Rapid evolution of a bacterial parasite during outbreaks in twoDaphnia populations
Authors:
Clara L. Shaw1,2*
Meghan A. Duffy1
1Department of Ecology & Evolutionary Biology,
University of Michigan, Ann Arbor, MI 48109
2Department of Biology, The Pennsylvania State
University, University Park, PA 16802
*Correspondence: shawclar@umich.edu
ORCID:
Shaw: 0000-0002-0176-8519
Duffy: 0000-0002-8142-0802
Keywords: parasite, Daphnia , Pasteuria ramosa , evolution,
matching alleles, epidemic
Acknowledgments: We thank Katie Hunsberger, Camden Gowler, Mary
Rogalski, Nina Wale, and Rebecca Bilich for assistance in field
collections and sample counting. We thank Aaron King, Ed Ionides, Tim
James, Mark Hunter and Alex Kondrashov for feedback on this manuscript
when it was in the form of a dissertation chapter. This work was
supported by grants to MAD from the US National Science Foundation
(DEB-1305836) and the Gordon and Betty Moore Foundation (GBMF9202; DOI:https://doi.org/10.37807/GBMF9202).
Author contributions: CLS and MAD jointly conceived of the study. CLS
collected and analyzed the data, with input from MAD. CLS wrote the
first draft of the manuscript; both authors revised the manuscript.
ABSTRACT
Myriad ecological and evolutionary factors can influence whether a
particular parasite successfully transmits to a new host during a
disease outbreak, with consequences for the structure and diversity of
parasite populations. However, even though the diversity and evolution
of parasite populations is of clear fundamental and applied importance,
we have surprisingly few studies that track how genetic structure of
parasites changes during naturally occurring outbreaks in non-human
populations. Here, we used population genetic approaches to reveal how
genotypes of a bacterial parasite, Pasteuria ramosa, change over
time, focusing on how infecting P. ramosa genotypes change during
the course of epidemics in Daphnia populations in two lakes. We
found evidence for genetic change – and, therefore, evolution – of the
parasite during outbreaks. In one lake, P. ramosa genotypes
structured by sampling date; in both lakes, genetic distance between
groups of P. ramosa isolates increased with time between
sampling. Diversity in parasite populations remained constant over
epidemics, though one epidemic (which was large) had low genetic
diversity while the other epidemic (which was small) had high genetic
diversity. Our findings demonstrate that patterns of parasite evolution
differ between outbreaks; future studies exploring the feedbacks between
epidemic size, host diversity, and parasite genetic diversity would
improve our understanding of parasite dynamics and evolution.
INTRODUCTION
Parasite genotypes vary in traits that
impact their fitness, including infectivity, virulence, and the ability
to persist in the environment
(Salvaudonet al., 2005; Refardt & Ebert, 2007; Rogalski & Duffy, 2020).
As epidemics progress, parasite population structure changes due to
selection on these traits, which can influence epidemic dynamics.
Understanding how parasite population structure and diversity change
during epidemics is important for public health, conservation, and our
fundamental understanding of parasitism. Despite this, aside from
studies on some humans pathogens (e.g., ebolavirus
(Park et al., 2015),
influenza (McCrone et
al., 2018), SARS-CoV-2
(Forster et
al., 2020; Lin et al., 2021)), few empirical studies have
analyzed changes in parasite population structure and diversity during
natural epidemics in wild systems, particularly in systems where
multiple ecological conditions can be explored (but see
(Zhan et al.,
2002; Eck et al., 2021)). This gap leaves us without a clear
understanding of parasite evolution during epidemics in wild hosts, or
how different ecological scenarios might influence changes in parasite
population structure and diversity.
Predicting how population
structure and diversity will evolve is challenging for several reasons
(Burdon, 1993). First, the
assemblage of hosts is an important selective force on parasites
(Gandon, 2004;
Koskella, 2014; Paplauskas et al., 2021). But, hosts could
impose selection that decreases diversity (if selection is directional)
or maintains and even increases it (if the most common parasite
genotypes are less successful and rare genotypes are more successful
(i.e., negative frequency dependent selection)). Second, the greater
ecology of a system, where there are myriad interactions between hosts,
parasites, and their biotic and abiotic environment, may impact
selection in epidemics in the wild
(Paplauskas et al.,
2021). Third, chance events could impact parasite genotype frequencies
due to bottlenecks and rapid changes in parasite population sizes
(Papkou et al.,
2016). Fourth, starting parasite diversity is an important determinant
of parasite evolution (Ecket al., 2021), but the amount of variation at the start of an
epidemic might vary substantially (e.g., if the epidemic started due to
the introduction of a few infected migrants vs. being triggered by
contact between the host population and large numbers of persistent
environmental transmission stages in a spore bank). Given the diversity
of factors that might influence parasite evolution and the paucity of
prior studies monitoring changes in parasite population structure and
diversity over the course of an outbreak, we still lack an understanding
of how parasite assemblages change over time, or even whether genetic
diversity tends to increase, decrease, or stay consistent during
epidemics.
We studied how genotypes of the wide-spread bacterial parasitePasteuria ramosa changed during two natural outbreaks inDaphnia dentifera hosts, important planktonic grazers in lakes.
Hosts become infected after consuming P. ramosa environmental
transmission stages (spores) floating in the water column. Infection
requires attachment of spores to the host esophagus
(Duneau et al.,
2011) as well as evasion of additional within-host processes that can
prevent parasite infection and proliferation after the attachment step
(Luijckx et al.,
2014). Resistance to spore attachment is governed by multiple alleles
at one locus in the host, giving rise to a matching-allele model of
infection (Routtu &
Ebert, 2015; Bento et al., 2017). After infection, the parasite
castrates its host (preventing a host genotype to which it is infective
from producing more susceptible progeny) and propagates itself within
the host hemolymph (Ebertet al., 1996). P. ramosa is an obligate killer, and
spores are only released from decaying host corpses
(Ebert et al.,
1996). These spores can go on to infect susceptible hosts, and thereby
extend an outbreak, or remain infective for many decades in lake
sediments
(Decaestecker et
al., 2004, 2007). Within an epidemic season, uninfected D.
dentifera hosts reproduce asexually, yielding many asexual clutches
(Smirnov, 2017), only
switching to sexual reproduction late in the fall towards the end of
epidemics (Duffyet al., 2008; Hite et al., 2017; Gowler et al.,
2021). Sexual offspring are enclosed in resting eggs that overwinter in
sediments. Therefore, host diversity during an epidemic is governed by
evolutionary forces acting on standing variation in hosts after sexual
offspring hatch in the spring. We hypothesized that parasite diversity
during an epidemic is similarly governed by evolutionary forces acting
on standing variation that had been seeded from the spore bank.
Though there is phenotypic evidence that P. ramosa evolution
occurs over the course of outbreaks
(Auld et
al., 2014; Paplauskas et al., 2021; Gowler et al.,
2022), we did not know how parasite genotype assemblages would change
within outbreaks. Parasite genotypic diversity could increase as
mutations occur, decrease if parasites adapt to low-diversity host
populations, or be maintained if different genotypes are favored through
time due to negative frequency dependent selection and/or by
reintroduction from the spore bank. We tracked genotypes of P.
ramosa that successfully infected hosts during outbreaks in two lakes
and used variable number tandem repeats (VNTRs) to assess how P.
ramosa genotype assemblages changed during natural outbreaks
(Mouton & Ebert,
2008; Andras & Ebert, 2013).
METHODS
We tracked epidemics in two lakes in southeast Michigan and collected
infected hosts from multiple timepoints in these epidemics to track
parasite diversity over time. The lakes, Little Appleton (Waterloo
Recreation Area) and Crooked Lake (Pinckney Recreation Area; known as
“Crooked-P” in other publications on this Daphnia -parasite
system to distinguish it from another Crooked Lake) were sampled every
two weeks from mid-July until mid-November 2017 by combining 3 plankton
tows (using a 12 cm Wisconsin net, 153 µm) from at least 10 m apart at
the deepest part of each lake. Two samples, each combining 3 plankton
tows from the deep basin, were collected by this method; the first
sample was used to assess infection prevalence and the second was used
to measure host density. For infection prevalence, subsamples from one
of these combined samples were taken and hosts were counted and visually
diagnosed for P. ramosa infection using a dissecting microscope
until at least 200 D. dentifera individuals were counted or until
the entire sample was processed. While processing, we collected infected
hosts and preserved them individually in 90% ethanol. Preserved
infected hosts were stored at -20°C until DNA extraction. The second
combined sample was preserved in 90% ethanol and later subsampled
volumetrically and counted under a dissecting microscope to assess host
density. Infected host density was calculated by multiplying infection
prevalence by host density at each sample date.
Genotyping parasites required DNA extraction from infected animals, PCR
amplification of VNTRs, and analysis of VNTR lengths via fragment
analysis. For DNA extraction, preserved infected animals were removed
from ethanol and placed in sterile microcentrifuge tubes. We used the
mericon bacteria plus DNA extraction kit (Qiagen, Hilden, Germany) to
extract DNA. The preserved infected animals were vortexed in 200 µL fast
lysis buffer with a battery-powered pestle to make an emulsion.
Emulsions were transferred to “pathogen lysis” tubes and vortexed for
10 minutes. These tubes were then centrifuged at 11,500 rpm. The
DNA-containing supernatant was removed and saved. We attempted to
amplify samples at 11 VNTR loci by PCR
((Mouton & Ebert,
2008; Andras & Ebert, 2013); Table S1). We carried out reactions in 10
µL volumes of 1X Qiagen multiplex mastermix (QIAGEN, Hilden, Germany),
10 nM forward primer with M13(-21) tail, 400 nM reverse primer, and 400
nM M13(-21) 6FAM-labeled forward primer or M13(-21) HEX-labeled forward
primer. The labeled primers allowed loci to be identified in fragment
analysis (Schuelke, 2000).
Amplification conditions were: 94°C (15 min), then 42 cycles of 94°C (30
s)/ 50°C (30 s)/ 72°C (1 min), and a final extension time at 72°C for 10
min. Following PCR, loci with distinct labels were combined and diluted
1:50 in molecular grade water. 1 µL diluted product was added into
capillary electrophoresis loading plates containing 1 µL Hi-Di formamide
and a LIZ500 or a ROX500 size standard (University of Michigan DNA
Sequencing Core). Fragment analysis was performed by the University of
Michigan DNA sequencing core. We used the software GeneMapper
(ThermoFisher Scientific) to read fragment lengths.
We analyzed parasite genotypes (combined alleles at all VNTR loci) to
quantify parasite diversity in the two lakes and to understand how
parasite genotype structure and diversity changed over time in
epidemics. In our analysis, we excluded three loci: Pr17 because it was
uniform across all samples and Pr3 and Pr7 due to poor amplification
(missing in 39.3% and 28.6% of samples respectively). This left 8 loci
in our analysis. We subsequently removed from analysis any samples
without amplification for at least 6 out of the 8 remaining VNTR loci.
Samples were assigned to the same multilocus genotype (MLG) if they were
identical at all loci (including loci with missing data ⎼ loci with
missing data were assumed to have null alleles). In several samples (3
of the 38 Little Appleton samples and 2 of the 42 Crooked samples),
multiple alleles were present for a given locus, indicating coinfection
with multiple P. ramosa genotypes. We thus created two datasets
for each lake: one using the alleles with the highest amplification in
coinfected animals (i.e., ignoring coinfection), and another that
included two MLGs within infected animals. For the dataset that included
coinfections, alleles that were secondary in amplification were assumed
to belong to the same coinfecting MLG; when only one allele was
amplified within a coinfected host at a particular locus, that allele
was assumed to be the allele for both coinfecting genotypes. Analysis
with this coinfection dataset yielded qualitatively similar results to
the dataset with a single genotype identified per host; we therefore
only present analysis from the latter.
Three metrics were used to quantify how parasite genotype diversity was
structured and changed over time: comparisons of Nei’s gene diversity
between lakes and sampling dates, analysis of molecular variance
(AMOVA), and Prevosti distances between parasite genotype assemblages at
different sampling dates within lakes. We first calculated Nei’s gene
diversity at each sample date for each lake
(Nei, 1973). This metric
measures the probability that two randomly drawn alleles for a given
locus will be different from each other
(Nei, 1973). We
bootstrapped values of Nei’s gene diversity, resampling 1000 times, and
centered confidence intervals around the observed values
(Marcon et al.,
2012). We used linear models to assess changes in gene diversity over
time, and we used a t-test to compare levels of gene diversity between
lakes. Second, to quantify the extent to which parasite genotypes
structured by sampling date, we constructed a Prevosti distance matrix
(i.e., the fraction of allelic differences out of all loci
(Wright, 1978) between all
parasite samples) and performed an AMOVA for each lake partitioning by
sample date (Excoffieret al., 1992). This analysis is analogous to an analysis of
variance where the variation that is partitioned are the genetic
distances between pairs of individuals. Significance of the partitioning
is then determined by randomly permuting the distance matrix (in our
case, 1000 times), each time calculating variance assigned to sample
dates to create a null distribution for comparison
(Excoffier et al.,
1992). Lastly, we calculated Prevosti distance (absolute genetic
distance) between parasite populations from different sample dates for
each lake (Prevosti et
al., 1975). In contrast to the Prevosti distance metric between
individuals described above, this metric measures the average difference
in allele frequencies over all loci between two assemblages (here,
sample dates). We used a linear model to test if genetic distance
between parasite assemblages was related to the amount of time that had
passed between sampling dates. All statistical tests were performed in R
version 4.0.3 (R Core Team,
2020). All population genetics calculations were computed using the R
package ‘Poppr’ version 2.9.3
(Kamvar et al.,
2014). All figures were constructed using the R package ‘ggplot2’
version 3.3.5 (Wickham,
2016).
RESULTS
The outbreak in Little Appleton was much larger than the outbreak in
Crooked in terms of both infection prevalence and infected host density
(Figure 1 A & B). Despite the substantially larger numbers of infected
hosts, both allelic and MLG diversity were lower in Little Appleton than
in Crooked with an average of 3.62 alleles per locus in Little Appleton
and 6.38 alleles per locus in Crooked; over the course of the outbreak
there were only 16 MLGs in Little Appleton (out of 38 samples) compared
to 26 MLGs in Crooked (out of 42 samples). Gene diversity remained flat
over time in both outbreaks (Little Appleton: F1,4=2.86,
P=0.17; Crooked: F1,4=0.13, P=0.74) though it was much
higher in Crooked (t=4.89, P=0.002; Figure 1 C & D). Interestingly, the
gene diversity at the first sample date from Little Appleton was much
higher than later in the outbreak and comparable to gene diversity in
Crooked (Figure 1 C & D). Parasite genotypes were structured by sample
date in Little Appleton, but not in Crooked (Table 1).