Nanopore‐Enabled Microbiome Analysis: Investigating Environmental and Host‐Associated Samples in Rainbow Trout Aquaculture

Microbiome sequencing is at the forefront of health management development, and as such, it is becoming of great interest to monitor the microbiome in the aquaculture industry as well. Oxford Nanopore Technologies (ONT) platforms are gaining popularity to study microbial communities, enabling faster sequencing, extended read length, and therefore, improved taxonomic resolution. Despite this, there is a lack of clear guidelines to perform a metabarcoding study, especially when dealing with samples from non‐mammalian species, such as aquaculture‐related samples. In this article, we provide general guidelines for sampling, nucleic acid extraction, and ONT‐based library preparation for both environmental (water, sediment) and host‐associated (gill or skin mucus, skin, gut content, or gut mucosa) microbiome analysis. Our procedures focus specifically on rainbow trout (Oncorhynchus mykiss) reared in experimental facilities. However, these protocols can also be transferred to alternative types of samples, such as environmental DNA (eDNA) monitoring from alternative water sources, or to different fish species. The additional challenge posed by the low biomass and limited bacterial diversity inherent in fish‐associated microbiomes is addressed through the implementation of troubleshooting solutions. Furthermore, we describe a bioinformatic pipeline starting from raw reads and leading to taxonomic abundance tables using currently available tools and software. Finally, we provide a set of specific guidelines and considerations related to the strategic planning of a microbiome study within the context of aquaculture. © 2024 The Authors. Current Protocols published by Wiley Periodicals LLC.


INTRODUCTION
Increasing evidence is pointing to the host and environmental microbiomes as key components involved in the general health status of an organism.This is also the case in the aquaculture field, where, supporting this hypothesis, several recent studies found that microbial dysbiosis, i.e., an imbalance in the microbiota composition, was correlated to distress or disease conditions in the host (Infante-Villamil et al., 2021).Several methods are available to investigate microbiomes.The most common culture-independent high throughput sequencing (HTS)-based microbiome studies start with sample collection, followed by nucleic acid extraction, and subsequently either PCR amplification and sequencing of a marker gene (metabarcoding) or untargeted (whole metagenome) sequencing.With metabarcoding, the chosen marker gene-e.g., 16S rRNA gene (Caporaso et al., 2012)-is used to estimate the relative abundance of the bacterial taxa present in the sample.To date, such studies were predominantly based on short-read sequencing (Illumina), but more recently there has been a shift towards long reads, using platforms such as Oxford Nanopore Technologies (ONT) or PacBio.ONT in particular offers the added advantages of a shorter sequencing time, portability, the possibility of real-time analysis, and feasibility for small-scale projects.Importantly, in the context of metabarcoding studies, longer reads translate to enhanced taxonomic resolution (Johnson et al., 2019).
Microbiome studies based on HTS have revolutionized the way microbial communities are studied thanks to the amount of information that can be gathered at once from a single sample.However, the field faces a substantial challenge in terms of reproducibility.Even with extensively studied sample types, finding a consensus on robust protocols remains elusive, hindering the ability to conduct meaningful meta-analyses (Thompson et al., 2017;Tourlousse et al., 2021).In the field of aquaculture, microbiome studies are still relatively new, and the field is lacking comprehensive protocols and guidelines for appropriate and effective sampling techniques, nucleic acid extraction, and preparation for sequencing.Additionally, from the few shotgun metagenomics studies in salmonids (Collins et al., 2021;Rasmussen et al., 2021;Riiser et al., 2019), it is now known that both the biomass and biodiversity of microbial communities in these host species are extremely low in comparison to those in mammals (Limborg et al., 2023).This observation introduces further challenges during library preparation, requiring specific considerations that will be addressed in this article.
Here we describe the methods we optimized in our laboratory to study microbiomes through 16S rRNA gene metabarcoding.We will mainly address methodologies appropriate to being coupled with Oxford Nanopore Technology (ONT) sequencing, but also easily adaptable to short-read Illumina sequencing.We particularly describe procedures related to environmental (water, sediment) and host-associated microbiomes, which were optimized by focusing on one of the most farmed fish species worldwide, the rainbow trout (Oncorhynchus mykiss).Specifically, we first provide step-by-step guidance for the collection of microbiomes residing within the biological barriers interfacing the host and its environment, including the skin, skin mucus, gill mucus, gut content and mucosa (Basic Protocols 1 and 2).After collection, a pre-processing and extraction method that proved to be effective for all kinds of samples mentioned is presented (Basic Protocol 3).Finally, quality controls and appropriate tests to perform prior library preparation with ONT kits are described (Basic Protocol 4).Resulting from these protocols, full-length (hypervariable regions V1 to V9) 16S rRNA gene amplicons are generated in a sufficient amount to undergo sequencing.While we focus on metabarcoding, the following protocols could potentially be adapted for metagenomics or metatranscriptomics studies with suitable modifications, such as omitting the PCR amplification and treating the sample with specific nucleases.The protocols are supplemented by best practices and guidance concerning the design of a microbiome study in the field of aquaculture, applicable to experimental facilities and field investigations.Among others, we provide practical considerations for laboratory procedures, instructions on the choice of relevant controls, and finally a bioinformatics workflow to obtain taxonomic abundance tables.NOTE: All protocols involving animals must be reviewed and approved by the appropriate Animal Care and Use Committee and must follow regulations for the care and use of laboratory animals.

ENVIRONMENTAL SAMPLE COLLECTION
Here we describe the optimized methods to collect environmental microbial samples in our experimental facilities.We will focus on two environmental sample types, namely water and sediment.
Generally, the collection of water is a straightforward procedure in experimental tank facilities, as it can usually be sampled from the surface or, if present, from the outlet of recirculating pumps.Since conditions and equipment availability can vary considerably when performing on-site sampling (e.g., in a fish farm), this step can be subject to modifications and can be easily modified to suit field sampling (see "Adaptations for field sampling" in the Critical Parameters section).
On the other hand, with most experimental tank designs, it is impractical or even not at all feasible to collect the sedimented organic matter from the tank's bottom compartment repeatedly and reproducibly over the course of an experimental trial.Thus, when interested in collecting and analyzing sediment from experimental fish tanks, it is necessary to place the suitable sampling equipment in the tank prior the start of the experiment.In our laboratory setup, we use a system based on the communicating vessels principle: a flexible rubber conduit is positioned to connect the tank's base and the exterior.The terminal end of the tube external to the tank is securely plugged, while the lower aperture remains unobstructed.Upon de-capping the upper end, the water pressure facilitates the ingress of sediment through the rubber conduit, resulting in a continuous outward efflux (Fig. 1).This method ensures that the samples are consistently collected from the same site, and additionally, neither the water nor the fish are disturbed during collection.

Water collection and filtration
1. Before starting, wear PPE and wipe the exterior part of the glass bottle with 70% (v/v) ethanol.
2. Collect 1 L water in a sterile glass bottle.Take notes of physical parameters of the water, at least temperature and depth at which the water was collected (e.g., surface level; 50 cm depth; tank bottom; etc.).
When collecting water from an aquaculture farm, alternative filtration strategies and considerations on other metadata to include can be found in the Critical Parameters subsection "Adaptation to field sampling".
3. Set up the vacuum filtration apparatus in a laminar flow hood to avoid capturing dust and air particles during filtration.
4. Place a 0.22-μm PVDF sterile filter on the membrane support plate using sterile surgical tweezers, then place and tighten the upper chamber.
5. Pour the water in the upper chamber and start the vacuum pump.Continue adding and filtering water until the desired sample volume is reached (e.g., 500 ml) or until the filter membrane clogs.6. Recover the filter using sterile tweezers and place it in a sterile Petri dish.
7. Using a sterile scalpel, section the filter as shown in Figure 1A.Place the sectioned filter in a 2-ml sterile tube.
Change tweezers, scalpel, and Petri dish for every sample.The filter membrane needs to be sectioned in order to facilitate nucleic acids extraction in the subsequent protocols.This step ensures that all the surface of the filter is equally accessible to the lysis buffer, allowing uniform distribution across the membrane.
At this point, the filter can be cut in halves to store separately as a backup sample.The following extraction method has also proven successful when using half a filter.
Important: in parallel, section and store a clean (not used) filter membrane, to use as a negative control in the following steps.
When immediate storage at −80°C is not possible, alternative storage conditions can be temporarily used (see the Critical Parameters subsection "Adaptation to field sampling")

Sediment collection
9. Uncap the rubber tube that was set up prior the start of the experimental trial for the purpose of sediment collection (see Basic Protocol 1, Introduction), and let the sediment flow out for a few seconds.
This step will ensure that the water and sediment previously present in the straw are discarded, and only fresh sediment is collected.
To ensure that it does not clog, diameter and placement of the rubber tube need to be optimized according to the texture of the sediment.
11.By this method, a mix of water and sediment is collected.Place the 50-ml tube in a rack in a vertical position to let the sediment accumulate on the bottom of the tube.
12. Discard the water on the top of the tube using a serological pipette and pipettor.
Thoroughly mixing the sample will ensure that the collection of a portion of sediment will be representative of the overall composition of the site of sampling.
14.With either a laboratory spoon or a pipette with a cropped 1-ml tip, transfer a fixed amount of sediment (at least 300 mg) into a sterile 2-ml tube.
Prepare replicates for each sample and collect some excess considering the possible need for backup material.The minimum collected amount should correspond to the sample starting quantity recommended by the nucleic acid extraction kit of choice.In this case, the recommended amount of sediment to collect for each replicate sample is 300 mg, which is the maximum required sample quantity for the MagMAX Microbiome Ultra nucleic acid isolation kit.
15. Snap freeze on dry ice and store immediately at −80°C until extraction.

of 29
Current Protocols When immediate storage at −80°C is not possible, alternative storage conditions can be temporarily used (see the Critical Parameters subsection "Adaptation to field sampling").

HOST-ASSOCIATED SAMPLE COLLECTION
For the collection of host-associated samples, we will mainly focus on the microbiomes that, together with the fish immune system, constitute the interface between the animal and the environment.In an experimental study that involves living animals, it is important to consider non-lethal sampling options to reduce the number of individuals that will be included (see "Non-lethal and non-invasive sampling options" in the Critical Parameters section).For this reason, here we will first describe two non-lethal sampling methods to collect gill mucus and skin mucus.Nevertheless, these non-lethal sample types will not always be suitable for the kind of study that is being conducted.In the Alternate Protocol, we describe other host-associated sample collection methods, which detail how to collect skin tissue, gut content, and gut mucosal samples.All these sample types were tested and optimized on rainbow trout but can also be transferred to other fish species.
Anesthetize or euthanize or the fish.For anesthesia, benzocaine concentration should be 75 to 100 mg/L in oxygenated water, and bath immersion should last for 2 to 5 min, until achievement of stage 4 to 5 anesthesia.In this case, after the procedure, the fish should be placed and monitored in a separate water basin until complete recovery, before being returned to the experimental tank.For euthanasia, bath immersion must be performed in water containing an overdose of benzocaine (>250 mg/L) for a minimum of 10 min.
Euthanasia is not necessary, as this is in principle a non-lethal sampling method.Consistency in the chosen method for anesthesia or euthanasia is necessary among all individuals in the same study.
2. Lay the fish down on a flame-sterilized surface.

With sterile, toothed dissecting forceps open the operculum to reveal the gill arches.
With tweezers, place a piece of medical wipe (Ivanova et al., 2021) on the gill arches and filaments to absorb the gills mucus (Fig. 1C).Repeat on both left and right gills of the fish using the same section of medical wipe.
Alternatively, a sterile swab can be used to collect the mucus by gently wiping it between the gill arches by a rotating motion (Fig. 1C).
Caution is required not to damage the gills during the procedure.
4. Place the medical wipe (or swab) in a sterile 2-ml tube.
Important: a clean section of medical wipe should also be stored similarly to be used as a control for the sampling procedure.
5. Place the tube on dry ice to snap freeze it.Store immediately at −80°C until extraction.
When immediate storage at −80°C is not possible, alternative storage conditions can be temporarily used (see the Critical Parameters subsection "Adaptation to field sampling") Skin mucus collection 6.With sterile tweezers, quickly lay a section of sterile medical wipe on the surface of the skin (Ivanova et al., 2021), being careful not to damage the skin.
Note that it is crucial to sample the skin mucus immediately, before dehydration of the skin mucus occurs.
The medical wipe should absorb the mucus instantly; otherwise, it might mean that the mucus has dehydrated, and little to no material will be collected by this method.The skin area to be sampled should be chosen according to the purpose of the study (e.g., if skin lesions or specific skin regions are the subject of the study, those areas should be sampled).
If the aim of the study implies comparison between groups subject to different experimental conditions, samples should be collected consistently from the same area.To ensure consistent sampling of the same area, it is crucial to select coordinates that are easily identifiable around specific anatomical landmarks (e.g., right-side lateral line).
7. After absorption, pick up the section of medical wipe with sterile tweezers and place it in a new 2-ml tube (Fig. 1D).
Alternatively, if a larger amount of material is required and the focus is not a particular skin section (e.g., skin lesions) but rather a more general assessment of skin mucus microbial communities, a cell scraper or the blunt end of a sterile scalpel can be run along the lateral line and the accumulated mucus can be sampled with a pipette and placed in a sterile 2 ml tube.
Important: a clean section of medical wipe should also be stored similarly to be used as a control for the sampling procedure.
8. Place the tube on dry ice to snap freeze it.Store immediately at −80°C until extraction.
When immediate storage at −80°C is not possible, alternative storage conditions can be temporarily used (see the Critical Parameters subsection "Adaptation to field sampling")

HOST-ASSOCIATED SAMPLE COLLECTION: ALTERNATIVE SAMPLE TYPES
The following procedure describes the collection of skin tissue and subsequently gut content and gut mucosal samples.In this protocol, we use the generic term "skin" to refer to the cutis, i.e., the combination of the epidermal and dermal layers.2. Euthanize the fish.For euthanasia, bath immersion must be performed in water containing an overdose of benzocaine (> 250 mg/L) for a minimum of 10 min.

Additional Materials
Consistency in the chosen method for euthanasia is necessary among all individuals in the same study.

of 29
Current Protocols 3. Lay the fish on a flame-sterilized surface.
4. Select and identify the fish side (right or left), skin site (e.g., mid-lateral line), and surface measurement (cm 2 ) of the skin portion that will be sampled.
5. The sampled skin area should be selected according to the purpose of the study.
As for the skin mucus sampling, the sampled region must be chosen on a precise anatomical area that is easily identifiable.This is crucial to ensure consistency in the sampling site across specimens in the same experiment.With a sterile, pointed scalpel, perform a ∼2 mm deep incision in the shape of a square (Fig. 1D) to excise the dermal and epidermal layer (Schmidt et al., 2021).
Usually, ∼1 cm 2 of cutis will collect a sufficient number of bacteria for 16S rRNA gene amplification.
6.With toothed dissecting forceps, firmly grab one corner of the incised skin square and pull tightly to lift the skin.The epidermal layer should easily detach from the muscular layer beneath.
If residues of white muscular tissues are still present on the dermis, gently scrape them to remove them from the upper skin layers.
7. Place the skin section in a sterile 2-ml tube.
8. Place the tube on dry ice to snap freeze it.Store immediately at −80°C until extraction.
When immediate storage at −80°C is not possible, alternative storage conditions can be temporarily used (see the Critical Parameters subsection "Adaptation to field sampling").

Gut content and mucosa collection
9. With sterile toothed dissecting forceps, firmly grab on the operculum and perform a long incision along the lateral line using surgical scissors.The incision should then curve down towards the vent.With tissue forceps, move the tissue to reveal the abdominal cavity.With new tweezers and scalpel or surgical scissors, perform an incision caudally to the pyloric caeca and at the vent to dissect the mid or distal intestine (MI and DI).
Although less common, if interested, also the anterior intestine (AI) and pyloric caeca (PC) can be sampled similarly, as it was shown that they also harbor specific microbiomes in salmonids (Nguyen et al., 2020).
10. Place the intestine on a sterile surface and use sterile scissors to isolate the portion of interest of the intestinal tube.The collected section (e.g., PC, AI, MI, DI) must be consistent for all samples.
The chosen size of the intestinal tract depends on the developmental stage of the fish; for early developmental stages (e.g., rainbow trout fry), it can be necessary to sample the whole intestinal tract to obtain enough initial biomass.
11.With a sterile scalpel, perform a longitudinal incision on the intestinal tube section (Fig. 1E).
12. Using tweezers, collect the luminal content (digesta) and place the sample in a sterile 2-ml tube.
Take care not to damage the intestinal mucosa while collecting the digesta.
13.If necessary, quickly rinse the open intestinal tube section in a sterile 60-mm Petri dish containing sterile PBS, then place it on a sterile surface with the luminal side facing up.
14.With a cell scraper or the blunt side of a new scalpel, gently scrape the mucosa and collect it in a new 2 ml tube.
15. Place both samples on dry ice to snap freeze them.Store immediately at −80°C until extraction.
When immediate storage at −80°C is not possible, alternative storage conditions can be temporarily used (see the Critical Parameters subsection "Adaptation to field sampling") BASIC PROTOCOL 3

SAMPLE PRE-TREATMENT AND NUCLEIC ACID EXTRACTION
One of the main challenges in microbiome research is the lack of a global standard in processing and analyzing samples, which in turn often impairs comparison and metaanalyses between studies (Thompson et al., 2017;Tourlousse et al., 2021).Furthermore, it is advised to use the same extraction kit among sample types included in the same study (Salter et al., 2014).
For this reason, here we propose a pre-treatment and extraction protocol combination that works for a broad range of aquaculture-derived samples.Most protocols for total microbial nucleic acid extraction include either mechanical cell disruption or enzymatic lysis to ensure that no bias is present against difficult-to-lyse bacteria.In our case, we use a combination of mechanical and enzymatic lysis that is suitable for all our sample types.The lysozyme incubation step is particularly useful when dealing with difficultto-lyse gram-positive bacteria.However, it is optional and can be omitted if preservation of RNA integrity in the sample is a priority.The extraction kit of choice is the MagMAX Microbiome Ultra nucleic acid isolation kit, which effectively purified all the sample types mentioned above with slight modifications.Other kits can also be used, taking into account the considerations described in the Critical Parameters subsection "Choice of extraction kit".

Mechanical lysis
1. Before starting, wear PPE and clean lab surfaces and equipment.Perform all steps in a laminar flow cabinet.
2. Weigh each sample (sediment, skin tissue, gut content or mucosa) up to 300 mg and move it to a sterile 2-ml tube.
Important: for water filters and mucus in absorptive medical wipes or swabs, this step is not necessary as they will be entirely processed for extraction.
3. Add a 5-mm sterile stainless-steel bead and 1 g of 100-μm zirconia beads directly in the 2-ml tube containing the sample.
Important: for water filters and mucus in absorptive medical wipes or swabs, the stainless-steel bead is not necessary.For these kinds of samples, half the quantity (500 mg) of zirconia beads is used.
4. Add 800 μl lysis buffer (from the MagMAX Microbiome Ultra nucleic acid isolation kit) to the sample tube.
If the kit of choice does not allow homogenization in the lysis buffer, perform this step in a small volume of PBS instead.
5. Vortex to mix the sample and beads.
6. Lyse in a mechanical disruptor at 25 Hz for 2 min or until completely homogenized.
For water filters, mucus in absorptive medical wipes or swabs, 1 min of mechanical disruption is deemed sufficient.
For skin samples, 5 min of homogenization could be necessary.Note that excessive homogenization times through bead-beating may lead to DNA shearing and fragmentation.
Lysozyme treatment and extraction 7. Centrifuge 5 min at 14,000 × g, room temperature, in a microcentrifuge.
This centrifugation step is intended to spin down the foam generated during the homogenization step, and it can be repeated in the presence of an excessive amount of foam.
The final concentration of lysozyme in the sample will correspond to 0.9 mg/ml.

Vortex the sample and spin down.
This step does not require high speed centrifugation.A quick spin for a few seconds will be sufficient to bring the content to the bottom of the tube.
10. Incubate at 37°C for 1 hr on a shaking thermal block.
12. Save 500 μl of supernatant for extraction, and load in a 96 deep-well plate.
13. Extract nucleic acids from the homogenate supernatant using the chosen nucleic acid extraction kit (e.g., MagMAX Microbiome Ultra nucleic acid isolation kit)

QUALITY CONTROL AND PREPARATION FOR 16S rRNA GENE SEQUENCING
Standard DNA quality control (QC) steps are followed before library preparation.The subsequent 16S rRNA gene V1 to V9 regions PCR amplification tests are aligned with and serve as a precursor to the library preparation protocol designed for the ONT 16S barcoding kit (SQK-16S024).The test endpoint PCR amplification should be performed for all the sample types in analysis (and controls).This step is particularly crucial when a relatively low bacterial biomass is expected, e.g., tissue-rich samples where it is inevitable to obtain a high amount of host DNA.Specifically for these samples, conducting a preliminary amplification test will enable the optimization of PCR conditions prior to library preparation.This is useful to determine the required ng of initial template, number of optimal PCR cycles, as well as the possible need of replicate PCR reactions to pool.Test the same DNA template amount, cycling conditions, reaction volume, and primers that will be later used for library preparation.
2. To assess purity, measure the absorbance of the purified DNA by loading 1 μl of the eluted sample on a microvolume spectrophotometer.Ratios of absorbance are used to estimate purity of nucleic acids over protein or contaminants.
3. Perform agarose gel electrophoresis to evaluate the integrity of the nucleic acidssee Voytas (2000) for detailed instructions.
This step is particularly important when longer amplicons (e.g., 16S rRNA gene V1 to V9) will be sequenced in the next steps, since excessive fragmentation could impair the PCR amplification step.
Gels with agarose 0.8% up to 2% can be used in this step, but lower percentages will be better for high-molecular weight genomic DNA separation.High molecular weight DNA will appear as a bright band at the top of the lane (Fig. 2A); however, if fragmentation occurred to some degree, a long smear will be visible along the lane (Fig. 2B).
4. Before moving on to library preparation, accurately measure the concentration of DNA in the samples with a fluorometric assay of choice, e.g., the Qubit dsDNA quantification assay.
The DNA samples can be diluted to a standardized concentration (see step 9) to simplify the preparation for the PCR step.
Test PCR amplification with selected 16S rRNA gene primers 5. Thaw the DNA samples to test.Among samples to test, include positive control (Microbial Community Standard) and negative controls from the extraction step.
Additionally, prepare at least one blank PCR reaction, where nuclease-free water is added instead of DNA template.
6. Dilute both primer oligonucleotides to a final concentration of 10 μM in nucleasefree water.Use the same primers sequences as for library preparation (for the purpose of testing, they can be deprived of 5' tags/barcodes/spacers).The most used couple of primers for full-length 16S rRNA gene sequencing are the following V1 to V9 primers, which are also the ones provided in the ONT 16S barcoding kit (SQK-16S024): 27F (5´-AGAGTTTGATCMTGGCTCAG-3´) 1492R (5´-CGGTTACCTTGTTACGACTT-3´) Important: the primers provided in the ONT SQK-16S024 kit contain the same sequence as described above; however, they are also flanked by additional sequences (barcodes) that differ for each sample to be multiplexed in the same library.For PCR optimization purposes, the above oligonucleotides without barcodes can be used.
7. In a sterile 2-mL tube, prepare the master mix for the PCR reactions as described in Table 1.Do not add the DNA template yet.
The master mix should be prepared by calculating the required volumes for all the samples included in the test.Prepare an additional volume of ∼10% to account for pipetting errors (e.g., if 20 samples need to be tested, prepare the master mix for 22 samples).
8. Aliquot the master mix in 0.2-ml PCR tubes.9. Add the DNA template for each sample (or an equal volume of nuclease-free water, in the control blank reaction) to the 0.2-ml tubes where the master mix has been aliquoted.
Although the recommended conditions from the ONT SQK-16S024 kit require 10 ng of starting DNA template, this can be excessively low when dealing with contamination from non-microbial DNA.As previously recommended for gilthead sea bream mucosal

of 29
Current Protocols samples (Toxqui-Rodríguez et al., 2023), the conditions can be adapted to host-DNA rich samples by using a higher (exceeding 500 ng) amount of template DNA.In Table 2, we collected examples of starting material yielding a satisfactory amount of amplicon for each sample type, which can be used as a starting point for optimization.
10. Set up the PCR reaction in a thermal cycler using the thermocycling conditions detailed in Table 3. 11.Use an aliquot of each sample to perform an agarose gel electrophoresis to visualize the PCR products.This step will assess possible failed reactions and specificity of amplification.

of 29
Current Protocols The expected band size of V1 to V9 16S rRNA gene amplicons is 1.5 kbp.The ideal agarose gel percentage for clear band separation is 1%.
Importantly, low-biomass samples may produce amplicon bands that are too faint or even not visible (Fig. 3A).In that case, it can be necessary to assess the presence of amplification by an alternative protocol (e.g., Bioanalyzer).Such alternative protocols (Bioanalyzer, TapeStation) are also suggested for high-throughput applications.
12. If DNA bands are bright and visible at the correct size (1.5 kbp), with no unexpected bands, the sample can be subject to library preparation with the ONT 16S metabarcoding kit (SQK-16S024 kit or chosen kit) without further optimization (Fig. 3B).
In the case of failed test amplifications, it is necessary to optimize the PCR conditions (see the Troubleshooting section).Failed PCR could also be due to inhibition.This can be assessed by an additional quantitative PCR assay with 16S universal primers (Support Protocol 1).

ASSESSMENT OF INHIBITION BY QUANTITATIVE PCR
If the last step of Basic Protocol 4 (test PCR amplification) failed, testing different DNA dilutions through a qPCR assay can help assess if failure is due to PCR inhibition.This protocol can help determine optimal DNA template dilution and number of PCR cycles needed for library preparation.Importantly, it must be taken into consideration that DNA polymerase and primers, as well as target region and amplicon size, are not the same that will be used during library preparation for ONT long-read sequencing.For this reason, this test will only give an indication on the optimal range of cycles or amount of PCR inhibition that may be occurring during amplification.
This Support Protocol can also be used to estimate the bacterial DNA proportion relative to the host DNA, allowing for the efficacy assessment of alternative sampling methods or host DNA exclusion protocols.To perform this additional estimation, an assay targeting the host DNA (e.g., rainbow trout) should be included.

Additional Materials (also see Basic Protocol 4)
Bacterial universal primer oligos, 10 μM each, e.g., primers in 2. Dilute the DNA samples in nuclease-free water to decreasing 10-fold concentrations (e.g., 1:1, 1:10, 1:100, 1:1000).a These primers were used in Support Protocol 1 to assess the possible presence of inhibition and to estimate the number of cycles needed for sufficient amplification.

of 29
Current Protocols 3. Choose universal 16S rRNA gene primers that can work for a qPCR assay.To perform this test, we recommend the 16S rRNA gene universal primers from the Earth Microbiome Project (EMP) (Thompson et al., 2017), in Table 4.
Other universal 16S rRNA gene primers can be used, following the general recommendations on the design of a typical SYBR green-based qPCR assay.Importantly, the amplicon size should not exceed 300 bp in length.Specific conditions to perform the qPCR will depend on the chosen primer set and polymerase.
4. Prepare the master mix by mixing the chosen primers and the SYBR assay master mix following the manufacturer's instructions.In a 0.2-ml optical plate, aliquot the prepared master mix and add the diluted DNA samples.Seal the plate with optical caps and perform qPCR using the real-time PCR thermal cycler.
5. After running the qPCR, import the data to the qPCR analysis software of choice (e.g., CFX Maestro, BioRad), and observe the amplification curves plotted in the logarithmic scale for each sample.
6. Assess the possible presence of PCR inhibition by calculating the amplification efficiency and by observing the Cq values between 10-fold dilutions.Choose the optimal DNA sample dilution accordingly.
Usually, PCR inhibitors will be diluted together with DNA, and less concentrated samples will display a lower inhibition level.
See the Understanding Results section to interpret the outcomes of the assay.
7. Determine the optimal number of cycles, which lies in the range where the amplification curve is in the exponential phase and the PCR amplification efficiency is constant (Ruiz-Villalba et al., 2021).
To obtain a maximum amount of amplicon to sequence, choose the number of cycles just prior to the curve reaching the plateau phase.

BIOINFORMATIC ANALYSIS FROM RAW FILES TO TAXONOMIC ABUNDANCE TABLES
Most bioinformatic tools are intended for processing of short read sequences, as Illumina sequencing platform has dominated the field until recent years.As such, tools to process Nanopore-derived long reads are being continuously developed and optimized, both by ONT and by third-party developers.For scientists with no command-line experience, a graphical user interface application from ONT exists, the EPI2ME data-analysis platform (https:// epi2me.nanoporetech.com), which takes FASTQ files and can run a multitude of pre-assembled data analysis pipelines, called workflows.Among them, the "Fastq 16S" workflow was designed specifically to accept FASTQ files as input, and to produce taxonomical tables featuring relative abundances.Alternatively, when more flexibility is needed to finely adjust the analysis parameters, custom pipelines can be constructed.Here we describe the pipeline in use in our laboratory, starting from raw reads to obtain taxonomic relative abundance tables.The pipeline we propose is based on Linux command-line scripts and it only requires a minimal experience with bash scripting.The pre-processing pipeline, from FAST5 files to polished and demultiplexed FASTQ files (steps 2 to 9), is also available on GitHub (https:// github.com/zrngiulia/ 16S_ONT_preprocessreads).

Necessary Resources Hardware
Computer equipped with a Linux-based operating system NVIDIA Graphic Processing Unit (GPU) 8. Perform a second quality control on the pre-processed, concatenated reads using fastqc.
9. If interested in an overview of how the pre-processing steps affected the quality of the original reads, place all fastqc reports in the same folder and run multiqc (Ewels et al., 2016) afterwards.
10. Perform taxonomical classification of the reads to the species level using Emu (Curry et al., 2022).As a database, either the default Emu database [based on rrnDB (Stoddard et al., 2015) and NCBI taxonomy (O'Leary et al., 2016)] or alternatives [e.g., SILVA (Quast et al., 2013)] can be used, which are already pre-built for use with Emu.Perform this step for each barcode.
Emu installation can be performed in a conda environment as detailed on GitLab (https: // gitlab.com/treangenlab/ emu).Similarly, easy-to-follow tutorials to export a database or to build a custom one are provided by the authors.
emu collapse-taxonomy path/to/abundance/tables <rank> 12.The taxonomy tables can then be imported to R statistical software (R Core Team, 2018) for subsequent data analysis and visualization.

REAGENTS AND SOLUTIONS
Preparation should be conducted in a laminar flow hood to avoid microbial contamination of the stock solution.

COMMENTARY Background Information
Since the advent of high-throughput sequencing (HTS), the scientific community's interest in investigating complex microbial networks has been rising, starting from human-colonizing microbiomes, and in more recent years extending to a multitude of fields, such as animal, plant, and environmental mi-crobiome studies.In particular, the opportunity to monitor a large number of microbes at the same time is naturally appealing in agriculture and the food production chain, and as expected, the fish farming industry has recently joined in the collective interest towards microbiome investigation as well (Llewellyn et al., 2014).In fact, shifts in the host and/or Current Protocols surrounding environment microbial communities' compositions have been related to exposure of fish to antibiotics (Almeida et al., 2021) or experimental phage therapy (Donati et al., 2022), pollutants (Bellec et al., 2022;Hano et al., 2021), diet variations (Serra et al., 2021) and prebiotics administration (Menanteau-Ledouble et al., 2022), hypoxia conditions (Gatesoupe et al., 2016), earlylife stress (Uren Webster et al., 2021), and to deteriorating health status (Bozzi et al., 2021;Brown et al., 2021;Tongsri et al., 2023;Vasemägi et al., 2017), with compelling possibilities for disease prevention, management, and informed decision making in the fish farm industries.The vast majority of the current literature on the subject was generated based on short reads (Illumina sequencing platform) up to 600 bp.Thus, with this platform, it was only possible to sequence fragments of the 16S rRNA gene encompassing a maximum of three variable regions (e.g., V1 to V3).These were subsequently being used to infer the identity of bacteria and their relative abundance in the community, although with different taxonomic assignment performance outcomes depending on the chosen regions (Ghyselinck et al., 2013).Third generation sequencing platforms, e.g., Oxford Nanopore Technology (ONT), allow long-read sequencing, virtually up to the length of the generated nucleic acid fragments.Consequently, full 16S rRNA gene sequencing (hypervariable regions V1 to V9) becomes possible, which in turn can provide a higher taxonomical resolution than sequencing of shorter variable regions (Johnson et al., 2019).When studying teleosts' microbiomes, additional considerations must be considered based on the unique nature of their associated communities.The total biomass and diversity of bacterial communities associated with salmonid species are lower compared to those of most well studied warm-blooded mammals (Limborg et al., 2023).This results in a low proportion of microbial DNA (related to the sample total nucleic acids content), and thus poses an additional challenge during library preparation.For this reason, in our sample collection protocols, we focused on the enrichment of microbial biomass by minimizing the collection of host tissue.Moreover, considering that a study might comprise two or more of the presented sample types, we tested and optimized an extraction method that proved to be effective for all sample types in this protocol albeit with slight modifications.Finally, we provide guidance for all the key steps leading to library preparation.That involves quality control of the extracted metagenomes, test amplification of 16S rRNA gene amplicons, and further troubleshooting tests related to the low microbial biomass issues referenced above.All tests are aimed to use the ONT 16S metabarcoding kit, which relies on the 27F and 1942R primer couple.However, they can easily be adapted to alternative primer pairs that can be coupled with ONT commercialized sequencing kits that are currently available.

Sample collection considerations
When designing an experiment, especially a longitudinal study, it is crucial to consider additional variables when establishing sampling days and times.First, unless it is the variable of interest, the time of the day for the sampling will generally need to be consistent throughout the experiment.For instance, previous studies have shown that the skin microbiota composition of rainbow trout exhibits circadian fluctuations and is profoundly affected by photoperiods (Ellison et al., 2021).Importantly, routine interventions, such as feeding, are going to impact the microbial quality of the entire system; therefore, sampling must always take place in a similar timeframe in relation to, e.g., feeding times.Similarly, the time elapsed from the last tank cleaning, or any other maintenance intervention, needs to be consistent among sampling points.Such measures must be taken to avoid external confounders that can add an unwanted source of variability among microbiome samples.

Non-lethal and non-invasive sampling options
Considerations about welfare and ethics are a prerequisite of any study involving living animals.All experimentations must comply with the laboratory animals care and welfare regulations standards (Directive 2010/63/EU, 2010).While we refer to the European Legislation, it is crucial for each scientist to verify and adhere to their local regulations, as they may differ based on jurisdiction.According to these regulations, individuals sacrificed for sample collection should be euthanized using humane methods.For fish, these generally involve an anesthetic overdose by immersion drugs, such as benzocaine hydrochloride or MS-222 (Tricaine Methane Sulfonate) solutions (Schroeder et al., 2021).However, when possible, the experimental design

of 29
Current Protocols should consider the principle of the three R's (replacement, reduction, and refinement) to minimize the distress on the animals involved and the number of animals involved and sacrificed in the study (Tannenbaum & Bennett, 2015).In this framework, the interest to monitor animal's health by obtaining samples with non-lethal methods is growing.Among the sample types in this article, alternatives to lethal samples are presented (e.g., gill mucus; skin mucus).Choosing to focus on non-lethal sample types not only minimizes the number of animals sacrificed in one experiment, but also, when coupled to individual tagging, it opens the possibility of collecting samples from the same individuals over time along the same study.This is particularly important when time-dependent processes, such as disease dynamics or exposure to treatments or stressors, are investigated.To collect such samples, the animal must be anesthetized or at the very least sedated.As any procedure can introduce unwanted sources of variation, it is advisable to be consistent with the chosen methods for anesthesia or euthanasia in the same study.Finally, environmental DNA (eDNA)-based studies are paving the way for future molecular investigations that further minimize the sacrifice of animals, both in research and in commercial settings.Sampling water or sediment represents one of the noninvasive sample collection methods with the potential to enable monitoring of the health status of an entire community, from microbial surveillance of human populations (Ko et al., 2022) to farmed animals (Peters et al., 2018).

Adaptation to field sampling
Although the protocols described here were optimized in experimental facilities, they can be transferred to field sampling.The first concern when transferring the protocol to the field is the potentially limited ability to snap-freeze and immediately preserve or process the samples.For this reason, when dry ice and −80°C preservation are not an option, cooling and transportation on ice or at 4°C can be implemented until long term storage at −80°C or downstream processing.The elapsed time between sampling and storage at −80°C must be consistent among all samples, as any variation may contribute to biases in the detection of microbial composition.For water samples specifically, given the potential challenges associated with filtration in the field, a widely accepted approach is to transport them on ice or at 4°C before processing for up to 24 hr (Perrin et al., 2019;van Rossum et al., 2015).Regarding other samples, which need to be transported in tubes, preservation solutions can be considered.For skin and mucus samples, when choosing a preservation solution, we recommend using a solution where the sample can be homogenized, and that can be carried over to extraction (e.g., DNA/RNA Shield, Zymo Research).This is because, in the case of preserving agents that need to be removed at the time of extraction, the act of separating and rinsing the sample from the solution could result in a variable dispersion and subsequent loss of bacteria within the solution.Concerning gut content and mucosal scrapes from rainbow trout specifically, previous reports showed that 96% ethanol is a suitable preservation method (Hildonen et al., 2019).
An additional concern about transferring this protocol to field samples arises from the expectation that a fish farm industry may exhibit higher overall microbial biomasses and a different degree of microbiome complexity.As such, adjustments might be required not only in the experimental design, but also in some of the sampling methods.For instance, water is expected to be more turbid and richer in waste particles, which might prematurely obstruct the membranes and therefore impair the vacuum filtration step.Sequential filtrations using membranes of reducing pore size could be necessary in that case to separate the suspended particles fractions, as recommended in protocols focusing on the analysis of microbiomes from rivers (Reddington et al., 2019).
Furthermore, the metadata that need to be collected will also vary from experimental facilities to an aquaculture farm.For instance, information on eventual water treatments (antibiotics, formalin), disease outbreaks, and mortality of the farmed biomass, will need to be collected at the time of sampling.

Choice of extraction kit
Several commercially accessible kits are available for extraction of total nucleic acid content from complex microbial communities.The choice between column-based and magnetic bead-based, or manual and automatic extraction methods depends on both the available equipment in the laboratory and the number of samples to process.Although some of the most used extraction kits for microbiome samples are column-based (e.g., PowerSoil DNA Isolation kit, MO BIO), when starting from environmental samples (water and sediment) Current Protocols and gut contents, magnetic bead-based extraction is commonly favored over column-based methods.This preference is due to the selective retention of nucleic acids on magnetic beads throughout the various steps, facilitating a more effective removal of contaminants and PCR inhibitors, which are commonly found in such samples (Schrader et al., 2012).The current recommendations from the EMP regarding DNA extraction also involve a magnetic bead-based process on an automated system (L.Marotz et al., 2018).Ideally, all samples should be processed at the same time to avoid batch effects as a confounder.Consequently, especially when dealing with a considerable number of samples, it is preferrable to use an automated extraction system over manual extraction.
Different extraction kits lead to significantly different yields and different biases in terms of bacterial taxa over-or underrepresentation.Therefore, when planning an investigation that aims to compare different sample types, it is recommended to use an extraction kit that works for all kinds of samples included in the study (Salter et al., 2014).In this protocol, we describe the application and adaptation of a magnetic bead-based extraction using an automatic purification system.The kit of choice in our case is the MagMAX Microbiome Ultra nucleic acid isolation kit, which proved to work effectively for extraction from all the previously mentioned sample types.
As a side note, in studies including also metatranscriptomics, it is advised to use a kit that preserves both DNA and RNA, followed by either DNase or RNase treatment, so that the same eluted sample can be used for both applications.

Control samples
As every processing phase is likely to introduce bias or contamination, controls must be included in every step.It is crucial to include both positive and negative controls that are appropriate for the system and methods of interest.
As a positive control, at least one aliquot of a known bacterial mock community must be used.This is commonly included to assess possible bias in the DNA isolation, in the marker gene amplification step, as well as in the final sequencing.The ideal extraction method should prove the same performance on bacteria with different cell wall compositions.Similarly, since non-proportional amplification can affect the final representation of the sample composition, the use of a known bacterial community will allow the optimization of the PCR step.An increasing number of commercial options are available to suit several types of samples and studies.As microbiome research is still mainly focusing on the human biomedical field, most of the mock bacterial communities that have been developed and are commercially available will be composed of human-associated strains (Highlander, 2013).However, the existing mock bacterial communities may not be relevant for research fields other than biomedicine.As aquaculture is a relatively new system for microbiome studies, future studies should aim to develop a custom mock community that includes the most relevant, prevalent, or challenging bacterial taxa in the sample type of interest (Colovas et al., 2022).
Most importantly, negative controls must be included to assess the presence of contamination during the pre-processing and extraction steps.Such controls are constituted of all the extraction reagents and must undergo the same treatments at the same time as the experimental samples.Negative controls also must be included in the sequencing runs, as they will constitute a baseline to identify putative contaminant OTUs during the bioinformatic analysis.
As a final note, the microbiome field has seen an increased interest in absolute quantification of copy numbers as opposed to the mere calculation of relative microbial abundance.This can be achieved by using a spikein synthetic community standard, which is added into the samples in a defined quantity prior to extraction, processing, and sequencing.Synthetic spike-ins have been developed for both 16S rRNA gene metabarcoding (Tourlousse et al., 2017) and for shotgun metagenomics (Hardwick et al., 2018) applications.The read number of spiked-in standards can subsequently be used as a reference to normalize read counts, or to calculate absolute microbial abundances.

Replicate samples
In the analysis of host-associated microbiomes, it is important to account for individual variation and expected complexity in the study design and the sample size estimation.For instance, wild populations are expected to display a much higher metagenomic complexity with respect to production animals (Alberdi et al., 2022).With increasing degree of expected complexity, a higher number of individuals must be included in the study.On

of 29
Current Protocols the other hand, when dealing with the study of a particular environment microbiome, it can be challenging to use "biological" replicates.For instance, a specific aquaculture facility, or river, might be the study subject.In that case, replicate sampling of the same environment, such as replicate filtration of different aliquots of water from the same source, can be performed.This strategy has been previously used in the study of river microbiomes (Urban et al., 2021).
Technical replicates can also be used in the amplification step.The Earth Microbiome Project (EMP) recommendation for 16S rRNA gene metabarcoding is to perform triplicate amplification reactions for each sample, and to later pool them before amplicon electrophoresis, quantification, and purification (Thompson et al., 2017).While such a protocol is appropriate for sequencing projects based on the Illumina short-read platforms, most studies using the ONT sequencing kit (SQK-16S024) perform one amplification reaction for each sample, since primers in the kit are already barcoded, and pooling is only possible if the same barcoded primers are used for each replicate.However, the kit provides more than one primer set; therefore, such an approach is a feasible option.Another common approach to technical replicates of amplification (mostly for short-read Illumina sequencing) is to prepare replicate PCR reactions and, instead of pooling them, sequence them separately (Alberdi et al., 2018).After separately sequencing PCR replicates, two analysis methods are possible, depending on the purpose of the study.An additive method, which will add all the detected taxa in the replicates, will maximize the detection of microbial diversity.A more restrictive (and less common) method of analysis will instead only keep the OTUs that appear in all sequenced replicates, therefore minimizing the false positives and experimental artifacts originating during library preparation.However, replicate PCR reactions of the same samples were often reported to yield different results in terms of detected taxa and richness, especially when the initial sample has a low bacterial biomass and a low amount of DNA template is used (Kennedy et al., 2014).Therefore, in cases where the template DNA amount is exceptionally low, the PCR stochasticity will be higher, and each PCR replicate will not represent the diversity of the original sample.The restrictive approach is thus not advised in such circumstances, as it could result in the loss of real diversity (Alberdi et al., 2018).

16S rRNA gene PCR troubleshooting
One of the most common issues with PCRbased library preparation, especially starting from host-associated samples, is an unsuccessful or suboptimal 16S rRNA gene amplification.There are distinct reasons why the PCR can fail, but in this section, we will mainly deal with problems related to the sampling methods described above, i.e., mixed hostmicrobial DNA.
First, the overall microbial biomass in the sample might be too low.Therefore, the DNA template quantity may be raised to increase the starting material for amplification, or the number of PCR cycles can be increased.Second, and intimately related with low bacterial biomass, is a sample where the host to bacteria nucleic acid ratio is too high.In some kinds of samples (e.g., entire fish gut sections), the host DNA can easily exceed 90% of the extracted material.In this case, adding the advised starting quantity of DNA template could still result in a failed amplification, because the majority of starting DNA derives from the host.Also in this case, increasing the starting material or raising the number of amplification cycles can be beneficial.On the other hand, a failed PCR could be caused by inhibition, and diluting the sample could solve the issue (Wang et al., 2017).A qPCR using universal 16S rRNA gene primers can be performed prior to library preparation and subsequently used as a starting point to identify the optimal template DNA dilution and to optimize the number of PCR cycles needed.
Previous studies have highlighted the challenges related to cross-kingdom amplification of host DNA when using universal primers targeting the bacterial 16S rRNA gene (Galkiewicz & Kellogg, 2008;Walker et al., 2020).Such lack of specificity can outcompete the amplification of target bacterial sequences and result in unexpected DNA bands among the products of test PCR reactions.Possible solutions to this issue include purification of the expected amplicon after sizeseparation on agarose gel (McDevitt-Irwin et al., 2019), changing the universal primers used in the study, or by using peptide nucleic acid (PNA) clamps to impair off-target amplification (Reigel et al., 2020).
Another issue with library preparation might be related to a subsequent step, i.e., the amplicon purification step.Most library preparation kits rely on magnetic-bead purification (e.g., AMPure XP beads, Beckman Coulter), which needs to yield an appropriate concentration to undergo sequencing.For instance, according to the ONT 16S barcoding kit, the final concentration of pooled barcoded libraries needs to be in the range of 50 to 100 ng in a volume of 10 μl for optimal performance.Thus, according to the desired degree of multiplexing, ranging from 1 to 24 samples for such ONT kit, the concentration of purified amplicons for each sample must exceed a minimum threshold (from 5 down to 0.2 ng/μl).However, it is likely that the purification step will result in the loss of a certain amount of amplicon, to the point that the final concentration will not be sufficient for sequencing.Strategies to solve this issue include preparing replicate PCR reactions to pool prior purification or testing alternative PCR purification kits to achieve a higher amplicon recovery rate.
In Table 2, we provide examples of average initial template DNA quantity, number of amplification cycles and replicate PCR reac-tions to pool that previously worked for our samples, producing sufficient amplicon to undergo sequencing.
Alternatively, adjustments to the sampling method could be required to try to minimize the collection of host tissue.However, there are also cases in which the tissue is important per se, e.g., when interested in bacterial infiltration in tissues (e.g., skin diseases) or in intracellular bacterial taxa.In these situations, the exclusion of host tissue is not possible, and thus additional protocols intended for exclusion of host DNA (Heravi et al., 2020;C. A. Marotz et al., 2018) can be taken in consideration.
Table 5 summarizes all possible troubleshooting solutions for library preparation that we mentioned in this section.

Understanding Results
The quality of the extraction step is assessed prior to library preparation.As described in Basic Protocol 4, this is

of 29
Current Protocols usually done in three phases.First, purity of the nucleic acids is evaluated by reading the absorbance values on a spectrophotometer.The ratio between nucleic acids and protein is generally assessed by 260/280 absorbance ratio.The purity over proteins is deemed satisfactory when the value is between 1.7 and 2.0.The 260/230 absorbance ratio can be used to estimate the purity of nucleic acids over salts or contaminants, and it is usually acceptable above 1.8.DNA integrity is assessed by agarose gel electrophoresis.A clear band of high molecular weight DNA should be visible in the agarose gel (Fig. 2A).If the genomic DNA appears as a long "smear" (Fig. 2B), this could indicate that DNA shearing occurred either during the extraction step or during a suboptimal preservation of the sample.This could impact the ability to amplify long amplicons (1.5 kbp) in subsequent steps, and as such, it might be a source of potential bias in the results.Accurately measuring the DNA concentration in the eluted sample using a fluorometric assay will be useful to evaluate the DNA yield for each sample.This is particularly important for low-biomass samples, as fluorometric assays are usually more sensitive and accurate to low DNA concentrations than measurements on a spectrophotometric device.Positive controls are important to assess the efficacy of the extraction method (see "Control samples" in the Critical Parameters section).
Next, a test PCR amplification is performed with universal bacterial primers targeting the V1 to V9 hypervariable regions of the 16S rRNA gene.The resulting PCR products are evaluated through gel electrophoresis.Under optimal conditions, a single bright, clear DNA band is expected to appear at 1.5 kbp size (Fig. 3B).A faint DNA band (Fig. 3A) may indicate that the PCR amplification step requires optimization (see Troubleshooting).If no bands are visible, or if increasing template and/or PCR cycles number does not solve the issue, PCR inhibitors, e.g., specific dissolved organic compounds commonly found in stool or environmental samples (Schrader et al., 2012) may be present in the sample.An additional qPCR test can be performed to assess this (see Support Protocol 1).
The results of such qPCR assay can be interpreted using a qPCR analysis software of choice.Calculation of the amplification efficiency (E= −1+10 (−1/slope) ) for each group of sample dilutions is usually performed automatically by the analysis software.The op-timal range is between values of 90% to 110%; if it exceeds 110%, this could indicate that the amplification is affected by inhibition.If the efficiency of the PCR is in the optimal range, each 10-fold dilution should amplify with a difference of ∼3.3 Cq values from one dilution to the next.If a moderate degree of inhibition is present, the undiluted samples will show a delay in their exponential amplification phase, while more diluted samples will display a better efficiency of amplification and excluding the highest concentration samples will improve the efficiency value.If that is the case, lower amounts of DNA template for library preparation are recommended, and a new endpoint PCR test (see Basic Protocol 4, "Test PCR amplification with selected 16S rRNA gene primers") can be performed with optimized conditions.If the efficiency of amplification is overall suboptimal, or no amplification is at all possible, application of additives to overcome inhibition (Schrader et al., 2012) can be considered.
After optimization of the PCR step, it is still possible that the amplicon quantity might be too low to undergo sequencing.A successful library preparation will yield an appropriate concentration of amplicon.As detailed in the Troubleshooting section, for the ONT 16S barcoding kit this means that, for each sample, the minimum concentration after amplicon purification should be from 5 ng/ μl down to 0.2 ng/ μl (depending on the samples to be multiplexed, 1 to 24).A test purification from the test PCR step can determine if the amplicon amount will be sufficient, or if multiple reactions will need to be prepared for each sample to be pooled prior purification and sequencing.
With the optimized conditions, the samples can finally undergo library preparation and sequencing by following the manufacturer's instructions from the kit of choice (e.g., SQK-16S024).Starting from a sample type of unknown complexity, it is important to perform a test sequencing run and data analysis to determine the minimum sequencing depth for the sample type and study purpose.For instance, the approximate number of reads that are needed to reach saturation (plateau) of the read rarefaction curves can be calculated and used as a target sequencing depth.Such curves can be simply plotted as number of features (taxa) over number of reads-e.g., using vegan (Dixon, 2003) in R.This knowledge will also help determining how many samples can be multiplexed together in the same run.For Current Protocols instance, if samples exhibit a relatively low complexity and rarefaction curves are saturated around 50,000 reads, it will be possible to run multiple libraries in the same run (e.g., all 24 barcodes in an SQK-16S024 kit).After sequencing, raw FAST5 files will be produced.Support Protocol 2 shows an example of a bioinformatics pipeline that can be used to retrieve and process the raw reads to produce clean and filtered FASTQ reads.The Emu taxonomic assignment algorithm is finally used to generate taxonomic abundance tables, which can be then imported to R Statistical Software for further analysis.A plethora of R packages were developed for microbiome datasets analysis, such as vegan (Dixon, 2003) and phyloseq (McMurdie & Holmes, 2013).When selecting the analysis tools, it is imperative to consider that, during statistical analysis, tests should match the compositional nature of microbiome datasets (Gloor et al., 2017).

Sampling procedures
Water sampling and filtration: up to 1 hr/sample (highly dependent on water turbidity and sampling volume).

Nucleic acid extraction
Sample homogenization and pre-treatment: 1 hr 30 min (dependent on number of samples to process).
DNA extraction: 1 hr for up to 96 samples (time relative to DNA extraction using an automated extraction system, e.g., KingFisher Flex).

Quality control and test amplification
DNA purity assessment: 10 s/sample.DNA electrophoresis: 30 min.Fluorometric quantification: 2 min/sample.Test 16S rRNA gene amplification: 3 hr (dependent on the number of PCR cycles).

Figure 1
Figure 1 Sampling procedures for microbiome analysis in experimental facilities, environmental (water, sediment) and rainbow trout-associated (gill, skin, gut).(A) Collection and vacuum filtration of water onto 0.22-μm PVDF membranes.(B) Separation by sedimentation and collection of organic sediment samples.(C) Collection of gills mucus by swabbing (left) or absorption (right).(D) Sampling procedure for skin mucus through absorption (left) or collection of the cutis layer (right).(E) Collection of distal gut content (left) and distal gut mucosal scraping (right).Sterile 60-mm Petri dishes (Falcon, cat.no.35007, or equivalent) Sterile 2-ml tubes (Eppendorf, cat.no.022363352, or equivalent) 50-ml sterile conical tubes Laboratory rack for 50-ml conical tubes 20-ml serological pipettes Pipettor Vortex Laboratory spoon Laboratory balance Dry ice

Figure 2
Figure2Different extraction methods from water samples lead to distinct degrees of fragmentation of genomic DNA.(A) Electrophoretic run of genomic DNA from 5 water samples (1-5) and controls extracted using a magnetic bead-based kit (MagMAX Microbiome Ultra nucleic acid isolation kit, Thermo Fisher Scientific, modified as described in the Basic Protocol 3).(B) Electrophoretic run of genomic DNA from 5 water samples (1-5) and controls extracted using a column-based kit (DNeasy PowerWater kit, Qiagen, modified with lysozyme pre-treatment).In both cases, 0.22-μm filters were used to filter water samples replicates in the same way.Two negative controls for extraction were used, i.e., 0.22-μm sterile filters (Cf) and extraction reagents only (Cr).An aliquot of ZymoBIOMICS microbial community standard (Zymo Research) was also extracted with both methods (Z).DNA was run on a 1.2% agarose ethidium bromide E-Gel (Thermo Fisher Scientific) for 30 min. 1 kb Plus DNA ladder (Invitrogen, cat.no.10787018) was used as a reference marker (M).Fragment size is indicated to the left of the marker lane.

Figure 3
Figure3Host-associated sample types, i.e., skin and skin mucus samples, necessitate different optimization of the PCR amplification step.(A) 5 rainbow trout skin samples (1-5) PCR amplification of the V1-V9 variable regions of the 16S rRNA gene.(B) 5 rainbow trout skin mucus samples (1-5) PCR amplification of the V1-V9 variable regions of the 16S rRNA gene.Both sample types were collected from the same skin region and subsequently extracted using the MagMAX Microbiome Ultra nucleic acid isolation kit, Thermo Fisher Scientific, modified as described in the Basic Protocol 3. A negative control for extraction was used, i.e., extraction reagents only (Cr).A 75 μl aliquot ZymoBIOMICS microbial community standard (Zymo Research) was also extracted in the same way (Z).Amplicons were generated using 50 ng of initial DNA template for each sample.30 PCR cycles were used.DNA was run on a 2% agarose ethidium bromide E-Gel (Thermo Fisher Scientific) for 30 min. 1 kb Plus DNA ladder (Invitrogen, cat.no.10787018) was used as a reference marker (M).Fragment size is indicated to the left of the marker lane.
Take notes of the filtered volume (ml) on each filter membrane.The established maximum volume should take into account the expected bacterial biomass or turbidity in the water.This can be estimated by a test filtration to assess at which volume the filter starts to clog, and how much time it takes to filter a water sample.The sample volume should also be feasible in terms of filtration time, as replicates of filtration are needed, and all samples should be filtered on the same day.

Table 1
Recommended Volumes and Concentrations of Each Component to Assemble a PCR to Test the Amplification of the 16S rRNA Gene, Variable Regions V1-V9

Table 2
Example of Average Template DNA Quantity, Number of Required Amplification Cycles, and PCR Reactions to Pool to Obtain Enough Amplicon for Sequencing a a Information is provided for each sample type in the protocol, as previously obtained in our laboratory, based on at least 5 replicates per sample type.
The cycling conditions recommended by the ONT protocol require a maximum of 25 cycles.Such conditions are generally able to generate the desired amplicons when using water or sediment samples.Nevertheless, for samples with lower relative microbial biomass (host-associated samples), it may be necessary to use a higher numberof PCR  cycles (e.g., 30 to 40 cycles).Refer to Table2for examples to use as a starting point for optimization.

Table 4
Universal Primer Oligonucleotide Sequences Targeting the Hypervariable Region V4 of 16S rRNA Gene a