Table of Contents

• Coverage
• Dataset Description
  • Methods & Sampling
  • Data Processing Description
  • BCO-DMO Processing Description
• Data Files
• Supplemental Files
• Related Publications
• Parameters
• Instruments
• Project Information
• Funding

Culturing:

Wild-type Bacillus subtilis bacteria (strain OI1084) were taken from −80°C frozen stock and streaked onto 1.5% agar plates prepared with Terrific Broth (TB, Sigma). Plates were incubated at 25°C for 24 hours, after which time a single colony from the plate was used to inoculate an overnight liquid TB culture at 30°C with shaking (200 rpm). The bacterial suspension was then subcultured (1.5 ml of cell culture into 60 ml of pre-warmed TB) and grown at 35°C and 200 rpm for 6 hours to mid-log phase (OD600 ≈ 0.2). Immediately prior to experiments, dense cell suspensions (∼ 1010 cells.ml−1) were prepared by centrifugation at 5000 g for 5 minutes, and the pellet was resuspended with 2 μl of fresh TB media.

Microfluidics and image analysis

Polydimethlysiloxane (PDMS) microfluidic channels were fabricated through soft lithography and plasma bonded to No. 1 thickness glass coverslips. The PDMS channels were thinly cast to ensure ample diffusion of oxygen to the bacterial suspension and prolonged cell activity. Dense cell suspensions were gently loaded into the microfluidic devices via pipette, and the channel inlet and outlet were sealed with wax to prevent residual flows. For all experiments, bacterial suspensions were imaged with brightfield illumination on an inverted microscope (Nikon Ti-E) using a sCMOS camera (Zyla 5.5, Andor Technology). Time-resolved velocity fields, u(x, t), of the bacterial suspensions were measured by performing Particle Image Velocimetry (PIV) using PIVLab implemented in MATLAB. The subsequent velocity fields were then lightly smoothed using a Guassian kernel with a standard deviation of one PIV pixel (1.73 μm) in space and one frame (9.5 ms) in time. Vorticity fields, ω(x, t), were computed from measured velocity fields using central differencing (MATLAB).

Bulk suspensions with decaying activity (quasi-2D)

To capture the bulk dynamics of dense bacterial suspensions in the absence of lateral wall effects, large microfluidic chambers (2 mm ×2 mm ×26 μm) were prepared, where the side length corresponds to ≈80 correlation lengths of the turbulent bacterial suspension. Imaging was performed with a 20 × objective (0.45 NA) at 105.5 fps. Bacterial activity was varied using a previously established approach. Briefly, the cell suspensions were imaged periodically over the course of 30 minutes, during which time the cell swimming speed naturally decayed due to oxygen depletion, where the decay rate of cell activity was controlled via the thickness of the PDMS device. Without the need to manipulate the cell suspension, this approach ensured that the bacterial concentration was constant across varying activity levels. Seven data sets were captured in total, which consisted of 6, 300 frames each (≈ 1 min per video) and a 2 − 3 min delay between acquisitions. Data analysis was restricted to the central portion of the chamber, ≈ 30 correlation lengths from the lateral boundaries.

Cell suspensions under varying confinement

To quantify POD modes for bacterial suspensions under varying degrees of confinement experiments from Wioland et al (2016). were replicated with minor modifications. Microfluidic devices (19 µm deep) were fabricated, which comprised a series of racetrack geometries connected by narrow inlets (30 µm wide). Nine racetrack widths were chosen between 30 µm ≤ W ≤ 171 µm in fractional increments of our measured characteristic vortex size (38 µm) taken as the minimum point in the spatial ACF in quasi-2D conditions (Fig. 1C, dotted red line). Cell suspensions were imaged at 40× (0.6 NA) magnification and 100 fps for 10 s per channel width. All confined microfluidic geometries were simultaneously loaded with the same cell suspension and imaged within the first 10 min of loading to ensure consistent bacterial activity. Experiments were 8 repeated across multiple days with freshly cultured bacteria to verify repeatability. The net flow ψ was calculated following the definition in [18]: ψ = |(Σu · eˆx) / (Σ||u||)|, where eˆx is the unit vector along the principal channel direction and the sum is over all PIV sub-windows over 5 s (500 frames) of video. ψ = 1 indicates uni-directional circulation flow around the racetrack, and ψ = 0 corresponds to a globally stationary suspension. The stream-wise flow profiles, F(y), were generated from the velocity field by averaging over time and space as F(y) = hu(x, y, t) · eˆxi(x,t) , where eˆx was chosen as defined in [18], such that F(y) is on average positive.

Flow fields were calculated using PIV (PIVLab, MATLAB), publicly available here. https://www.mathworks.com/matlabcentral/fileexchange/27659-pivlab-particle-image-velocimetry-piv-tool-with-gui . Box sizes of 32 pixel then 16 pixel (50%) overlap were used.

The subsequent velocity fields were then lightly smoothed using a Guassian kernel with a standard deviation of one PIV pixel (1.73 μm) in space and one frame (9.5 ms) in time.

An example script of the POD analysis is included in the submitted data files - this makes uses of the in-built "svd" function in MATLAB, and should be compatible with all versions.

Primary data file set to a csv representation of POD Flow Field Averages from each of the seven experiments.

Spaces removed from parameter names in primary data file (csv).

NSF Award Abstract:
Drifting photosynthetic microbes in surface ocean waters carry out nearly half of global carbon (C) fixation, both supporting the marine food web and reducing atmospheric carbon dioxide (CO2) levels. The fate of C in ocean ecosystems is controlled by myriad individual interactions within a highly interconnected planktonic food web, the sheer complexity of which has hindered predictive understanding of global C cycling. Chemical cues govern microbial interactions, and during infection, marine viruses manipulate the metabolism of phytoplankton and bacteria, facilitating the release of dissolved organic matter from infected cells. This research aims to determine how viral metabolic reprogramming of and organic matter release from intact, infected phytoplankton influences microbial interactions and C cycling. The interdisciplinary, collaborative nature of the project will enable direct training of two postdoctoral researchers, one graduate student, and undergraduate students in viral ecology, microfluidics, and metabolomics. An educational outreach program that engages middle school students in hands-on, high speed imaging of microbes will be expanded, and the project will culminate in a three-day workshop to advance the application of microfluidic devices and mass spectrometry analyses in microbial ecology.

The overarching hypothesis behind this research is that viral infection alters the chemical landscape of intact, infected picophytoplankton cells, attracting neighboring chemotactic bacteria and protistan zooplankton, and altering C flux pathways. To test this idea, a series of linked multi-scale laboratory-based experiments will be run to 1) Characterize the response of diverse model marine microbes to dissolved organic matter (DOM) released from intact, virus-infected picophytoplankton using microfluidics-based chemotaxis assays, 2) Identify key viral-derived DOM compounds eliciting chemotactic responses using stable isotope labeling, metabolomics analyses, and chemotaxis assays, and 3) Quantify micron-scale cross-trophic encounter dynamics and evaluate their impact on bulk-scale C cycling using liter-scale measurements of C dynamics linked to high spatiotemporal resolution live imaging of microbial food webs. The ultimate goal of the project is to develop a mechanistic understanding of the role of intact, virus-infected cells in oceanic C cycling.