Design on What Was Actually Done

Visual summary (first-order)

• Hosts: Germ-free mice (C57BL/6) as recipients. Use same sex/age to reduce variance.
• Donors: Two donor microbiotas with contrasting community states (e.g., healthy vs dysbiotic / Donor A vs Donor B).
• Reciprocal arms: (1) Donor A -> GF mice; (2) Donor B -> GF mice; (3) GF mice colonized with A then moved to B-host environment (habitat swap); (4) GF mice colonized with B then moved to A-host environment; (5) GF-only controls and mock-transplant controls.
• Sampling: High-frequency fecal sampling (days 0,1,3,7,10,14,21,28, then weekly to day 56) and host readouts (histology, immune phenotyping) at pre-specified endpoints.

Design rationale and references (compact)

Reciprocal-transplant gnotobiotic designs separate the effects of microbial input (legacy) and host habitat selection; classic zebrafish↔mouse reciprocal transplants established that host habitat reshapes donor communities while preserving some lineage-level signals and producing conserved host responses, so replicate that logic with dense longitudinal sampling to measure dynamics and host outcomes Rawls 2006: host habitat selection.

Use dense temporal sampling and dynamical modeling (e.g., module-based Bayesian gLV frameworks) for forecasting and keystone identification — recent gnotobiotic studies show high-value in dense sampling with perturbations and module inference to discover keystone groups and predict dynamics MDSINE2 2025: dynamics & keystones.

Expect microbial context–dependent immune outcomes: taxa interactions can be epistatic for host immunity (e.g., specific bacteria modulate colitis only in certain community contexts) so include immune readouts and co-colonization/knockout arms where feasible Huang et al. 2022: epistasis in immunity.

Detailed experimental plan (stepwise)

1. Donor selection and characterization (weeks –2 to 0):
   • Choose two donor fecal samples representing contrasting states (healthy-like vs dysbiotic or two species-different communities). Deeply profile donors by shotgun metagenomics and metabolomics before transplantation to create baseline multi-omic maps.
   • Culture-archive isolates where possible to allow later defined-community reconstruction (useful to mechanistically test keystones detected by dynamical models) — supported by prior work using isolates to recapitulate some host responses Rawls 2006: isolates for mechanistic tests.
2. Recipient cohorts and arms (Day 0 = colonization):
   • GF recipients per arm: recommend n=6–8 mice/arm (allowing for attrition and per-arm inference; see plotted suggestion above).
   • Arms: A->GF, B->GF, GF only (sham), reciprocal habitat-swaps (see step 4), and optional co-colonization arms if testing epistasis/immune effects.
3. Colonization protocol:
   • Standardize donor inocula by wet-weight and cell concentration (e.g., normalized 16S copy number) to reduce inoculum-size confounds; deliver by oral gavage (single or staggered 2–3 doses in 48 h) following published gnotobiotic practices.
   • House colonized groups in separate sterile isolators to prevent cross-contamination; use sentinel PCR checks weekly to confirm gnotobiotic integrity.
4. Reciprocal habitat swap (core experimental manipulation):
   • At day 14–21 post-colonization (after initial equilibration), transfer a subset of A->GF mice into cages/environment previously occupied by B-derived microbiota (or transplant environmental microbiota components) and vice versa to test host habitat-driven reassembly; alternately perform cross-housing vs cecal content transplant into conventionalized recipients to separate environmental/host factors.
   • Implement cross-housing only with strict transfer protocols to avoid direct cross-contamination between donor lines (clean transfer of soiled bedding or cage swap protocols under sterile conditions).
   • Collect dense longitudinal samples around the swap (days –3,0,+1,+3,+7,+14,+21 post-swap) to capture immediate and medium-term dynamics. Dense sampling is critical: prior studies used high-frequency sampling to infer dynamics and modules MDSINE2 2025: dense sampling benefits.
5. Perturbations and controls (recommended):
   • Include at least one small perturbation (e.g., short dietary change or low-dose antibiotic pulse) in a subset of animals to reveal interaction responses and enable dynamical inference; MDSINE2 and other frameworks use perturbations to infer edges/keystoneness robustly MDSINE2 2025: perturbation use.
6. Sampling & readouts (longitudinal + endpoints):
   • Microbiome: frequent fecal pellets for 16S rRNA amplicon sequencing plus absolute quantification by universal 16S qPCR; collect schedule: pre-colonization, daily first 3 days, then 3×/week to week 3, then weekly to week 8 (example plotted timeline above).
   • Metagenomics & metabolomics: shotgun and fecal metabolomics at baseline, pre-swap, 3 days post-swap, and endpoint to map functional shifts and metabolites relevant to host immune signaling.
   • Host: immune phenotyping (flow cytometry of MLN, Peyer’s patches, lamina propria), histology of gut segments, cytokine ELISAs, and host transcriptomics (bulk RNA-seq) at endpoints and important intermediate timepoints; this is crucial because microbial context modulates immunity (epistasis) Huang et al. 2022: immunity readouts.
7. Data analysis plan (prioritize dynamical inference):
   • Use compositional-aware pipelines for sequencing preprocessing (DADA2 or similar), and include absolute abundance calibration with qPCR to enable dynamical models based on absolute counts.
   • Fit dynamical models with modular structure (e.g., MDSINE2 or Bayesian modular gLV) to infer interaction modules, keystone modules, and forecast post-swap dynamics; bench against simpler gLV and differential-abundance analyses MDSINE2 2025: analysis recommended.
   • Integrate host transcriptomics and immune phenotypes with microbial modules and metabolites using multivariate correlation/causal-inference frameworks to test whether module shifts predict host readouts.
8. Follow-up mechanistic validation:
   • Use isolate consortia or synthetic communities representing keystone modules (inferred by dynamics) to re-test colonization and host effects in simplified gnotobiotic mice (this mirrors the isolates/defined-community approach from previous work) Rawls 2006: defined communities.

Key controls, limitations, and blindspots (explicit)

• Gnotobiotic mouse results may not generalize to conventional mice or humans; host genetics and environment strongly influence habitat selection and immune responses (limitation of gnotobiotic models) Rawls 2006: generalizability caution.
• Dense longitudinal sampling and absolute abundance calibration are essential to reduce compositional/measurement biases and to enable reliable dynamical inference (MDSINE2 demonstrates this necessity) MDSINE2 2025: sampling importance.
• Epistatic interactions between microbes can yield counterintuitive immune outcomes; including co-colonization and immune phenotyping helps identify such nonlinear effects and avoid misattribution Huang et al. 2022: epistatic immune effects.

Immediate deliverables & experimental checklist

• Donor QC: shotgun metagenomes & metabolomes, frozen glycerol stocks, isolate cultures where feasible.
• Protocol SOPs: colonization, cross-housing/swap, sterile handling, sample collection/metadata schema.
• Sequencing & qPCR plan: amplicon + absolute qPCR for every fecal sample; select shotgun + metabolomics timepoints.
• Data analysis pipeline: DADA2 & absolute calibration → MDSINE2-style modular dynamical inference → integrate host phenotypes.

Concise confidence & next steps

This plan synthesizes reciprocal-transplant logic (host habitat vs microbial legacy), dense longitudinal dynamical inference best-practices, and immune-context sensitivity; implement pilot with n=6–8 per arm and dense sampling, then scale and perform mechanistic follow-ups with isolates/defined consortia. Key falsifiers: if donor lineages remain unchanged after habitat swap, or dynamical models fail to predict post-swap trajectories, the habitat-selection vs legacy hypotheses will need revision Rawls 2006: falsification criteria.