Experimental inheritance of antibiotic acquired dysbiosis affects host phenotypes across generations

Abstract

Microbiomes can enhance the health, fitness and even evolutionary potential of their hosts. Many organisms propagate favorable microbiomes fully or partially via vertical transmission. In the long term, such co-propagation can lead to the evolution of specialized microbiomes and functional interdependencies with the host. However, microbiomes are vulnerable to environmental stressors, particularly anthropogenic disturbance such as antibiotics, resulting in dysbiosis. In cases where microbiome transmission occurs, a disrupted microbiome may then become a contagious pathology causing harm to the host across generations. We tested this hypothesis using the specialized socially transmitted gut microbiome of honey bees as a model system. By experimentally passaging tetracycline-treated microbiomes across worker ‘generations’ we found that an environmentally acquired dysbiotic phenotype is heritable. As expected, the antibiotic treatment disrupted the microbiome, eliminating several common and functionally important taxa and strains. When transmitted, the dysbiotic microbiome harmed the host in subsequent generations. Particularly, naïve bees receiving antibiotic-altered microbiomes died at higher rates when challenged with further antibiotic stress. Bees with inherited dysbiotic microbiomes showed alterations in gene expression linked to metabolism and immunity, among other pathways, suggesting effects on host physiology. These results indicate that there is a possibility that sublethal exposure to chemical stressors, such as antibiotics, may cause long-lasting changes to functional host-microbiome relationships, possibly weakening the host’s progeny in the face of future ecological challenges. Future studies under natural conditions would be important to examine the extent to which negative microbiome-mediated phenotypes could indeed be heritable and what role this may play in the ongoing loss of biodiversity.

Full text

Frontiers in Microbiology 01 frontiersin.org
Experimental inheritance of
antibiotic acquired dysbiosis
aects host phenotypes across
generations
Vienna Kowallik
1
*, Ashutosh Das
2,3 and
Alexander S. Mikheyev
1,2*
1 Okinawa Institute of Science and Technology, Tancha Onna-son, Okinawa, Japan, 2 Australian
National University, Canberra, ACT, Australia, 3 Chattogram Veterinary and Animal Sciences
University, Khulshi, Chattogram, Bangladesh
Microbiomes can enhance the health, fitness and even evolutionary potential
of their hosts. Many organisms propagate favorable microbiomes fully or
partially via vertical transmission. In the long term, such co-propagation
can lead to the evolution of specialized microbiomes and functional
interdependencies with the host. However, microbiomes are vulnerable to
environmental stressors, particularly anthropogenic disturbance such as
antibiotics, resulting in dysbiosis. In cases where microbiome transmission
occurs, a disrupted microbiome may then become a contagious
pathology causing harm to the host across generations. We tested this
hypothesis using the specialized socially transmitted gut microbiome of
honey bees as a model system. By experimentally passaging tetracycline-
treated microbiomes across worker ‘generations’ we found that an
environmentally acquired dysbiotic phenotype is heritable. As expected,
the antibiotic treatment disrupted the microbiome, eliminating several
common and functionally important taxa and strains. When transmitted,
the dysbiotic microbiome harmed the host in subsequent generations.
Particularly, naïve bees receiving antibiotic-altered microbiomes died
at higher rates when challenged with further antibiotic stress. Bees with
inherited dysbiotic microbiomes showed alterations in gene expression
linked to metabolism and immunity, among other pathways, suggesting
effects on host physiology. These results indicate that there is a possibility
that sublethal exposure to chemical stressors, such as antibiotics, may
cause long-lasting changes to functional host-microbiome relationships,
possibly weakening the host’s progeny in the face of future ecological
challenges. Future studies under natural conditions would beimportant to
examine the extent to which negative microbiome-mediated phenotypes
could indeed beheritable and what role this may play in the ongoing loss
of biodiversity.
KEYWORDS
microbiome, antibiotics, honey bees, experiments, dysbiosis, transgenerational
eects
TYPE Original Research
PUBLISHED 01 December 2022
DOI 10.3389/fmicb.2022.1030771
OPEN ACCESS
EDITED BY
Erick Motta,
University of Texas at Austin, UnitedStates
REVIEWED BY
Guan-Hong Wang,
Chinese Academy of Sciences (CAS), China
Margaret Thairu,
University of Wisconsin-Madison,
UnitedStates
*CORRESPONDENCE
Vienna Kowallik
vienna.kowallik@forento.uni-freiburg.de
Alexander S. Mikheyev
alexander.mikheyev@anu.edu.au
SPECIALTY SECTION
This article was submitted to
Microbial Symbioses,
a section of the journal
Frontiers in Microbiology
RECEIVED 29 August 2022
ACCEPTED 24 October 2022
PUBLISHED 01 December 2022
CITATION
Kowallik V, Das A and Mikheyev AS (2022)
Experimental inheritance of antibiotic
acquired dysbiosis aects host phenotypes
across generations.
Front. Microbiol. 13:1030771.
doi: 10.3389/fmicb.2022.1030771
COPYRIGHT
© 2022 Kowallik, Das and Mikheyev. This is
an open-access article distributed under
the terms of the Creative Commons
Attribution License (CC BY). The use,
distribution or reproduction in other
forums is permitted, provided the original
author(s) and the copyright owner(s) are
credited and that the original publication in
this journal is cited, in accordance with
accepted academic practice. No use,
distribution or reproduction is permitted
which does not comply with these terms.

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 02 frontiersin.org
Introduction
e Anthropocene provides many novel selection pressures
on organisms, such as climate change and the application of
agrochemicals and antibiotics (Sánchez-Bayo and Tennekes, 2017;
Cavicchioli etal., 2019). Organisms respond in various ways to
these pressures, ranging from the evolution of resistance to
extinction. When animals are exposed to nutritional disturbance
(e.g., by chemicals), in addition to potential direct eects on the
organism itself, their gut microbiome may beaected. Dwelling
at the interface between host epithelia and the external
environment, microbial symbionts (microbiomes) can aect host
health by inuencing traits such as nutrition, immunity and
behavior (Round and Mazmanian, 2009; Flint et al., 2012;
Tremaroli and Bäckhed, 2012). Microbial communities can
change rapidly in composition or in gene-expression patterns
when responding to ecological forces. erefore, a microbiome
can extend host evolutionary potential and may facilitate rapid
host acclimation to environmental change (Alberdi etal., 2016;
Henry et al., 2021). Specic gut microbial communities can
provide hosts with novel functions, such as mediating insecticide
resistance (Kikuchi etal., 2012; Wang etal., 2020) or promoting
tolerance to thermal stress (Zare etal., 2018; Zhang etal., 2019;
Raza etal., 2020). Such microbial rescue eects have the potential
to stabilize host dynamics and may explain population persistence
in changing environments (Mueller etal., 2020). Due to the wide
range of functional benets they provide, microbiomes are oen
tightly curated by the host, for example by management and
vertical transmission between generations (Foster etal., 2017;
Rosenberg and Zilber-Rosenberg, 2021). In general, transmission
of microbiomes across generations will transmit the community
and its associated functions – which may bepositive or negative
for the host depending on the conditions.
Indeed, a microbiome is not always benecial for the host.
Some organisms even completely lack it (Hammer etal., 2019) and
the functional benet provided by a microbiome may also
be dependent on environmental conditions. For example
experiments in mice show that adapted microbiomes eciently
harvest energy from food but causing obesity in recipient
individuals when being transferred (Turnbaugh et al., 2006).
While such eciency may bebenecial under food restriction, it
could lead to health problems in times of plenty. Importantly,
evolved cooperation between hosts and symbionts can result in
wide reciprocal functional inter-dependencies. In such cases,
disturbances to the microbiome can compromise host health and
development by, e.g., loss of important microbiome-mediated
functions, or microbial production of harmful substances as a
response to environmental change (Littman and Pamer, 2011;
Soen, 2014). As a result, vertical transfer of such sub-optimal
microbiomes could compromise host health transgenerationally.
Hypothetically, in extreme cases, a host population that is unable
to escape a mal-adapted microbiome may face extinction.
Dysbiotic (dened by a loss of benecial microbes, expansion
of pathobionts or loss of diversity of the healthy, homeostatic gut
condition (Petersen and Round, 2014)) parental microbiomes can
aect the microbiome composition and phenotypes of ospring
across systems. For example, female mice inoculated with
antibiotic-disturbed microbiomes will transfer this dysbiosis to
the ospring causing enhanced colitis (Schulfer etal., 2018). In
sh, chemical exposure causes dysbiosis which persists in F1
ospring with correlating intestinal problems (Chen etal., 2018)
and even result in alterations in the F2 intestinal epigenome,
transcriptome and morphology (Guzman, 2021). Diet induced
microbiome changes modulate transgenerational cancer risk in
mice (Poutahidis etal., 2015). In addition, another interesting
study in ies showed antibiotic-mediated depletion of a
commensal bacterial genus can cause non-Mendelian,
transgenerational inheritance of a stress-induced phenotype
(Fridmann-Sirkis etal., 2014).
By their design, antibiotics pose particular threats to
microbiomes. Antibiotic pollution is omnipresent in ecosystems
due to heavy usage in medicine and agriculture (Kraemer etal.,
2019) and they are known to decrease microbial diversity, to
compromise host-microbiome interactions, to weaken immune
system homeostasis (Modi etal., 2014) and impair colonization
resistance (Bäumler and Sperandio, 2016). Still so far, the focus in
most studies on stress factor eects on microbiomes usually lays
on immediate eects during an individual’s life (Francino, 2016),
and in such cases direct eects of stressors on the host cannot
clearly be disentangled from indirect eects via a disturbed
gut microbiome.
Here, weset out to examine whether the deleterious eects of
a disrupted microbiome can persist transgenerationally, using
honey bees as a tractable model system. Honey bee microbiomes
are socially transmitted between worker ‘generations’, whereby
newly eclosed workers acquire microbiomes from their colony-
mates and the direct hive environment. While this is a dierent
vertical transmission approach from the classical parent-to-
ospring one, it was successfully leading to strong co-evolution
between corbiculate bees and their microbiomes (Koch etal.,
2013; Kwong et al., 2017). e adult honey bee microbiome
consists of ~8 bacterial phylotypes that are involved in key
biological functions such as nutrition, digestion, and immunity
(Engel etal., 2016; Emery etal., 2017; Kešnerová etal., 2017;
Raymann and Moran, 2018). Because young adults emerge from
pupation without a microbiome, they can reliably beinoculated
with a microbiome of choice in the lab (Powell etal., 2014; Zheng
etal., 2018; Kowallik and Mikheyev, 2021). us, it is possible to
serially transfer microbiomes across worker ‘generations’ to study
how microbial changes in response to environmental stressors
aect host phenotypes and health. In addition, honey bees are
important pollinators and are exposed to diverse chemicals in the
agricultural landscape as well as by beekeepers. It could beshown
that antibiotics have strong eects on the honey bee microbiome
(Powell etal., 2021; Tian etal., 2012; Moullan etal., 2015; Li etal.,
2017; Raymann etal., 2017; Baoni etal., 2021; Jia etal., 2022) and
that such dysbiosis can even be experimentally transferred
between workers (Jia etal., 2022).

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 03 frontiersin.org
In our study weused controlled lab experiments passaging
microbiomes aected by antibiotics from one worker cohort to the
next and examined mediated eects on host physiology by exposing
naïve bees receiving these microbiomes to high levels of antibiotic
stress. is design allowed us to isolate changes in the microbiome
from host responses and from environmental changes. Wefound
that the microbiome was disturbed aer antibiotic exposure leading
to compositional and functional changes. ese were both
transmitted to subsequent host generations, leading to some changes
in host gene expression and to high mortality under stress.
Materials and methods
To test how honey bee microbiomes respond under antibiotic
pressure and how this aects host phenotypes across generations,
we conducted experiments in which microbiomes were
transferred over two host cycles (worker “generations”) under
sub-lethal chemical administration. In the third cycle, to examine
whether past chemical exposure aects host survival, weapplied
lethal levels of the chemicals to which prior “generations” had
been exposed. We quantied changes in both host gene
expression and microbial composition using RNA-seq and 16S
amplicon sequencing, respectively.
Experimental setup
e rst experiment (Figure1) was conducted in February/
March 2019 at Australian National University in Canberra,
Australia. See also the Supplementary information for more
methodological details. e same, chemically untreated Apis
mellifera ligustica colony was used throughout the whole
experiment to avoid host genetic background changes. Westarted
with a cohort of microbiome depleted individuals of the same age
in each cycle. Late-stage pupae (dark eyes but lacking movement)
were carefully removed from brood frames and allowed to develop
under sterile conditions in the lab. Workers eclosing within 24 h
were randomly distributed into six cages (three independent cages
per treatment with ~25 bees/cage) and provided with lter-
sterilized 0.5 M sucrose solution (Supplementary Figure S1).
When all bees were distributed, the sucrose feeders were replaced
with sterile sucrose or antibiotic-infused sucrose. We used a
tetracycline hydrochloride concentration previously published in
a honey bee microbiome study (450 μg tetracycline / mL sucrose
(Raymann etal., 2017)). Concurrently, 10 nurse bees from the
same hive were surface sterilized, and their dissected hindguts
were macerated in 1:1 PBS/sucrose solution, mixed with gamma-
irradiated bee bread (previously collected from colonies from the
same apiary and then sterilized with 35kGY) and equally
distributed across all six cages. On the following day, the
remaining food was discarded and the microbiome feeding
method was repeated for a second time for 24 h using again 10
nurse bee guts. On both days, small amounts of the microbiome
pools were kept for later determination of the start microbiome.
Aer the inoculation period the bees received sterile pollen and
sucrose with or without antibiotics. Daily, the tetracycline solution
was freshly prepared, dead bees were removed and fresh sucrose
and sterile bee bread were oered ad libitum. Bees were
maintained under these conditions for 6 days in cycle one and 10
days in cycle two, dierences due to the need to have enough
pupae of the same age and hive background ready for the next
cycle. However, the aim was to provide enough time that the
microbiome can befully established. Wepreviously experienced
that when newly emerged bees receive a microbiome pool for 48 h,
they show the full adult bee microbiome in composition and
abundance aer 7 days (Kowallik and Mikheyev, 2021). It is also
known that under natural conditions, adult bees get colonized
within the rst 2 days aer emergence which is followed by rapid
establishment within 4 to 6 days post-eclosion (Powell etal., 2014).
We therefore gave a minimum of 6 days to allow inoculation,
internal growth and establishment of the microbiome. For
microbiome transfer in cycle two the newly emerged bees received
the microbiome from the previous cycle to mimic
FIGURE1
Design of the main experiment. Pupae emerge in the lab and are first inoculated for 48 h with a natural microbiome from hive siblings. Three cages
per treatment were used. Throughout cycle 1 and 2, bees are continuously fed with sterile pollen and sucrose containing tetracycline or not
(control). These exposed and control microbiome communities get passed to the following cycle of lab-emerged bees (cage to cage transfer). In
cycle 3, bees that received control or pre-exposed microbiomes are kept naïve toward the chemical until they are administered high doses of
tetracycline at the end.

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 04 frontiersin.org
generation-spanning microbiome transmission. For this, three
bees in each cage were sacriced, surface sterilized, and their
dissected hindguts were mixed with sterile pollen and
administered to one bee cage of the next cycle for 48 h (cage to
cage transfer provided three independent cage replicates per
treatment). Wealways kept small amounts of these transfer pools
for later sequencing. All other surviving bees in each cage at the
end of cycle 1 and 2, as well as small amounts of the gut transfer
pools were snap-frozen in liquid nitrogen and stored in a − 80° C
freezer until further processing. In the beginning of the third
cycle, control and exposed microbiomes from the previous cycle
were transferred again to newly emerged bees as stated above.
However, in cycle three, all cages received sterile food without
toxins for 6 days. On day six, three individuals per cage were
collected and snap frozen to examine the established microbiome
community (“cycle 3 before stress”) at this time point.
Subsequently, all cages were then challenged with a high dose of
the stressor (20 mg tetracycline per mL sucrose), a concentration
identied to cause 50% mortality in 24 h (LD50) during a pilot
study (see Supplementary Methods). Due to the high mortality in
the “exposed microbiome” cages, wecounted survival aer 20 h,
with the surviving bees (“cycle 3 aer stress”) being snap-frozen
and stored at −80° C until further extractions.
We calculated the survival proportion for each day of the
experiment before high stress application and plotted the mean of
the three cages for both treatments for each cycle with standard
deviations. To compare the control and tetracycline treatment
weperformed two-sided Fisher’s exact tests on alive/dead count
data of the three cages for each day. For statistical analysis of the
nal survival data aer high stress application, weused a Bayesian
logistic regression approach to examine eects of past chemical
exposure on survival in the face of lethal stress levels. To account
for between-cage heterogeneity within treatments, we rst
estimated mortality levels for each cage regardless of treatment
(survival ~ cage) using the brms package (Bürkner, 2017).
We chose standard minimally informative priors and veried
adequate model performance using diagnostic plots and statistics
provided by the package. Wethen tested the hypothesis that cage
mortality coecients were the same in control vs. experimental
treatment, using the brms hypothesis function, which computes
the posterior distribution of the dierence between Bayes factor
levels in the contrast. is approach parallels planned linear
contrasts in regression analysis. In addition, weconducted a
non-parametric analysis using two-sided Fisher’s exact tests on
alive/dead count data (altogether 53 control-gut and 47
tetracycline-gut individuals).
Mechanisms underlying phenotypic
eects of tetracycline-exposed
microbiome transfer
To exclude leover tetracycline or derived by-products inside
the transferred guts as proximal drivers of stress-induced
mortality weran an additional control experiment. In March
2021in Okinawa Japan, westarted the experiment as described
before by graing pupae. Experimental procedures were generally
identical to the previous experiment. Aer sterile emergence, bees
were distributed equally to eight cages with ~28 bees each.
Microbiome transfer from nurse bees of the same hive was done
as before. Four cages received tetracycline and the other sterile
food only. Aer 6 days, the volume of four macerated guts (one
more to account for any loss in the lter) per cage was ltered
using a 0.2 um syringe lter to remove microbial cells. Aer
surface-sterilizing and dissecting 20 nurse bees from the same
colony, wepooled the hindguts to receive a healthy microbiome
pool as base for the next cycle’s bees. is pool was equally split
into eight parts, and each got mixed with the ltered gut solution
of one cage from cycle 1 (Figure2). For the next cycle, this resulted
in four cages of microbiome + ltered control (supernatant of
cycle 1 bee guts receiving sterile food) and four cages of
microbiome + ltered tetracycline-exposed (supernatant of cycle
1 tetracycline-exposed guts) solution. All bees received sterile
food for 6 days and high tetracycline dose on day six. Aer 15 h,
mortality was recorded. e same statistical approach as described
above was used by applying Bayesian logistic regression and
Fisher’s exact tests (N = 4 cages; altogether 60 control-lter-gut and
50 tetracycline-lter-gut individuals).
Extractions and sequencing
For extractions of bees from the rst experiment weused the
Qiagen AllPrep PowerFecal DNA/RNA Kit on abdomens of
frozen bees. Every bee was rst rinsed with ethanol and three
subsequent rinsing steps in sterile water to clean the surfaces and
then the whole abdomen or the microbiome transfers were
processed following the recommended settings of the protocols,
including bead beating using the Geno/Grinder®. DNA was
eluted in 30 μl TE buer. For 16S sequencing weexamined the
microbial community composition of 75 samples. ese were
one sample of start microbiome composed of the nurse
microbiome pool (day 1 and day 2 pooled together), six nurse
bees from the same hive as natural controls, two ZymoResearch
Mock DNA controls, 12 microbiome transfer pools (one for each
cage being composed of three pooled guts) for the cycle to cycle
microbiome transfers in the beginning of cycle 2 and 3 (=24
pools together). In addition, wesequenced 54 individual bee
abdomen from four dierent time points during the experiment
(end of cycle 1 (9 control, 3 tetracycline), end of cycle 2 (9
control, 9 tetracycline), cycle 3 before high tetracycline
application (9 control, 9 tetracycline) and aer (8 control, 1
tetracycline)). Weaimed to sequence three individuals per cage
and time point, however, as the number of sampled individuals
relied on the numbers of bees surviving, minus the ones used for
gut transfer and sometimes a dissection may have gone wrong
or a bee escaped, weended up with fewer numbers of sequenced
samples in some cases.

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 05 frontiersin.org
DNA of samples was submitted to DNA Sequencing Section
at the Ramaciotti Centre for Genomics in Sydney Australia.
Library preparation was performed based on Illumina protocol
with 25-μl reactions. Illumina barcoded primers (Klindworth
etal., 2013) were used to create a single amplicon of approximately
460 bp encompassing the V3-V4 region of bacterial 16S
rRNA. Samples were pooled to equimolar concentration and
sequenced on Illumina MiSeq v3 2 × 300 bp platform. Reads were
demultiplexed on the basis of barcode sequences, allowing for
one mismatch.
16S amplicon sequence analysis
Demultiplexed reads were processed using QIIME2 version
2019.1 (Bolyen et al., 2019), denoising of the fastq les was
performed using the denoise-paired command from the DADA2
soware package (Callahan etal., 2016), wrapped in QIIME2,
including removal of chimeras using the “consensus” method.
Decreased quality scores (below 20) of the sequences at the
beginning to remove primers and end were truncated (trim-
le-f = 17, tr im-le-r = 21, trunc-len-f = 275, trunc-len-r 225). is
resulted in a remaining overlap of ~40 bases in merged sequences.
e result is an amplicon sequence variant (ASV) table, a higher-
resolution analog of the traditional OTU table. For taxonomic
assignment, the QIIME2 q2-feature-classier plugin (Bokulich
etal., 2018) and the Naïve Bayes classier (Wang etal., 2007),
which wetrained with our primers previously, were used on the
SILVA release 132 (Quast etal., 2013; Yilmaz etal., 2014).
All following graphical and statistical comparisons were
performed in R using the phyloseq package (McMurdie and
Holmes, 2013). In short, we rst removed all non-bacterial
sequences, mitochondrial and chloroplast sequences, and ASVs
not present in any sample (likely artifacts) from the datasets using
the subset_taxa and prune_taxa functions. Weplotted rarefaction
curves of all samples using the ranacapa function ggrare
(Kandlikar etal., 2018) on the minimum sample depths (12,351
reads). Alpha diversity of the rareed samples was explored by
plotting Observed species numbers and Shannon’s diversity index.
Pairwise, two-sided Wilcoxon rank sum tests were used to test for
A
B
C
D
E
FIGURE2
Control experiment with filtered gut solution. Emerged bees with transferred natural nurse microbiome are raised in four cages with control or
tetracycline diet for 6 days (A). On day six a bee gut pool for each cage is prepared as done in the first experiment and filtered to exclude microbes
but to keep all potential tetracycline and derivates potentially present in the guts (B). A microbiome pool of hive sibling guts is generated to allow a
healthy background microbiome for newly emerged bees for the next cycle (C). This microbiome pool is equally split and each part gets mixed
with the filtrate of one control or tetracycline cage (D). The bees receive sterile food for 6 days and are exposed to high tetracycline stress in the
end and mortality is recorded (E).

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 06 frontiersin.org
signicant alpha diversity dierences between treatments in each
cycle. As rarefying sample counts is not recommended, unless
necessary, (McMurdie and Holmes, 2014) weconverted data to
proportions for normalization purposes. On these proportions,
non-metric Multidimensional Scaling (NMDS) was performed on
Bray-Curtis distances for ordination plots.
To test for variation within groups, weused the betadisper
function in the Vegan package, version 2.5–5in R on the Bray-
Curtis distance matrix on proportion data to calculate distances
to group centroids per treatment for each cycle. Subsequently,
output was plotted as ordination for visualization and permutest
was run for each cycle to check for homogeneous distribution of
samples across the two treatments. Multifactor permutational
multivariate analysis of variance (PERMANOVA) on Bray-Curtis
distances with 999 permutations using the ADONIS function
were performed to test for eects of experimental factors on the
gut community. As we sequenced single bees as well as
microbiome transfers aer each cycle, werst tested whether there
is a dierence according to method for each treatment. Wealso
tested for cage eects in the data set in each cycle and treatment.
In addition, wecompared each treatment against the respective
controls for all 3 cycles. Finally, wetested whether the microbiomes
of each treatment changed across cycles. For taxonomic
visualization weplotted the relative abundances of all genera
accounting for at least 1% of the abundance across treatments and
cycles. We then extracted the seven dominant taxa from the
rareed sample set and plotted their individual, total abundances
across cycles with subsequent two-sided Wilcoxon rank sum tests
between treatments and the respective controls. To further
investigate response variation in species as well as ASV level (also
see Supplemental results for more details), wepooled all cycles
aer checking that no cycle-specic dierences could beobserved
and extracted the abundant species for each core genus (>1,000
reads) and plotted their abundances across the two treatments.
We used online megablast against the full NCBI Nucleotide
collection database on abundant ASVs (>1,000 reads) for each
genus for better taxonomic resolution (sequences and alignment
output in supplements). Similarly, we also plotted the total
abundances of ASVs across the two treatments.
RNA-sequencing and analysis
To understand the molecular basis of physiological eects that
the microbiome’s antibiotic treatment history has on hosts,
weconducted RNA-sequencing of six honey bees in cycle 3 before
high stress application. Wesequenced one individual per cage
(three per treatment), comparing bees with tetracycline-stressed
and control microbiomes.
For RNA library preparation, the QIAseq® Stranded mRNA
Select Kit was used following the standard protocol. Sequencing
was done on a Nextseq2000 with V2 75 cycles (75-bp Single
Read). Reads were quantied using kallisto (Bray etal., 2016) with
the honey bee transcriptome (version Amel_HAv3.1) as a
reference, using default parameters. e R package DESeq2 was
used to normalize and determine which genes were dierentially
expressed among control and treatment samples, setting the
control group as reference to becompared against. Genes were
considered dierentially expressed at an FDR adjusted value of p
<0.05. To visualize the dierences in expression prole between
the samples, the plotPCA function in DESeq2 was used to generate
principal component analyses. MA plots visualizing base-2 log
fold-change (LFC) (y-axis) versus normalized mean expression
(x-axis) in the tetracycline treatment against the control were
plotted using the ggmaplot function on previously shrinked eect
sizes using the lfcShrink function for better visualization and
ranking of genes. To study the amount by which each of the
signicantly dierent determined genes deviates in a specic
sample from the gene’s average across all samples wecreated a
heatmap using the pheatmap function on regularized logarithm
rlog() transformed data. Gene ontology (GO) enrichment analysis
of the signicantly dierentially expressed genes were carried out
using GOstats, GSEABase and Category R packages (Falcon and
Gentleman, 2007). Biological processes associated with these GO
terms were summarized and visualized using REVIGO (Supek
etal., 2011).1 e semantic similarity was measured using the
Resnik’s measure (SimRel) (Resnik, 1999) and the threshold used
was C = 0.7 (medium). e results were then used to produce a
scatter plot using the ggplot2 package in R.
Results
Microbiomes aect bee immunity and
survival under high toxin stress
Bee guts were transferred three times to new hosts aer
exposure to sub-lethal doses of tetracycline. Bee survival
during the 3 cycles showed higher mortality under tetracycline
in all cycles in comparison to respective control
(Supplementary Figure S2). At the end of these transfers, in cycle
3, naïve recipient bees were given lethal doses of the tetracycline.
Survival was compared between bees receiving chemical-exposed
microbiomes and those receiving unexposed control microbiomes.
e microbiomes with previous tetracycline exposure signicantly
decreased the survival of the host bees (Bayes Factor (BF)
comparing survival in control vs. dysbiotic treatments 95% CI
-26.84 – −4.38) (−34% survival, p < 0.001, Fisher’s exact test on
alive/dead count data) (Figure3).
We further experimentally investigated if the microbiome
itself or rather tetracycline residues inside the transferred guts
aected the bee survival. Wefound no support for the latter
hypothesis, as the ltered gut solutions did not decrease survival
under high stress (BF 95% CI -4.14 – 3.88) (Fisher exact test,
p = 0.64).
1 http://revigo.irb.hr

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 07 frontiersin.org
Tetracycline aects the bacterial
community composition
Challenging bees with tetracycline over two cycles
(“worker generations”), aected microbial community
composition. We examined the gut microbial community
composition of 54 individual bees from four dierent time
points during the experiment as well as six hive nurse bees, the
start microbiome, 12 microbiome transfer samples and two
mock DNA controls. e V3-V4 region of the bacterial
16SrRNA gene was amplied and sequenced on the Illumina
MiSeq platform, generating an average of 30,462 reads per
sample (range, 14,253 to 65,293). e total number of ASVs
was reduced from 1717 to 460 aer ltering out mitochondria,
chloroplasts, artifacts and reads not assigning to the kingdom
Bacteria. e two mock community control DNA samples
(ZymoResearch cat D6306) sequenced in this study showed no
qualitative dierences compared to expected theoretical
proportions provided by the mock community manufacturer
(Supplementary Figure S3). ASVs matching non-mock taxa
belonged to honey bee core symbionts but accounted for only
0.23% of the abundance, representing neglectable cross-
contamination during library preparation or sequencing.
Rarefaction plots on the minimum sample count
(Supplementary Figure S4) show quickly reaching converged
lines in all samples, indicating sucient depth. Weobserved
no signicant dierences between whole bee and microbiome
transfer samples for control as well as tetracycline treatments
(PERMANOVA; control: p = 0.52, R2 = 0.03, F = 0.75;
tetracycline: p = 0.55, R
2
= 0.03, F = 0.72). Based on these results
wecontinued analyzing the transfer and bee samples together.
Microbial alpha diversity was much lower in the tetracycline
treated individuals at all time points, as measured with the
Shannon index (Figure 4) and numbers of observed species
(Supplementary Figure S5). is eect could be seen using
Non-metric Multidimensional Scaling (NMDS) with tetracycline-
treated samples being distinct from control samples (Figure4).
PERMANOVA on Bray-Curtis distances identied tetracycline-
stressed microbiomes as being signicantly dierent from controls
(cycle 1: p = 0.003, F = 31.2, R2 = 0.71; cycle 2: p < 0.001, F = 62.5,
R2 = 0.74; c ycle 3; p < 0.001, F = 41, R2 = 0.72). Treatments did show
signicant eects on groups dispersion in cycle 1 (permutest;
p < 0.001, F = 11.4), and 3 (p = 0.01, F = 8.5) but not in cycle 2
(p = 0.64, F = 0.19) (Supplementary Figure S6) indicating that high
dispersion may aect the PERMANOVA statistical output.
At the end of the rst cycle, several bacterial core genera
disappeared from guts of antibiotic-fed bees, namely Frischella,
Bartonella, Snodgrassella and Commensalibacter (Figure5). e
abundances of almost all core symbionts were signicantly
aected by tetracycline (Supplementary Figure S9 and
Supplementary Table S1 for stats). On a ner scale, weobserved
in several bacterial species some ASVs being susceptible to
antibiotic treatment and getting eliminated, while others were
unaected or even increased in relative abundance
(Supplementary Figure S10).
Tetracycline aected microbial
communities aect host gene expression
We sequenced mRNA of one honey bee per cage (three per
treatment and control respectively) in cycle three before high
stress application, with an average of 99.5 million (min 4.8 million,
max 567 million) raw reads. While most of these reads mapped to
bees, the pathogen Nosema could be detected as a higher
percentage of the control reads (0.35, 0.95, 0.11 percent aligned)
in comparison to the tetracycline treated bees (0, 0, 0.04 percent
aligned) in the taxonomy analysis of NCBI on the submitted raw
reads. e pseudoalignment rates of the samples were 64 土
5.1% (s.d.).
Dierential gene expression analysis showed that receiving the
antibiotic-disturbed microbiomes aects host gene expression.
Altogether 30 genes were signicantly dierently expressed
(p > 0.05) aer FDR adjustment for multiple comparisons
(Figure6). Surprisingly, only three genes were down-regulated
and are mainly involved in lipid metabolism such as phospholipase
A2-like (LOC724436) and fatty acyl-CoA reductase 1
(LOC724560). Some of the up-regulated genes have likely
functions in immunity such as apidermin 1 (GeneID_551367) or
lysozyme-like (LOC113218576), transport activities, e.g., NPC
FIGURE3
Past chemical exposure of a microbiome can aect future host
survival. The Bayes factor dierence between treatment and
control groups measures whether survival in treatments was
higher (positive axis) or lower (negative) under a high dose of
tetracycline relative to the respective control group.95%
posterior distribution confidence intervals lying outside zero are
highlighted by asterisks. Transferring tetracycline pre-exposed
guts (N = 3 cages; altogether 53 control-gut and 47 tetracycline-
gut individuals) negatively aected bee health under high stress,
while transferring filtered gut solution together with a healthy
adult microbiome (N = 4 cages; altogether 60 control-filter-gut
and 50 tetracycline-filter-gut individuals) did not aect the
survival.

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 08 frontiersin.org
intracellular cholesterol transporter 2 (LOC724386) and
metabolism such as lipase member H-A (LOC727193) or
chitooligosaccharidolytic beta-N-acetylglucosaminidase
(LOC725178) (Figure 6 and Supplementary Figure S11). As
wesequenced only three individuals per treatment it is important
to be cautious about generalizations. However, while the
tetracycline-pre-treated-gut bees showed more within-group
variation in their expression proles comparison to the control
(Supplementary Figure S12), the signicantly dierent genes
showed relatively similar expression patterns within the two
groups although both coming from three individual cage
communities (Figure6). See additional information such as the
lists of up- and down regulated genes with information on gene
description, GO term and beebase IDs in the Github folder.
Discussion
Considering the worldwide increase in variety and abundance
of anthropogenic stresses together with the loss of biodiversity
(Barnosky etal., 2011, 2012), there is urgent need to understand
all potentially contributing eects. is includes consideration of
interactions between organisms. How microbiomes aect host’s
responses to such selection remains underexplored (Cavicchioli
etal., 2019). Associated microbial symbionts and their functional
relationships with their hosts are sensitive to disturbance. Given
that microbiomes are vertically inherited, wholly or in part, in
many organisms, any changes in composition and associated
second-order eects on organismal health may bepropagated
across generations.
FIGURE4
Gut microbial community composition responds to tetracycline treatment. Alpha- and beta-diversity as well as taxonomy show tetracycline
leading to a strong dysbiosis, decreasing several taxa. Alpha diversity Shannon index accounts for abundance and evenness of ASV in samples.
Pairwise Wilcoxon rank sum tests were used for statistical comparisons between treatments and controls (*** < p 0.001; ** < p 0.01) (cycle 1: W = 36,
p = 0.004; cycle 2: W = 144, p < 0.001; cycle 3 before stress: W = 81, p < 0.001). NMDS on Bray–Curtis dissimilarity which considers presence/absence
as well as abundances of ASVs, represents compositional dierences between samples (beta diversity). Stress of NMDS was 0.069. Ellipses
represent 95% confidence intervals around treatment centroids.

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 09 frontiersin.org
Here, we used controlled lab experiment to show that
deleterious eects of antibiotics on the microbiome can bepassed
across generations and aect host health decoupled from any
direct toxicity of the antibiotic.
Antibiotics reduce microbiome diversity
on genus- species- and strain level
Consistent with the direct action of antibiotics on bacteria,
weobserved substantial changes in the honey bee gut community
aer tetracycline exposure. While in previous studies, antibiotics
were shown to aect the honey bee microbiome (Powell etal.,
2021; Tian et al., 2012; Moullan et al., 2015; Li et al., 2017;
Raymann etal., 2017; Baoni etal., 2021; Jia etal., 2022), it rarely
led to the total collapse of bacterial species as weobserved in our
design. At the end of the rst cycle, four bacterial genera
disappeared from guts of antibiotic-fed bees (Figure 5 and
Supplementary Figure S9). In general, it may be dicult to
compare dierent studies as they dier in methodology. Also,
honey bees used in the studies may dier genetically, in their
surrounding environment and likely in their colony’s chemical
exposure histories which can aect the microbial strain
composition (Tian etal., 2012; Ellegaard and Engel, 2019; Wu
etal., 2021). Low antibiotic intake (10 ug/mL) aer emergence did
not show to aect the later establishment of the microbiome in
honey bees (Jia et al., 2022). However, we used previously
published higher concentrations (Raymann etal., 2017) which
were administered immediately aer emergence, which could
aect the uncolonized gut environment and the overall response
of a microbiome. Cox etal. introduced early life as “critical
developmental window” when antibiotics have greatest impact on
FIGURE5
Taxonomy of bacterial genera across the 3 cycles (cycle three before high stress application) with at least 1% relative abundance across samples
(everything else is combined to “others”) shows several taxa disappearing under tetracycline.

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 10 frontiersin.org
the gut microbiome of mice, leading to lasting metabolic
consequences (Cox etal., 2014). Such time-dependent response
dierence has also been demonstrated in honey bees (Motta and
Moran, 2020), which eclose largely without microbes and then
acquire them from the surrounding environment and nestmates
(Powell etal., 2014). Aer the high tetracycline exposure in cycle
three, the microbiome composition did not change
(Supplementary Figure S7). For the tetracycline-treated microbial
communities this could beexplained by the fact that weselected
for antibiotic resistant strains in the previous cycles, however
wealso did not observe changes in the controls. While this seems
surprising considering the extreme eects of lower antibiotic
dosages in the cycles before, this may becaused by the fact that
werstly applied tetracycline to fully colonized adults in cycle
three and secondly that wesampled 20 h aer antibiotic exposure
and DNA sequencing will also capture dead material. Other
studies also detected a more prominent eect of antibiotics on the
honey bee gut community several days aer treatment was
stopped which may bea result of a delayed eect of the antibiotic
(Raymann etal., 2017). In addition, while 16S sequencing has
limitations when it comes to ne-scale taxonomic identication
(Ellegaard and Engel, 2016), wefound extensive response variation
at the generic, species, and ASV levels (Supplementary Figure S10).
is is consistent with other studies that found eects of antibiotics
(Raymann etal., 2018) and other pesticides (Cuesta-Maté etal.,
2021) vary across bee gut bacterial species and strains. ese data
together with the increase in resistance genes in antibiotic exposed
bee microbiomes (Tian etal., 2012; Sun etal., 2022) indicate
adaptation to chemical selection factors.
Negative eects of antibiotic-disturbed
microbiomes can betransferred to
following generations
In general, perturbations of a healthy gut environment can
aect gene expression, protein activity, and the overall metabolism
of a host associated gut microbiota (Franzosa et al., 2015).
Antibiotic exposure causes dysbiosis, with eects on host health
(Francino, 2016; Neuman et al., 2018), the resistome (genes
involved in resistance responses), and gut bacterial diversity (Li
etal., 2019; Xu etal., 2020). Wefound that short-term dysbiosis
could betransferred to subsequent worker bee generations which
is in line with previous experiments in honey bees and ies (Ourry
et al., 2020; Jia et al., 2022). In our experiment, tetracycline
disrupted the normally stable bee gut community, which did not
recover over subsequent generations even aer antibiotic
administration was ceased. In cycle three only Bartonella could
recover in some samples, while the other antibiotic-aected
genera appeared permanently eliminated from the community
(Figure5). is transmitted dysbiosis was likely the reason of the
higher mortality under subsequent tetracycline stress at the end
of the experiment in naïve bees that inherited the disturbed
microbiome (Figure3).
AB
FIGURE6
Dierential gene expression of genes in naïve bees that received tetracycline-exposed in comparison to individuals that received control
microbiomes. MA plots show the dierential expression of the tetracycline-gut against the control-gut treatment (n = 3 (one bee per cage for both
treatments respectively)) (A). The x axis shows the average expression over the mean of normalized counts, and the y axis shows the gene-wise
dispersion estimate’s shrunken log2 fold change. Red and blue points indicate significant up- or downregulation (FDR ≤ 0.05 determined by
DESeq2) of individual genes. Heatmap on rlog() transformed data shows the expression dierence of each significantly dierent gene in a specific
sample from the gene’s average across all samples. In addition the gene descriptions are shown (B).

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 11 frontiersin.org
Feeding macerated honey bee guts to other bees is an
established method of microbial transfer in laboratory studies
(Powell etal., 2014; Zheng etal., 2018; Kowallik and Mikheyev,
2021). However, since bees do not defecate in captivity, toxins such
as tetracycline could conceivably accumulate in the hindgut. It is
therefore imaginable that small amounts of le tetracycline or
derivates may negatively aect the health of the following worker
bee generation. We excluded this possibility by an additional
experiment transferring tetracycline-exposed guts ltered to
remove bacteria and seeing no eects on mortality (Figures2, 3).
is supports the interpretation that the detrimental eect was
indeed caused by the disturbed microbial community. In general,
we cannot exclude that we also transmitted non-bacterial
pathogens during microbiome transfer in our design which may
aect host health. For the fungal pathogen Nosema a potential
correlation has been reported between infection load and gut
microbiome structure (Rubanov etal., 2019). However, wedo not
see a higher Nosema load in the antibiotic treated bees in our
RNA data but rather the opposite. In addition, none of the
signicant genes in our design are common pathogen-response
genes. e humoral immunity in honeys bees involves synthesis
of antimicrobial peptides (AMPs) from which abaecin, apidaecin,
defensin and hymenoptaecin usually respond to bacterial, viral and
fungal infection (Evans et al., 2006; Chaimanee et al., 2012;
Flenniken and Andino, 2013; Doublet etal., 2017).
Gut bacteria function as a protective barrier, enhancing
nutritional provisioning and aecting the host immune system
across animal systems (Hooper et al., 2012; Tremaroli and
Bäckhed, 2012; Kamada et al., 2013) including honey bees
(Kešnerová et al., 2017; Raymann and Moran, 2018).
Administration of antibiotics has been shown to reduce gene
expression of antimicrobial peptides in bees (Li etal., 2017; Motta
etal., 2022). We observed a signicant up-regulation of genes
having functions in immunity, biotic responses, carbohydrate
metabolism and transport for all kind of molecules (e.g., metal
ion, sodium ion, sterol transport) in bees receiving dysbiotic
microbiomes (see Figure6 and Supplementary Figure S11). Only
three genes showed to bedown-regulated which were mainly
involved in lipid metabolism. As our cycle three bees did not
consume tetracycline themselves, we can conclude that the
dierential gene expression was most likely caused by the
microbial community changes. Changes in community structure
such as those observed in our study can alter the provided
microbiome function such as provision of nutrients or removal of
toxic metabolites across systems (Willing etal., 2011). In general,
interactions between symbionts can be as important as the
individual species in gut microbiomes, therefore the eects of a
disturbed microbiome go far beyond the loss of functions
attributable to single taxa (Gould etal., 2018). In our design, a
disturbed cross-talk between host and microbiome could have
aected host gene expression as the host may have had to
compensate for missing functions. However, as wesequenced only
one bee per cage and the expression of bees receiving tetracycline
pre-exposed microbiomes shows higher within treatment
variation than the control (Supplementary Figure S12) weshould
becautious with generalizations.
The honey bee as model system
Previous work characterized the honey bee microbiome and
developed methods such as articial microbiome transmission
(Engel etal., 2013; Powell etal., 2014; Kwong and Moran, 2015).
Webuilt on this foundation using honey bees as a model to study
stress-induced, microbiome-mediated eects on subsequent
generations. In our experiments weperformed a purely vertical
microbiome transfer between individuals, a rate at the extreme
end of a continuum of strategies. While in most systems microbes
are acquired both vertically and horizontally, high rates of vertical
transfer are typical in honey bees (Engel and Moran, 2013).
We did not provide the opportunity to recruit dierent strains
through the environment or social contact inside the hive which
could have led, for instance, to some recovery from the dysbiotic
state induced by tetracycline or could have led to colonization of
opportunistic pathogens. Although a previous study did not nd
that honey bees with antibiotic-induced dysbiosis recovered their
microbiomes to a healthy state when being put back to the hive
environment and that they also suered from higher mortality in
this natural environment compared to the control (Raymann
etal., 2017). Beside chemically induced changes to the microbiota,
even communities in our control treatment were also gradually
changing in the lab. For instance, weobserved an increase of
Bartonella abundance in all treatments in comparison to hive
nurse siblings and the starting microbiome pool
(Supplementary Figure S8). ese changes likely reect lab
adaptations and emphasize the need to run proper lab controls in
microbiome experiments (Arora etal., 2020), but also a need to
run more natural experiments in the future. Additionally, the high
tetracycline dosage over two worker generations may not reect
natural conditions, though mimicking nature was not our intent.
Controlled laboratory experiments such as microbiome
transplants, provide the most convincing insights into functional
host-microbiome relationships (Greyson-Gaito etal., 2020). ey
are invaluable because they can simplify the complexity and
disentangle factors to achieve fundamental understanding which
is still lacking in the eld. However, these experiments trade
control for natural complex conditions, which is important for
drawing ecological and evolutionary conclusions (Carrier and
Reitzel, 2017).
In addition to being a tractable model for microbiome research,
honey bees are important pollinators in natural and agricultural
ecosystems (Hung et al., 2018). ey are exposed to diverse
agricultural chemicals including those applied to plants making up
their diet but also the ones used by beekeepers to prevent infection or
suppress parasites (Ortiz-Alvarado etal., 2020). Antibiotics have been
experimentally demonstrated to disturb the core microbial bee
microbiome, lowering diversity on species and strain level and leading
to negative health eects (Raymann etal., 2017, 2018; Powell etal.,

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 12 frontiersin.org
2021; Jia etal., 2022). Facilitated by social transmission between
workers, changes in the microbiome could theoretically quickly go to
xation in a population. Indeed, antibiotic resistance genes have
accumulated in bacterial symbionts in managed honey bee colonies,
demonstrating long-term impacts with unknown consequences (Tian
etal., 2012; Ludvigsen etal., 2017; Daisley etal., 2020; Piva etal.,
2020). Considering the social-vertical transfer of the microbiome
between worker generations in honey bee colonies with the fact that
chemicals including antibiotics accumulate and persist in the hive
environment over longer periods (Martel etal., 2006), the damage on
the bee microbiome could theoretically go beyond one individual’s
health aecting a whole population. In mice, diet-induced progressive
loss of taxonomic diversity is cumulative over generations and
indicate that taxa driven to low abundance are ineciently transferred
to the next generation, and are at increased risk of becoming extinct
within an isolated population making this change eventually
irreversible (Sonnenburg et al., 2016). is suggests that
multigenerational environmental exposure could indeed cause a
stable transgenerational alteration of organism physiology via
the microbiome.
Conclusion
Co-evolved microbiomes can oer a range of benets to their
hosts and vice versa. However, under disturbance this picture may
change, and the dependent partner could suer negative
consequences. While it is oen dicult to disentangle cause and
consequences of chemical-induced microbiome disruption on host
health, weprovide evidence that a disturbed microbiome and its
mediated eects on host phenotypes can get transmitted across
generations in a lab environment. is “dark side” of a specialized,
vertically transferred microbiome could, likewise as negative
mutations, theoretically go into xation aecting the health of a
whole population if no refreshing is possible. is is particularly true
if the whole population is aected by chemical stress, for example in
an agricultural context. For instance, agrichemical degradation of
microbiomes may bea plausible, silent factor underlying global
insect declines. Future studies would beimportant to examine the
extent to which negative microbiome-mediated phenotypes are
really heritable in the eld. Examining whether such heritable
dysbiosis has the potential to threaten host populations or which
potential rescue mechanisms may play a role to prevent such
scenario under natural conditions would be relevant to further
understand organism health and conservation.
Data availability statement
e datasets presented in this study can befound in online
repositories. e names of the repository/repositories and
accession number(s) can befound at: https://www.ncbi.nlm.nih.
gov/, PRJNA863631. All data les and codes used for processing
and analysis can be found following this link: https://github.com/
kowallik/inheritance_microbiome_disturbance. RNA and 16S raw
reads are available under NCBI Bioproject PRJNA863631.
Author contributions
VK and AMwrote the manuscript. AMprocessed raw RNA
sequences. VK designed research, conducted experiments,
processed 16S sequences, and analyzed 16S, RNA and survival
data. AD extracted samples and prepared RNA libraries. All
authors contributed to the article and approved the
submitted version.
Funding
A fellowship provided by the German Research Foundation
(DFG) (KO 5604/1–1) and the Okinawa Institute of Science and
Technology (OIST) supported VK’s research. AMwas funded by
a Future Fellowship from the Australian Research Council
(FT160100178).
Acknowledgments
We are deeply grateful to Tony Xu and Robert Lawless for
beekeeping – their expertise, hard work and motivation helped us
tremendously. In addition, wethank Dr. Amy Paten from CSIRO
for her help with weighing and aliquoting the antibiotics for our
daily use.
Conflict of interest
e authors declare that the research was conducted in the
absence of any commercial or nancial relationships that could
beconstrued as a potential conict of interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their aliated
organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or
claim that may be made by its manufacturer, is not guaranteed or
endorsed by the publisher.
Supplementary material
e Supplementary material for this article can befound online
at: https://www.frontiersin.org/articles/10.3389/fmicb.2022.1030771/
full#supplementary-material

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 13 frontiersin.org
References
Alberdi, A., Aizpurua, O., Bohmann, K., Zepeda-Mendoza, M. L., and
Gilbert, M. T. P. (2016). Do vertebrate gut metagenomes confer rapid ecological
adaptation? Trends Ecol. Evol. 31, 689–699. doi: 10.1016/j.tree.2016.06.008
Arora, J., Mars Brisbin, M. A., and Mikheyev, A. S. (2020). Eects of microbial
evolution dominate those of experimental host-mediated indirect selection. PeerJ.
8:e9350. doi: 10.7717/peerj.9350
Baoni, L., Alberoni, D., Gaggìa, F., Braglia, C., Stanton, C., Ross, P. R., et al.
(2021). Honeybee exposure to veterinary drugs: how is the gut microbiota aected?
Microbiol Spectr. 9:e0017621. doi: 10.1128/Spectrum.00176-21
Barnosky, A. D., Hadly, E. A., Bascompte, J., Berlow, E. L., Brown, J. H.,
Fortelius, M., et al. (2012). Approaching a state shi in Earth’s biosphere. Nature 486,
52–58. doi: 10.1038/nature11018
Barnosky, A. D., Matzke, N., Tomiya, S., Wogan, G. O. U., Swartz, B., Quental, T. B.,
et al. (2011). Has the Earth’s sixth mass extinction already arrived? Nature 471,
51–57. doi: 10.1038/nature09678
Bäumler, A. J., and Sperandio, V. (2016). Interactions between the microbiota and
pathogenic bacteria in the gut. Nature 535, 85–93. doi: 10.1038/nature18849
Bokulich, N. A., Kaehler, B. D., Rideout, J. R., Dillon, M., Bolyen, E., Knight, R.,
et al. (2018). Optimizing taxonomic classication of marker-gene amplicon
sequences with QIIME 2’s q2-feature-classier plugin. Microbiome. 6:90. doi:
10.1186/s40168-018-0470-z
Bolyen, E., Rideout, J. R., Dillon, M. R., Bokulich, N. A., Abnet, C. C.,
Al-Ghalith, G. A., et al. (2019). Reproducible, interactive, scalable and extensible
microbiome data science using QIIME 2. Nat. Biotechnol. 37, 852–857. doi: 10.1038/
s41587-019-0209-9
Bray, N. L., Pimentel, H., Melsted, P., and Pachter, L. (2016). Near-optimal probabilistic
RNA-seq quantication. Nat. Biotechnol. 34, 525–527. doi: 10.1038/nbt.3519
Bürkner, P. C. (2017). Brms: an R package for Bayesian multilevel models using
Stan. J. Stat. Sow. 80, 1–28. doi: 10.18637/jss.v080.i01
Callahan, B. J., McMurdie, P. J., Rosen, M. J., Han, A. W., Johnson, A. J. A., and
Holmes, S. P. (2016). DADA2: high-resolution sample inference from Illumina
amplicon data. Nat. Methods 13, 581–583. doi: 10.1038/nmeth.3869
Carrier, T. J., and Reitzel, A. M. (2017). e Hologenome across environments and
the implications of a host-associated microbial repertoire. Front. Microbiol. 8:802.
doi: 10.3389/fmicb.2017.00802
Cavicchioli, R., Ripple, W. J., Timmis, K. N., Azam, F., Bakken, L. R., Baylis, M.,
et al. (2019). Scientists’ warning to humanity: microorganisms and climate change.
Nat. Rev. Microbiol. 17, 569–586. doi: 10.1038/s41579-019-0222-5
Chaimanee, V., Chantawannakul, P., Chen, Y., Evans, J. D., and Pettis, J. S. (2012).
Dierential expression of immune genes of adult honey bee (Apis mellifera) aer
inoculated by Nosema ceranae. J. Insect Physiol. 58, 1090–1095. doi: 10.1016/j.
jinsphys.2012.04.016
Chen, L., Lam, J. C. W., Hu, C., Tsui, M. M. P., Wang, Q., Giesy, J. P., et al. (2018).
Peruorobutanesulfonate exposure causes durable and transgenerational Dysbiosis
of gut microbiota in marine Medaka. Environ. Sci. Technol. Lett. 5, 731–738. doi:
10.1021/acs.estlett.8b00597
Cox, L. M., Yamanishi, S., Sohn, J., Alekseyenko, A. V., Leung, J. M., Cho, I., et al.
(2014). Altering the intestinal microbiota during a critical developmental window has
lasting metabolic consequences. Cells 158, 705–721. doi: 10.1016/j.cell.2014.05.052
Cuesta-Maté, A., Renelies-Hamilton, J., Kryger, P., Jensen, A. B., Sinotte, V. M.,
and Poulsen, M. (2021). Resistance and vulnerability of honeybee (Apis mellifera)
gut bacteria to commonly used pesticides. Front. Microbiol. 12:717990. doi: 10.3389/
fmicb.2021.717990
Daisley, B. A., Pitek, A. P., Chmiel, J. A., Gibbons, S., Chernyshova, A. M., Al, K. F.,
et al. (2020). Lactobacillus spp. attenuate antibiotic-induced immune and microbiota
dysregulation in honey bees. Commun Biol. 3, 1–13. doi: 10.1038/
s42003-020-01259-8
Doublet, V., Poeschl, Y., Gogol-Döring, A., Alaux, C., Annoscia, D., Aurori, C.,
et al. (2017). Unity in defence: honeybee workers exhibit conserved molecular
responses to diverse pathogens. BMC Genomics 18:207. doi: 10.1186/
s12864-017-3597-6
Ellegaard, K. M., and Engel, P. (2016). Beyond 16S rRNA community proling:
intra-species diversity in the gut microbiota. Front. Microbiol. 7.
Ellegaard, K. M., and Engel, P. (2019). Genomic diversity landscape of the honey
bee gut microbiota. Nat. Commun. 10:446. doi: 10.1038/s41467-019-08303-0
Emery, O., Schmidt, K., and Engel, P. (2017). Immune system stimulation by the
gut symbiont Frischella perrara in the honey bee (Apis mellifera). Mol. Ecol. 26,
2576–2590. doi: 10.1111/mec.14058
Engel, P., James, R. R., Koga, R., Kwong, W. K., McFrederick, Q. S., and
Moran, N. A. (2013). Standard methods for research on Apis mellifera gut
symbionts. J. Apic. Res. 52, 1–24. doi: 10.3896/IBRA.1.52.4.07
Engel, P., Kwong, W. K., McFrederick, Q., Anderson, K. E., Barribeau, S. M.,
Chandler, J. A., et al. (2016). e bee microbiome: impact on bee health and model
for evolution and ecology of host-microbe interactions. MBio 7, e02164–e02115.
doi: 10.1128/mBio.02164-15
Engel, P., and Moran, N. A. (2013). e gut microbiota of insects – diversity in
structure and function. FEMS Microbiol. Rev. 37, 699–735. doi:
10.1111/1574-6976.12025
Evans, J. D., Aronstein, K., Chen, Y. P., Hetru, C., Imler, J. L., Jiang, H., et al.
(2006). Immune pathways and defence mechanisms in honey bees Apis mellifera.
Insect Mol. Biol. 15, 645–656. doi: 10.1111/j.1365-2583.2006.00682.x
Falcon, S., and Gentleman, R. (2007). Using GOstats to test gene lists for GO term
association. Bioinformatics 23, 257–258. doi: 10.1093/bioinformatics/btl567
Flenniken, M. L., and Andino, R. (2013). Non-specic dsRNA-mediated antiviral
response in the honey bee. PLoS One 8:e77263. doi: 10.1371/journal.pone.0077263
Flint, H. J., Scott, K. P., Louis, P., and Duncan, S. H. (2012). e role of the gut
microbiota in nutrition and health. Nat. Rev. Gastroenterol. Hepatol. 9, 577–589. doi:
10.1038/nrgastro.2012.156
Foster, K. R., Schluter, J., Coyte, K. Z., and Rako-Nahoum, S. (2017). e
evolution of the host microbiome as an ecosystem on a leash. Nature 548, 43–51.
doi: 10.1038/nature23292
Francino, M. P. (2016). Antibiotics and the human gut microbiome: Dysbioses and
accumulation of resistances. Front. Microbiol. 12:6. doi: 10.3389/fmicb.2015.01543
Franzosa, E. A., Hsu, T., Sirota-Madi, A., Shafquat, A., Abu-Ali, G., Morgan, X. C.,
et al. (2015). Sequencing and beyond: integrating molecular ‘omics for microbial
community proling. Nat. Rev. Microbiol. 13, 360–372. doi: 10.1038/nrmicro3451
Fridmann-Sirkis, Y., Stern, S., Elgart, M., Galili, M., Zeisel, A., Shental, N., et al. (2014).
Delayed development induced by toxicity to the host can beinherited by a bacterial-
dependent, transgenerational eect. Front. Genet. 5:27. doi: 10.3389/fgene.2014.00027
Gould, A. L., Zhang, V., Lamberti, L., Jones, E. W., Obadia, B., Korasidis, N., et al.
(2018). Microbiome interactions shape host tness. PNAS 115, E11951–E11960. doi:
10.1073/pnas.1809349115
Greyson-Gaito, C. J., Bartley, T. J., Cottenie, K., Jarvis, W. M. C., Newman, A. E.
M., and Stothart, M. R. (2020). Into the wild: microbiome transplant studies need
broader ecological reality. Proc. Biol. Sci. 287:20192834. doi: 10.1098/rspb.2019.2834
Guzman, E. (2021). Transgenerational dierences in gut microbiota population
and epigenetic responses of host intestinal epithelial cells.
Hammer, T. J., Sanders, J. G., and Fierer, N. (2019). Not all animals need a
microbiome. FEMS Microbiol. Lett. 366:fnz117. doi: 10.1093/femsle/fnz117
Henry, L. P., Bruijning, M., Forsberg, S. K. G., and Ayroles, J. F. (2021). e
microbiome extends host evolutionary potential. Nat. Commun. 12:5141. doi:
10.1038/s41467-021-25315-x
Hooper, L. V., Littman, D. R., and Macpherson, A. J. (2012). Interactions between
the microbiota and the immune system. Science 336, 1268–1273. doi: 10.1126/
science.1223490
Hung, K. L. J., Kingston, J. M., Albrecht, M., Holway, D. A., and Kohn, J. R . (2018).
e worldwide importance of honey bees as pollinators in natural habitats. Proc. R.
Soc. B Biol. Sci. 285:20172140. doi: 10.1098/rspb.2017.2140
Jia, S., Wu, Y., Chen, G., Wang, S., Hu, F., and Zheng, H. (2022). e pass-on eect
of tetracycline-induced honey bee (Apis mellifera) gut community Dysbiosis. Front.
Microbiol. 12:781746. doi: 10.3389/fmicb.2021.781746
Kamada, N., Chen, G. Y., Inohara, N., and Núñez, G. (2013). Control of pathogens
and pathobionts by the gut microbiota. Nat. Immunol. 14, 685–690. doi: 10.1038/
ni.2608
Kandlikar, G. S., Gold, Z. J., Cowen, M. C., Meyer, R. S., Freise, A. C., Kra, N. J.
B., et al. (2018). Ranacapa: an R package and shiny web app to explore environmenta l
DNA data with exploratory statistics and interactive visualizations. F1000Res 7:1734.
doi: 10.12688/f1000research.16680.1
Kešnerová, L., Mars, R. A. T., Ellegaard, K. M., Troilo, M., Sauer, U., and Engel, P.
(2017). Disentangling metabolic functions of bacteria in the honey bee gut. PLoS
Biol. 15:e2003467. doi: 10.1371/journal.pbio.2003467
Kikuchi, Y., Hayatsu, M., Hosokawa, T., Nagayama, A., Tago, K., and Fukatsu, T.
(2012). Symbiont-mediated insecticide resistance. Proc. Natl. Acad. Sci. 109,
8618–8622. doi: 10.1073/pnas.1200231109
Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast, C., Horn, M., et al.
(2013). Evaluation of general 16S ribosomal RNA gene PCR primers for classical
and next-generation sequencing-based diversity studies. Nucleic Acids Res. 41:e1.
doi: 10.1093/nar/gks808
Koch, H., Abrol, D. P., Li, J., and Schmid-Hempel, P. (2013). Diversity and
evolutionary patterns of bacterial gut associates of corbiculate bees. Mol. Ecol. 22,
2028–2044. doi: 10.1111/mec.12209

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 14 frontiersin.org
Kowallik, V., and Mikheyev, A. S. (2021). Honey bee larval and adult microbiome
life stages are eectively decoupled with vertical transmission overcoming early life
perturbations. MBio 12, e02966–e02921. doi: 10.1128/mBio.02966-21
Kraemer, S. A., Ramachandran, A., and Perron, G. G. (2019). Antibiotic pollution
in the environment: from microbial ecology to public policy. Microorganisms. 7:180.
doi: 10.3390/microorganisms7060180
Kwong, W. K., Medina, L. A., Koch, H., Sing, K. W., Soh, E. J. Y., Ascher, J. S., et al.
(2017). Dynamic microbiome evolution in social bees. Sci. Adv. 3:e1600513. doi:
10.1126/sciadv.1600513
Kwong, W. K., and Moran, N. A. (2015). Evolution of host specialization in gut
microbes: the bee gut as a model. Gut Microbes 6, 214–220. doi:
10.1080/19490976.2015.1047129
Li, J. H., Evans, J. D., Li, W. F., Zhao, Y. Z., DeGrandi-Homan, G., Huang, S. K.,
et al. (2017). New evidence showing that the destruction of gut bacteria by antibiotic
treatment could increase the honey bee’s vulnerability to Nosema infection. PLoS
One 12:e0187505. doi: 10.1371/journal.pone.0187505
Li, J., Rettedal, E. A., van der Helm, E., Ellabaan, M., Panagiotou, G., and
Sommer, M. O. A. (2019). Antibiotic treatment drives the diversication of the
human gut Resistome. Genomics Proteomics Bioinformatics 17, 39–51. doi: 10.1016/j.
gpb.2018.12.003
Littman, D. R., and Pamer, E. G. (2011). Role of the commensal microbiota in
normal and pathogenic host immune responses. Cell Host Microbe 10, 311–323. doi:
10.1016/j.chom.2011.10.004
Ludvigsen, J., Porcellato, D., L’Abée-Lund, T. M., Amdam, G. V., and Rudi, K.
(2017). Geographically widespread honeybee-gut symbiont subgroups show locally
distinct antibiotic-resistant patterns. Mol. Ecol. 26, 6590–6607. doi: 10.1111/
mec.14392
Martel, A. C., Zeggane, S., Drajnudel, P., Faucon, J. P., and Aubert, M. (2006).
Tetracycline residues in honey aer hive treatment. Food Additives Contaminants
23, 265–273. doi: 10.1080/02652030500469048
McMurdie, P. J., and Holmes, S. (2013). Phyloseq: an R package for reproducible
interactive analysis and graphics of microbiome census data. PLoS One 8:e61217.
doi: 10.1371/journal.pone.0061217
McMurdie, P. J., and Holmes, S. (2014). Waste not, want not: why rarefying
microbiome data is inadmissible. PLoS Comput. Biol. 10:e1003531. doi: 10.1371/
journal.pcbi.1003531
Modi, S. R., Collins, J. J., and Relman, D. A. (2014). Antibiotics and the gut
microbiota. J. Clin. Invest. 124, 4212–4218. doi: 10.1172/JCI72333
Motta, E. V. S., and Moran, N. A. (2020). Impact of glyphosate on the honey bee
gut microbiota: eects of intensity duration, and timing of exposure. mSystems.
5:e00268-20. doi: 10.1128/mSystems.00268-20
Motta, E. V. S., Powell, J. E., and Moran, N. A. (2022). Glyphosate induces immune
dysregulation in honey bees. Animal Microbiome. 4:16. doi: 10.1186/
s42523-022-00165-0
Moullan, N., Mouchiroud, L., Wang, X., Ryu, D., Wil liams, E. G., Mottis, A., et al.
(2015). Tetracyclines disturb mitochondrial function across eukaryotic models: a
call for caution in biomedical research. Cell Re p. 10, 1681–1691. doi: 10.1016/j.
celrep.2015.02.034
Mueller, E. A., Wisnoski, N. I., Peralta, A. L., and Lennon, J. T. (2020). Microbial
rescue eects: how microbiomes can save hosts from extinction. Funct. Ecol. 34,
2055–2064. doi: 10.1111/1365-2435.13493
Neuman, H., Forsythe, P., Uzan, A., Avni, O., and Koren, O. (2018). Antibiotics in
early life: dysbiosis and the damage done. FEMS Microbiol. Rev. 42, 489–499. doi:
10.1093/femsre/fuy018
Ortiz-Alvarado, Y., Clark, D. R., Vega-Melendez, C. J., Flores-Cruz, Z.,
Domingez-Bello, M. G., and Giray, T. (2020). Antibiotics in hives and their eects
on honey bee physiology and behavioral development. Biology Open.
9:bio053884. doi: 10.1242/bio.053884
Ourry, M., Lopez, V., Hervé, M., Lebreton, L., Mougel, C., Outreman, Y., et al.
(2020). Long-lasting eects of antibiotics on bacterial communities of adult ies.
FEMS Microbiol. Ecol. 96:aa028. doi: 10.1093/femsec/aa028
Petersen, C., and Round, J. L. (2014). Dening dysbiosis and its inuence on host
immunity and disease. Cell. Microbiol. 16, 1024–1033. doi: 10.1111/cmi.12308
Piva, S., Giacometti, F., Marti, E., Massella, E., Cabbri, R., Galuppi, R., et al. (2020).
Could honey bees signal the spread of antimicrobial resistance in the environment?
Lett. Appl. Microbiol. 70, 349–355. doi: 10.1111/lam.13288
Poutahidis, T., Varian, B. J., Levkovich, T., Lakritz, J. R., Mirabal, S., Kwok, C., et al.
(2015). Dietary microbes modulate transgenerational cancer risk. Cancer Res. 75,
1197–1204. doi: 10.1158/0008-5472.CAN-14-2732
Powell, J. E., Carver, Z., Leonard, S. P., and Moran, N. A. (2021). Field-realistic
Tylosin exposure impacts honey bee microbiota and pathogen susceptibility, which
is ameliorated by native gut probiotics. Microbiol. Spectr. e00103–e00121. doi:
10.1128/Spectrum.00103-21
Powell, J. E., Martinson, V. G., Urban-Mead, K., and Moran, N. A. (2014). Routes of
acquisition of the gut microbiota of Apis mellifera. Appl. Environ. Microbiol. 80,
7378–7387. doi: 10.1128/AEM.01861-14
Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., et al. (2013).
e SILVA ribosomal RNA gene database project: improved data processing and
web-based tools. Nucleic Acids Res. 41, D590–D596. doi: 10.1093/nar/gks1219
Raymann, K., Bobay, L. M., and Moran, N. A. (2018). Antibiotics reduce genetic
diversity of core species in the honeybee gut microbiome. Mol. Ecol. 27, 2057–2066.
doi: 10.1111/mec.14434
Raymann, K., and Moran, N. A. (2018). e role of the gut microbiome in health
and disease of adult honey bee workers. Current opinion in insect. Science 26,
97–104. doi: 10.1016/j.cois.2018.02.012
Raymann, K., Shaer, Z., and Moran, N. A. (2017). Antibiotic exposure perturbs
the gut microbiota and elevates mortality in honeybees. PLoS Biol. 15:e2001861. doi:
10.1371/journal.pbio.2001861
Raza, M. F., Wang, Y., Cai, Z., Bai, S., Yao, Z., Awan, U. A., et al. (2020). Gut
microbiota promotes host resistance to low-temperature stress by stimulating its
arginine and proline metabolism pathway in adult Bact rocera dorsalis. PLoS Pathog.
16:e1008441. doi: 10.1371/journal.ppat.1008441
Resnik, P. (1999). Semantic similarity in a taxonomy: an information-based
measure and its application to problems of ambiguity in natural language. JAIR 11,
95–130. doi: 10.1613/jair.514
Rosenberg, E., and Zilber-Rosenberg, I. (2021). Reconstitution and transmission
of gut microbiomes and their genes between generations. Microorganisms. 10:70.
doi: 10.3390/microorganisms10010070
Round, J. L., and Mazmanian, S. K. (2009). e gut microbiome shapes intestinal
immune responses during health and disease. Nat. Rev. Immunol. 9, 313–323. doi:
10.1038/nri2515
Rubanov, A., Russell, K. A., Rothman, J. A., Nieh, J. C., and McFrederick, Q. S.
(2019). Intensity of Nosema ceranae infection is associated with specic honey bee
gut bacteria and weakly associated with gut microbiome structure. Sci. Rep. 9:3820.
doi: 10.1038/s41598-019-40347-6
Sánchez-Bayo, F., and Tennekes, H. A. (2017). Assessment of ecological risks of
agrochemicals requires a new framework. Environ. Risk Assess. Remediation. 1,
20–28. doi: 10.4066/2529-8046.100025
Schulfer, A. F., Battaglia, T., Alvarez, Y., Bijnens, L., Ruiz, V. E., Ho, M., et al.
(2018). Intergenerational transfer of antibiotic-perturbed microbiota enhances
colitis in susceptible mice. Nat. Microbiol. 3, 234–242. doi: 10.1038/
s41564-017-0075-5
Soen, Y. (2014). Environmental disruption of host–microbe co-adaptation as a
potential driving force in evolution. Front. Genet. 20:5. doi: 10.3389/
fgene.2014.00168
Sonnenburg, E. D., Smits, S. A., Tikhonov, M., Higginbottom, S. K.,
Wingreen, N. S., and Sonnenburg, J. L. (2016). Diet-induced extinctions in the gut
microbiota compound over generations. Nature 529, 212–215. doi: 10.1038/
nature16504
Sun, H., Mu, X., Zhang, K., Lang, H., Su, Q., Li, X., et al. (2022). Geographical
resistome proling in the honeybee microbiome reveals resistance gene transfer
conferred by mobilizable plasmids. Microbiome. 10:69. doi: 10.1186/
s40168-022-01268-1
Supek, F., Bošnjak, M., Škunca, N., and Šmuc, T. (2011). REVIGO summarizes
and visualizes long lists of gene ontology terms. PLoS One 6:e21800. doi: 10.1371/
journal.pone.0021800
Tian, B., Fadhil, N. H., Powell, J. E., Kwong, W. K., and Moran, N. A. (2012). Long-
term exposure to antibiotics has caused accumulation of resistance determinants in
the gut microbiota of honeybees. MBio 3, e00377–e00312. doi: 10.1128/
mBio.00377-12
Tremaroli, V., and Bäckhed, F. (2012). Functional interactions between the
gut microbiota and host metabolism. Nature 489, 242–249. doi: 10.1038/
nature11552
Turnbaugh, P. J., Ley, R. E., Mahowald, M. A., Magrini, V., Mardis, E. R., and
Gordon, J. I. (2006). An obesity-associated gut microbiome with increased capacity
for energy harvest. Nature 444, 1027–1031. doi: 10.1038/nature05414
Wang, G. H., Berdy, B. M., Velasquez, O., Jovanovic, N., Alkhalifa, S.,
Minbiole, K. P. C., et al. (2020). Changes in microbiome confer multigenerational
host resistance aer sub-toxic pesticide exposure. Cell Host Microbe 27,
213.e7–224.e7. doi: 10.1016/j.chom.2020.01.009
Wang, Q., Garrity, G. M., Tiedje, J. M., and Cole, J. R. (2007). Naïve Bayesian
classier for rapid assignment of rRNA sequences into the new bacterial taxonomy.
Appl. Environ. Microbiol. 73, 5261–5267. doi: 10.1128/AEM.00062-07
Willing, B. P., Russell, S. L., and Finlay, B. B. (2011). Shiing the balance: antibiotic
eects on host–microbiota mutualism. Nat. Rev. Microbiol. 9, 233–243. doi: 10.1038/
nrmicro2536

Kowallik et al. 10.3389/fmicb.2022.1030771
Frontiers in Microbiology 15 frontiersin.org
Wu, J., Lang, H., Mu, X., Zhang, Z., Su, Q., Hu, X., et al. (2021). Honey bee genetics
shape the strain-level structure of gut microbiota in social transmission. Microbiome
9:225. doi: 10.1186/s40168-021-01174-y
Xu, L., Surathu, A., Raplee, I., Chockalingam, A., Stewart, S., Walker, L., et al.
(2020). e eect of antibiotics on the gut microbiome: a metagenomics analysis of
microbial shi and gut antibiotic resistance in antibiotic treated mice. BMC
Genomics 21:263. doi: 10.1186/s12864-020-6665-2
Yilmaz, P., Parfrey, L. W., Yarza, P., Gerken, J., Pruesse, E., Quast, C., et al. (2014).
e SILVA and “all-species living tree project (LTP)” taxonomic frameworks. Nucleic
Acids Res. 42, D643–D648. doi: 10.1093/nar/gkt1209
Zare, A., Johansson, A., Karlsson, E., Delhomme, N., and Stenberg, P. (2018).
The gut microbiome participates in transgenerational inheritance of low-
temperature responses in Drosophila melanogaster. FEBS Lett. 592, 4078–4086.
doi: 10.1002/1873-3468.13278
Zhang, B., Leonard, S. P., Li, Y., and Moran, N. A. (2019). Obligate bacterial
endosymbionts limit thermal tolerance of insect host species. PNAS 116,
24712–24718. doi: 10.1073/pnas.1915307116
Zheng, H., Steele, M. I., Leonard, S. P., Motta, E. V. S., and Moran, N. A. (2018).
Honey bees as models for gut microbiota research. Lab Anim. 47, 317–325. doi:
10.1038/s41684-018-0173-x