Molting Alters the Microbiome, Immune Response, and Digestive Enzyme Activity in Mud Crab (Scylla paramamosain)

ABSTRACT Molting is a crucial lifelong process in the growth, development, and reproduction of crustaceans. In mud crab (Scylla paramamosain), new exoskeleton, gills, and appendages are formed after a molting, which contributes to a 40 to 90% increase in body weight. However, little is currently known about the associations between molting and the dynamic changes of microbiota and physiological characteristics in mud crabs. In this study, the effects of molting on changes of the microbiome, immune response, and digestive enzyme activities in mud crabs were investigated. The results showed dynamic changes in the abundances and community compositions of crab-associated microbiota harboring the gills, subcuticular epidermis, hepatopancreas, midgut, and hemolymph during molting. Renewed microbiota was observed in the gills and midgut of crabs at the postmolt stages, which seems to be related to the formation of a new exoskeleton after the molting. A significant positive correlation between the expression of two antimicrobial peptide (AMP) genes (SpALF5 and SpCrustin) and the relative abundance of two predominant microorganisms (Halomonas and Shewanella) in hemolymph was observed in the whole molt cycle, suggesting that AMPs play a role in modulating hemolymph microbiota. Furthermore, digestive enzymes might play a vital role in the changes of microbiota harboring the hepatopancreas and midgut, which provide suitable conditions for restoring and reconstructing host-microbiome homeostasis during molting. In conclusion, this study confirms that molting affects host-associated microbiota and further sheds light on the effects on the immune response and the digestive systems as well. IMPORTANCE Molting is crucial for crustaceans. In mud crab, its exoskeleton is renewed periodically during molting, and this process is an ideal model to study the effects of host development on its microbiota. Here, multiple approaches were used to investigate the changes in microbial taxa, immune response, and digestive enzyme activity with respect to molting in mud crab. The results found that a renewed microbiota was generated in the gills and midgut of crab after a molt. A significant positive correlation between changes in the relative abundances of microbes (such as Halomonas and Shewanella) and the expression of AMP genes (SpALF5 and SpCrustin) was observed in the hemolymph of crabs during the whole molt cycle, suggesting the modulation of hemolymph microbes by AMPs. Furthermore, the digestive enzymes were found to participate in the regulation of microbiota in hepatopancreas and midgut, consequently providing a suitable condition for the restoration and reconstruction of host-microbiome homeostasis during the molting. This study confirms that molting affects the microbial communities and concomitantly influences the immune and digestive systems in mud crabs. This is also the first time the homeostasis of the host and microbiome, and the associations between molting and physiological characteristics in crustaceans, have been revealed.

IMPORTANCE Molting is crucial for crustaceans. In mud crab, its exoskeleton is renewed periodically during molting, and this process is an ideal model to study the effects of host development on its microbiota. Here, multiple approaches were used to investigate the changes in microbial taxa, immune response, and digestive enzyme activity with respect to molting in mud crab. The results found that a renewed microbiota was generated in the gills and midgut of crab after a molt. A significant positive correlation between changes in the relative abundances of microbes (such as Halomonas and Shewanella) and the expression of AMP genes (SpALF5 and SpCrustin) was observed in the hemolymph of crabs during the whole molt cycle, suggesting the modulation of hemolymph microbes by AMPs. Furthermore, the digestive enzymes were found to participate in the regulation of microbiota in hepatopancreas and midgut, consequently providing a suitable condition for the restoration and reconstruction of host-microbiome homeostasis during the molting. This study confirms that molting affects the microbial communities and concomitantly influences the immune and digestive systems in mud crabs. This is also the first time the homeostasis of the host and microbiome, and the associations width, were measured at pre (Ds)-and postmolt (0 to 15 days) stages. The Ds crabs are those with a crack on their exoskeleton (Fig. 1) and molt within 1 to 2 days. The 0-day group indicates mud crabs finishing their molt within 8 h. The bodyweight of mud crabs at 0 days was significantly increased compared to those at Ds (with an average increase of 66.70%, P , 0.05; Fig. 2A). The carapace length and width of crabs increased 20.16% and 18.86%, respectively, at 0 days compared with those at Ds (P , 0.05; Fig. 2B and C). A gradual decrease in hepatopancreas index (HI) also occurred during the whole postmolt stages, with the highest and lowest values of 0.08 (Ds) and 0.03 (15 days), respectively (Fig. 2D).
Hemolymph conditions and expression of immune-related genes. The average number of hemocytes at the intermolt stage was 3.9 Â 10 6 cells/ml (15 days) (Fig. 3A). A significant increase in hemocyte density was observed only at 3 days (8.5 Â 10 6 cells/ ml) (P , 0.05; Fig. 3A). Compared to the Ds, the relative ROS level increased by 1.5-fold at 0 days and 6 days (P , 0.05), and this value reduced to the basal level at 3 days and 15 days (Fig. 3B). The total antioxidant capacity (TAOC) in hemolymph showed a 56.00% decrease at 0 days compared to Ds (P , 0.05), which remained at a low value of 0.13 mmol/liter at 15 days (Fig. 3C).
Transcript levels of two AMPs, SpALF5 and SpCrustin, significantly increased by 3.3fold and 8.5-fold at 0 days compared to Ds (Fig. 3D). The expression of SpALF5 was slightly low, whereas SpCrustin was highly expressed in the hemolymph of crabs from 3 days to 15 days compared to Ds. The expression of two phagocytosis-related genes, SpLAMP and SpRab5, was significantly increased at 0 days compared to Ds (P , 0.05) and decreased to the basal level from 3 days to 15 days (Fig. 3E). The highest expression of SpCTL-B and SpproPO was found in crabs between 3 days and 6 days after molt (Fig. 3F).
Changes in microbial abundance and composition during molting. The highest microbial diversity was found in the culturing seawater and crab foods, as determined by the Shannon, Chao1, observed species, and PD whole tree indexes (see Fig. S1 in the supplemental material). Principal coordinate analysis (PCoA) revealed that the  The numbers 0d, 3d, 6d, and 15d represent the sampled crabs at different days after a molt. Significant differences (P , 0.05) are represented by different letters. microbiota was tissue specific and stage related during the molting process (Fig. 4). The bacterial community structure in hemolymph was separated from those in other tissues (Fig. 4). However, the microbiotas between subcuticular epidermis and hepatopancreas were similar among different molt stages (Fig. 4). The shared microbial groups (at the genus level) in gills, subcuticular epidermis, midgut, hemolymph, and hepatopancreas were 62, 48, 32, 16, and 17, respectively (Fig. S2).
In gills, the highest microbial abundance was in the crabs at Ds (1.1 Â 10 8 cells/g), and the lowest was at 0 days (3.1 Â 10 6 cells/g). There was a gradual increase from 0 days to 15 days, with the highest density of 3.4 Â 10 7 cells/g at 15 days (Fig. 5A). Based on the 16S rRNA gene sequencing results, the Venn diagram showed that a total of 36 bacterial genera disappeared after molt (Ds versus 0 days), while 41 genera were newly detected at 0 days compared to Ds crabs (Fig. S2A). The microbial diversity was lowest at 0 days, and the relative abundance of major bacteria also showed a distinct pattern compared to other molt stages. The relative abundances of unclassified Rhodobacterales, unclassified Saprospiraceae, Vibrio, and Flavobacterium were significantly decreased after molting (Ds versus 0 days) (P , 0.05), whereas the relative abundances of Halomonas and Devosia significantly increased from Ds to 0 days (P , 0.05) before decreasing from 3 days to 15 days (Fig. 6, Table S3A).
In the subcuticular epidermis, the lowest abundances of microorganisms were at 0 days (7.2 Â 10 5 cells/g) and 6 days (6.5 Â 10 5 cells/g) (Fig. 5B). The highest microbial abundance (6.8 Â 10 6 cells/g) was found at 15 days, followed by 3 days (2.8 Â 10 6 cells/g) and Ds (2.1 Â 10 6 cells/g). The microbial diversity reached the highest level at Ds and maintained at the similar lower values between 0 days and 15 days (Fig. S1). In total, 135 bacterial groups disappeared at 0 days compared to Ds. Of these bacterial groups, 43, 13, and 7 genera were recovered at 3 days, 6 days, and 15 days, FIG 4 Principal coordinate analysis (PCoA) separates the crab microbiota with different tissues and molt stages. Different symbols and colors indicate different crab tissues and molt stages, respectively. G, gill; Se, subcuticular epidermis; Mi, midgut; Hemo, hemolymph; H, hepatopancreas; Sw, seawater; Ds, double-shelled crab. The numbers 0d, 3d, 6d, and 15d represent the sampled crabs at different days after a molt. The distances were determined using the Bray-Curtis method with relative abundances of microorganisms at the genus level. Three samples are sequenced for each group. respectively (Fig. S2B). Twenty-five genera, including Cetobacterium, Clostridium, Myroides, Lactococcus, Clostridium, and Lewinella, were newly detected at 0 days (Fig. 6, Table S3B).

DISCUSSION
This study shows that various tissues of mud crabs harbored distinct relative abundance and communities of microorganisms, which confirms the hypothesis that hostassociated microorganisms are tissue specific (34). Because the gills, subcuticular epidermis, and gut of mud crabs are in direct contact with the surrounding seawater, these organs may accumulate beneficial and harmful substances (including microorganisms) from the environment (19,35). The symbiotic microorganisms harboring hosts' bodies seem to change their diversity and community composition at different developmental stages, such as molting and metamorphosis (20,22,23). In this study, great differences in mud crab's morphometric characteristics (body weight, carapace length, width, and HI), the activity of digestive enzymes, immune response (the expression of immune-related genes), and colonized microbiota (abundance and community composition) were observed at 0 days postmolt compared to those at premolt (Ds) and 15 days postmolt.
The microbial density in the gills dramatically decreased to the lowest level at 0 days, which gradually recovered during the postmolt period. Notably, the relative abundance of some potential pathogens, such as Flavobacterium, Pseudoalteromonas, and Vibrio, decreased after molting, suggesting that molting plays an essential role in removing pathogenic bacteria from the gills of mud crabs. In mud crab, gills are responsible for gaseous exchange, osmolyte transport, nitrogenous excretion, acid-base balance, volume regulation between the body and external environment, and protection against infection and also act as a potential hematopoietic organ (36)(37)(38)(39). The highest microbial cell density was found in the gills of crabs at the premolt stage, which might be related to the direct contact with the surrounding environment of gills during respiration, water absorption, and microbial filtration.
Microbial numbers decreased to the lowest in midgut at 0 days (3.8 Â 10 7 cells/g) but recovered at 3 days (4.2 Â 10 8 cells/g) to reach levels that are similar to those at 15 days (4.2 Â 10 8 cells/g) and Ds (6.1 Â 10 8 cells/g), which could be attributed to the observation that crabs cease feeding during molting (40). At a period of 3 to 15 days postmolt, an increase in the relative abundance of microbiota (e.g., Photobacterium, Escherichia, Vibrio, and Arcobacter) in the midgut was observed, which might come from the surrounding seawater and feed (22,41). Interestingly, SourceTracker analysis showed that some microorganisms in the midgut were from other tissues, such as gills, hemolymph, hepatopancreas, and subcuticular epidermis (Fig. 8). Taken together, microorganisms originated from the surrounding environment and other tissues are important sources of gut microbiota in mud crabs, as previously reported (15,16,20,29).
However, the core microbiota in these mud crab tissues did not change during the whole molt cycle, although some microorganisms are separated from the old tissues. The translocation and reconstruction of the core microbiota in different tissues cooccur with the molting process. These events are also regulated by the physiological response of the host (through the expression of immune-related genes and modulation of digestive enzyme activity) and environmental factors (foods and seawater), which finally maintain the homeostasis between crabs and their associated microbiota (42).
The results of this study revealed an upregulation in the expression of genes involved in antimicrobial response (SpALF5 and SpCrustin), phagocytosis (SpLAMP and SpRab5), agglutination (SpCTL-B), and melanization (SpproPO) in the hemolymph of mud crabs at 0 days compared with those at Ds, which indicates the role of molting in activating host immune response. Antimicrobial peptides (AMPs) are essential immune molecules that protect hosts against exogenous microorganisms. Hemocytes of healthy Pacific oysters could express basal levels of two AMPs (defensins and prolinerich peptides), which indicates that hemolymph microorganisms stimulate the generation of these AMPs to boost host immunity (43). Moreover, the results obtained previously showed the high expression of SpALF5 and SpCrustin genes in the hemolymph of mud crab challenged with Vibrio parahaemolyticus (34,44,45). In the current study, the expression of AMPs (SpALF5 and SpCrustin) correlated positively and significantly with the relative abundances of Halomonas and Shewanella during molting (Fig. 7A), which indicates the role of AMPs in modulating the abundance of these two bacterial genera in mud crab hemolymph. As previously reported, AMPs can induce the transcription of immune effectors that protect the host against bacterial invasion through IMD and Toll signaling pathways (46,47) and their downstream molecules (i.e., Dorsal, Relish, and Caudal) to maintain homeostasis. However, Halomonas and Shewanella genera are opportunistic pathogens (29,(48)(49)(50). For example, exposure to Microcystis aeruginosa increased the relative abundance of pathogenic microorganisms such as Halomonas and Shewanella in the zebrafish intestine, which enhanced inflammatory response and produced detrimental effects (51). Shewanella spp. are occasionally isolated from diseased aquatic animals such as fish and shrimps and may facilitate the increase in the hemolytic activity of V. parahaemolyticus in a time-dose-temperaturedependent manner (52,53). Furthermore, hemocyte agglutination and phagocytosis are important in immune defense against pathogens, with increased ROS levels involved in eliminating bacteria pathogens in crabs and shrimps (54,55). At 0 days after molting, the significant increase in SpCTL-B (a C-type lectin gene), two phagocytosisrelated genes, SpLAMP and SpRab5, coupled with the increased ROS level, indicates an immune response by the crabs to invasion by the exogenous bacteria or proliferation of indigenous bacteria. This observation is probably explained by the important role of hemolymph in the host immune response. The microbiota colonizing the hemolymph are tightly controlled by host-microbiome homeostasis immunity (29), which is confirmed here by the stability of microbial abundances during a molt cycle in the hemolymph of mud crabs. A positive correlation between total hemocyte counts (THC) and the relative abundance of unclassified Caulobacteraceae was observed in present study.
Caulobacteraceae are a family of nonfilamentous bacteria that form stalks or zoogloeal masses of ferric hydroxide (56), which indicates that the Caulobacteraceae found in hemolymph live a syntrophic or epibiotic lifestyle by colonizing or attaching themselves to the surface of hemocytes, as reported previously (29). In the digestive system, the interactions between digestive enzymes and the abundance of microorganisms are versatile. Digestive enzymes are produced and secreted by the host to digest fats, proteins, and carbohydrates in foods to provide nutrients for the growth and development of the host (57). Gut microorganisms can further use digestible and/or indigestible debris (e.g., dietary fibers) as carbon and energy sources (19,58). In the present study, the highest activity of digestive enzymes was found 3 days postmolt, which possibly relates to the feeding habit of mud crabs during a molt cycle. Molting is an extremely energy-consuming process that lasts 1 to 2 days, with crabs staying still in their shelters and not feeding during and immediately after molting (40). Crabs might only start to eat at 3 days postmolt; thus, the higher digestive enzyme activity facilitates more efficient assimilation of nutrients. Interestingly, in fish and other vertebrates, gut microbiota (such as Clostridium, Actinomyces, and Ruminococcus) are capable of synthesizing digestive enzymes (such as cellulase, xylanase, pectinase, and/or b-glucanase) (59-62). Here, changes in the relative abundance of microbiota in the hepatopancreas and midgut of mud crabs are correlated with digestive enzyme activities during the whole molt period. For example, in the hepatopancreas, the relative abundances of Arcobacter and Methylobacterium were significantly related to the activity of chitinase and amylase (Fig. 7B). The results agree with the previous findings that Arcobacter spp. can produce chitinase and lysozyme and participate in the biodegradation of shrimp shells (63). Methylobacterium spp. can produce cellulase, pectinase, lipase, amylase, and chitinase (64). The bacterial genera Bacteroides, Clostridium, and Pseudoalteromonas are significantly correlated with amylase activity in the midgut of mud crabs during molting (Fig. 7C), which is consistent with the results of previous studies that many members of these genera can produce amylase (65)(66)(67). Collectively, the results of this study suggest that digestive enzymes could regulate the microbiota in the hepatopancreas and midgut of mud crab during molting, which provides a suitable condition for the restoration and reconstruction of host-microbiome homeostasis in crab molting.
In conclusion, this study confirms the hypothesis that molting affects the microbiota in the tissues of mud crabs. Renewed microbiota was generated in gills and midgut after a molt, and dynamic changes of crab-associated microbiota are observed in a molt cycle. This study suggests that the molting would further alter the immune response and the digestive systems in mud crab as well. Future studies using metagenomics, metatranscriptomics, and/or culturomics are needed to identify the potential functions of the host-associated microorganisms during molting in mud crabs.

MATERIALS AND METHODS
Ethics statement. The mud crabs used in this study were obtained from a local aquaculture farm in Xiamen, Fujian, China. The animals were processed according to the Regulations for the Administration of Affairs Concerning Experimental Animals established by Guangdong Provincial Department of Science and Technology on animal use and care. The experiments were approved by the Institutional Animal Care and Use Committee of Shantou University.
Mud crab aquaculture and environmental parameters of the seawater. In general, male crabs (100 6 20 g), which were healthy with no signs of disease, had complete appendages, and no material attachments on their body surfaces, were used. Before the experiments, crabs were cultured in a "crab house" aquaculture system equipped with a continuous recirculating seawater system. The seawater used was drawn from nearby natural seawater, which was then passed through a filter system containing activated carbon and silica sand, with the seawater's daily exchange capacity maintained at 600 to 800% in volume. Crabs were individually kept in the units of the crab house (40 by 25 by 18 cm) to avoid fighting or being killed during molting. The crab house was cleaned once daily to remove crab feces and residual feeds. Crabs were fed daily with a mixture of Ruditapes philippinarum, Cipangopaludina cahayensis, and Sinonovacula constricta at a total amount of 10 to ;20 g per day. During the culturing period, the average temperature (24.6°C), salinity (20.9 %), pH (7.9), and nitrate (35.4 mg/liter) and ammonium concentrations (0.05 mg/liter) in the seawater were relatively stable (see Fig. S4 and S5 in the supplemental material), and the dissolved oxygen concentration was maintained at 8.15 6 0.36 mg/liter every day. Molting was observed and recorded three times daily (at 8:00, 16:00, and 24:00, respectively). Most of the crabs were molted at two time intervals, including October 19 to 31 and November 7 to 11 in 2016 (Fig. S4). In this study, five experimental groups were set up, and six crabs were used in each group. The five groups included premolt crabs (double-shelled crabs, or Ds), 0 days postmolt, 3 days postmolt, 6 days postmolt, and 15 days postmolt.
Morphometric characteristics. The body weight, carapace length, and width of each crab were measured at the beginning (when the crabs were introduced to the crab house) and end (the time of sample collection) of each experiment. The hepatopancreas index (HI) was calculated according to the formula hepatopancreas index (hi) = total hepatopancreas weight/bodyweight.
Sample collection and microbial cell enumeration. The sampling procedure was performed in a sterile laminar flow hood, with the surface of each crab disinfected and wiped three times with 75% (vol/vol) ethanol. Hemolymph was withdrawn via the arthropodal membrane (at the bases of the walking legs) using a 1-ml disposable syringe into a tube containing precooled sterile anticoagulant buffer (450 mM NaCl, 100 mM glucose, 26 mM citric acid, 30 mM sodium citrate, pH 4.6) at a ratio of 1:1. Two hundred microliters of the anticoagulant-hemolymph mixture was fixed with 4 ml of sterile formalin solution (20 g/liter NaCl, 30 ml/liter formalin) for 2 h and stored at 4°C for cell enumeration (three replicates). The gills, subcuticular epidermis (a thin tissue that attached to the inner part of the shell), and midgut were collected into 2-ml sterile tubes with separate flame-sterilized scissors, which were then cut into pieces (,0.2 cm 2 ). Pieces of food (including R. philippinarum, C. cahayensis, and S. constricta) used for feeding the mud crabs were collected and cut into small pieces (,0.2 cm 2 ). Tissue samples were quickly weighed, and 0.1 g of each sample was preserved in 500 ml of 1Â phosphate-buffered saline (PBS)-ethanol (vol/vol, 1:1) before being stored at 220°C until further analyses. One liter of culturing seawater was filtered through a sterile filter with a 0.22-mm mesh membrane (GPWP04700; Millipore, USA) using a vacuum pump, and the microbial cells from the culturing seawater were collected by direct filtering of the water. Half of the membranes were stored at 220°C for DNA extraction, and the other half were fixed with sterile formalin solution for 2 h and stored at 4°C for cell counting.
The anticoagulant-hemolymph-formalin mixture was processed as described previously (29). For gills, subcuticular epidermis, midgut, and crab food, the preserved samples were thawed and vortexed for 30 min (30). The mixture was diluted to the appropriate volume with sterile formalin solution (20 g/liter NaCl, 30 ml/liter formalin) and fixed for 2 h. The microbial mixture was captured by a 0.2-mm mesh membrane (GTBP02500; Millipore, USA).
All the collected samples containing microbial cells were stained with SYBR green I solution (1:40 vol/ vol SYBR green I in 1Â Tris-EDTA buffer) for 20 min. After staining, the dye was removed and the membrane was immersed in 25 ml of 10% glycerin. Cells were counted at Â1,000 magnification using a fluorescence microscope (ECLIPSE 90i; Nikon, Japan) with a blue filter as described elsewhere (68,69). Each sample was extracted and counted in triplicates.
DNA extraction, PCR amplification, and sequencing. The total microbial DNA of hemolymph, gills, subcuticular epidermis, midgut, hepatopancreas, and crab foods were extracted using the PowerFecal DNA isolation kit (MO BIO Laboratories, USA) according to the manufacturer's protocols. The hypervariable V3-V4 regions of the bacterial 16S rRNA genes were amplified by PCR using primers 341F and 806R (Table S1). PCRs were performed in triplicates with a 20-ml mixture containing 4 ml of 5Â FastPfu buffer, 2 ml of 2.5 mM deoxynucleoside triphosphates, 5 mM forward and reverse primers, 0.4 ml of FastPfu DNA polymerase (TransStart, China), and 10 ng of template DNA and made up to the required volume with sterile water. The PCR conditions were 95°C for 5 min, 27 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 45 s, and a final extension at 72°C for 5 min. The PCR products were then extracted from the 1% agarose gels and purified with a DNA gel extraction kit (Axygen, Hangzhou, China) according to the manufacturer's instructions. The concentration of the purified DNA was quantified using NanoDrop 2000 spectrophotometer. Equal molar amounts of each purified PCR product were pooled and sequenced on an Illumina HiSeq 2500 platform (2 Â 250 bp) according to standard protocols.
Determination of total hemocyte counts, reactive oxygen species, and total antioxidant capacity in hemolymph. Approximately 20 ml of hemolymph was placed in a hemocytometer for total hemocyte count using an inverted phase-contrast microscope (Carl Zeiss Tessar Axio Lab, Germany). All hemocytes in both the top and bottom fields (1 by 1 mm) of the hemocytometer were counted. One milliliter of diluted hemolymph was used to measure ROS levels with a reactive oxygen species assay kit (Beyotime Biotechnology, China). Briefly, hemolymph was centrifuged at 800 Â g for 20 min at 4°C to isolate hemocytes, followed by washing with 1Â PBS and resuspended with 200 ml of 1Â PBS containing 10 mM DCFH-DA. The mixture was incubated (protected from light) for 20 min at 37°C, after which the fluorescence intensity was measured at 488 and 525 nm using an Infinite 200 PRO NanoQuant microplate reader (Tecan, Switzerland). Total antioxidant capacity was determined using a commercial test kit (number A015-2-1) according to the manufacturer's instructions (Nanjing Jiancheng Bioengineering Institute, China). The optical density was measured at 405 nm using a microplate reader.
Determination of hepatopancreas digestive enzyme activity. The hepatopancreas of each crab was weighed and homogenized with 1Â PBS before being centrifuged at 3,000 Â g for 10 min at 4°C. The supernatant was used to measure enzymatic activities of lipase, amylase, and trypsin using commercial test kits (numbers A054-2-1, C016-1-1, and A080-2-2, respectively) (Nanjing Jiancheng Bioengineering Institute, China) according to the manufacturer's instructions. An increase of 0.001 per min is defined as one active unit (U). Chitinase activity was measured using an enzyme-linked immunosorbent assay kit (CK-EN95851; Shanghai Guchen Biotechnology, China) according to the manufacturer's instructions.
RNA extraction and RT-qPCR. The expression of six immune-related genes representing the four major types of innate immunity in mud crab were measured, including those of antimicrobial response (SpALF5 and SpCrustin), phagocytosis (SpLAMP and SpRab5), agglutination (SpCTL-B), and melanization (SpproPO) (44,(70)(71)(72)(73)(74). The hemocytes of crabs at various molt stages were collected by centrifuging hemolymph at 800 Â g for 20 min at 4°C. Total RNA from hemocytes was extracted using TRIzol reagent (Ambion, USA). First-strand cDNAs were synthesized using the PrimeScript RT reagent kit with gDNA Eraser (TaKaRa, Dalian, China) according to the manufacturer's instructions. The quantitative real-time PCR (RT-qPCR) analysis used the SYBR Premix Ex Taq II kit (Perfect Real Time) (TaKaRa, Dalian, China) by following the manufacturer's instructions on a LightCycler 480 (Roche, USA). The RT-qPCR total reaction volume was 20 ml containing 10 ml of SYBR Premix Ex Taq II, 2 ml of the 4-fold diluted cDNA, 0.8 ml (10 mM) each of the forward and reverse primers, and 6.4 ml of ultrapure water. The primers used for the six immune-related genes (including SpALF5, SpCrustin, SpLAMP, SpRab5, SpCTL-B, and SpproPO) are in Table S1. The amplification procedure included a denaturation step of 95°C for 30 s, followed by 40 cycles of 95°C for 5 s, 60°C for 20 s, and then a melting curve analysis from 65°C to 95°C (74,75). Triplicate samples were used for all experiments. The relative transcript level of each gene was determined with the b-actin gene as an internal control (75).
Microbial community composition and SourceTracker analysis based on the 16S rRNA genes. The raw sequence data analysis used a previously described procedure (29,30). After removing the chimeric sequences, 39,750 effective sequences, on average, were randomly sampled for each sample to maintain the same sequencing depth (Table S2). The sequences were clustered into operational taxonomic units (OTUs) at a 97% sequence similarity cutoff using the Uparse package (version 7.0.1001) (76), and all OTUs were assigned to the Greengenes database by the QIIME software pipeline. For the SourceTracker analysis, the OTU table derived from quality filtering and OTU picking of the 16S rRNA gene sequences was used as the input file. This tool compares the community profiles in the "source" groups (i.e., gill, hepatopancreas, hemolymph, subcuticular epidermis, culturing seawater, and crab foods) with those in the "sink" group (i.e., midgut) by using Bayesian methods to calculate the contribution of each source to the sink. The SourceTracker analysis was performed using the default parameters: rarefaction depth, 1,000; burn-in, 100; restart, 10; and alpha (0.001) and beta (0.01) Dirichlet hyperparameters.
Statistical analysis. The statistical significance between two samples of crab physiology, hepatopancreas digestive enzymes, hemolymph conditions, expression of immune-related genes, and microbial abundances was analyzed using the Mann-Whitney U test (SPSS 22.0). Metastat analysis with Fisher's exact test was performed in R software (version 3.2.0) (77) to determine the differences in the relative abundances of the microbiota of crab at different molt stages. The PCoA plot was performed in R using Bray-Curtis distances with the relative abundance of microorganisms at the genus level. Alpha diversity indices (i.e., Shannon, Chao1, observed species, and PD whole tree) were calculated by alpha_diversity.py in QIIME. The Mantel test was performed in R with the relative abundance of microorganisms at the genus level and measured immune-related genes/digestive enzymes at different molt stages. The differences were considered statistically significant at a P value of ,0.05. The LDA effect size (LEfSe) analysis was performed with an LDA score threshold of .2.0.
Data availability. The 16S rRNA gene sequences of this study were deposited in the Sequence Read Archive of the National Center for Biotechnology Information (NCBI) under the accession number PRJNA640924.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.