Low-Biomass Maternal-Infant Microbiome Studies: Breast Milk and Meconium

Early-life microbial colonization is closely associated with the development of the infant immune system, metabolism, and gut ecosystem. Studying how maternal microbial communities contribute to early infant exposure requires careful analysis of challenging biospecimens, including human breast milk and first-pass neonatal meconium. Both matrices can contain relatively little microbial DNA compared with higher-biomass samples such as maternal feces, but they present different analytical problems.

Human milk can combine limited microbial biomass with substantial maternal DNA, while first-pass meconium may contain very little detectable bacterial material and can be particularly sensitive to contamination, collection timing, and postnatal exposure. These characteristics make early-life microbiome studies vulnerable to over-interpretation if sequence detection is treated as proof of viable colonization or maternal-to-infant transmission. This guide provides maternal-infant cohort researchers, developmental biologists, biotechnology teams, and CRO project managers with a framework for contamination-aware sampling, evidence grading, paired cohort design, and sequencing method selection for species-level microbiome research.

Key Takeaways

• Breast Milk and Meconium Present Different Low-Biomass Problems: Human milk can be strongly influenced by host DNA, while first-pass meconium often contains very limited and highly variable microbial biomass.
• DNA Detection Is Not Equivalent to Colonization: Detecting microbial sequence reads does not by itself establish microbial viability, persistent infant colonization, prenatal presence, or maternal-to-infant transmission.
• Collection Timing Is a Biological Variable: Birth-to-meconium collection time, feeding exposure, antibiotic exposure, lactation stage, storage, and processing intervals can materially affect early-life microbiome observations.
• Species Sharing Is Not Proof of Vertical Transmission: Detecting the same species in a mother and infant can identify a candidate source relationship, but strain-resolved genomic evidence is generally required for stronger transmission inference.
• Controls Must Reflect the Actual Collection Workflow: Collection blanks, extraction negatives, library controls, and source-tracking samples serve different purposes and should not be treated as interchangeable.
• Different Sequencing Methods Answer Different Questions: Short-region 16S sequencing, shotgun metagenomics, and reduced-representation 2bRAD-M differ in taxonomic resolution, host-DNA sensitivity, genomic information content, and sample requirements.
• Published 2bRAD-M Studies Support Difficult-Sample Feasibility: Benchmark experiments have included total DNA inputs down to 1 pg, degraded DNA, and separate high-host-background tests. These observations should be interpreted as published feasibility evidence rather than universal sample guarantees.

The Biological and Analytical Challenges of Early-Life Samples

What Makes Breast Milk Microbiome Studies Difficult?

Human milk is biologically complex. In addition to microbial material, it contains maternal epithelial cells, immune cells, lipids, proteins, and other components that can influence DNA extraction and downstream library preparation. Previous studies have reported microbial concentrations on the order of 10⁵–10⁶ cells per milliliter in some milk samples, while the proportion of human DNA can exceed 90% depending on the specimen and workflow.

For untargeted shotgun metagenomics, a high host-DNA fraction can substantially reduce the number of reads available for microbial analysis. This does not mean that shotgun sequencing is inherently unsuitable for milk, but it makes effective microbial depth highly sample-dependent. Host-depletion strategies may improve sequencing efficiency in some workflows, although cell-selective depletion can introduce taxonomic bias and may behave differently after freezing, storage, or repeated freeze-thaw exposure.

What Makes First-Pass Meconium Different?

First-pass meconium is not simply a low-biomass version of infant stool. Microbial biomass can be extremely low and highly variable, and some specimens may contain insufficient bacterial material for standard microbiome analysis. In a multi-center study of very preterm infants, Klopp et al. demonstrated substantial variation in bacterial load among meconium samples and highlighted the importance of contamination controls and collection timing (Klopp et al., 2022).

Importantly, first-pass meconium is collected after birth. Microbial DNA detected in these samples can potentially reflect early postnatal acquisition, environmental exposure, low-level contamination, transient DNA, or other biological sources. Therefore, first-pass meconium sequencing alone cannot establish the existence of a resident prenatal microbiome. Kennedy et al. emphasized this distinction when evaluating low-biomass fetal microbiome evidence and the contamination risks that can generate apparently convincing microbial signals (Kennedy et al., 2023).

Sampling Timing and Metadata Are Part of the Biological Design

Early-life microbiome studies are unusually sensitive to time. A microbial profile obtained shortly after birth may not represent the same biological state as a sample collected several hours or days later. For first-pass meconium, the interval between birth and sample collection should therefore be recorded and incorporated into the analytical design rather than treated as routine administrative metadata.

Important variables can include:

• Birth-to-sample interval: Postnatal microbial acquisition can begin rapidly, so collection timing affects how early-life signals are interpreted.
• Feeding before collection: Breast milk, formula, or other feeding exposure can alter the infant gut environment and introduce additional microbial sources.
• Perinatal antibiotic exposure: Maternal or infant antimicrobial treatment can alter community establishment and should be captured as a study covariate.
• Delivery mode and environment: Vaginal delivery, Caesarean delivery, NICU exposure, home environment, and other factors may influence early microbial acquisition.
• Lactation stage: Colostrum, transitional milk, and mature milk should not automatically be treated as microbiologically equivalent.
• Collection-to-freezing interval: Storage temperature, preservative use, and delay before freezing can affect DNA recovery and community composition.

These variables should be specified before sequencing so that technical and biological differences are not confused during downstream interpretation.

Sample-Specific Contamination Landscapes and Control Design

In low-biomass sequencing, contaminant DNA introduced during collection, extraction, library preparation, or the laboratory environment can approach or exceed the authentic microbial signal. Controls should therefore represent the actual workflow rather than follow a universal checklist. For broader principles, review our low-biomass microbiome study design guide.

Maternal areolar skin, for example, should not automatically be labeled a negative control. It may represent a genuine potential microbial source for milk or infant exposure and is therefore better treated as a source-tracking specimen. Likewise, a diaper blank is useful only if the study actually collects meconium from diapers.

Contamination-aware statistical methods such as decontam can help identify candidate contaminants by examining their prevalence in negative controls or relationships with DNA concentration (Davis et al., 2018). These methods should complement, not replace, inspection of controls, batch structure, prevalence, and biological context.

Designing a Paired Maternal–Infant Cohort

A study investigating only breast milk and first-pass meconium can characterize those two matrices, but it provides limited information about the source of infant microbes. If the research objective is maternal–infant microbial sharing or transmission, the sampling framework should include potential maternal reservoirs and longitudinal infant follow-up whenever feasible.

Longitudinal sampling is especially important because a sequence detected once may represent transient exposure rather than stable gut colonization. Repeated detection over time, increasing microbial biomass, and consistent ecological patterns can provide stronger evidence of persistence.

The Maternal–Infant Evidence Ladder: From Detection to Transmission

Early-life microbiome studies benefit from separating several evidence levels that are sometimes treated as interchangeable. A species name in a sequencing table is the beginning of biological interpretation, not the final proof of colonization or transmission.

1. Level 1: Sequence Detection — Microbial reads or taxonomic markers are detected in the specimen. This establishes only that sequence information compatible with the taxon was recovered.
2. Level 2: Contamination-Aware Detection — The signal is evaluated against collection blanks, extraction negatives, library controls, prevalence patterns, and other technical evidence. This increases confidence that the observation is not primarily explained by background contamination.
3. Level 3: Quantitative Confirmation — qPCR, ddPCR, or another validated quantitative method confirms that the target microbial signal is reproducibly measurable. Quantitative DNA detection still does not establish viability.
4. Level 4: Persistence or Viability Evidence — Repeated longitudinal detection, increasing microbial biomass, culture where feasible, or another appropriate activity/viability measurement provides stronger support for persistent colonization than a single DNA observation.
5. Level 5: Strain-Resolved Source Tracking — Genetic evidence demonstrates that maternal and infant organisms are the same or closely related strains. This can support transmission inference when combined with sampling chronology and appropriate alternative-source controls.

This hierarchy is particularly important for claims of maternal-to-infant transmission. Detecting Bifidobacterium longum in both maternal and infant samples identifies species sharing, but it does not establish that the same strain moved from the mother to the infant. Strain-level studies such as Asnicar et al. used shotgun metagenomic SNVs and pangenome information to investigate maternal–infant transmission at substantially higher genomic resolution (Asnicar et al., 2017).

Sequencing Method Selection for Breast Milk and Meconium

Short-Region 16S rRNA Sequencing

Short-region 16S sequencing remains useful for sensitive bacterial community screening. It generally requires less microbial genomic material than untargeted whole-metagenome sequencing and is less affected by host-DNA sequencing overhead because microbial marker genes are selectively amplified.

Its principal limitation is taxonomic resolution. Short-region 16S frequently provides genus-level information, while species-level classification varies according to the targeted region, organism, read length, reference database, and classifier. This makes 16S appropriate for many broad community questions but potentially insufficient when the research hypothesis depends on distinguishing closely related species. For standard marker-gene workflows, see our 16S/18S/ITS amplicon overview and microbial diversity analysis solutions.

Whole-Metagenome Shotgun Sequencing

Shotgun metagenomics provides substantially broader genomic information than marker-gene sequencing. When effective microbial coverage is sufficient, it can support species taxonomy, functional gene profiling, strain analysis, and other genome-wide applications.

For low-biomass breast milk and meconium, however, library construction and microbial sequencing efficiency can become limiting. Hou et al. directly compared 16S, whole-metagenome sequencing, and 2bRAD-M in a maternal–infant cohort (Hou et al., 2025). In that particular study and workflow, WMS library preparation failed for all 30 first-pass meconium specimens because of low DNA input. Among 34 breast-milk samples subjected to WMS, average sequencing output exceeded 100 million reads per sample, but only three samples retained more than 0.5 million microbial reads after host filtering.

These results illustrate the technical difficulty of WMS in that cohort, but they should not be interpreted as universal failure rates for every breast-milk or meconium project. Sample biomass, extraction, preservation, host background, library preparation, and sequencing strategy can all affect performance. Host depletion is also not an obligatory component of shotgun metagenomics. Whether it is useful should be evaluated according to the host fraction, sample integrity, biological endpoint, and potential depletion-associated bias. For suitable projects, explore our shallow shotgun metagenome sequencing and broader microbiome sequencing services.

2bRAD-M Reduced Metagenomics

2bRAD-M uses type IIB restriction enzymes to generate short genomic tags and analyzes a reduced representation of microbial genomes. Sun et al. introduced the approach for species-resolved profiling of bacteria, archaea, and fungi and evaluated it under several difficult-sample conditions (Sun et al., 2022).

Published benchmark experiments included total DNA inputs down to 1 pg, severely fragmented DNA, and separate tests involving high host-DNA backgrounds. These experimental conditions should not be combined into a universal claim that every 1-pg sample with extreme host contamination will perform identically.

Hou et al. subsequently applied 2bRAD-M to maternal breast milk, maternal feces, and infant meconium. In that study, the method generated species-level profiles from the low-biomass breast-milk and meconium specimens that were challenging for WMS. Method concordance with WMS was assessed primarily using maternal fecal samples, where both methods generated sufficient microbial data for comparison. The milk and meconium results therefore provide evidence of technical feasibility, whereas the fecal dataset provides stronger direct method-comparison evidence.

These characteristics make 2bRAD-M analysis for microbiome research a potential option when the primary endpoint is species-level taxonomy from low-input, degraded, or host-rich samples. It should not be treated as a substitute for deep shotgun metagenomics when strain-resolved transmission, unrestricted gene discovery, mobile elements, or genome reconstruction are required. For broader difficult-sample planning, see our low-biomass and host-rich microbiome solutions.

What Each Method Cannot Establish

Method selection becomes clearer when research teams define not only what a platform can provide, but also what the resulting data cannot legitimately prove.

A Practical Maternal–Infant Study Design Workflow

Step 1: Define the Biological Claim Before Selecting the Assay

Decide whether the project aims to characterize milk communities, describe first-pass meconium, identify shared maternal–infant species, investigate persistence, or test strain-level transmission. These endpoints require progressively stronger evidence.

Step 2: Capture Timing and Exposure Metadata

Record collection timing, feeding, antibiotic exposure, delivery mode, hospitalization, lactation stage, sample preservation, and other factors that could alter microbial acquisition or recovery.

Step 3: Design Controls Around the Actual Collection Workflow

Include collection blanks where contamination can be introduced, extraction negatives, appropriate library controls, and positive controls. Treat maternal skin, stool, or vaginal samples as biological source candidates rather than generic negatives.

Step 4: Select the Sequencing Method from the Primary Deliverable

Use short-region amplicons for broad bacterial screening, reduced-representation profiling when species-level taxonomy is the central goal in difficult samples, or shotgun sequencing when direct genomic or strain-level information is required and effective microbial coverage can be achieved.

Step 5: Separate Discovery from Stronger Biological Claims

A species detected during exploratory sequencing may warrant quantitative confirmation, longitudinal follow-up, or strain-resolved analysis before being described as a persistent colonizer or maternally transmitted organism.

Summary: Match the Strength of the Claim to the Strength of the Evidence

Breast milk and first-pass meconium provide valuable windows into early-life microbial exposure, but their low microbial biomass makes them especially sensitive to technical contamination and biological over-interpretation. Reliable maternal–infant microbiome research therefore requires more than generating a taxonomic table.

• Record sampling time and exposure metadata because early infant microbial communities can change rapidly after birth.
• Use controls that reflect the real sampling and laboratory process rather than applying a fixed checklist to every cohort.
• Interpret first-pass meconium as an early postnatal sample, not automatic evidence of a prenatal resident microbiome.
• Distinguish DNA detection, quantitative confirmation, viability, persistence, and transmission as separate evidence levels.
• Use species sharing for source hypothesis generation, while reserving stronger transmission claims for strain-resolved and longitudinal evidence.
• Select sequencing around the required endpoint: community screening, species-level profiling, functional genomics, and strain transmission are different technical problems.

The strongest maternal–infant study is therefore one in which sampling chronology, controls, paired maternal sources, infant follow-up, sequencing resolution, and biological interpretation are aligned before the first sample is processed.

Frequently Asked Questions (FAQ)

References

1. Hou S, Jiang Y, Zhang F, et al. Unveiling early-life microbial colonization profile through characterizing low-biomass maternal-infant microbiomes by 2bRAD-M. Frontiers in Microbiology. 2025;16:1521108. [DOI: 10.3389/fmicb.2025.1521108]
2. Kennedy KM, de Goffau MC, Perez-Muñoz ME, et al. Questioning the fetal microbiome illustrates pitfalls of low-biomass microbial studies. Nature. 2023;613(7945):639–649. [DOI: 10.1038/s41586-022-05546-8]
3. Klopp J, et al. Meconium Microbiome of Very Preterm Infants across Germany. mSphere. 2022;7(1):e00808-21. [DOI: 10.1128/msphere.00808-21]
4. Sun Z, Huang S, Zhu P, et al. Species-resolved sequencing of low-biomass or degraded microbiomes using 2bRAD-M. Genome Biology. 2022;23:36. [DOI: 10.1186/s13059-021-02576-9]
5. Asnicar F, Manara S, Zolfo M, et al. Studying Vertical Microbiome Transmission from Mothers to Infants by Strain-Level Metagenomic Profiling. mSystems. 2017;2(1):e00164-16. [DOI: 10.1128/mSystems.00164-16]
6. Davis NM, Proctor DM, Holmes SP, Relman DA, Callahan BJ. Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data. Microbiome. 2018;6:226. [DOI: 10.1186/s40168-018-0605-2]

For Research Use Only (RUO). Not for use in diagnostic procedures.