The environment influences the microbiota composition in the excreted feces of S. frugiperda larvae
As a new invasive species, the gut microbiota of the S. frugiperda has not been well studied. To explore the impact of the environment on the microbiota of S. frugiperda larvae, as well as the representativeness of microbiota at different sampling places, we sampled from cornfields in four provinces (Hubei, Yunnan, Hainan, Henan) of China (Fig. 1a) at ecologically significant sites, including bulk soils (BS) in cornfields, the excreted feces (EF) from active 3rd -5th instar larvae, and gut content from the midgut. Within the midgut, we differentiated the microbiota into those within the gut contents and those attached to the gut mucosa, which we named gut feces (GF) and gut mucosa (GM) respectively.
The results showed that all bacteria found in the gut of the S. frugiperda can be detected in the soil. The environment surrounding the larvae provides the resource for their gut organisms. Specifically, the Sphingobacterium constitute 2.46% of the soil and 30.37% of the bacteria in the excreted soil (Fig. 1b, Figure S1). However, they make up only about 0.7% of the gut contents (GM and GF). This indicates the connection between the exposure of S. frugiperda larvae and host selection. Similarly, genera such as Flavobacterium, Empedobacter, and Acinetobacter are also in the same state, originating from the soil and enriching in the excreted feces, then disappearing in the gut contents. It is worth noting that these microbes are environmentally specific enrichments. For example, Sphingobacterium is enriched only in the larval feces from the Hubei samples (EF: 58.4 ± 17.4%, Figure S1a), Acinetobacter is enriched in samples from Henan (EF: 10.0 ± 6.84%, Figure S1b) and Yunnan (EF: 10.5 ± 2.54%, Figure S1c), and Empedobacter is enriched in the Hainan samples (EF: 58.4 ± 17.4%, Figure S1). Conversely, Klebsiella is enriched in the excreted feces (EF: 25.7%) but persists in the gut contents (GF: 25.7%, GM: 12.3%, Fig. 1b). The most interesting bacterium is Enterococcus, which comes from the soil (1.5%). However, it may have a strong survival ability in the gut content (GF: 50.4%, GM: 65.4%) and is not excreted (EF: 0.85%). The same phenomenon was also observed in the alpha diversity. Significant differences in alpha diversity were found between the soil and the feces excreted by insects from four sites (Figure S2 a-b), however, no significant differences in Shannon diversity were observed for the gut contents (GF and GM) (Figure S2c-d). This indicated the microbes in excreted feces are significantly influenced by the environments.
The host determines the stability of the microbiota in the gut contents of S. frugiperda larvae
Subsequently, to explore the stability and variation of different sampling sites, we conducted statistical tests on the microbial composition between four sampling sites and found significant differences between samples from different sites (Adonis: R2 = 0.129, p = 1e-04, Fig. 1c). The microbial composition of BS and EF was particularly different among the four different locations (Adonis: R2BS = 0.79, pBS=1e-04, R2EF = 0.72, pEF = 1e-04, Figure S3a-b). While significant differences in microbial composition of the gut contents (GF, GM) remained across different sites (pGF = 1e-04, pGM = 1e-04), the their proportion of explained variance (R2) were considerably lower than BS and EF (R2GF = 0.45, R2GM = 0.45, Figure S3c-d). We then compared the differences between the microbiota of EF, GF, and GM from four places. We found significant differences in the similarities within the EF group at different sites compared to EF, GF, and GM (p < 2.22e-16, Fig. 1d), and the similarity within the GF group was significantly lower than within the EF group (p = 2e-05, Fig. 1d), and there were no differences in the similarities between GF and GM and within the GM group (p = 0.69, Fig. 1d). This further indicates that the microbiota in the excreted feces are significantly influenced by the environment than those in the gut contents, and there is almost no difference between the gut feces and gut mucosa-attached microbiota, which are much less affected by the environment, further highlighting the role of host selection. Therefore, we believe that, like other insects, the fecal microbiota of S. frugiperda cannot represent the microbiota colonizing within the gut; the microbial composition of the gut contents can represent the overall gut microbiota, and there is no difference between the microbiota in the gut feces and gut mucosa.
PVC film is degraded by the microbiota in the larvae’s gut, rather than by the gut itself
Those ecological findings allowed us to understand the microbiota stability and representativeness of different sites of S. frugiperda larvae; therefore, based on the original experiment, we only included gut feces (GF) obtained from the midgut to represent the gut microbial composition. Previous studies measuring physiological indicators proved that larvae could survive on PVC film; however, there was no conclusive evidence of plastic degradation indicating whether the decomposition occurred due to the gut microbiota or the gut of the insect itself. Therefore, we collected plastic film fragments from the excreted feces, including those fed PVC film (PVC + Larvae), and PVC film with antibiotics (Anti + PVC), and set standard PVC film as control. Then we used Advanced Polymer Chromatography (APC, Waters, China) to measure changes in the number-average molecular weight (Mn) and the weight-average molecular weight (Mw). Compared to the control and PVC with antibiotics groups, the Mn of PVC film without antibiotics treatment group significantly increased (p < 0.01, Fig. 2a) by 24.6%. The Mw also increased but not significantly (p > 0.05, Fig. 2b), which can be explained by the degradation of short-chain polymers in the plastic. This result indicates that the S. frugiperda larvae gut cannot directly degrade PVC polymers, however, the microbiota living there can. Subsequently, we further detected the degradation products from the standard PVC film (PVC), the larvae fed only corn leaves (Corn), and PVC film fragments recovered after larvae were fed PVC (PVC + larvae) to distinguish whether they originated from the microbial metabolites, plastic itself or from the plasticizers. The GC-MS results showed that the PVC film in the larval gut was effectively degraded, producing smaller organic molecules. In the larvae fed with PVC without antibiotics treatment group (PVC + larvae), a total of 7 new substances were detected (labeled as compounds 1 to 7, Fig. 3c), which were not detected in the standard PVC film, including 2-ethyl-1-hexanol, 2-ethylhexanoic acid, 2-nonanol, adipic acid, 3-hydroxydodecanoic acid, hexadecenoic acid, and octadecenoic acid. Compounds 8 and 9 were identified as two known plasticizers, namely dioctyl adipate and dioctyl phthalate. Based on the structures of the compounds, compounds 1, 2, and 4 may originate from the degradation of plasticizers (compounds 8 or 9, Fig. 3c), and we observed a decrease in the abundance of compound 8 after digestion by the larval microbiota, indicating that this plasticizer can be utilized by the gut microbiota and may serve as a major nutritional source within the gut. Additionally, compounds 3, 5, 6, and 7 could be potential PVC degradation products. Overall, this indicates that the gut microbiota of the larvae can degrade PVC film through various microbial pathways, including the degradation of plasticizers and PVC itself, but the gut of the larvae itself does not play a role.
Feeding PVC leads to inconsistent changes in the gut microbiota of S. frugiperda and T. molitor
Having confirmed that the gut microbiota of S. frugiperda larvae can indeed degrade PVC film, we included a dataset from T. molitor that degrades pure PVC to broaden the data resources for PVC-degrading gut microbiota and to exclude the influence of degrading PVC plasticizers. Since the sequencing regions of the 16S rRNA gene and analysis pipelines of the two datasets were inconsistent, direct comparison was challenging. To solve that we applied the same pipeline and database to reanalyze and delve deeper into these two datasets (see Methods). Consistent with the results in T. molitor, we observed substantial differences between the PVC group and the Corn group in S. frugiperda, with the R2 of 0.75 (Fig. 3a), higher than that observed in T. molitor (R2 = 0.49, Fig. 3b). This suggests that the gut microbiota of insects undergoes significant changes after being fed PVC to adapt to this novel "food". Unlike S. frugiperda larvae collected from the wild, the guts of our lab-reared larvae, which were fed exclusively on corn leaves, were occupied by a kind of unclassified Enterobacteriaceae (Fig. 3c), suggesting that this bacterium is capable of efficiently degrading corn leaves.
In contrast to T. molitor, the Shannon diversity significantly increased in the PVC group of S. frugiperda (p = 0.1, Figure S4a) but decreased in the PVC group of T. molitor (p = 0.17, Figure S4b). This might be explained by inconsistent feeding durations between the two experiments and the enrichment of Enterobacteriaceae in the Corn group of S. frugiperda, which inhibits the growth of other bacteria, with this inhibition disrupted in the PVC group. At the same time, we noted that the gut microbiota of both insects are partially similar, with numerous microbes belonging to Proteobacteria (primarily Enterobacteriaceae) and Firmicutes. In T. molitor, Actinobacteriota were not detected; instead, there were considerable numbers of Tenericutes (currently termed Mycoplasmatota) and Patescibacteria, though the microbes in this phylum disappeared after the PVC feeding (Fig. 3c-d). This likely indicates the differences in dietary preferences between the two insects, one eating corn leaves and the other eating bran. Following PVC feeding, the microbiota enriched in the two insects differed completely, with the most notable increase in S. frugiperda being Enterococcus, which surged from 4.7 ± 3.7% to 37.0 ± 19.8%. Additionally, there was enrichment of Ochrobactrum (from 0.1 ± 0.2% to 3.4 ± 0.7%), Klebsiella (from 1.4 ± 0.6% to 1.7 ± 0.4%), Falsochrobactrum (from 0% to 2.80 ± 1.40%), Microbacterium (from 0.03 ± 0.05% to 2.23 ± 1.55%), and Sphingobacterium (from 0.21 ± 0.30% to 2.55 ± 1.54%) (Fig. 3c). On the contrary, in T. molitor, significant enrichments were seen in Spiroplasma and Lactococcus, with Lactobacillus also showing notable enrichment in one of the parallel groups (Fig. 3d). These results indicate that the gut microbial compositions of the two insects are completely different, and the bacteria enriched after consuming PVC are also completely different.
Functional redundancy discovered in the gut microbiota of two PVC-fed insect larvae
Despite the distinct microbial compositions of the two insects, both of them are capable of degrading PVC, leading us to investigate further if they perform similar functions after feeding with PVC. Consequently, we performed PICRUSt2 to match the ASV sequences with the closest reference genomes to predict the metabolic functions of the gut microbiota in both insects. Through KEGG (Kyoto Encyclopedia of Genes and Genomes) database annotations, we identified 566 KEGG orthologs (KOs) significantly enriched in the PVC group of S. frugiperda (p < 0.01, Log2Foldchange > 2, Fig. 4a), mapping to 42 pathways. In T. molitor, 293 KOs were significantly enriched in the PVC group (p < 0.01, Log2Foldchange > 2, Fig. 4b), mapping to 39 pathways. We selected these pathways from the total identified and computed the overall enrichment across these pathways. Although the species of gut microbiota enriched in PVC differ between the two insects, the enriched metabolic functions show considerable redundancy. We discovered 23 enriched pathways common to both insects, with 19 pathways associated with metabolism (Fig. 4a-b). The enrichment of the other glycan degradation pathway might indicate the ability to degrade macromolecules.
To further identify the enzymes related to PVC degradation, we systematically summarized the plastic-degrading enzymes reported in the literature along with their Enzyme Commission (EC) numbers, and classified them into five categories: peroxidases, depolymerases, esterases, lipases, and cutinase. Subsequently, we used the same methodology above to predict the abundance of ECs in each sample, conducted the differential analysis using Deseq2, and extracted these five classes of enzymes based on text search. In S. frugiperda, we detected 32 enzymes, and for T. molitor, 33 enzymes were found. Carboxylesterase (EC 3.1.1.1) and Sialate O-acetylesterase (EC 3.1.1.53), NADH peroxidase (1.11.1.1), and Dye decolorizing peroxidase (1.11.1.19) were significantly enriched in the PVC group of S. frugiperda (Fig. 4c, p < 0.05). It is noteworthy that EC 1.11.1.1 and EC 1.11.1.19 were also observed to be enriched in T. molitor (Fig. 4d).
Since we mapped ASVs to the reference genome and estimated the gene abundance of the overall samples through the gene content of the reference genome, we traced back the microbes encoding these enzymes. We discovered that in S. frugiperda, EC 1.11.1.19 is primarily present in Microbacterium (ASV18) and Flavobacterium (ASV33), while EC 1.11.1.1 is mainly contributed by Enterococcus (ASV2, ASV4) (Dataset S2). Conversely, in T. molitor, these enzymes are mainly associated with ASVs belonging to Glutameicibacter and Lactobacillus (Dataset S3). Likewise, in the PVC group of S. frugiperda, there was a notable enrichment of Enterococcus ASVs (ASV2 and ASV4). The abundance of ASV2 rose from 2.59 ± 0.04% to 26.6 ± 0.69%, and ASV4 from 1.91 ± 0.01% to 10.3 ± 0.13% (Figure S5). The increase in abundance of ASV2 was twice that of ASV4, suggesting its greater potential for PVC degradation. Furthermore, in the genus Microbacterium, all eight ASVs showed an increase in abundance, especially ASV12, which increased from 0.2–0.4%. Similarly, in the genus Sphingobacterium, two branches showed an upward trend, with the most notable being ASV11, whose abundance increased from 0.1–2.8% (Figure S5). These results indicate the potential of these gut microbes and enzymes in the degradation of PVC.
NAD-dependent oxidoreductase isolated from E. casseliflavus EMBL-3 can dechlorinate and degrade PVC
The identified potential degradative enzymes and their encoding microbes leaded us to further verify these results in lab. Particularly interested in EC 1.11.1.1 encoded by Enterococcus ASV2, we isolated and cultivated microbes from the feces of S. frugiperda larvae and identified a strain of Enterococcus after several screenings (Figure S6a). By blasting the complete 16S rRNA gene sequence (Table S1) with the rRNA/ITS databases at NCBI, the isolate was confirmed to be Enterococcus casseliflavus (99.08%). Upon constructing a phylogenetic tree with our ASVs, we found that this bacterium was closer to ASV2 (Fig. 5a). This means that the bacterium we isolated had a low abundance in the gut of S. frugiperda fed on corn leaves, but its abundance significantly increased in the PVC group, and it was twice that of another Enterococcus strain, ASV4. However, the results of liquid culture experiments using PVC film (PVC and plasticizers) as the sole carbon source indicated that the strain E. casseliflavus did not solely exhibit significant degradative activity (Figure S6b).
To further investigate the gene content of this bacterium, we sequenced its whole genome using 2nd-generation sequencing and named this bacterium Enterococcus casseliflavus EMBL-3. By whole genome annotation using Prokka25, we found a gene was annotated as EC 1.11.1.1 in E. casseliflavus EMBL-3 and extracted the corresponding sequence (Table S2). Through protein sequence blast in the uniprot database, we identified this protein as NAD-dependent Oxidoreductase (NDO). Then it was expressed and purified in vitro (Fig. S6c). The activity of the enzyme (9.8 U/mgprot) was assessed by using peroxidase activity assay in accordance with the manufacturer’s instructions (Beijing Solarbio Science & Technology Co.,Ltd., Beijing, China). To validate the degradation ability of the enzyme on PVC, we set a 2 mL reaction system with 200 ug enzyme and 200 mg pure PVC powder (without plasticizer) and measured molecular weight change and dechlorination of the pure PVC powder (Method S7). The molecular weight of NDO-treated PVC decreased significantly than those untreated. The Mn decreased by 12.02% (from 79.34 kDa to 69.81 kDa, Fig. 5b) and the Mw decreased by 14.07% (from 164.04 kDa to 140.97 kDa, Fig. 5c). This result indicated a depolymerization of PVC polymer chains by NDO. A dechlorination of 0.74 mg/L was detected in the NDO-treatment system compared to the control system (Fig. 5d). NADH, which can be used as a reducing agent to provide hydrion and electron, may contribute to the dechlorination process. So we added NADH into the reaction system. When 2 mM NADH was added in combination with NDO, a distinct dechlorination activity was detected, with the dechlorination amount detected significantly increasing to 6.48 mg/L, while NADH alone did not have dechlorination activity on PVC (Figure S6d). This indicates that NDO works on PVC polymer by dechlorination. Although, we did not detect any degradation intermediates with a carbon chain shorter than 20 in the reaction system (Figure S7), the changes in polymer molecular weight due to dechlorination did not sufficiently account for the observed changes in the molecular mass of the polymer. Therefore, under several ideal assumptions, we performed a series of calculations on dechlorination amounts and changes in molecular weight (Method S8). Results demonstrated that 6.48 mg/L dechlorination had minimal impact on molecular weight, indicating that dechlorination cannot explain the molecular weight changes. The enzyme further cleaves long-chain PVC polymers beyond dechlorination, producing polymers with more than 20 carbons and resulting in decreased molecular weight of the PVC polymers. Overall, we demonstrated that this enzyme can act on pure PVC polymers, simultaneously dechlorinating and cutting long-chain PVC polymers.