The Intestinal Microbiota of Hermetia illucens Larvae Is Affected by Diet and Shows a Diverse Composition in the Different Midgut Regions

The demand for food of animal origin is expected to increase by 2050. Since traditional protein sources for monogastric diets are failing to meet the increasing demand for additional feed production, there is an urgent need to find alternative protein sources. The larvae of Hermetia illucens emerge as efficient converters of low-quality biomass into nutritionally valuable proteins. Many studies have been performed to optimize H. illucens mass rearing on a number of organic substrates and to quantitatively and qualitatively maximize the biomass yield. On the contrary, although the insect microbiota can be fundamental for bioconversion processes and its characterization is mandatory also for safety aspects, this topic is largely overlooked. Here, we provide an in-depth study of the microbiota of H. illucens larval midgut, taking into account pivotal aspects, such as the midgut spatial and functional regionalization, as well as microbiota and nutrient composition of the feeding substrate.

ABSTRACT

The larva of the black soldier fly (Hermetia illucens) has emerged as an efficient system for the bioconversion of organic waste. Although many research efforts are devoted to the optimization of rearing conditions to increase the yield of the bioconversion process, microbiological aspects related to this insect are still neglected. Here, we describe the microbiota of the midgut of H. illucens larvae, showing the effect of different diets and midgut regions in shaping microbial load and diversity. The bacterial communities residing in the three parts of the midgut, characterized by remarkable changes in luminal pH values, differed in terms of bacterial numbers and microbiota composition. The microbiota of the anterior part of the midgut showed the highest diversity, which gradually decreased along the midgut, whereas bacterial load had an opposite trend, being maximal in the posterior region. The results also showed that the influence of the microbial content of ingested food was limited to the anterior part of the midgut, and that the feeding activity of H. illucens larvae did not significantly affect the microbiota of the substrate. Moreover, a high protein content compared to other macronutrients in the feeding substrate seemed to favor midgut dysbiosis. The overall data indicate the importance of taking into account the presence of different midgut structural and functional domains, as well as the substrate microbiota, in any further study that aims at clarifying microbiological aspects concerning H. illucens larval midgut.

IMPORTANCE The demand for food of animal origin is expected to increase by 2050. Since traditional protein sources for monogastric diets are failing to meet the increasing demand for additional feed production, there is an urgent need to find alternative protein sources. The larvae of Hermetia illucens emerge as efficient converters of low-quality biomass into nutritionally valuable proteins. Many studies have been performed to optimize H. illucens mass rearing on a number of organic substrates and to quantitatively and qualitatively maximize the biomass yield. On the contrary, although the insect microbiota can be fundamental for bioconversion processes and its characterization is mandatory also for safety aspects, this topic is largely overlooked. Here, we provide an in-depth study of the microbiota of H. illucens larval midgut, taking into account pivotal aspects, such as the midgut spatial and functional regionalization, as well as microbiota and nutrient composition of the feeding substrate.

INTRODUCTION

The black soldier fly (BSF), Hermetia illucens (Diptera: Stratiomyidae), is a true fly that occurs worldwide in tropical and temperate regions. Adults of this insect have never attracted interest because they do not approach humans, do not bite, and are not known to be pathogen vectors. On the contrary, BSF larvae have been the object of intense research efforts because of their remarkable utility for humans that takes advantage of their feeding regime of generalist detritivore (1). In particular, BSF larvae are largely used in forensic entomology to estimate human postmortem interval (2, 3), but the major promising potential of the voracious BSF larvae is their use as efficient bioconverters (4–6). Indeed, BSF larvae can be reared in mass cultures on a very wide variety of organic waste (e.g., crop and food processing residues, food waste, manure, and feces), leading to the conversion of low-quality material into valuable biomass. The latter is exploitable for the isolation of bioactive compounds (e.g., antimicrobial peptides, chitosan, and degrading enzymes), biodiesel production, and as feed or feed ingredients (mainly for their content of high-quality proteins and lipids) for poultry, aquaculture, and livestock (1, 6). The production of BSF larvae is technically simple, cost effective, and environmentally sustainable (1, 6). However, the rate of waste recycling and the final value of the biomass obtained depend on the rearing strategy, in terms of feeding substrate composition, feed consumption rate, and environmental parameters (i.e., temperature, humidity, and photoperiod) (1). For this reason, many research efforts now focus on the characterization of nutrient and micronutrient content of BSF larvae in response to different rearing conditions and substrates, in order to optimize biomass yield and quality (1, 6–10).

Safety aspects concerning the microbiological loads of intermediate and final products of bioconversion processes are also crucial, especially when BSF larvae are exploited for feed applications. In principle, this issue can be approached by classical food microbiology methods to establish whether a product meets the recommendations imposed by current hygiene criteria. On the other hand, an in-depth characterization of BSF larval microbiota and the factors that influence its composition is particularly important. Microbiota composition is known to impact insect health and performance and has to be considered in the effort to optimize biomass yield (11). In addition, the analysis of the microbiota could allow the identification of bacterial species with peculiar and unique characteristics, such as the capacity to degrade complex substrates (e.g., cellulose, hemicellulose, and lignin) (12) or xenobiotics. These microorganisms, or even the enzymes responsible for the degradation, could be isolated and exploited at the industrial level for waste recycling and bioremediation. Populations of gut bacteria able to compete with pathogens or to act as probiotics could be boosted for the improvement of BSF larva performances and bioconversion efficiency or may be used in other animal hosts with similar purposes. Moreover, the study of the BSF microbiota has strong potential to contribute to the global problem of the identification of new antimicrobials. Indeed, BSF larval feeding activity is able to reduce the bacterial load of substrates and, importantly, this capacity is not accompanied by the accumulation of pathogens in the larval gut (13–16). Such evidence implies the presence of potent antimicrobial effectors produced by BSF larvae and their intestinal microbiota. It should be pointed out that the latter is, in turn, implicated in the maintenance of gut homeostasis and supports gut immune functions (11, 17, 18).

A few studies on microbiota of BSF larvae have already been performed (19, 20). Rearing substrate and insect development stage have a significant impact on the overall composition of the microbial community (19, 20). A very critical issue that these preliminary microbiological surveys have not taken into account is the high complexity of the gut of fly larvae. In fact, this organ, and in particular the midgut, shows peculiar regional structural and functional features associated with changes of luminal pH (21–23). The differences in gut morphology and epithelial architecture along the different intestinal tracts of some insects are in fact accompanied by remarkable differences in physiological, metabolic, and immune features that impact microbiota composition (24–26). These complex relationships have been exhaustively described for the fruit fly Drosophila melanogaster (Diptera: Drosophilidae) (17, 27–30).

In all insects, the digestive tract is divided into three regions with different embryonic origins and peculiar morphological and functional features, as follows: an initial tract, the foregut; a midgut where digestion and absorption occur; and a final hindgut where water, salts, and other molecules are absorbed prior to elimination of the feces. Even though a detailed morphofunctional description of the H. illucens larval midgut is lacking, it is expected that, as in other nonhematophagous brachycerous Diptera, discrete regions with peculiar pH values can be recognized along the midgut, and that each distinct midgut region possesses its own features, at both structural and functional levels, and a peculiar resident microbiota (17, 24–30).

In the present work, we analyzed the effects of different diets and their microbial community on the midgut microbiota of BSF larvae, and the impact of the insect feeding activity on the diet microbiota. Most importantly, we analyzed the different tracts of the BSF larval midgut separately and highlighted the need of having future research on the BSF larval midgut that considers each midgut domain independently.

RESULTS

Determination of the pH values of the midgut lumen content.

Since luminal pH is a good marker for midgut regionalization in flies (21–23), we evaluated how the pH of the lumen content of BSF larvae changed along the midgut, in order to have a clear identification of the regions into which this organ could be subdivided. For this purpose, last-instar H. illucens larvae were fed with diet containing one of two pH indicators, bromophenol blue or phenol red. The color of the luminal content of larvae fed with diet containing bromophenol blue was clearly visible through the isolated epithelium (Fig. 1A). The anterior region of the midgut presented a blue color, indicating that its luminal content has a pH ≥4.6. Then, a marked change was observed, since the middle region turned yellow, revealing that its lumen has a pH ≤3. Moving toward the posterior midgut, the color gradually turned blue. Bromophenol blue turns at pH values between 3.0 and 4.6, and thus differences in the pH values of the anterior and posterior midgut contents could not be evidenced. Figure 1B shows the gut isolated from a larva fed with diet containing phenol red, a dye that turns yellow at a pH ≤6.8 and fuchsia at a pH ≥8.2. Since the anterior and the middle regions of the midgut presented a golden yellow color, whereas the posterior midgut content appeared fuchsia, it is possible to state that the luminal content of the anterior and middle regions have an acidic pH and that of the posterior has an alkaline pH. The evidence obtained with phenol red supported and completed results obtained with bromophenol blue. In conclusion, the luminal content of the midgut of H. illucens larvae presents different pH values; the anterior region has an acid luminal content, the middle region presents a strongly acid pH (pH ≤ 3), and the posterior region has an alkaline luminal content. These three regions are separated by transition zones, in which the pH values gradually change (Fig. 1A and B). Taking this evidence into account, we could easily distinguish three main regions of the larval midgut of H. illucens, a fundamental aspect to correctly isolate midgut samples for the analyses reported below (Fig. 1C).

FIG 1.

Determination of pH value in BSF larval midgut lumen (A and B) and definition of the midgut portions for the microbiota analysis (C). In panels A and B, the anatomy of the BSF larval gut is visible. The short foregut is followed by a very long midgut. The beginning of the hindgut (which extends out of the field of view) is easily recognizable by the insertion of Malpighian tubules (MT), structures involved in excretion in insects and that deliver the primary urine into hindgut lumen. The whole gut isolated from H. illucens larvae fed with diet containing bromophenol blue (A) or phenol red (B) pH indicators shows the presence of different pH values along the midgut lumen. (C) Image of the midgut, which is subdivided in a relatively short and thick anterior midgut, a middle midgut characterized by an enlarged highly acidic portion (stomach), and the posterior midgut. Bars of different colors highlight the positions of the cuts for the isolation of the portions used for microbiota analyses. Bars, 2 mm.

Insect performances on different diets.

The microbiota analyses were performed on larvae reared on the following three different feeding substrates: standard diet, an optimal diet for fly larvae rearing (31), veg mix diet, containing a mixture of fruits and vegetables, and fish diet, based on fish meal (see Materials and Methods for detailed composition). First, we evaluated the performances of the BSF larvae on these substrates. The maximum weight reached before pupation by BSF larvae reared on standard diet was significantly higher than that of larvae reared on the other two diets (Table 1) (F_[2,12] = 15.50, P = 0.0005, df = 14; one-way analysis of variance [ANOVA]). There was also a trend in the increase of larval period duration (F_[2,12] = 12.00, P = 0.0014, df = 14). This was particularly evident for the larvae reared on the fish diet, which showed doubled developmental time and almost halved maximum weight compared to larvae grown on the standard diet (Table 1).

TABLE 1.

Length of BSF larval cycle and maximum weight at pupation for the different diets used in this study^a

Diet	Larval period (days)b	Maximum wt (mg)b	Day of sample collection for microbiota analysis
Standard	18 ± 1 (5) A	218 ± 8 (5) A	16
Veg mix	24 ± 2 (5) A	195 ± 5 (5) B	22
Fish	36 ± 3 (5) B	173 ± 3 (5) C	30

^aData are expressed as mean ± standard error, with number of experiments in parenthesis. For each experiment, at least 20 larvae were monitored for development time and weight.

^bDifferent letters denote statistical differences (one-way ANOVA).

Evaluation of relative bacterial counts in the different regions of BSF larval midgut.

The bacterial loads in different midgut regions of H. illucens larvae (Fig. 1C) were determined by quantitative reverse transcription-PCR (qRT-PCR) on RNA samples in order to narrow in the analysis on live bacteria. The results demonstrate that the profile of the relative bacterial counts in the different midgut regions was similar for the three diets. In particular, while anterior and middle midgut had comparable bacterial loads, bacterial loads were higher in the posterior portion (Fig. 2) (standard F_[2,12] = 8.869, n = 5, P = 0.0043, df = 14; veg mix F_[2,12] = 295.51, n = 5, P < 0.0001, df = 14; fish F_[2,12] = 33.882, n = 5, P < 0.0001, df = 14; one-way ANOVA). We observed a statistically significant interaction between the effects of diet and midgut region on bacterial load (F_[4,36] = 17.601, P < 0.0001), which was significantly affected by both the considered independent variables (diet F_[2,36] = 23.339, P < 0.0001; midgut region F_[2,36] = 137.170, P < 0.0001).

FIG 2.

Relative quantification of bacterial load by qRT-PCR in the different tracts of the midgut of BSF larvae reared on different diets. The values reported are the mean ± standard error (n = 5 for each sampling point containing pools of 5 midgut portions each) of the relative expression of the 16S rRNA gene normalized to that of the HiRPL5 gene (see “qRT-PCR for relative bacterial load determination” in Materials and Methods). Different letters denote significant differences for each diet (one-way ANOVA).

Microbiota composition in the different regions of BSF larval midgut and diet substrates.

We analyzed the microbiota by 16S rRNA gene sequencing, starting from cDNA obtained from RNA samples, in order to consider communities of live bacteria. A total of 2,175,325 high-quality reads were analyzed, with an average of 29,000 reads/sample. Our study also included analysis of the microbiota of the feeding substrates immediately after preparation (fresh diet) and after BSF larva feeding (conditioned diet). This is particularly important because these larvae feed and develop inside the food substrate, which is not daily and completely renewed but to which fresh substrate is periodically added. The anterior part of the midgut was always characterized by a high microbial diversity (P < 0.05) that progressively decreased going from the anterior to the posterior part (Fig. 3), and this trend happened regardless of the diet. The microbiota of the feeding substrate showed a strong impact in shaping the midgut microbiota in larvae fed with standard or fish diet, at least in the first regions of the midgut (Fig. 4); in contrast, the microbiota of larvae fed with the veg mix diet was not found in the midgut (Fig. 4). The posterior part always showed a significantly different microbiota compared with those of the middle and anterior part of the midgut, as determined by multivariate ANOVA (MANOVA) based on Bray Curtis distance (standard F_[2,12] = 24.945, P < 0.001; veg mix F_[2,12] = 46.287, P < 0.001; fish F_[2,12] = 16.968, P < 0.001), and the composition of the microbiota in this region reflected a strong selection of the species that were present in the food substrate, an aspect of particular extent for fish diet (Fig. 4 and 5). The composition of the microbiota determined a clear differentiation of the samples according to both midgut portion and diet (Fig. 5). Indeed, a significant effect of both diet type and midgut region was found by MANOVA for both the independent variables (diet F_[2,36] = 57.047, P < 0.001; midgut region F_[2,36] = 39.256, P < 0.001) and for the interaction between them (F_[4,36] = 19.540, P < 0.001). Fish diet microbiota seemed to have the strongest effect on the gut microbiota, leading to a higher abundance of Proteobacteria taxa in the posterior tract of the midgut, while Firmicutes taxa prevailed in the anterior and middle tract (Fig. 4A). In contrast, the midgut of BSF larvae fed with standard and veg mix diets were more similar and characterized by higher levels of Bacteroidetes (Fig. 4). Indeed, the midgut of larvae fed with the fish diet showed significantly higher weighted UniFrac distances from the standard and veg mix diets, compared to the distance between standard and veg mix, in all three portions (Fig. S1). Although the larvae feed and develop inside the diet, the data show that BSF larvae do not significantly alter the microbiota composition of the substrate, except for an increase in Lactobacillus population in the veg mix diet (Fig. 4B). A complete list of the taxa identified is reported in Tables S1 to S4.

FIG 3.

Microbial diversity. Box plot showing the number of observed OTU in the different samples, as detected by high-throughput sequencing of the 16S rRNA gene. Boxes represent the interquartile range (IQR) between the first and third quartiles, and the line inside represents the median (second quartile). Whiskers denote the lowest and the highest values within 1.5× IQR from the first and third quartiles, respectively. Different letters indicate a significant difference (P < 0.05), as obtained by pairwise Wilcoxon’s tests. “Fresh diet” and “conditioned diet” refer to the analysis of the microbiota of the feeding substrates just after preparation and after larval feeding, respectively.

FIG 4.

Incidence of the major bacterial taxonomic groups. The stacked bar chart shows the relative abundances of bacterial phyla (A) and genera (B) identified in the midgut and diet samples analyzed. The order of the taxa in each bar is the same provided in the legend. Values are the average of 5 replicates. Genera and phyla with an abundance of <2% in at least 5 samples are summed up and shown as “others.”

FIG 5.

Heat plot based on microbiota composition at genus level. Hierarchical Ward’s linkage clustering based on the Spearman’s correlation coefficient of the microbial taxa abundance. Column bar is color coded according to the type of diet and the midgut region. Row bar is colored according to the taxa assignment at phylum level. The color scale represents the scaled abundance of each variable, denoted as Z-score, with red indicating high abundance and blue indicating low abundance.

DISCUSSION

Despite the great and exponentially increasing interest in BSF larvae for bioconversion (4–6) and bioremediation (32), several aspects concerning the biology of this insect are still neglected. Surprisingly, there is still a paucity of information on its intestinal microbiota (11), an issue that should be instead considered a priority for an organism that can be used for such purposes. A recent review on the microbial community associated with BSF (11) highlights knowledge gaps and provides suggestions on aspects that still need to be unraveled rather than presenting a summary of the available data.

First, none of the few studies on the BSF intestinal microbiota have taken into account the correlation between the different regions of the midgut of this insect and the microbiota. In this paper, we provide evidence that discrete regions can be recognized along the midgut of BSF larvae, as clearly demonstrated by the differences in the luminal pH (Fig. 1). The anterior region is characterized by an acid luminal content, followed by a strongly acidic middle region and an alkaline posterior tract. These data are partially in accordance with previous reports on nonhematophagous brachycerous Diptera. Indeed, in the larvae of Musca domestica (Diptera: Muscidae), three main segments can be identified. The anterior and the posterior midgut are characterized by a slightly acidic luminal pH, while the middle midgut presents a very low pH in the lumen (21), which is generated by the so called “copper cells,” a distinctive cell type present in the acidic segment of the midgut of flies (23, 33–35). The midgut of D. melanogaster larvae presents distinct regions as well (23, 35), with different pH of the luminal content. The anterior segment and the anterior part of the posterior segment is between neutral to mildly alkaline, while the middle segment is highly acidic, and the posterior part of the posterior segment is highly alkaline (23). The differences of the pH in fly midgut regions are associated with peculiar physiological, immune, and microbiological features (22, 26–28, 30).

Here, we have demonstrated that in BSF larvae, the presence of different midgut regions associates with differences in microbial density and composition. We have observed that each tract is characterized by a different bacterial load, which is higher in the posterior compared to the anterior midgut. Interestingly, microbial diversity has an opposite trend, since it gradually decreases along the midgut, suggesting that a selection of fewer taxa takes place. A simple explanation may be a reduced flow rate of luminal content to the posterior region due to the possible presence of sphincters or epithelium folding. Alternatively, or in addition, most bacteria are killed in the anterior and middle region, and only a selection of the initial microbiota proliferates in the posterior midgut using the available nutrients, thus leading to higher numbers. This process of selection may result by the balanced combination of extreme pH values in the middle region of the midgut and the activity of antimicrobial peptides, lysozymes, and digestive enzymes produced and secreted by midgut cells into the lumen of the anterior and middle midgut (17, 21, 27, 36, 37).

To understand whether and how food affects the microbial communities that colonize the digestive tract of BSF larvae, we have examined dietary substrates that strongly differ in terms of nutrient composition. In particular, each of the three diets was characterized by a different protein/carbohydrate ratio, a parameter that has been demonstrated to impact on the gut microbiota (38–40) and insect performances (41–43). Indeed, we have detected differences in BSF larval development on the different diets. A major novelty introduced by our study is the characterization of the microbiota of the dietary substrates, an aspect that was previously overlooked (11) and that could strongly affect the composition of the bacterial community of the midgut. In addition, we have studied the influence of feeding activity of BSF larvae on dietary substrates. A comparative analysis of the results shows that diet composition plays a major role in shaping the diversity of the midgut microbiota. Similarly, the microbiota present in the diet influences the composition of the microbiota resident in the anterior/middle tracts of the midgut and less strongly influences the one occurring in the posterior, which presents a very narrow selection of the species in the food substrate. Interestingly, BSF larvae do not have detrimental effects on the microbiota of the substrates on which they feed and develop. They are not able to significantly change the bacterial community of the standard and fish diet substrates, and, although an increase of a specific population (i.e., Lactobacillus) occurs in veg mix substrate, these bacteria are known as nonpathogenic for their potential probiotic properties for humans (44–46), and some species are involved in detoxification of pesticides and xenobiotics in humans and insects (47–50). This evidence is in contrast with previous claims about the capacity of BSF larvae to change the microbiota of substrates and, in particular, to reduce pathogenic bacteria of substrates (1, 11), but it is a valuable trait for an organism that has to be mass-reared for bioconversion and bioremediation on a variety of substrates.

The differences found in the microbiota of larvae fed on different diets could reflect their physiological performances and bioconversion efficiency, and the posterior midgut, where the resident microbiota results from a selection of microbes present in previous midgut tracts, may have a relevant contribution in nutrient conversion and thus in energy harvest and overall fitness. Standard and veg mix diets are associated with an overall similar microbiota composition, both leading to increased levels of Bacteroidetes taxa in the midgut, bacteria known as glycan degraders because of the presence of polysaccharide utilization loci in their genome (51). Members of the genera Sphingobacterium and Dysgonomonas are particularly abundant, likely reflecting a remarkable potential for complex polysaccharide degradation, and are worthy to be isolated and explored for biotechnological purposes. Bacteroidetes taxa have been identified as core members of the gut microbiome in many Drosophila species across the globe and also in those of other insects, including termites and honeybees (52), and several have xylanases directly involved in hemicellulose digestion (53, 54). On the other hand, the fish diet apparently induces a more putrefactive environment, with a microbiota severely dominated by Proteobacteria (Fig. 4A), mainly Providencia species (Fig. 4B), which are highly transmitted vertically throughout the insect life cycle (11) but which can also be pathogens of many organisms, including humans and insects (55). On the basis of the above consideration, the fish diet may induce a gut dysbiosis, which may contribute to the reduced performance that we detected for BSF larvae reared on fish diet compared to those reared on the other two feeding substrates. These data, along with a previous study performed on the same insect (7), suggest that unbalanced diets with a high protein/carbohydrate ratio content are not optimal for BSF larvae rearing.

Despite the great potential of H. illucens larvae (see the introduction for details), information on its microbiota is surprisingly very limited. Apart from a recent study on mycobiota (56), only two studies have previously examined the microbiota of H. illucens larvae. In the first study (19) (Table 2) the microbiota of the entire guts from larvae reared on three different feeding substrates were investigated. In the second one (20) (Table 2) the microbiota analysis was performed on whole larvae. The differences in the experimental samples analyzed make it difficult to compare the results from those studies and, for the same reason, results from previous studies and the present study. Moreover, both studies completely overlooked the bacterial communities present in the feeding substrates, which we demonstrated can affect midgut microbiota composition. Nevertheless, as summarized in Table 2, a limited comparison can be done. In Zheng et al. (20), larvae were reared on a diet with a composition very similar to the standard diet used in this study, and the major phyla that characterize the microbiota match (both considering each midgut tract separately or the average value of the different tracts). This evidence, along with the differences associated with the microbiota of larvae reared on different substrates, suggests that diet composition had a role in shaping bacterial communities. In particular, when diets were very unbalanced (i.e., cooked rice and fish diet), the diversity of microbial communities decreased compared to those in nutritionally more balanced diets. In those unbalanced diets, Proteobacteria was the major group identified, whereas in all other cases, Bacteroidetes was one of the dominant phyla (Table 2). Interestingly, our data (Table 2) demonstrate that the overall gut microbiota does not mirror the microbiota composition of each tract, confirming the relevance of working with each tract separately.

TABLE 2.

Short summary of the data on microbiota composition of H. illucens larvae from present work and published studies^a

Authors and yr (reference) or source	Sample	Feeding substrate	Major phyla	%b
Jeon et al., 2011 (19)	Larval gut	Food waste	Bacteroidetes	67.4
			Proteobacteria	18.9
			Firmicutes	9.4
			Fusobacteria	2.0
			Actinobacteria	1.9
		Cooked rice	Proteobacteria	54.0
			Firmicutes	47.3
			Unclassified	3.5
		Calf forage	Proteobacteria	31.1
			Actinobacteria	24.6
			Firmicutes	23.5
			Bacteroidetes	20.5
Zheng et al., 2013 (20)	Whole larvae	Gainesville dietc	Bacteroidetes	54.4
			Firmicutes	20.0
			Proteobacteria	16.0
			Actinobacteria	9.0
Present study	Larval midgut	Standard diet	Bacteroidetes	41.5 (A: 65.9, M: 54.4, P: 41.1)
			Proteobacteria	28.2 (A: 25.9, M: 33.7, P: 25.2)
			Firmicutes	13.6 (A: 4.7, M: 5.6, P: 30.4)
			Actinobacteria	3.9 (A: 3.2, M: 5.3, P: 3.1)
		Veg mix	Bacteroidetes	65.4 (A: 85.2, M: 61.2, P: 49.8)
			Proteobacteria	19.1 (A: 12.2, M: 28.9, P: 16.2)
			Firmicutes	15.7 (A: 2.0, M: 28.9, P: 16.2)
			Actinobacteria	11.6 (A: 0.1, M: 3.8, P: 30.8)
		Fish diet	Proteobacteria	55.5 (A: 37.1, M: 30.8, P: 98.6)
			Firmicutes	43.0 (A: 59.1, M: 68.6, P: 1.4)

^aIn all of the studies the microbiota composition was obtained by 16S rRNA gene sequencing. Estimated on the basis of the histogram presented in references 19 and 20.

^bOnly percentages of >1% are reported. For the present study, the % reported is the average of the percentages in the three different midgut portions (A, anterior; M, middle; P, posterior) that are specified in parenthesis.

^cGainesville diet is composed of 20% corn meal, 30% alfalfa meal, and 50% wheat bran, saturated with water.

Our study focused on the effect of midgut morphofunctional regionalization in shaping the residing microbiota. Future work on microbiota in the hindgut of H. illucens larvae is also needed, although the establishment of a stable bacterial community in the hindgut of insect larvae is problematic (due to the molts during the larval period that involve the removal of the cuticle lining the hindgut epithelium) and often requires the presence of special structures that provides a stable environment for bacterial colonization (57), structures that have never been reported for H. illucens larvae.

In conclusion, the presence of different midgut domains, diet composition, and diet microbiota have a nonnegligible effect on BSF larval microbial ecology. These factors and their interdependence will play a major role for a proper exploitation of the biotechnological uses of this insect.

MATERIALS AND METHODS

Insect rearing.

BSF eggs were collected from a colony established in 2015 at the University of Insubria (Varese, Italy) and were maintained in a humid chamber at 27°C until hatching. The eggs were laid on a petri dish (9 × 1.5 cm) with the experimental diet. Three diets were used in the current study: standard diet for Diptera (standard), a diet containing fruits and vegetables (veg mix), and a diet based on fish feed (fish). Standard diet (31) was composed of wheat bran (50%), corn meal (30%), and alfalfa meal (20%) mixed in a 1:1 ratio of dry matter/water (approximately 13% protein; protein/carbohydrate ratio, 1:1). Veg mix diet was composed by seven fruits and vegetables (apple, banana, pear, broccoli, zucchini, potato, and carrot) in equal quantity, appropriately minced (approximately 1% protein; protein/carbohydrate ratio, 1:9). Fish diet was composed of fish meal (FF type; Mazzoleni SpA, Bergamo, Italy), mixed in a 1:1 ratio of dry matter/water (approximately 35% protein, no carbohydrates). Percentages were calculated on diet weight, including water. The values in parenthesis concerning protein and carbohydrate content were estimated based on data available on the web for standard and veg mix diet, whereas for the fish diet, they were reported in the product data sheet. Nipagin (methyl 4-hydroxybenzoate) was added to the diet administered to larvae for the first 4 days after hatching to avoid mold growth (an 18% [wt/vol] stock solution in absolute ethanol was prepared; 20 μl of this stock solution was added to each gram of veg mix diet, whereas 1 ml of a 1.7% [vol/vol] dilution in water of the stock solution was added to each gram of standard and fish diet). Four days after hatching, 300 larvae were placed in a plastic container (16 × 16 × 9 cm), and fed ad libitum with the three experimental diets described above, without nipagin. The larvae were maintained at 27.0 ± 0.5°C, 70% ± 5% relative humidity, in the dark. Fresh diet was added every 2 days, until larvae reached the last larval instar. Five independent rearing groups were set up for each diet. Random samples of 30 individuals were weighed every 2 days. For each experimental diet, the sampling and the annotation of the larval weight were made in triplicate. Before weighing, the larvae were washed in tap water to remove diet matter from their body and then wipe dried. The weights were recorded until 25% of insects reached the pupal stage. Last-instar, actively feeding larvae were used for the measurement of midgut lumen pH and microbiota analyses.

Determination of pH in the midgut lumen with colorimetric indicators.

The presence of different pH in the midgut lumen of H. illucens larvae was assessed using phenol red and bromophenol blue, two pH indicators that assume different colorations at different pH values. Bromophenol blue is yellow at pH values lower than or equal to 3.0 and blue at pH higher than or equal to 4.6; phenol red is yellow at a pH lower than or equal to 6.8 and fuchsia at pH higher than or equal to 8.2, with a gradual color transition for intermediate values. H. illucens larvae were fed ad libitum with standard diet until they reached the last instar, as described above. Larvae with a weight ranging between 180 and 200 mg were selected and transferred to plastic containers on standard diet with added 0.2% (wt/wt) bromophenol blue or phenol red. After 24 h, the larvae were removed from the diet, placed in a plastic tube, and anesthetized on ice with CO₂. The guts were isolated, and the coloration of the midgut content was evaluated by means of a stereomicroscope.

Collection of midgut and diet samples and RNA extraction.

Last-instar larvae were washed with 70% ethanol in autoclaved distilled water and then dissected with the help of a stereomicroscope, under a horizontal-flow hood, using sterile tweezers and scissors to avoid cross-contaminations of the samples. Each midgut was isolated in autoclaved 1× phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na₂HPO₄, and 1.4 mM KH₂PO₄; pH 7.4) in a sterile petri dish (5.5 × 1.3 cm). Once collected, the midgut was divided into three districts, the anterior, middle, and posterior regions (see Results and Fig. 1). For the dissection of each larva, a new petri dish was used, and tweezers and scissors were washed with 70% ethanol in water. For each diet, pools of five midgut regions samples for each of the five replicates of insect rearing were collected in a cryovial, immediately put into TRIzol reagent (Life Technologies, Carlsbad, CA), and kept at −80°C until total RNA extraction, which was performed according to the manufacturer’s instructions. Briefly, after homogenization with Eppendorf fitting pestles to lyse samples in TRIzol reagent, total RNA was precipitated with isopropanol, washed with ethanol, and suspended in RNase-free water. Samples of fresh (before administration to larvae) and conditioned diets (on which larvae have fed) were also immediately put into TRIzol reagent and kept at −80°C until total RNA extraction. Ten samples of both fresh and conditioned diets were collected for each of the 5 experimental replicates on the 3 different feeding substrates.

RNA concentration was assessed by measuring the absorbance at 280 nm with a Varioskan Flash Multimode Reader (Thermo Scientific, Waltham, MA), and sample purity was evaluated by assessing the 260/280-nm absorbance ratio. Total RNA preparations were then treated with Turbo DNase I (Life Technologies) according to the manufacturer’s instructions, and RNA quality was checked by electrophoresis on 1% agarose gel.

qRT-PCR for relative bacterial load determination.

Total RNA was isolated as described above. The relative bacterial load in the three midgut regions (n = 5 for each sampling point containing pools of 5 midgut portions each) was quantified by normalization of the relative expression of the 16S rRNA gene (16S rRNA forward primer, ACTCCTACGGGAGGCAGC; 16S rRNA reverse primer, ATTACCGCGGCTGCTGGC) to that of the ribosomal protein L5 gene of H. illucens (HiRPL5). The primers used for HiRPL5 (HiRPL5 forward primer, AGTCAGTCTTTCCCTCACGA; HiRPL5 reverse primer, GCGTCAACTCGGATGCTA) were designed on conserved regions of RPL5 in other insect species and their sequences checked by sequencing the PCR product. Changes in relative bacterial loads were measured by one-step qRT-PCR (58–60), using the SYBR green PCR kit (Applied Biosystems, Carlsbad, CA), according to the manufacturer’s instructions, using the primers reported above. Relative gene expression data were analyzed using the 2−^ΔΔCT method (61–63). Expression data were normalized, taking into account the differences in the areas of the cross-section of the different intestinal tracts (81,000 ± 7,300 μm², 250,000 ± 17,200 μm², and 46,000 ± 1,700 μm² for the anterior, middle, and posterior midgut, respectively; n = 10 for each tract) by dividing the threshold cycle (C_T) values (for both 16S rRNA and HiRPL5 transcripts) by the area of the cross-section of the corresponding midgut tract. The areas were calculated using the diameter of the lumen of each midgut tract, obtained by direct measurement on the micrographs of different cross-sections acquired from semithin cross-sections of BSF larval midguts stained with crystal violet and basic fuchsin, prepared for light microscopy analysis (64). For validation of the ΔΔC_T method the difference, between the C_T value of 16S rRNA and the C_T value of HiRPL5 transcripts [ΔC_T = C_{T(16S rRNA)} − C_T(HiRPL5)] was plotted versus the log of 2-fold serial dilutions (200, 100, 50, 25, and 12.5 ng) of the purified RNA samples. The plot of log total RNA input versus ΔC_T displayed a slope lower than 0.1 (y = 1.3895 − 0.0137x, R² = 0.0566), indicating that the efficiencies of the two amplicons were approximately equal.

Analysis of the microbiota and bioinformatics of the 16S rRNA gene sequencing data.

After extraction, 400 ng of RNA were reverse-transcribed into cDNA with random primers using RETROscript (Life Technologies), according to the manufacturer’s instructions. The midgut microbiota was assessed by sequencing of the amplified V3 to V4 region of the 16S rRNA gene as recently described (65). Demultiplexed forward and reverse reads were joined by using FLASH (66). Joined reads were quality trimmed (Phred score < 20), and short reads (<250 bp) were discarded by using Prinseq (67). High-quality reads were then imported in QIIME1 (68). Operational taxonomic units (OTU) were picked through a de novo approach, and UCLUST method and taxonomic assignment were obtained by using the RDP Classifier and the Greengenes database (69), following a previously reported pipeline (65). To avoid biases due to different sequencing depth, OTU tables were rarefied to the lowest number of sequences per sample. Statistical analyses and visualization were carried out in the R environment (https://www.r-project.org). Alpha diversity analysis was carried out in QIIME on rarefied OTU tables. Kruskal-Wallis and pairwise Wilcoxon tests were used to determine significant differences in alpha diversity parameters, weighted UniFrac distance, or OTU abundance. Permutational multivariate analysis of variance (nonparametric MANOVA) based on a Bray-Curtis distance matrix was carried out to detect significant differences in the overall microbial community composition among the different parts of the midgut or as affected by the type of diet, by using the adonis function in the R vegan package.

Statistical analysis.

Data were analyzed using Prism version 6.0b (GraphPad Software Inc., San Diego, CA, USA) software using one-way ANOVA with Tukey’s multiple-comparison test to compare bacterial load and parameters of larval performances within any single diet treatment. Two-way ANOVA analysis followed by Bonferroni’s post hoc tests, when significant effects were observed (P value < 0.05), was carried out on bacterial load as affected by different diet treatments and different midgut traits. When necessary to meet assumptions of normality, transformation of data was carried out. Levene’s test was carried out to test the homogeneity of variance.

Accession number(s).

The 16S rRNA gene sequences produced in this study are available at the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI), under the accession number SRP149894.