of 40
In situ single-cell activities of microbial populations revealed by spatial
transcriptomics
Daniel Dar
1,2
, Nina Dar
1
, Long Cai*
1
, and Dianne K. Newman*
1,2
1
Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA,
USA.
2
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA.
Corresponding authors: dkn@caltech.edu, lcai@caltech.edu
Lead contact: dkn@caltech.edu
Abstract
Microbial populations and communities are heterogeneous, yet capturing their diverse activities has
proven challenging at the relevant spatiotemporal scales. Here we present par-seqFISH, a targeted
transcriptome-imaging approach that records both gene-expression and spatial context within microscale
assemblies at a single-cell and molecule resolution. We apply this approach to the opportunistic bacterial
pathogen,
Pseudomonas
aeruginosa
, analyzing ~600,000 individuals across dozens of physiological
conditions in planktonic and biofilm cultures. We explore the phenotypic landscape of this bacterium and
identify metabolic and virulence related cell-states that emerge dynamically during growth. We chart the
spatial context of biofilm-related processes including motility and kin-exclusion mechanisms and identify
extensive and highly spatially-resolved metabolic heterogeneity. We find that distinct physiological states
can co-exist within the same biofilm, just a few microns away, underscoring the importance of the
microenvironment. Together, our results illustrate the complexity of microbial populations and present a
new way of studying them at high-resolution.
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted February 25, 2021.
;
https://doi.org/10.1101/2021.02.24.432792
doi:
bioRxiv preprint
Introduction
Life exists in context. Cells within microbial populations and communities are typically closely
associated with one another in multicellular biofilms, whether found within infected tissues, attached to
diverse surfaces, or forming assembla
ges in the d
eep sea (Costerton et al.,
1987; Flemming and Wuertz,
2019). Natural microbiota and infectious bacteria generally exist in biofilm aggregates that are on the
order of several dozen microns and which can contain many interacting species (DePas et al., 2016; Mark
Welch et al., 2016; Schaber et al., 2007). Despite the ubiquity of the biofilm lifestyle in both natural and
manmade habitats, understanding what life is like within it for individual microbes has proven highly
challenging. While single-cell level activities have been tracked at high spatial resolution using a variety
of approaches in diverse contexts (Chadwick et al., 2019; Hatzenpichler et al., 2020; Jorth et al., 2019),
we have been unable to resolve the hundreds if not thousands of concurrent activities that characterize
microbial life at relevant spatiotemporal scales.
What we understand about microbial life literally has
been limited by our ability to see.
Despite this limitation, it has become clear in recent years that extreme phenotypic heterogeneity defines
the microbial experience (Ackermann, 2015; Evans et al., 2020). This is as true for isogenic populations
as it is for complex biofilm communities. Clonemates sampled from the same environment often display
significant differences that are thought to result from stochastic gene-expression and variable
environmental factors (Ackermann, 2015; Schreiber and Ackermann, 2019; Schreiber et al., 2016). The
detection of phenotypic diversity even in seemingly well-mixed environments such as chemostats (Kopf
et al., 2015; Schreiber et al., 2016) also serves as a powerful reminder that life at the microscale may
inhabit far more diverse niches than are readily apparent. Phenotypic diversity has been rationalized as
providing microbes with a fitness advantage in an unpredictable world (Ackermann, 2015; Veening et al.,
2008). In addition, specialized functions have been proposed to underpin collective interactions such as
division of labor (Ackermann, 2015; Armbruster et al., 2019; Diard et al., 2013; Rosenthal et al., 2018).
However, little is still known about the range of possible cellular phenotypic states and their roles in most
biological processes.
What triggers such phenotypic plasticity, and are there underlying “rules” that govern any patterns that
may exist at the microscale? In sessile communities, bo
th clonal or multispecies, biological activities give
rise to changing chemical gradients that create a range of local microenvironments (Stewart, 2003;
Stewart and Franklin, 2008). Furthermore, spatial organization enables different conflicting metabolic
states or species to co-exist via physical separation, increasing the potential for diversity and allowing for
new interactions to emerge (Bocci et al., 2018; Evans et al., 2020; Kotte et al., 2014; Nadell et al., 2016;
Wolfsberg et al., 2018). Indeed, natural communities often contain many interacting species that assemble
into intricate spatial structures. These microscale assemblies can promote interactions between species
and represent a key ecosystem feature (Cordero and Datta, 2016; Nadell et al., 2016). Yet a wide gulf—
limited by technology—still separates such observations from a coherent conceptual framework to
explain the rules governing microbial ecology.
Recent advances in imaging methods provide a means to chart the physical associations between different
species in natural environments (Mark Welch et al., 2016; Shi et al., 2020; Tropini et al., 2017; Wilbert et
al., 2020). However, interpreting these maps rema
ins challenging without additional functional
information on the physiological states and activities of relevant community members. In contrast, recent
adaptions of eukaryotic single-cell RNA-sequencing (scRNA-seq) approaches provide a powerful means
of exploring the phenotypic landscape of planktonic bacteria (Blattman et al., 2020; Imdahl et al., 2020;
Kuchina et al., 2021). However, these approaches do not preserve the spatial context of analyzed cells and
are therefore limited in their capacity to address single and multispecies biofilms. Thus, a major gap exists
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted February 25, 2021.
;
https://doi.org/10.1101/2021.02.24.432792
doi:
bioRxiv preprint
in our ability to account for both spatial and functional complexity, limiting progression toward a high-
resolution understanding of microbial life.
Single-molecule fluorescence in situ hybridization (FISH) based technologies have been used to measure
gene-expression directly within native tissues, recording both spatial and functional information.
However, while these methods have shed important light on single-cell heterogeneity they have been
traditionally limited to measuring the expression of only a few genes at a time (Choi et al., 2014; Femino
et al., 1998; Raj et al., 2008; So et al., 2011). In addition to this limited throughput, single-gene
measurements do not provide a means to capture coordinated cellular responses—the molecular
“fingerprint” of multiple biological
activities that
underpin distinct physiologi
cal states. R
ecent adva
nces
in combinatorial mRNA labeling and sequential FISH (seqFISH) now allow for hundreds and even
thousands of genes to be analyzed within the same sample at a sub-micron resolution (Chen et al., 2015;
Eng et al., 2019; Lubeck et al., 2014). Until now, seqFISH has been used in mammalian systems to
expose the physical organization of cell states within tissues (Chen et al., 2015; Eng et al., 2019; Lubeck
et al., 2014; Moffitt et al., 2018; Shah et al., 2016). We reasoned that the high spatial resolution of these
modern transcriptome-imaging techniques also had the potential to illuminate the microscale organization
of microbial populations and communities.
In this study, we adapted and further developed seqFISH for studying bacteria, measuring the expression
of hundreds of genes within individual cells while
also capturing their spatial context. We utilized
Pseudomonas aeruginosa
planktonic and biofilm populations to demonstrate how different cellular
functions are coordinated in time and space. Our proof-of-concept work illustrates how the ability to
observe transcriptional activities at the microscale permits insights into the spatiotemporal regulation and
coordination of critical life processes, enabling hitherto unrecognized, transient physiological states to be
identified and new hypotheses to be generated. These findings represent the tip of the iceberg and the
opportunities for discovery our approach enables promis
e to reveal new insights
about the rules governing
microbial ecology.
Results
A sequential mRNA-FISH framework for studying bacterial gene-expression.
Combinatorial mRNA labeling requires that each measured mRNA molecule be individually resolved.
However, this is much more challenging in bacteria due to the small size of their cells, as many different
mRNA molecules occur in close proximity and cannot be resolved using standard fluorescent microscopy.
We therefore utilized a nonbarcoded seqFISH approach (Lignell et al., 2017).
In seqFISH, target mRNAs are first hybridized with a set of primary, non-fluorescent probes, which are
flanked by short sequences uniquely assigned per gene (Figure 1A). Specific genes can be turned “ON”
via a secondary hybridization with short fluorescently labeled “readout” probes, complementary to the
gene-specific flanking sequences (Figure 1A). Several genes can be measured at once using a set of
readout probes labeled with different fluorophores (Methods). Importantly, these short fluorescent readout
probes can be efficiently stripped and washed away from the sample without affecting the primary probes
(Shah et al., 2018) (Figure 1A). Thus, once expression is measured, fluorescence can be turned OFF and a
new set of genes can be measured by introducing a new set of readout probes (Figure 1B). This 2-step
design allows for potentially hundreds of genes to be measured sequentially, one after the other in the
same sample, using automated microscopy (Figure 1B). The individual gene mRNA-FISH data can be
combined into spatially resolved multigene profiles at the single-bacterium level (Figure 1B).
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted February 25, 2021.
;
https://doi.org/10.1101/2021.02.24.432792
doi:
bioRxiv preprint
Due to the diffraction limit and the small size of bacteria, mRNA-FISH fluorescent signals (appearing as
spots within cells) can contain overlapping mRNA molecules that cannot be spatially resolved in standard
microscopes. Thus, counting the number of spots within a bacterial cell severely underestimates
expression levels. This problem can be overcome by integrating the fluorescent intensity per spot, which
scales linearly with the number of mRNAs. Fluorescent intensity can be converted to discrete mRNA
counts by measuring the characteristic intensity of a single transcript. This analog to digital conversion
approach has been shown to provide a wide dynamic range in bacteria (Skinner et al., 2013; So et al.,
2011).
We developed seqFISH in the study of
Pseudomonas aeruginosa
, an opportunistic human pathogen and a
severe cause of morbidity and mortality in cystic fibrosis (CF) patients (Bhagirath et al., 2016; Malhotra
et al., 2019). We generated a probe library targeting a set of 105 marker genes that capture many core
physiological aspects of this pathogen (Tables S1-S2). These included genes involved in biosynthetic
capacity (ribosome and RNA-polymerase subunits), anerobic physiology (fermentation and denitrification
pathways), stress responses (oxidative and nutrient limitation), cellular signaling (c-di-GMP), biofilm
matrix components, motility (flagella and T4P), all major quorum-sensing (QS) systems, as well as
multiple antibiotic resistance and core virulence factors. In addition, to control for false positives, we
designed probes targeting three different negative control genes that do not exist in
Pseudomonas
(Figure
S1).
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted February 25, 2021.
;
https://doi.org/10.1101/2021.02.24.432792
doi:
bioRxiv preprint