing the Plataspidae, the Urostylididae, and other groups (19, 28, 49, 54). In these species, plausibly, after young nymphs have orally acquired the symbiotic bacteria and established infection in the posterior midgut, the constricted region is closed and degenerated during the subsequent nymphal development, thereby reinforcing the isolation of the posterior midgut for symbiosis. This unique configuration of the alimentary tract, which requires complete food absorption in the anterior region, must have been enabled by the sap-sucking lifestyle of the stinkbugs.

Fig. 8.

Dissected alimentary tracts of diverse stinkbugs fed with water supplemented with a food coloring. (A and B) C. punctiger (Coreidae). (C and D) A. sordidus (Coreidae). (E and F) T. hemipterus (Rhyparochromidae). (G and H) P. exiguus (Rhyparochromidae). (I and J) D. baccarum (Pentatomidae). (K and L) P. lewisi (Scutelleridae). (M and N) M. gracilicorne (Dinidridae). (O and P) A. triguttulus (Cydnidae). (Q and R) E. putoni (Acanthosomatidae). (S and T) M. punctatissima (Plataspidae). Adult insects were fed with water supplemented with Congo red for 3 d and then dissected. Arrows indicate the constricted region, of which red ones in D and T highlight reduction of the constricted region and consequent structural disconnection between the M3 and the M4B. Abbreviations are as in Fig. 1.

Evolutionary Origin of Constricted Region.

Our finding that the sorting function of the constricted region is conserved among the diverse stinkbugs suggests that the novel intestinal organ emerged in their common ancestor, which sheds new light on the evolutionary origin and relevance of the gut-associated bacterial symbiosis in the stinkbugs. Insects of the order Hemiptera, which consists of such higher taxa as the Sternorrhyncha (aphids, coccids, whiteflies, and psyllids), the Auchenorrhyncha (cicadas, spittlebugs, leafhoppers, treehoppers, and planthoppers), the Coleorrhyncha (moss bugs), and the Heteroptera (water bugs, water striders, bedbugs, and stinkbugs), are characterized by their needle-like mouthpart specialized for exploiting liquid food sources (55) (Fig. S1A). Almost all members of the Sternorrhyncha, the Auchenorrhyncha, and the Coleorrhyncha are plant sap feeders and obligatorily dependent on bacteriocyte-associated microbial symbionts (19) (Fig. S1A), which provide essential amino acids and other nutrients deficient in their host’s plant sap diet (8, 56, 57). In the Heteroptera, by contrast, whereas predatory species such as water bugs and water striders possess neither bacteriocytes nor microbial symbionts, the majority of plant-sucking species represented by diverse stinkbugs harbor crypt-associated symbiotic bacteria in the posterior midgut (19–21) (Fig. S1B). Considering that all predatory taxa such as the Nepomorpha (water bugs) and the Gerromorpha (water striders) constitute the basal lineages whereas all plant-sucking stinkbugs with the midgut symbiotic bacteria are restricted to the derived group Pentatomomorpha (58–60) (Fig. S1 B and C), it is conceivable, although speculative, that the bacteriocyte-associated microbial symbiosis was lost in the predatory ancestor of the Heteroptera, and the gut-associated bacterial symbiosis evolved in an ancestor of the Pentatomomorpha in association with the evolutionary transition from predatory lifestyle to plant-sucking lifestyle (61). Here we propose a hypothesis that the acquisition of the novel intestinal organ, the constricted region, which functionally isolates the posterior midgut for symbiosis and ensures colonization of specific bacteria therein, might be relevant to the diversity and prosperity of the stinkbugs as a major group of herbivorous insects.

Fig. S1.

Evolution of symbiotic systems in the Hemiptera. (A) Phylogeny of higher taxa in the Hemiptera (55). (B) Phylogeny of higher taxa in the Heteroptera (58). (C) Phylogeny of higher taxa in the Pentatomomorpha (59, 60). Presence/absence of bacteriocytes, endocellular symbionts, midgut crypts, and gut symbionts are indicated in colors. Common insect names and their feeding habits are also shown. *There are exceptional taxa that have evolved bacteriocytes/endocellular symbionts secondarily (71–73). ^†There are several groups, for example the Asopinae, that have lost midgut crypts/gut symbionts secondarily. Presence/absence of the midgut crypts and constricted region in the Pentatomomorpha are based on our observations (Fig. 8) and histological observations by Miyamoto (49).

Conclusion and Perspective.

In conclusion, we unveiled a unique configuration of the alimentary tract in the stinkbugs. The alimentary tract is, although structurally stretching from mouth to anus, divided in the middle by a tiny organ for symbiont sorting, at which food materials and nonsymbiotic bacteria are blocked and only symbiotic bacteria are selectively allowed to colonize the posterior midgut region. The sorting organ functionally partitions the stinkbug’s midgut into the anterior region for digestion and absorption and the posterior region specialized for symbiosis. The selective passing of the symbiotic bacteria and the functional isolation of the posterior midgut probably ensure the establishment of a specific bacterial association, the stable maintenance of a large amount of beneficial bacteria, and the proper control over the bacterial population in the stinkbug gut symbiosis. Symbiont mutants of flagella-related genes tend to be rejected at the sorting organ, highlighting that not only the host’s control but also the symbiont’s phenotype is involved in the sorting process at the host–symbiont interface.

In diverse animals including insects, fish, mammals, and others, specific microbial consortia within their alimentary tract, especially those in the posterior region such as the large intestine of humans, are generally involved in a variety of biologically important traits including growth, health, immunity, and so on (62–66). In the stinkbugs, the peculiarity resides in the conspicuous morphological specialization of the posterior symbiotic midgut as well as the highly specific symbiotic microbiota usually consisting of a single dominant bacterial species (20, 21). On the grounds that microbial nutritional provisioning is generally important for plant sap-feeding insects (8, 56, 57) and that the ancestral heteropterans were predatory and devoid of microbial symbionts (58, 61), it is conceivable, although speculative, that the constricted region for symbiont sorting and midgut partitioning may represent an evolutionary innovation with which the stinkbugs are currently prosperous as a major plant-sucking insect group embracing many agricultural pests (16). What molecular, cellular, and physiological mechanisms underlie the selective symbiont sorting at the constricted region is still an open question and deserves future studies.

Most animals (namely, large-sized metazoans, except for those with a highly parasitic lifestyle) have a mouth for feeding, an anus for defecation, and a gut connecting them for digestion and absorption. However, some animals intimately associated with microbial symbionts represent striking deviations from this general rule. For example, tubeworms gathering around hydrothermal vents at the oceanic floor lack mouth, anus, and gut. By extending fan-like gills, the animals deliver hydrogen sulfide from surrounding spring water to huge amounts of chemoautotrophic symbiotic bacteria harbored in their body, thereby acquiring energy and nutrition without feeding (67). Juvenile kleptoplastic mollusks such as Elysia chlorotica, so-called solar-powered sea slugs, accumulate functional chloroplasts from food algae within the gut lining cells, and adult animals survive without feeding for as long as 10 mo, solely depending on photosynthetic products of the cyanobacterium-derived “stolen” organelles (68). Although not so drastic as these amazing cases, the stinkbugs commonly found in our backyard also exhibit an extraordinary symbiosis-associated modification of their alimentary tract, which highlights the general relevance of animal–microbe symbiosis to organismal adaptation and diversification.

Materials and Methods

Insects, Bacterial Strains, and Plasmids.

The R. pedestris strain used in this study was originally collected from a soybean field in Tsukuba, Ibaraki, Japan and maintained in the laboratory. The insects were reared in Petri dishes (90 mm in diameter, 20 mm high) at 25 °C under a long-day regimen (16 h light, 8 h dark) and fed with soybean seeds and distilled water containing 0.05% ascorbic acid (DWA). Bacterial strains and plasmids used in this study are listed in Table S2.

Table S2.

Bacterial strains and plasmids used in this study

Bacteria or plasmid	Description	Source
Burkholderia symbiont
RPE75	Spontaneous Rfr mutant of wild-type Burkholderia symbiont RPE64	39
RPE225	GFP-labeled mutant of Burkholderia symbiont RPE75	38
E. coli
DH5α	F- Φ80dlacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17 (rK−, mK+) phoA supE44 λ- thi-1 gyrA96 relA1	Toyobo
WM3064	thrB1004 pro thi rpsL hsdS lacZΔM15 RP4–1360 Δ(araBAD)567 ΔdapA1341::[erm pir(wt)]	70
DH5αλpir	F- Φ80dlacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17 (rK−, mK+) phoA supE44 λ- thi-1 gyrA96 relA1 λpir	74
PIR1	F- Δlac169 rpoS(am) robA1 creC510 hsdR514 endA recA1 uidA(ΔMluI)::pir-116	Invitrogen
Other bacteria
P. putida AHU1890	Soil bacterium: Gram negative, Proteobacteria (Gammaproteobacteria), motile	Hokkaido University
B. subtilis IAM12118T	Soil bacterium: Gram-positive, Firmicutes, motile	RIKEN BRC
Plasmids
pRL27	Kmr; mini-Tn5 plasposon (oriR6K) delivery vector	69
pURR25	Mini Tn7KsGFP, GFP driven by Plac (PA1/04/03) promoter, mobilizable oriTIncPα, suicide oriRR6Kγ; Apr (bla)	75
pUX-BF13	Tn7 transposase genes tnsABCDE, mobilizable oriTIncPα, suicide oriRR6Kγ; Apr (bla)	75
pVSV208	Cmr, rfp (DsRed.T3[DNT])	76
pBBR122	Broad host range vector: Cmr, Kmr	77

Inoculation of Burkholderia Symbiont.

The Burkholderia symbiont strains were grown at 30 °C to an early log phase in yeast–glucose (YG) medium [0.5% (wt/vol) yeast extract, 0.4% (wt/vol) glucose, and 0.1% (wt/vol) NaCl] supplemented with an adequate antibiotic on a gyratory shaker at 150 rpm. Colony-forming unit values were estimated by plating the culture media on YG agar plates (1.5% agar in YG medium) containing an adequate antibiotic. The cultured symbiont cells were harvested by centrifugation, suspended in DWA, and adjusted to 10⁷ cfu/mL. Newly molted second instar nymphs were deprived of DWA overnight, which made the insects thirsty and willing to ingest the symbiont-containing DWA. The thirsty nymphs were fed with the symbiont-containing DWA for 24 h, which was subsequently replaced by normal sterile DWA. Two days after the third instar molt (approximately 5 d after inoculation), their symbiotic organs were dissected and examined for infection with the Burkholderia symbiont by diagnostic PCR as described (22, 32).

Inoculation of Food Colorings, GFP-Labeled Burkholderia Symbiont, and RFP-Labeled E. coli.

To visualize the food passage in the alimentary tract, the nymphs were fed with DWA supplemented with the food colorings (0.05% wt/vol) listed in Table S3 for 3 d and dissected for visual inspection under a dissection microscope. To elucidate the relationship between food passage and symbiont colonization, second instar nymphs were fed with DWA supplemented with 0.05% (wt/vol) Congo red and 10⁷ cfu/mL of the GFP-labeled Burkholderia symbiont. Six hours after inoculation, the nymphs were dissected and their alimentary tracts were examined under a light and fluorescence microscope (Axiophot; Carl Zeiss). Coinoculation of the GFP-labeled Burkholderia symbiont and the RFP-labeled E. coli (Table S2) was conducted to evaluate symbiont selection in the stinkbug–Burkholderia symbiosis. The respective bacterial strains were grown to an early log phase, harvested, diluted, and mixed to be 10⁷ cfu/mL each in DWA. Six hours after inoculation, the nymphs were dissected and their alimentary tracts were observed under the fluorescence microscope and a confocal laser scanning microscope (LSCM Pascal5; Carl Zeiss).

Table S3.

Food colorings used in this study

Color	Name	Chemical formula	Molecular weight	CAS no.
Red	Congo red	C32H22N6Na2O6S2	696.66	573-58-0
Scarlet	New coccine	C20H11N2Na3O10S3	604.47	2611-82-7
Yellow	Gardenia yellow*	C20H24O4	328.40	27876-94-4
		C44H64O24	976.97	42553-65-1
Purple	Purple sweet potato color*	C32H39O20Cl	779.03	16727-02-9
		C22H23O11Cl	498.90	6906-39-4
Blue	Brilliant blue FCF	C37H34N2Na2O9S3	792.86	3844-45-9
Green	Gardenia yellow*	C20H24O4	328.41	27876-94-4
		C44H64O24	976.97	42553-65-1
	Gardenia blue*	C17H24O10	388.37	24512-63-8
		C11H14O5	226.23	6902-77-8
Brown	Kaoliang color	C15H10O5	270.24	520-36-5

^*Mixture of two color components.

Transmission Electron Microscopy.

From second instar nymphs, the constricted regions were dissected in 0.1 M phosphate buffer [25 mM NAH₂PO₄ and 75 mM Na₂HPO₄ (pH 7.2)] containing 2.5% (wt/vol) glutaraldehyde using fine forceps. The dissected tissues were prefixed in the fixative at 4 °C overnight and postfixed in 2% (wt/vol) osmium tetroxide at 4 °C for 60 min. After dehydration through an ethanol series, the tissues were embedded in an epoxy resin (Epon 812; Okenshoji Co., Ltd). Ultrathin sections were made on an ultramicrotome (EM UC7; Leica), mounted on copper meshes, stained with uranyl acetate and lead citrate, and observed under a transmission electron microscope (H-7600; Hitachi). For inspection of the flagellar arrangement and structure of the symbiont mutants, exponentially growing bacterial cells were negatively stained with 1% (wt/vol) uranyl acetate on copper meshes and observed under the transmission electron microscope.

Histology for Light Microscopy.

The Epon-embedded constricted regions were processed into semiultrathin sections at 0.3 μm on the ultramicrotome, mounted on glass slides, and stained with toluidine blue. The dissected constricted regions were also fixed in sterilized PBS [137 mM NaCl, 8.1 mM Na₂HPO₄, 2.7 mM KCl, and 1.5 mM KH₂PO₄ (pH 7.5)] containing 4% formaldehyde for 1 h at room temperature. After washing with PBS several times, the fixed tissues were then dehydrated through an ethanol series, embedded in a hydroxyethylmethacrylate resin (Technovit 7100; Kulzer), sectioned at 2 µm on a microtome (RM2165; Leica), mounted on glass slides, and stained with hematoxylin and periodic acid–Schiff reagent. These stained tissue sections were observed under a light microscope.

Random Insertion Mutagenesis of Burkholderia Symbiont.

For random Tn5 mutagenesis, we used pRL27, a plasmid carrying a hyperactive Tn5 transposase gene adjacent to an insertion cassette comprising two inverted Tn5 termini bracketing a kanamycin-resistant gene kan and a R6K pir-dependent origin (69). The E. coli donor strain WM3064, which is diaminopimelic acid auxotroph and carries pRL27 (70), was grown on LB medium [1% (wt/vol) tryptone, 0.5% (wt/vol) yeast extract, and 1% (wt/vol) NaCl] containing 0.3 mM diaminopimelic acid, and the plasmid was transferred to the Burkholderia symbiont strain RPE75 by conjugation. Burkholderia transconjugants containing the integrated Tn5 cassette were selected on YG agar plates containing 30 μg/mL kanamycin.

Screening for Motility Mutants of Burkholderia Symbiont.

The transconjugants from the primary selection plates were individually stabbed into semisolid YG agar [0.4% (wt/vol) agar and 10 µg/mL kanamycin in YG medium] in 96-well microtiter dishes and incubated at 27 °C for 24–48 h. The transconjugants that grew but did not form halo-shaped colonies were presumed as motility mutants. Absence of motility of transconjugants resulting from the first screen was confirmed by a second semisolid agar assay. Biofilm-deficient mutants were screened by a crystal violet staining assay as described (42) and subsequently subjected to the semisolid agar assay. The transconjugants that formed no halo in semisoft agar and exhibited no visible motility in liquid medium under a phase-contrast microscope were categorized as “nonmotile mutants.” Meanwhile, the transconjugants that formed no halo in semisoft agar but showed visible motility in liquid medium were designated as “altered-motility mutants.”

Inoculation Assay of Motility-Deficient Mutants.

Second instar nymphs were fed with DWA containing each of the motility-deficient mutants, and 2 d after the third instar molt, the symbiotic midgut sections M4B and M4 were dissected, subjected to DNA extraction, and analyzed by diagnostic PCR as described (23). To quantitatively evaluate the infection process of the motility-deficient mutants, the alimentary tracts were dissected from the second instar nymphs 48 h after inoculation and carefully cut into the sections M1, M2, M3, M4B, M4, and hindgut in PBS. After rinsing in sterilized PBS, each of the alimentary sections was homogenized and spread on YG agar plates containing 30 µg/mL kanamycin for cfu counting. As a control, the wild-type symbiont strain RPE75 was inoculated.

Detection of Transposon Insertion.

Whether a single Tn5 cassette was inserted in the genome of the symbiont mutants was examined by Southern blot hybridization. Bacterial genomic DNA samples (around 3 µg) were digested with NcoI, electrophoresed on 0.67% (wt/vol) agarose gels, and blotted onto Hybond N⁺ membranes (GE Healthcare) using the BacuGene XL Vacuum Blotting system (GE Healthcare). Hybridization was performed using an AlkPhos Direct Labeling and Detection System (GE Healthcare). A 1,761-bp hybridization probe was prepared from pRL27 by digestion with SalI, which was hybridized to the blot at 55 °C overnight.

Identification of Transposon Insertion Site.

Because the origin of replication (R6K ori) is within the Tn5 cassette, transposon insertions along with the flanking genomic DNA can be cloned in pir+ E. coli and sequenced to determine the insertion site in the genome of the Burkholderia symbiont. Total DNA was prepared from 1 mL of YG liquid culture grown overnight as described (32). Then, 2.5 µg of genomic DNA was digested with NcoI, self-ligated, and transformed into One ShotPIR1 Chemically Competent E. coli cells (Invitrogen) with selection on LB agar plates containing 100 μg/mL kanamycin. Selected colonies were picked and cultured overnight in kanamycin-containing LB liquid medium, from which plasmid DNA was extracted using QIAprep Spin Miniprep Kit (Qiagen). The insert DNA in the plasmid was sequenced with the primers tpnRL17-1 (5′-AAC AAG CCA GGG ATG TAA CG-3′) and tpnRL13-2 (5′-CAG CAA CAC CTT CTT CAC GA-3′) (69). The sequence was then compared with the protein sequence databases using the BlastX search.