Division of labor and growth during electrical cooperation in multicellular cable bacteria

Significance Cable bacteria form centimeter-long, multicellular filaments whose energy metabolism involves cooperation among cells that separately perform oxidation of the electron donor and reduction of the electron acceptor. This cooperative division of labor is facilitated via long-range electrical currents that run from cell to cell along a network of conductive fibers. Here we show that biomass synthesis shows a surprising asymmetry along the filament: only the cells oxidizing the electron donor conserve energy for growth, while the other cells reduce electron acceptors without biosynthesis. Our study hence provides insights into the physiology of an unconventional chemolithotroph, which forms a multicellular electrically connected system with unique functional differentiation, integration, and coordination.


Culturing conditions
(1, 2). The sediment was sieved (500 µm) to remove fauna, homogenized, and subsequently re-packed in polycarbonate cores (inner diameter 5.2 cm) as described before (1, 3). The sediment cores were submerged in artificial seawater (salinity of 32) and incubated in the dark for several weeks until an active cable bacteria population developed. The seawater was kept at 20 °C and bubbled with air to maintain 100% air saturation throughout the incubation.

Microsensor profiling
Microsensor profiling (O2, H2S, and pH) was performed to monitor the geochemical fingerprint and thus the developmental state of the cable bacteria population (4). The data were also used to discriminate between the oxic and suboxic zones in the sediment at the time of sampling. Microsensors were purchased from Unisense A/S (Denmark), connected to a four-channel Microsensor Multimeter (Unisense), and fixed in a two-dimensional micro-profiling system (Unisense) that enabled stepwise movement of the sensors. The software SensorTrace PRO (Unisense) was used to control the movement of the microsensors and log sensor signals. A general-purpose reference electrode (REF201 Red Rod electrode; Radiometer Analytical, Denmark) was used as reference during the pH measurements.

Stable isotope probing experiments
Two separate incubation experiments were conducted to quantify the assimilation of carbon and ammonium by cable bacteria. The first took place in March 2017 and used sediment cores amended with 13 C-labeled bicarbonate ( 13 C-DIC) and 15  Stock solutions used to amend the sediment cores were prepared as previously described (5). In March 2017, one stock solution was prepared by dissolving 62 mM 13 (5). In both cases the artificial seawater contained no Mg and Ca ions to avoid precipitation of Mg 13 CO3 and Ca 13 CO3 as well as no bicarbonate and ammonium ions as to not dilute the label. The salts (NaH 13 CO3, Na 13 CH3 13 CH2 13 COO and 15 NH4NO3) used for preparing the stock solutions were purchased from Sigma-Aldrich.
Labelling of the sediment cores was done by first inserting three sub-cores (inner diameter 1.2 cm) into the core without disturbing the sediment, and subsequently by injecting 500 µL of the labelled stock solution into each sub-core in ten 50 µL injections.
To ensure homogeneous spread of the label throughout the sediment, the syringe needle was inserted to a depth of 5 cm, and the liquid was released while slowly moving the needle upwards. The use of the sub-cores ensured that the label was spread within a wellconstrained volume. One of the sub-cores was used to retrieve cable bacteria, while the other two were used for porewater analyses.
After label addition the cores were incubated for 24 h at 20°C to allow label assimilation by the cable bacteria. This incubation time was chosen because the doubling time was found to be around 20h (5,6). In March 2017, this incubation was done in a dark container with no overlying water added on top of the sediment cores. The thin water film at the sediment surface was therefore in direct contact with air, which could have resulted in the dilution of the 13 C-DIC label close to the sediment-water interface due to CO2 exchange. In December 2017, exchange with atmospheric CO2 was avoided by performing the incubation in a sealed container filled with artificial seawater. For the cores amended with 13 C-DIC and 15 N-NH4 the artificial seawater was labelled with 13 C-DIC and 15 N-NH4 as well to ensure equal 13 C and 15 N labelling of the porewater and overlying water. This difference did not alter the results obtained from the March 2017 incubation.

Filament extraction
Clusters of cable bacterium filaments were picked under a microscope with fine glass hooks custom-made from Pasteur pipettes. Filaments were retrieved separately from the oxic (0-2 mm depth) and the middle of the suboxic (5-10 mm depth) zone of the sediment.
The filaments were washed several times (> 3) in Milli-Q water (Millipore, The Netherlands) to eliminate precipitation of salt as described before (5), transferred onto polycarbonate filters (pore size 0.2 µm; Isopore, Millipore, The Netherlands) pre-coated with a 5 nm thin gold layer, and air-dried in a desiccator for at least 24 h.

Scanning Electron Microscopy (SEM)
The polycarbonate filters were imaged with a scanning electron microscope (JEOL Neoscope II JCM-6000, Japan) to identify areas suitable for NanoSIMS analysis. This was done under a 0.1-0.3 mbar vacuum and a high accelerating voltage (15 kV) using a backscattered electron detector.

NanoSIMS analysis
Nano-scale secondary ion spectrometry (nanoSIMS) analysis was performed with the NanoSIMS 50L instrument (Cameca, France) to assess the assimilation of 13 C-bicarbonate or 13 C-propionate and 15 N-ammonium by individual cells of cable bacteria. Fields of view (FOV) selected through SEM were pre-sputtered with Cs + -ions until secondary ion yields stabilized. Subsequently the primary Cs + -ion beam (current: 1-2 pA, energy: 16 keV, spot size: 130 nm, dwell time: 1 ms/pixel) was scanned over the FOV (areas between 10×10 µm and 20×20 µm in size) while detecting secondary ions 12 C -, 13 C -, 12 C 14 N -, 12 C 15 Nand 31 P -. To increase the overall signal the same FOV was imaged multiple times  frames), and the resulting ion count images were aligned and accumulated.
NanoSIMS data were processed using the Matlab-based software Look@NanoSIMS (7). After alignment and accumulation of the measured planes, regions of interest (ROIs), which corresponded to individual cable bacteria cells or segments of cable bacterium filaments (comprising between 5-10 cells), were drawn manually using the 12 C 14 N − ion count image. The ROIs drawn around segments of cable bacteria were used to analyse differences between filaments extracted from different redox zones whereas the ROIs drawn around individual cells were used to analyse differences within cells belonging to the same filament. For each ROI, the ROI-specific 13 C atom fraction was calculated using the total 12 C − and 13 C − ion counts accumulated over all ROI pixels. Similarly, the ROI-specific 15 N atom fraction was calculated from the total 12 C 14 N − and 12 C 15 N − ion counts accumulated over all ROI pixels. ROIs were excluded from the final analysis if their 13 C or 15 N atom fraction varied significantly among measured planes.
The isotope data are presented as atom fractions, x( 13 C) and x( 15 N), or as excess atom fractions (also referred to as enrichment), x E ( 13 C) = x( 13 C) s -x( 13 C) ref  filaments is presented as the ion count ratio 31 P/( 12 C+ 13 C) calculated from the total ion counts of P and C determined in the filament ROIs. This quantity should therefore be only interpreted as a relative measure of P that is comparable among cells and filaments analysed in this study.

Pore water analyses
The 13 C-labelling of the porewater dissolved inorganic carbon pool (DIC) was measured as previously described (9). Because of the limited porewater volume in the sampled subcores, these analyses could not be performed separately for the oxic and anoxic zones.
When possible, the handling was done under CO 2 -free conditions (N 2 atmosphere) to minimize exchange with atmospheric CO 2 . Under CO 2 -free conditions, the top 3 cm of the sub-cores were sliced off and transferred into a 50 mL Greiner tube. The sediment was then centrifuged at 3000 rpm for 10 minutes. Again under anoxic conditions, the supernatant was retrieved and filtered over 0.45 µm pore size filters. Following filtration, 0.3 mL, 0.5 mL or 0.7 mL of the filtered porewater were injected into helium-flushed (5 min, flush rate of 70 mL min -1 ) air-tight septum-capped vials (12 mL) that contained four drops of 85% To calculate the total length of the filament on the electrode (used for cyclic voltammetry measurements) bright field microscopy (Zeiss Axioplan 2, Germany) was used. The length of the filaments as well as the average cell length (3 µm) was determined by subsequent image analysis in ImageJ.

Sediment manipulation experiment
First, the presence of an active population of cable bacteria in the sediment core was confirmed by measuring vertical profiles of O2, pH, H2S and electric potential (EP) using microsensors (Fig. S5). Subsequently, the overlying water was purged for several minutes with N2 gas to induce anoxia, and then purged with air for several minutes to re-establish oxic conditions. After each step, EP profiles were measured. Afterwards, the sediment was cut horizontally few mm below the oxic-suboxic boundary (5 mm depth) with a thin nylon thread (60 µm diameter), which disrupted the electron transport by the cable bacteria (11) and resulted in a decrease of the electric field in the sediment (12). EP depth profiles were measured 2, 15 and 30 minutes after the cutting with the top 5 mm sediment layer left in place. Subsequently, the top sediment layer was removed, exposing previously suboxic sediment to an oxygenated water column, and depth-profiles of EP were measured after 2, 15 and 30 minutes. At the end, the overlying water was subjected to another cycle of induced anoxia and re-oxygenation while depth-profiling EP after each step. Sediment cutting was done by fixing a ring with a height of 5 mm on top of the core liner and slowly pushing the sediment up with a plunger until the top of the sediment reached the top of the ring. Once the sediment was in place, a nylon thread was passed through the sediment guided by the slit between the core liner and the ring. Afterwards the slit was sealed with a water-proof adhesive tape to allow stable conditions during EP measurements. The sediment slice was removed by removing the adhesive tape around the ring and sliding the ring off the sediment core in one swift motion.
Electric potential (EP) was measured using an electric potential microelectrode (EPM) built at Aarhus University (Denmark) (13). The REF201 Red Rod electrode was used as a reference during the EP measurements. The EPM and the reference electrode were connected to a custom-made millivoltmeter with a resistance of >10 14 Ω.

Phase contrast microscopy
To observe the movement of cable bacteria with a phase contrast microscope special slides were constructed that allowed for a stable oxygen front. The slides were constructed by gluing slabs of microscope slide glass onto a microscope slide, creating a chamber in the centre of the slide (30 mm x 8 mm x 2 mm). Sediment was then placed in this chamber with a cover glass on top, and the narrow gap between the slide and cover glass was flooded with artificial seawater to remove loose sediment particles. This created a clear division between sediment and glass, and cable bacteria could move into the space left between the chamber and edge of the cover glass (14). The movement of the bacteria within the chamber was analysed in a phase contrast microscope at 20°C. A ZEISS Observer Z1 (Zeiss, Göttingen, Germany) inverted microscope with a PALM automated stage and a 40x phase contrast objective was used.
Quantification of specific assimilation rates of C and N Assimilation of carbon. Specific rates of carbon assimilation were estimated assuming that the cells assimilated carbon from two 13 C-labelled carbon sources, C1 and C2, and that for both carbon sources the increase in the cellular carbon content, C, followed first-order kinetics, dC/dt = kC1C + kC2C, where kC1 and kC2 denote the specific assimilation rate of carbon from the respective source. Both kC1 and kC2 represent the amount of C assimilated per unit time normalized to the C content of the cell, and are therefore expressed in units of d -1 (mol C (mol C) -1 d -1 ). These assumptions imply that the excess 13 C atom fraction of a cell, x E ( 13 C), changes in time according to the differential equation where x E ( 13 C)C1 and x E ( 13 C)C2 denotes the excess 13 C atom fraction of the respective carbon source.
For the 13 C-bicarbonate incubation we assumed that the 13 C-enriched porewater DIC was the only carbon source, and that there was no uptake of propionate due to its low basal concentration in the porewater. Additionally, we assumed that the 13 C enrichment of the DIC pool, x E ( 13 C)DIC, was constant during the incubation and equal to the value measured after the incubation. Based on these assumptions, the differential equation (1) is The solution to this differential equation implies that the increase in the 13 C enrichment of the cell approaches the enrichment of the DIC in an inverted exponential fashion, i.e., evolves in time, t, as Using this expression, the specific rate of inorganic carbon assimilation by a cable bacterial cell or a filament segment, kDIC, was therefore calculated from its 13 C enrichment measured at the end of the 13 C-bicarbonate incubation as For the 13 C-propionate incubation, we assumed that the cable bacteria assimilated carbon from two sources: the added 13 C-enriched propionate as well as the porewater DIC.
Since the added propionate far exceeded the propionate present in the porewater, the 13 C enrichment of the propionate, x E ( 13 C)P, was assumed to be constant during the incubation and equal to the value in the added stock (0.98). In contrast, the 13 C enrichment of the DIC pool, x E ( 13 C)DIC, was assumed to increase with time during the incubation due to the oxidation of the added 13 C-labeled propionate to CO2 by the microbial community in the core. That this occurred was confirmed by the significant 13 where denotes the propionate oxidation rate constant. For the given 13  Taking into account these results, the differential equation (1) for The solution to this differential equation implies that the 13 C enrichment of a cell assimilating carbon from these two sources will increase according to the function where K = kP + kDIC. How this expression was used to calculate the specific rate of propionate assimilation, kP, is described below.
Assimilation of nitrogen. Specific rates of ammonium assimilation were estimated assuming that ammonium was the only nitrogen source, and that the increase in the cellular nitrogen content, N, followed first-order kinetics, dN/dt = kNH4N, where kNH4 denotes the specific assimilation rate of ammonium. This assumption implies that the excess 15 N atom fraction of a cell, x E ( 15 N), changes in time according to the differential equation where x E ( 15 N)NH4 is the excess 15 N atom fraction of the porewater ammonium.
For both incubations ( 13 C-bicarbonate and 13 C-propionate) we assumed that the 15 Using this expression, the specific ammonium assimilation rate by a cable bacterial cell was therefore calculated from its 15 N enrichment measured at time t as Discrimination between the rates of inorganic carbon and propionate assimilation.
The purpose of the incubation with the added 13 C-enriched propionate was to quantify the cell-specific propionate uptake rate, kP. However, because the incubation with the 13 Cenriched DIC clearly showed that the cells assimilated inorganic carbon, and because there was a significant increase in the 13 C labelling of the DIC pool due to the conversion of 13 C-propionate to 13 C-CO2, the 13  Overall, we assumed that the coupling between the assimilation of DIC and NH4 was the same during both the 13 C-bicarbonate and 13 C-propionate incubations. Thus, for a given 15 N enrichment of a cell from the 13 C-propionate incubation, we first determined its specific ammonium assimilation rate constant kNH4 from equation (10). Then we used the rC/N factor determined from the results of the 13 C-bicarbonate incubation and predicted the corresponding rate constant of inorganic carbon assimilation as kDIC = rC/N×kNH4. Using this rate constant, we then found kP such that the 13 C enrichment predicted by equation (7) was equal to the measured value. Since an analytical solution to equation (7) for kP does not exist, we found this solution numerically.

Statistical Analysis
In the isotope labelling experiments, there were four biological replicates (cores) and each core had two categorical treatments (oxic and suboxic), providing eight samples in total.
All samples showed a non-normal distribution and unequal variances, while sample sizes were also variable (n between 24 and 94, Table S1). To evaluate the effect of the different treatments, a non-parametric Kruskal-Wallis test was performed, followed by a Dunn's post-hoc test to assess which sample pairs were statistically different. The R package "dunn.test" was used for pairwise comparisons. To avoid family-wise error rates, while retaining sufficient statistical power, the Holm-Bonferroni p-adjustment was used (15).  S1. Images of 13 C and 15 N assimilation by cable bacteria. Representative images of the 13 C atom fraction (a-d), 15 N atom fraction (e-h), 12 C 14 Nion counts (i-l) and secondary electrons (m-p) in filaments from the different incubations and redox zones: filaments incubated with 13 Cbicarbonate, and retrieved from the oxic zone (a, e, i, m) and from the suboxic zone (b, f, j, n); filaments incubated with 13 C-propionate, and retrieved from the oxic zone (c, g, k, o) and from the suboxic zone (d, h, l, p). Scale bars are 3 µm. A filament segment was defined as a region of interest (ROI). The counts of secondary ions 12 C -, 13 C -, 12 C 14 Nand 12 C 15 Nwere accumulated over the ROI pixels to calculate the average 13 C and 15 N atom fractions per filament.   The combination of these profiles showed the presence of an active cable bacteria population. The oxygen depth profile was used to determine the oxygen penetration depth. The EP data were used to calculate the difference in electric potential (ΔEP) shown in Fig. 3. (c) EP profiles were measured in an intact core with the conditions in the overlying water sequentially altered from oxic (light grey) to anoxic (red) and back to oxic (blue). Each profile corresponds to an average of two replicate profiles. (d) Sediment cutting experiment: EP profiles were measured 2 min (black), 15 min (red) and 30 min (blue) after the sediment was cut horizontally with a thin wire at a depth of 5 mm. Each profile is a single measurement. (e) Removal of the cut sediment layer: EP profiles were measured 2 min (black), 15 min (red) and 30 min (blue) after the top 5 mm of sediment was removed and O2 became available to the cable bacteria close to the new sediment-water interface (indicated by dashed line). Each profile is a single measurement. (f) EP profiles were measured 60 min after the top 5 mm of sediment was removed and the conditions in the overlying water were sequentially altered from anoxic (red) to oxic (blue) conditions. Each EP profile corresponds to an average of two replicate profiles. In panels (d, e, f) the original EP profile (light gray) is included for comparison. When necessary the profile was shifted downwards to reflect the correct position of the sediment-water interface.    Table S1. Variability of carbon assimilation rates among individual filaments of cable bacteria. Shown are the total number of filaments measured per replicate sediment core, and the corresponding amounts and percentages of filaments classified as 'inactive' (kDIC=0-0.01 d -1 ), 'minimally active' (kDIC=0.01-0.1 d -1 ), and 'active' (kDIC>0.1 d -1 ). For both incubations ( 13 Cbicarbonate and 13 C-propionate) most of the filaments retrieved from the oxic zone were inactive and only a small percentage was minimally active. Activity of the filaments from the suboxic zone was higher but still a large fraction of the filaments showed no or only a minimal activity. The average inorganic carbon assimilation rates are given separately for the active and minimally active filaments from both treatments. The standard deviation and the corresponding coefficient of variation illustrate considerable variability among individual filaments within the active and minimally active sub-populations. The ratio between the assimilation rates of propionate and inorganic carbon, kP/kDIC, shows that for filaments from the suboxic zone of the sediment cores labelled with 13 C-propionate only a small part (~7%) of the assimilated 13 C originates directly from the 13 C-propionate, whereas most of it originates from the 13 C-DIC produced through 13 Cpropionate remineralization (see Methods). In contrast, for filaments from the oxic zone the contribution of 13 C-propionate to the total carbon uptake is much larger (22-36%, calculated as kP/(kP+kDIC)). Table S2. Variability of carbon assimilation rates among cells within individual filaments of cable bacteria from the suboxic zone. Shown are statistics for the eight individual filaments retrieved from the suboxic zone from the bicarbonate incubation marked with different colours in Fig. 2a, including the distance (in mm and in number of cells) between the first and last segment measured along the filament, the total number of cells measured, the mean and standard deviation of 13 C atom fractions of all measured cells within a filament, the mean and standard deviation of the assimilation rate constant and the corresponding coefficient of variation within that filament. Data show that variability within filaments is clearly lower than variability among filaments (compare with Table S1).
Movie S1 (separate file). Time-lapse taken with a phase contrast microscope showing the sudden change in the direction of motion of cable bacteria. Two types of morphologically different cable bacteria are seen; very thin cable bacteria and cable bacteria with a larger diameter. The bottom cable bacterium with a large diameter shows a sudden reversal of movement even though the oxygen front was stable. Note that within the sediment enrichment used for nanoSIMS only the cable bacteria with a large diameter were present (diameter ~3-4 µm).
Dataset 1 (separate file). Document with each 13 C atom fraction, 15 N atom fraction and 31 P/C value for each filament segment. These values were used to construct Fig. 1, SI Appendix table S1 and Fig. S2.
Dataset 2 (separate file). Document with each 13 C atom fraction for each single cell from the each followed filaments. These values were used construct Fig. 2 and SI Appendix table S2.