Strategies for understanding dynamic, personalized profiles of host-derived proteins and microbes from human stool
Measuring host proteins through noninvasive stool-based assays opens new avenues for characterizing states of gastrointestinal health. However, the extent to which these proteins vary over time and between healthy subjects is poorly characterized. Here, we characterize technical and biological sources of variability in mass spectrometry-based measurements of host proteins in stool. We identify the proteins that most vary over time within an individual, and among different individuals. Finally, we examine and compare temporal and inter-individual variation in host protein and bacterial taxonomic profiles of the same fecal specimens. To address these issues, five self-reported healthy individuals were each sampled eight times over four weeks. First, we demonstrate that mass spectrometry-based identification and label-free quantification of stool proteins exhibit non-significant variability (p>0.05) between both technical and preparative replicates for a subset of 78 proteins, supporting the utility of this method for biomarker measurement. Second, although 13 human stool proteins varied significantly in relative abundance over time within individuals, 58 proteins varied significantly (at least four-fold) between subjects. The average pair-wise difference between individuals was greater than the average within-subject difference for both the proteome and microbiome datasets (p<0.0001). Fecal host proteins, like the traditional fecal protein marker, calprotectin, unambiguously pointed to innate and adaptive immune responses. For example, one subjects fecal protein profile suggested a sub-clinical inflammatory state. From these data, we conclude that host-centric protein measurements in stool reveal a wide range of variation during states of apparent health, and add a valuable complementary insight into host-microbiota relationships. IMPORTANCEHuman proteins in stool hold untapped potential for characterizing gastrointestinal health. To fully harness this potential and create a baseline of healthy stool protein abundances and identifications, it will be important to establish the extent to which these proteins might vary in the absence of disease. This study quantifies the major sources of variation in stool protein abundance data. We assessed technical, preparative, temporal, and inter-subject variability of human protein abundances in stool and found that among these sources, differences between subjects accounted for the greatest amount of variation, followed by temporal differences, and then technical factors. Our paired microbiome analysis found matching patterns of temporal and inter-subject variability. By characterizing multiple variance parameters in host stool protein abundances, our analysis helps to contextualize a wide range of future disease-focused stool studies as well as elucidate host-microbe interactions.