Micronutrient supplements with iron promote disruptive protozoan and fungal communities in the developing infant gut

TL;DR: In this article, the authors analyzed the bacterial, protozoal, fungal and helminth communities of stool samples collected from children that had previously been recruited to a cluster randomized controlled trial of micronutrient supplementation in Pakistan.

Abstract: Supplementation with micronutrients, including vitamins, iron and zinc, is a key strategy to alleviate child malnutrition. However, adverse events resulting in gastrointestinal disorders, largely associated with iron, has resulted in ongoing debate over their administration. To better understand their impact on gut microbiota, we analysed the bacterial, protozoal, fungal and helminth communities of stool samples collected from children that had previously been recruited to a cluster randomized controlled trial of micronutrient supplementation in Pakistan. We show that while bacterial diversity was reduced in supplemented children, vitamins and iron may promote colonization with distinct protozoa and mucormycetes, whereas the addition of zinc ameliorates this effect. In addition to supplements, residence in a rural versus urban setting is an important determinant of eukaryotic composition. We suggest that the risks and benefits of such interventions may be mediated in part through eukaryotic communities, in a manner dependent on setting.

Summary

Introduction

• Malnutrition is a global health crisis with 149 million children stunted and 45 million children wasted under the age of five years 1, 2 .
• Previous studies have demonstrated the role of gut microbiota in malnutrition, with microbiome immaturity (bacterial communities that are underdeveloped with respect to age) representing a key factor in disease development 5, 6 .
• Indole-producing gut bacteria were found to confer protection against Cryptosporidium infection, while deworming treatments targeting helminth endemic communities reduced abundance of protective Clostridiales 9, 10 .
• While beneficial, supplementation, especially with iron, may also promote unintended pathogen growth, particularly where the host is unable to restrict micronutrient bioavailability 21 .
• The trial was designed to investigate the impact of micronutrient powders (MNP) containing vitamins and iron with or without zinc on growth and morbidity, and has shown an excess of significant diarrheal and dysenteric episodes among children receiving MNPs 26 .

Description of cohort

• A total of 80 children (160 paired stool samples at 12 and 24 months of age) from all three supplementation arms in the parent cRCT 26 (control (n=24), MNP (n=29), and MNP with zinc (n=27)) conducted in Sindh, Pakistan were selected based on sample availability for inclusion in this study (Supplementary Fig. 1 ).
• The cohort includes children from both urban (Bilal colony) and rural (Matiari district) study sites (Fig. 1a ).
• Children were stratified by weight-for-length zscores (WLZ) at 24 months into a reference WLZ (WLZ >-1) or undernourished (WLZ < -2) group.
• The WLZ growth trajectories of the children selected as the reference WLZ group approximately tracked the upper 50th percentile of the original cohort, while the undernourished group started around the lower 50th percentile and gradually dropped over time ending at the bottom 80 th percentile of the cohort (Fig. 1b ).
• This drop in the WLZ of the undernourished children was driven by poor weight gain (Supplementary Fig.

The developing infant gut is colonized by complex eukaryotic communities

• The authors applied 18S rRNA amplicon sequencing to profile the eukaryotic communities in all 160 stool samples.
• From these the authors identified a total of 859 eukaryotic OTUs (median 66; Supplementary Table 1 ), which included 438 protozoan, three helminth and 418 fungal OTUs (Fig. 2a ).
• Fungi, dominated by Mucoromycota and Ascomycota, accounted for 71% of all reads.
• Carriage of multiple species was common in both the reference WLZ and undernourished groups, with on average 18 and 19 OTUs per child at each time point, and a maximum of 91.
• A high detection of gregarines, typically considered parasites of invertebrates, has not previously been reported in the human gut.

MNPs may destabilize microbial interactions in undernourished infants.

• Microbial interaction networks were constructed to define significant taxonomic co-occurrences (Fig. 4 ).
• The authors found that interactions, calculated as edges per node, increased with age irrespective of treatment, nutritional status or place of residence, which reflects the development of more complex communities as the child matures (Fig. 4a ).
• The greatest change, with a 2.5-fold increase, was noted in children in the MNP arm, which had the fewest taxon interactions at 12 months but achieved parity with the control and MNP with zinc groups by 24 months.
• Within control groups, the microbial networks of undernourished infants and those within the reference WLZ group had similar levels of connectivity, with non-significant differences in degree distribution and betweenness centrality scores.
• Low subject numbers precluded us from successfully generating networks at 24 months, where numbers of microbial taxa are greater.

Discussion

• Malnutrition, both undernutrition and obesity, are associated with altered bacterial compositions, where in the former, underdeveloped bacterial communities have the capacity to induce weight loss 6, 28 .
• An important cause of infant mortality and childhood malnutrition, Cryptosporidium infection is attributed to unsafe drinking water and inadequate sanitation often associated with rural settings 26, 31 .
• While deficiency in zinc has been associated with impaired immune function and susceptibility to enteroinfections 44 , supplementation in the context of enteric pathogens was shown to have parasite-specific outcomes.
• The authors did not detect this bacterium in their data, possibly due to exclusion of diarrheal samples.
• The impact of micronutrient supplementation also extended to the structure of the microbial communities.

Micronutrient interventions may impact undernourished children differently in the context of a

• Prevalence was nearly two-fold higher at the rural site, consistent with their findings for Cryptosporidium, emphasizing the need for location-specific investigations of the effects of micronutrient supplements.
• In addition to potential intraspecies variation, their detection of high sequence diversity in Cryptosporidium OTUs specifically, and eukaryotic taxa in general, may be exaggerated by a high proportion of non-overlapping amplicon reads, a consequence the authors have attempted to minimize through manual curation.
• Furthermore, the prediction from their model that complex cross-kingdom interactions may influence gut bacterial composition, provides a valuable framework to dissect the direct and indirect effects of eukaryotic infections or nutritional interventions on the maturing gut microbiome.
• Given the current debate over the use of MNP supplementation and its role in gastrointestinal disorders, such a framework is expected to play a key role in identifying scenarios where MNP supplementation may require more cautious thinking.

Conclusion

• This study demonstrates that micronutrient powders impact the infant microbiota, with potentially destabilizing effects driven through the promotion of specific organisms during early stages of microbiome development.
• These findings are of relevance to micronutrient supplementation strategies, especially those targeting vulnerable children in low resource settings.

Figures

Fig. 5. Graphic representation of the cross-associations among demographic variables, 838 micronutrient supplementation and microbiota over time. Interconnected arrows indicate the tested 839 cross-correlated paths between nodes of: place of residence (site), supplementation, and the 840 composite measures of bacterial and eukaryotic OTUs detected at 12 and 24 months, collapsed as 841 latent PLS-scores. Negative correlations are indicated in pink and positive in blue. Arrow thickness 842 is weighted by the effect size of the direct path coefficients as indicated in Supplementary Table 843 5. Significance of direct paths, *p < 0.05, **p < 0.01, ***p < 0.0001. OTUs that loaded positively 844 (>0.4) or negatively (<-0.4) within each PLS-score are listed within boxes. PLS, partial least 845 square; OTUs, operational taxonomic units. 846 847 848

Fig. 2. Eukaryotic communities in the gut are diverse and impacted by micronutrient 780 supplementation and place of residence. (a) Phylogenetic tree representing eukaryotic taxa 781 detected in children. Branches are coloured by phylum and bars represent the prevalences of OTUs 782 in the cohort. Named organisms represent those detected in more than 5% of samples with a 783 minimum of 100 reads. (b) Prevalences of protozoan (left), and specifically gregarine (middle) or 784 coccidian (right) OTUs detected in children at 12 and 24 months of age. Prevalences are subdivided 785 by nutritional group in bottom graphs, where shaded regions denote binned numbers of OTUs 786 identified per sample. (c) Rarefaction curves comparing the mean protozoan and fungal species 787

Fig. 3. Bacterial microbiota change with age and supplementation. (a) Relative abundances of 801 bacterial phyla in 12 (top) and 24 (bottom) month old children based on 16S data. Samples are 802 sorted by the proportion of Firmicutes along the horizontal axis. (b) Rarefaction curves comparing 803 mean species richness by age group, micronutrient supplementation, nutritional status and place 804 of residence (study site). Shaded regions represent standard errors and the dotted lines denote the 805 read depth at which significance was tested. (c) Non-metric multidimensional scaling of bacterial 806 compositions in samples based on Bray-Curtis dissimilarities. Samples are coloured by age and 807 ellipses represent 95% confidence intervals. Arrows indicate the direction of cosines of the top 10 808 bacterial OTUs significantly correlated with the ordination axes, and are scaled by their strength 809 of correlation (r2). (d) Mean DESeq2-transformed abundance of Actinobacteria and Firmicutes 810 grouped by nutritional status and treatment. (e) Compositional variance among samples grouped 811 by supplementation arm and age measured as distances to centroid, based on NMDS of weighted 812 Unifrac dissimilarity scores. *p < 0.05, **p < 0.01, ***p < 0.001 (f) Venn diagram showing the 813 numbers of bacterial taxa with significantly increased or decreased abundance, as indicated by 814 arrows, in supplemented groups relative to the control group. The pairs of numbers within brackets 815 refer to taxa at 12 and 24 months of age respectively, and select taxa are listed in boxes. (g) 816

Fig. 1. Participant characteristics. (a) Level of childhood undernutrition in Pakistan and the 771 surrounding regions. Latest country data was retrieved from www.who.int/data/gho/indicator-772 metadata-registry/imr-details/27 on Feb 1, 2021. Urban and rural places of residence of the 773 participants are indicated. (b) Weight-for-length z-scores of children recruited into clinical trial 774 NCT00705445 during the first 24 months of life. Median and quantile values are shown, with 775 medians for participants profiled in current study indicated by red (undernourished) and black 776 (reference WLZ) lines. 777 778

Fig. 4. Supplementation influences microbial interactions. (a) Density of microbial interactions, 822 calculated as significant correlations among microbiota (edges) normalized by the numbers of taxa 823 (nodes), by nutritional status, supplementation arm and place of residence (site). Lighter and darker 824 hues represent samples from 12 and 24 months respectively. (b) Proportions of significant 825 microbial interactions occurring cross-kingdom, within indicated sample groups. (c) Degree 826 distribution and (d) betweenness centrality scores of microbial networks in 12 month old children 827 grouped by nutritional status and supplementation arm. (e) Graphic representations of 828

Table 1. Participant characteristics. Categorical values are presented as n (%), continuous 762 variables show the mean and 95% confidence intervals. Premature birth was defined as 763 gestational age < 37 months. Initiation of breastfeeding was reported for the period prior to 764 recruitment into the study. 765 766

Full text

1
Title: Micronutrient supplements with iron promote disruptive protozoan and fungal
1
communities in the developing infant gut
2
3
Ana Popovic
1,2
, Celine Bourdon
3,4
, Pauline W. Wang
5,6
, David S. Guttman
5,6
, Sajid Soofi
7
,
4
Zulfiqar A. Bhutta
4,7
, Robert H. J. Bandsma
3,4
, John Parkinson
1,2,8,
* and Lisa G. Pell
4
5
6
1
Program in Molecular Medicine, Hospital for Sick Children Research Institute
7
2
Department of Biochemistry, University of Toronto, Toronto, Ontario, Canada
8
3
Division of Gastroenterology, Hepatology and Nutrition, Hospital for Sick Children, Toronto,
9
Ontario, Canada
10
4
Centre for Global Child Health, Hospital for Sick Children, Toronto, Ontario, Canada
11
5
Department of Cell & Systems Biology, University of Toronto, Toronto, Ontario, Canada
12
6
Centre for the Analysis of Genome Evolution & Function, University of Toronto, Toronto,
13
Ontario, Canada
14
7
Center of Excellence in Women and Child Health, the Aga Khan University, Karachi, Pakistan
15
8
Department of Molecular Genetics, University of Toronto, Toronto, Ontario, Canada
16
17
*To whom correspondence should be addressed:
18
john.parkinson@utoronto.ca
19
20
Keywords: Eukaryotic microbiota; Parasites; Malnutrition; Micronutrient Supplementation;
21
Microbiome
22
.CC-BY 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted July 9, 2021. ; https://doi.org/10.1101/2021.07.06.451346doi: bioRxiv preprint

2
Abstract
23
24
Supplementation with micronutrients, including vitamins, iron and zinc, is a key strategy to
25
alleviate child malnutrition. However, adverse events resulting in gastrointestinal disorders,
26
largely associated with iron, has resulted in ongoing debate over their administration. To better
27
understand their impact on gut microbiota, we analysed the bacterial, protozoal, fungal and
28
helminth communities of stool samples collected from children that had previously been recruited
29
to a cluster randomized controlled trial of micronutrient supplementation in Pakistan. We show
30
that while bacterial diversity was reduced in supplemented children, vitamins and iron may
31
promote colonization with distinct protozoa and mucormycetes, whereas the addition of zinc
32
ameliorates this effect. In addition to supplements, residence in a rural versus urban setting is an
33
important determinant of eukaryotic composition. We suggest that the risks and benefits of such
34
interventions may be mediated in part through eukaryotic communities, in a manner dependent on
35
setting.
36
37
38
.CC-BY 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted July 9, 2021. ; https://doi.org/10.1101/2021.07.06.451346doi: bioRxiv preprint

3
Introduction
39
Malnutrition is a global health crisis with 149 million children stunted and 45 million children
40
wasted under the age of five years
1,2
. With increased vulnerability to infection, undernourished
41
children are at elevated risk of death, not least from diarrheal diseases
3,4
. Previous studies have
42
demonstrated the role of gut microbiota in malnutrition, with microbiome immaturity (bacterial
43
communities that are underdeveloped with respect to age) representing a key factor in disease
44
development
5,6
. Beyond bacterial communities, parasites such as hookworm, Cryptosporidium and
45
Entamoeba have also been associated with severe diarrheal disease and intestinal malabsorption
7,8
.
46
However, much less is known regarding the role of other, potentially commensal, eukaryotic gut
47
microbes in undernutrition. Of particular interest is their ability to interact with and alter bacterial
48
communities. For example, indole-producing gut bacteria were found to confer protection against
49
Cryptosporidium infection, while deworming treatments targeting helminth endemic communities
50
reduced abundance of protective Clostridiales
9,10
. Mouse studies further showed that helminths
51
and protozoa influence bacterial communities by modulating the host immune system
9,11,12
. While
52
the number of published gut microbiome studies have increased rapidly over the last decade, few
53
have explored the composition of eukaryotic gut communities and their potential interactions with
54
bacteria. Previously, we applied 18S rRNA and internal transcribed spacer (ITS) sequence surveys
55
to systematically characterize eukaryotic microbiota in severely malnourished Malawian children,
56
and identified a high prevalence of protozoa, including commensals and pathobionts
13
. We
57
furthermore associated Blastocystis colonization with increased gut bacterial diversity.
58
59
Global health programs targeting vulnerable child populations include the use of micronutrient
60
supplements, consisting of vitamins as well as essential minerals zinc and iron, that have been
61
.CC-BY 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted July 9, 2021. ; https://doi.org/10.1101/2021.07.06.451346doi: bioRxiv preprint

4
demonstrated to improve growth and reduce morbidity
14-16
. Such supplements are thought to
62
address deficiencies that can impair immune responses to infectious pathogens and impact gut
63
bacterial communities
17-20
. While beneficial, supplementation, especially with iron, may also
64
promote unintended pathogen growth, particularly where the host is unable to restrict
65
micronutrient bioavailability
21
. For example, it has been shown that surplus iron promotes the
66
growth of enteropathogens and induces intestinal inflammation in infants
22,23
. Furthermore, while
67
known to reduce the duration of childhood diarrheal episodes, zinc supplementation has been
68
associated with increased duration of Entamoeba histolytica infections
24,25
.
69
70
In an attempt to understand the impact of micronutrient supplementation on the complex
71
interactions between eukaryotic and bacterial microbiota in the maturing infant gut and health, we
72
performed 18S rRNA and 16S rRNA amplicon surveys on stool samples obtained at 12 and 24
73
months of age from 80 children, previously recruited as part of a cluster randomized trial conducted
74
in Pakistan. The trial was designed to investigate the impact of micronutrient powders (MNP)
75
containing vitamins and iron with or without zinc on growth and morbidity, and has shown an
76
excess of significant diarrheal and dysenteric episodes among children receiving MNPs
26
.
77
Microbial profiles were analysed in the context of supplementation, nutritional status, age and
78
place of residence (i.e., urban or rural) to reveal a complex landscape of associations with microbial
79
diversity, as well as specific taxa.
80
81
Results
82
Description of cohort
83
.CC-BY 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted July 9, 2021. ; https://doi.org/10.1101/2021.07.06.451346doi: bioRxiv preprint

5
A total of 80 children (160 paired stool samples at 12 and 24 months of age) from all three
84
supplementation arms in the parent cRCT
26
(control (n=24), MNP (n=29), and MNP with zinc
85
(n=27)) conducted in Sindh, Pakistan were selected based on sample availability for inclusion in
86
this study (Supplementary Fig. 1). The cohort includes children from both urban (Bilal colony)
87
and rural (Matiari district) study sites (Fig. 1a). Children were stratified by weight-for-length z-
88
scores (WLZ) at 24 months into a reference WLZ (WLZ >-1) or undernourished (WLZ < -2) group.
89
Subject characteristics are summarized in Table 1. The WLZ growth trajectories of the children
90
selected as the reference WLZ group approximately tracked the upper 50th percentile of the
91
original cohort, while the undernourished group started around the lower 50th percentile and
92
gradually dropped over time ending at the bottom 80
th
percentile of the cohort (Fig. 1b). This drop
93
in the WLZ of the undernourished children was driven by poor weight gain (Supplementary Fig.
94
2).
95
96
The developing infant gut is colonized by complex eukaryotic communities
97
We applied 18S rRNA amplicon sequencing to profile the eukaryotic communities in all 160 stool
98
samples. We generated a total of 11,639,233 paired 18S rRNA amplicon sequence reads (median
99
70,642) of which 4,386,494 could be classified as a eukaryotic microbe (median 22,932;
100
Supplementary Table 1). From these we identified a total of 859 eukaryotic OTUs (median 66;
101
Supplementary Table 1), which included 438 protozoan, three helminth and 418 fungal OTUs (Fig.
102
2a). Fungi, dominated by Mucoromycota and Ascomycota, accounted for 71% of all reads. The
103
most abundant were species in the Candida-Lodderomyces clade, Saccharomyces, and taxa
104
increasingly associated with rare but fatal infections known as mucormycoses: Rhizomucor,
105
Actinomucor and Lichtheimia. Alveolates accounted for 25% of reads, with
106
.CC-BY 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted July 9, 2021. ; https://doi.org/10.1101/2021.07.06.451346doi: bioRxiv preprint