Preparation of the experimental organisms
The algal prey, Chlamydomonas reinhardtii (CC1690), was grown in liquid BG11 medium (Rippka et al. 1979), and D. magna was cultured in ADaM medium (Klüttgen et al. 1994). Both were cultured in a sterile temperature-controlled chamber at 23 ± 1 °C on a 14:10 h light/dark cycle under 20 µmol m− 2 s− 1 illumination, with constant stirring and aeration. D. magna were kept at a density of one individual per 10 mL and fed with saturating amounts of C. reinhardtii (105 cells/mL) each day, and the medium was refreshed once a week. N- and P-limited C. reinhardtii cultures were prepared with liquid nitrogen and phosphate-free BG11 medium (Rippka et al. 1979), respectively.
Grazing Experiment
Three different C. reinhardtii cultures (cultures grown in nutrient-balanced, N-limited or P-limited media) were used to feed the D. magna for seven days (Fig. 1). The prey was centrifuged and re-suspended with an appropriate amount of D. magna culture medium before being fed to the D. magna. All three experimental groups were constructed in triplicate in 1 L PC bottles and incubated in the sterile temperature-controlled chamber mentioned above. The D. magna were kept at a density of one individual per 10 mL and they were fed with saturating amounts of nutrient-balanced, N-limited, or P-limited C. reinhardtii cells (105 cells/mL), each day. For measuring the clearance and ingestion rates, triplicate 150 mL PC bottles were prepared for the three experimental groups (nutrient-balanced, N-limited, and P-limited), and the medium and bottles were renewed every day to avoid the influence of any remaining algae in the bottles. In addition, in these experiments the newly born neonates were removed from the culture and counted. To avoid cell aggregation or settlement, the cultures were gently agitated manually 2 to 3 times a day. As a control for the grazing experimental groups and to calculate the ingestion rate, another three groups were prepared in triplicate using the same concentration and type of C. reinhardtii but no D. magna. The calculation of ingestion and clearance rate was based on the previous reported method (Zhang et al. 2017).
Flow Cytometry Analysis
To determine the bacterial cell abundance inside the liquid algal cultures, filtrate were collected from the three different experimental groups before and after the grazing experiment via filtration through a 1 µm-pore-size filter. The filtrate were then stained with SYBR Green I solution at a ratio of 10:1 (the SYBR Green I solution was 1:1000 diluted with Milli-Q water; Molecular Probes) and incubated at 37°C in the dark for 1 h (Marie et al. 1997). The bacterial cell abundance was then examined using the Becton-Dickson FACSCalibur flow cytometer described above (details are provided in the Supporting Information).
Construction Of The Axenic Culture
In one series of experiments, C. reinhardtii and D. magna were separately made into sterile cultures, using antibiotics, as described in previous studies (Kan and Pan 2010, Macke et al. 2017b). For the sterile C. reinhardtii culture, R medium containing a cocktail of antibiotics (ampicillin, carbendazim, and cefotaxime (Sigma, Germany)) was used to obtain pure C. reinhardtii colony. The ampicillin and cefotaxime were used at final concentrations of 500 µg/mL and 100 µg/mL, respectively. As these antibiotics can be heat-inactivated, they were added to the agar medium after it was autoclaved and immediately prior to pouring the plates. In contrast, carbendazim is heat stable but only barely soluble, and so this was added to the agar medium (to a final concentration of 40 µg/mL) before it was autoclaved and then the solution was mixed well before the plates were poured (Kan and Pan 2010). After 14 days cultivation of plates in the sterile temperature-controlled chamber, the pure algal colonies in the antibiotics-containing agar plates were then inoculated into autoclaved liquid BG11 medium and the existing of bacteria was examined with a Becton-Dickson FACSCalibur flow cytometer. The level of microbial contamination was also examined at the end of the feeding experiment.
Bacteria-free eggs were obtained by disinfecting eggs, from the normally fed D. magna, through exposing them to 0.25% ampicillin (Sigma, Germany) for 30 min. Part of the antibiotic-treated eggs was crushed with a pestle and filtered through a 0.22 µm membrane for PCR assessment of remaining bacteria (Li et al. 2019b). After rinsing with sterile ADaM to remove ampicillin, the eggs were transferred to a sterile six-well plate for hatching. The axenic grazing experiment was conducted, where the axenic C. reinhardtii was used as prey. At the end of the grazing experiment, all the D. magna (thirty individuals in total) in each experimental group (Normal, N-limited, P-limited, Germ-free Normal, Germ-free N-limited, and Germ-free P-limited) were used for the measurement of body length.
Nutrient Analyses
Before the beginning of the grazing experiment, samples of C. reinhardtii that had been grown in the different conditions were collected for the analysis of cellular carbon, nitrogen, and phosphorus. Samples were taken from the respective culture bottles by filtering 15 to 25 mL of each culture onto pre-combusted (i.e., at 550°C for 5 h) GF/C glass-fiber filters. After the seven-day grazing experiment or following 6-h starvation, one D. magna from each experimental group was transferred to a pre-combusted 25 µm GF/C filter for determination of elemental composition (C and N), and another D. magna of similar body length and weight as the first, was collected for the phosphorus measurement. Cellular carbon and nitrogen in both the D. magna and C. reinhardtii were measured with a CHNS (carbon, hydrogen, nitrogen, sulphur) elemental analyzer (FlashSmart CHNS, Thermo Scientific Inc. Massachusetts, USA) according to previously described protocol (Zhang et al. 2015). The amount of phosphorus (in the form of orthophosphate) was analyzed manually following acidic oxidative hydrolysis with 1% HCl (Grasshoff et al. 2009) using a spectrophotometer at a wavelength of 880 nm, with a detection limit of 0.5 µmol/L.
Gut extraction of D. magna
For the purpose of molecular investigation, triplicate 1 L PC bottles were prepared for the three experimental groups (nutrient-balanced, N-limited, and P-limited) with 80 individuals raised in each bottle. At the end of the seven-day grazing experiment, the gut of all D. magna was extracted with sterilized (i.e., autoclaved and 70% ethanol steeped) dissection tweezers (Regine 5, Switzerland) in a sterile Petri dish under a stereomicroscope. Before each gut extraction procedure, tweezers were flame-sterilized and rinsed with 70% alcohol. Each of the extracted guts from the various experimental groups, was placed into a 1.5 mL sterile Eppendorf tube and dissociated into a cell suspension according to previous report (Li et al. 2019a). The cell suspension was then filtered through a 0.22 µm polycarbonate membrane (EMD Millipore, Billerica, MA, USA) with addition of 500 µL RNA protect reagent (Qiagen, Germany). In order to assess the potential operation contamination, the tweezers and Petri dishes used to prepare the cell suspension were rinsed with water and this was then filtered through another a 0.22 µm membrane for the detection of contamination. Therefore, a total of eighteen filters were used to collect the cell suspension from the gut and the contamination separately. All the filters were preserved in sterile 1.5 mL Eppendorf tubes and stored at -80 °C until RNA extraction.
Detection of microbial polyphosphate
Ten adult D. magna from each experimental group (i.e., nutrient-balanced, N-limited, or P-limited), were placed in 100 mL sterile ADaM medium to empty their gut, and their fecal pellets were collected by filtering the medium through a 2.0 µm polycarbonate membrane (EMD Millipore, Billerica, MA, USA). The membrane was sonicated for 30 s to release any bacteria that were attached to the fecal pellets into the suspension. The fecal detritus was removed via centrifugation at 4,000 g for 5 min, and the supernatant was used for the detection of microbial polyphosphate (polyP). To measure microbial polyP in zooplankton and alage cultures, the culture was firstly filtered through a 3 µm membrane to remove the alage and large particles. Then the filtrate was used for the detection of microbial polyP according to a previous report (Kulakova et al. 2011). In brief, the suspended bacterial cells (in a 96-well plate) were stained with 25 mM Tris/HCl at pH 7.0 containing 500 µg/mL DAPI for 10 min, and the level of fluorescence was measured using a Flex Station 3 multimode microplate reader with excitation and emission filters of 420 nm and 550 nm, respectively (Molecular Devices, Sunnyvale, CA, USA). The microbial protein was then further quantified as described previously (Binks et al. 1996), and the fluorescence intensity of microbial polyP was expressed as relative fluorescence units (r.f.u.) per mg of total cellular protein.
Dna Extraction And Pcr Amplification Of 16s Rrna Gene
The validation of bacterial contaminant and bacteria-free eggs was achieved through DNA extraction and PCR amplification of the 16S rRNA gene. Total genomic DNA was extracted from the filters of dissection tools rinse water, and from randomly sampled D. magna germ-free eggs, using a PureLink Genomic DNA kit (Invitrogen, ThermoFisher Scientific Corp., Carlsbad, CA, USA). Due to that failures of gut extraction happens occasionally, different number of D. magna guts were collected from normal (10), N-limitation (7) and P-limitation (12) experimental groups. Each of these guts was placed into tubes individually for amplification of the 16 s rRNA gene. The extracted DNA was then eluted into 100 µl Tris-EDTA (TE) buffer for PCR amplification. These 29 gut microbial communities were amplified and sequenced as described previously (Liu et al. 2017).
Analysis Of 16s Rrna Gene
The sequenced contig reads between 135 and 152 bp were preserved, and primers as well as low-quality reads were removed with FASTX-Toolkit (Pearson et al. 1997). Reads with an average Phred score < 25 were discarded, as were reads with any consecutive runs of low-quality bases > 3. The lowest quality score allowed was 3, the minimum of continuous high-quality bases was 75% of whole read length, and the maximum number of ambiguous bases was 0 (Pan et al. 2019). Chimeras were identified and removed using UCHIME (Edgar et al. 2011). The remaining high-quality sequences were merged according to the experimental treatments, and the taxonomically assignment was processed with Silva database (version 123) by following the previous reported method (Li et al. 2018). Finally, sequences were clustered into OTUs with a 97% sequence similarity cutoff. To get overall gut community distribution pattern within each experimental treatment, the OTUs were normalized with the sample number prior to further analyses. The results were further used in a Lda (linear discriminant analysis) effective size (LEfSe) analysis, which is typically used to reveal the taxonomic differences between experimental groups.
Rna Isolation And Metatranscriptomic Sequencing
The filters collected during the various experiments were briefly thawed on ice and the RNA protection solution was removed as previously described (Xu et al. 2013). In brief, the filters were transferred to a new 0.7-ml tube with a pinhole in the bottom. This was placed on top of a 1.5-ml centrifuge tube, and the residual RNA protection reagent was removed from the filters when the two tubes were centrifuged at 1000 rpm for one min. RNA extraction was achieved with the Totally RNA isolation kit (Ambion Inc, Germany) according to the manufacturer’s protocol. The Turbo DNA-free DNase kit (Ambion Inc, Germany) was used to remove the remaining DNA, and the MicroPoly (A) Purist kit (Ambion Inc, Germany) was used to isolate mRNA; both of these kits were used according to the manufacturer’s instructions. A Nanodrop spectrophotometer (Nanodrop Technologies, Wilmington, USA) was used to examine the purity of the extracted RNA, and the RNA BR Assay kit (Life Technologies, Invitrogen, Germany) in conjunction with a Qubit® 2.0 flurometer was utilized to estimate the concentration. A Ribo-Zero Magnetic kit (Epicentre, Madison, WI) was used to remove any remaining ribosomal RNA, after which the TruSeq mRNA Library Preparation kit (San Diego, CA, USA) was used to construct the Illumina sequence library. The pooled mRNA from each triplicate were barcoded and sequenced with an Illumina HiSeq2500 sequencer (Novogene Co., Ltd., China), generating between 131.3 and 207.1 million 150 bp paired-end reads per replicate.
Disentangling Partner Reads From The Holobiont System
According to the barcode, the sequencing data were assigned to nine experimental groups (Normal, N-limitation and P-limitation). The quality control of sequenced reads was performed as described in previous reports (Gong et al. 2018, Li et al. 2018). In addition, the reads that belong to different parts of the holobiont (i.e., D. magna and its gut microbiota), were separated by applying a previously reported method (Meng et al. 2018). In brief, the genome and previously published RNA-seq datasets of D. magna (Orsini et al. 2016) were downloaded to a local server to construct a host reference library, and the bacterial fractions of the Tara Oceans meta-genomic gene catalogue (OM-RGC) and non-redundant (nr) database were extracted with the blastdbcmd program (Camacho et al. 2009) to build a microbiota reference library. The SRC_c software (Marchet et al. 2018) was then used to map the metatranscriptomic data either to the host or to the gut microbiota with indexed k-mers set to 32 and suggested default similarity s value (50%).
Reads Assembly And Downstream Analysis
After separation of the metatranscriptomic data, the reads were assembled into longer transcripts, separately, using Trans-ABySS v2.0.1 (Robertson et al. 2010) with multiple k-mer sizes from 32 to 92, and a step of 4. Transdecoder (v5.3.0) (Haas et al. 2013) was used to predict the open reading frames (ORFs) of the assembly result. The annotation of ORFs was achieved using DIAMOND (v0.9.21.122) (Buchfink et al. 2015) against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database and the nr database, with the following parameters: blastp; k parameter = 1 ; and an e-value = 10− 7. For calculation of the coverage information of ORFs, reads were mapped back to the ORFs using Bowtie 2.2.9 (Langmead and Salzberg 2012) and SAMtools v1.9 (Li et al. 2009). The differentially expressed genes (DEGs) between experimental groups were calculated according to a previous report (Li et al. 2019a), using the edgeR package in R (Robinson et al. 2010). The DEGs were defined with the criteria of |log2 (fold change)| > 1 and p-value < 0.05 shown in the comparisons between experimental groups. In addition, the genes encoding microbial butyrate synthesis were also identified using the specific database (Vital et al. 2014).
Gene Expression Validation
For each sample, HiScript® III RT SuperMix for qPCR (+ gDNA wiper) (Vazyme Biotech, Nanjing, China), was used for the reverse transcription of extracted RNA (500 ng). No reverse transcription (RT) control was used in the qPCR experiment for the detection of the possible remaining DNA in the extracted RNA. After the synthesis of cDNA, 1 µL (nearly 50 ng) from each cDNA sample was used for qPCR with a Fast start Universal SYBR Green Master mix kit (Roche, Germany) in a LightCycler 384 device (Roche, Germany). The thermocycling conditions were as follows: an initial hold at 50℃ for 2 min and at 95℃ for 10 min followed by 45 cycles of 95℃ for 15 s and 60℃ for 1 min. All reactions were performed in triplicate. The relative amount of mRNA was determined using the 2−ΔΔCt method, and the 16S rRNA gene was selected as a reference for normalization of the gut microbe genes. The primers used to target specific genes in the gut microbiota and D. magna were as previously described (Li et al. 2019a) and they are listed in Table S1.
Statistical Analyses
For the ingestion rate, reproduction, and final body length, data were presented as the mean ± SD derived from the biological replicates. Student’s t-tests (two-tailed) were conducted with significance levels of p < 0.05. Sloan’s neutral community model (NCM) was constructed to evaluate the contribution of neutral processes in D. magna’s gut community structure (Sloan et al. 2006). R software was used to perform the analysis. In this analysis, Nm is an estimate of dispersal between communities while the R2 determines the overall fit to the neutral community model (Chen et al. 2017). Canonical correspondence analysis (CCA) was performed using the PAST 3.0 software.