Microbial Communities Associated With Stunted Growth Syndrome In Penaeus Vannamei Farming

Purpose: Stunted/slow growth syndrome is one of the yield-limiting diseases in Penaeus vannamei farming. Limited information is available on the aetiology of this condition, which needs to be studied to devise prophylactic measures to minimise the production losses. Amongst the factors that inuence this condition, microbial communities in the growing environment play an important role. This study aimed at understanding major microbial associations of affected and healthy pond waters through shotgun metagenomics. Method: The water samples were ltered through vacuum ltration to extract suspended microbes. Subsequently, DNA was isolated from the ltrate using PowerSoil® DNA Isolation Kit. Libraries prepared from isolated DNA were sequenced using the shotgun metagenomic method on the Illumina HiSeq platform. The microbial proling and their functional prediction of the shotgun metagenome sequences were carried out using stand-alone versions of Kaiju, OmicsBox respectively. Results: The taxonomic classication results revealed that species of Oceanospirillum, and vibrio were high in the disease sample, while Rhodobacteraceae bacterium and Neptunomonas were high in the healthy sample. The alpha diversity analysis showed slightly higher diversity in the healthy sample compared to the disease infected. The taxonomic biomarkers for healthy and infected states reported in previous studies were also observed in this study. The major functional associations of both the healthy and infected groups include amino acid transport and metabolism, cell wall/membrane/envelope biogenesis, and energy production and conversion. Conclusion: The study identied major taxonomical and functional associations of ponds affected and unaffected with stunted growth syndrome. These associations signicantly varied between the samples, indicating dysbiosis of the microbial proles in the pond waters. This dysbiosis could be a potential cause for the manifestation of stunted growth syndrome. Microbial associations along with other pond environmental factors need to be further explored for an in-depth understanding of stunted growth syndrome.


Introduction
Pond water microbiota plays an important role in nutrient cycling, maintenance of water quality, and health of the farm-reared shrimp . Signi cant correlations were reported between rearing-water bacterial communities and the intestinal microbiota of shrimps in a bio oc-based culture system (Cardona et al., 2016). There has been increasing evidence on the role of gut microbiota on important functions like immunity, health regulation, and nutrient absorption in shrimps ( Paci c white shrimp farming across the world is facing several impediments, mainly due to bacterial and viral diseases. Economic losses due to the diseases in India accounted for an annual loss of US$ 1.02 B (Patil et al., 2021). Stunted growth syndrome (SGS) is among one of the most commonly observed diseases contributing to considerable production losses. Genetics, environment, and many infectious agents especially the microsporidian Enterocytozoon hepatopenaei (EHP) were reported to play a role in the manifestation of this disease (Kooloth et al., 2021, Rajendran et al., 2016). However, the exact cause of this multifactor syndrome is unclear, suggesting a more in-depth investigation of all underlying factors.
Microbial communities of bacteria, archaea, viruses play a signi cant role in pond water quality and shrimp diseases. The study of the ambient microbiota is essential for establishing causal relationships for diseases. However, a major proportion of these microbes is culture-independent and could not be studied through conventional laboratory approaches. In such cases, metagenomics is a game-changer that enables researchers to study the microbial pro les on a larger scale (Quince et al., 2017, Forbes et al., 2017. Metagenomics applications in aquaculture include microbial pro ling, diversity analysis, identi cation of antibiotic resistance genes (ARGs), novel and potential pathogens, microbial associations with bio ocs, and probiotics (Martínez et al. Both Water and animal samples were collected from the pond experiencing SGS (Disease) and normal ponds (Healthy). Animal samples were tested negative for known pathogens like White spot syndrome virus (WSSV), Infectious hypodermal and hematopoietic necrosis (IHHNV), and Enterocytozoon Hepatopenaei (EHP) (Fig. 1).
Sample preparation, DNA isolation, and sequencing Water samples of 10 litres stored at 4°C were used for microbial DNA isolation. Five litres of water from infected and healthy samples were centrifuged at 5000g for 15 min at 4°C. The supernatant was collected in a sterile container. Pellet was suspended in 50 ml sterile seawater and stored at 4°C until use.
The supernatant was ltered through a 0.22 µ lter paper by vacuum ltration to collect suspended microbes. The ltrate was scrapped and collected in a sterile 2 ml microcentrifuge tube. DNA was isolated from 0.25g of microbial samples using the PowerSoil® DNA Isolation Kit (Qiagen). Sequencing libraries were prepared using the NEBnext ultra DNA library preparation kit. The quality of the library was checked using an Agilent TapeStation. Libraries were sequenced using the Illumina HiSeq2500 platform and Paired-end reads of 250 base pairs were generated for both samples.

Taxonomic classi cation
Raw reads obtained from the sequencer were subjected to different shotgun metagenomic data processing methods like quality assessment and control, taxonomic classi cation, assembly, differential abundance analysis, functional annotation, diversity analysis, virulence, and resistant gene predictions. Quality control tool FastQC v0.11.8 (Andrews et al., 2010) was used to obtain an overview of the quality metrics of the raw reads. These reads were trimmed for low quality and adapter content using trimmomatic v0.39 (Bolger et al., 2014) and retained sequences having a minimum length of 50 bases.
The quality trimmed reads were subjected to taxonomic classi cation using the standalone version of Kaiju v. 1.7.3 (Menzel et al., 2016). Kaiju was run with default parameters against the latest prokaryotic non-redundant protein database along with fungi and microbial eukaryotes (nr+euk). Microbial diversity analysis was carried out using R library 'vegan' (Oksanen et al., 2013) with a threshold of 0.1% minimum abundance to estimate Shannon, Simpson, Inverse Simpson and sher alpha.

Metagenome assembly and annotation
Simultaneously, quality trimmed reads were uploaded to the metagenomics pipeline in OmicsBox (BioBam, 2019) software for functional annotation. OmicsBox implementation of metaSpades v.3.15.2 (Nurk et al., 2017) for metagenome assembly, prodigal (Hyatt et al., 2010) for open reading frames (ORF) prediction and EggNOG (Jensen et al., 2008) mapping for annotation of orthologous groups of genes were chosen in this study. Predicted ORFs were searched against the virulence factor database (VFDB) (Chen et al., 2005) to identify candidate virulent genes with an e-value set at 1e-6. Venn diagram was plotted using the online tool Venn diagram plotting tool (http://bioinformatics.psb.ugent.be/webtools/Venn/).

Data availability
The nextgen sequencing data of this study are deposited in the SRA repository of Genbank with the accession number PRJNA686191

Sequence statistics
Shotgun sequencing of the samples generated 8,963,000 and 8,445,407 raw sequence reads for both the healthy and disease samples respectively. The quality trimming ltered around 30% of the reads from both the samples resulting in high quality reads for further processing (Table 1). Around 70 % of the reads have been retained post QC, with minimum and maximum read lengths of 50 and 250 for both the healthy and disease groups respectively. GC content of the healthy and diseased samples was found to be 54% and 45% respectively.
Complete taxonomic classi cation lists along with their corresponding read counts are given in supplementary table 1.
The diversity indices to assess the species richness within and between samples were calculated. As shown in table 2, the estimates for Shannon index, Simpson index and sher alpha were slightly higher for healthy samples compared to diseased samples, which indicates richer diversity in the healthy sample than that of the diseased sample.

Metagenome assembly and annotation
Metagenome assembly carried out with metaSPAdes assembler revealed a total of 16,315 and 8,503 contigs with an N50 value of 5,823 and 12,482 for healthy and diseased groups respectively (Table 3). A total of 91,003 ORFs for healthy and 62,136 for disease sample was predicted from the assembled contigs using the ORF prediction tool Prodigal.
The EggNOG annotations were assigned to 74,324 predicted ORFs of healthy and 54,077 of disease samples. Out of which 19,575 (healthy) and 17,139 (disease) were assigned with proper GO annotations (Supplementary Table 2). The genes were majorly involved in metabolism in both the samples, especially in amino acid transport & metabolism, and energy production & conversion, followed by cellular processing such as defence mechanism, cell wall, membrane and envelope related genes (Fig. 4). Search for antibiotic resistance genes (ARGs) resulted in hits for 147 sequences from healthy samples and 161 sequences from disease samples (Fig. 5). KEGG pathway map IDs were assigned to 29,055 and 22,510 sequences and toxin-antitoxin systems to 328 and 224 sequences for healthy and disease samples respectively. In addition, 197 sequences of healthy and 147 sequences of diseased samples were found to be secretion systems genes of type I, II, III, IV, VI, and VIII.

Virulence factors
A total of 66224 and 53127 hits were generated for healthy and diseased samples respectively from virulence gene prediction through blast against the VFDB database. On considering the hits with a minimum identity percentage of 80% and query coverage of 40% a total of 451 hits for healthy and 169 hits for infected samples remained on ltering. On further ltering for the best match with a 100% query coverage and a minimum of 80% identity, 18 and 25 putative virulence genes were identi ed for healthy and diseased samples respectively. The blast results of the predicted ORFs against the VFDB database are provided in Supplementary Table 3.

Discussion
The optimal quality of the rearing water is essential for the proper growth and development of aquaculture animals, if not managed properly, would act as a source for exerting biotic and abiotic stress and subsequently cause disease (Li et al., 2021). The composition and stability (Dysbiosis) of microbial colonies present in the pond waters are considered to play a major role in disease manifestations (Tello et al., 2020). In the case of shrimp, microbes present in pond waters enter through injury, feed, gills, mouth, and antennal gland (Aguirre-Guzmán et. al., 2010al., , Liu et al., 2021 and may act as bene cial or detrimental to the host. In recent years, alterations in microbial composition were reported to be the cause of several diseases; however, there are no reports on stunted growth syndrome as of date.
Here, we conducted whole metagenomic pro ling for two different shrimp ponds in which one of them was infested with stunted growth syndrome. The absence of known disease-causing pathogens like WSSV, IHHNV and EHP the commonly reported associations to growth retardedness necessitated this study.
Results revealed proteobacteria, bacteroidetes, actinobacteria and rmicutes, the commonly reported dominant phyla in the aquaculture environment (He et al., 2020, Fan, andLi, 2019), are found to be dominant in both samples. Families viz., Rhodobacteraceae and pseudoalteromonadaceae are dominant in the healthy sample, whereas oceanospirillaceae, avobacteriaceae, and vibrionaceae are dominant in diseased condition.
A recent study on white faeces syndrome reported a high abundance of Rhodobacteraceae in healthy shrimps compared to slow-growing shrimps and their surroundings (Wang et al., 2020). Other studies reported Rhodobacteraceae and Flavobacteriaceae as the potential taxonomic indicators of shrimp health (Wang et al., 2020;Xiong et al., 2017).
The Oceanospirillum is one of the common inhabitants of marine environments (Leonard et al., 2000). It has been identi ed to be potentially pathogenic to paci c white shrimp, oysters and humans (Horodesky The species-level search revealed a signi cant difference between healthy and disease samples. The healthy sample showed a high abundance of Rhodobacteraceae bacterium EhC02, while the disease sample was dominated by Oceanospirillum sanctuarii, Oceanospirillum maris, Fluvicola sp., and Vibrio aphrogenes.
Several studies reported lower microbial diversity in disease states in comparison with healthy groups. Here, we observed a similar pattern with regard to alpha diversity indices of SGS samples, even though the difference is marginal. This provides empirical evidence for the bene cial effects of more diverse microbial colonies in the habitat.
The functional pro le revealed that genes related to transcriptional regulators, major facilitator superfamily (MFS), proteins conserved in bacteria, acetyltransferase and ABC transporters were found among the top 10 abundant genes in both the samples, in addition to these few antibiotic resistancerelated genes such as resistance nodulation cell division gene was also observed. Here again, the EggNOG annotation also revealed that the functional pro le of the diseased sample was associated more with Oceanospirillales and Vibrionales.

Conclusion
In conclusion, the ambient microbiota showed a considerable difference between healthy and diseased conditions, and changes were observed in both taxonomic and functional pro les. This dysbiosis in the microbiome and its functional pro le may be a contributing factor to disease. On the contrary, though the dominant bacteria may not be directly associated with disease, still could be a potential threat when the animal is under stress with their opportunistic disease-causing nature. Future work with better coverage and study of environmental parameters along with biotic composition may provide conclusive results in identifying the causal relationships between microbiota and stunted growth disease. The ndings of this study will help in understanding the aetiology of stunted growth syndrome.   Table 3: Metagenome assembly and annotation statistics Figure 1 The above image shows the variation in size of shrimps from the same pond that had been affected by stunted growth.   Distribution of COG categories among healthy and disease samples

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.