LIMNOLOGY
and

OCEANOGRAPHY: METHODS

Limnol. Oceanogr.: Methods 12, 2014, 519–529
© 2014, by the American Society of Limnology and Oceanography, Inc.

Prokaryotic cells separated from sediments are suitable for
elemental composition analysis
Marc Neveu,1* Amisha T. Poret-Peterson,1 Zarraz M.P. Lee,2 Ariel D. Anbar,1,3 and James J. Elser2
1

School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85287, USA
School of Life Sciences, Arizona State University, Tempe, AZ 85287, USA
3
Department of Chemistry and Biochemistry, Arizona State University, Tempe, AZ 85287, USA
2

Abstract
Cell-sediment separation methods can potentially enable determination of the elemental composition of
microbial communities by removing the sediment elemental contribution from bulk samples. We demonstrate
that a separation method can be applied to determine the composition of prokaryotic cells. The method uses
chemical and physical means to extract cells from benthic sediments and mats. Recovery yields were between
5% and 40%, as determined from cell counts. The method conserves cellular element contents to within 30%
or better, as assessed by comparing C, N, P, Mg, Al, Ca, Ti, Mn, Fe, Ni, Cu, Zn, and Mo contents in Escherichia
coli. Contamination by C, N, and P from chemicals used during the procedure was negligible. Na and K were
not conserved, being likely exchanged through the cell membrane as cations during separation. V, Cr, and Co
abundances could not be determined due to large (>100%) measurement uncertainties. We applied this method
to measure elemental contents in extremophilic communities of Yellowstone National Park hot springs. The
method was generally successful at separating cells from sediment, but does not discriminate between cells and
detrital biological or noncellular material of similar density. This resulted in Al, Ti, Mn, and Fe contamination,
which can be tracked using proxies such as metal:Al ratios. With these caveats, we present the first measurements, to our knowledge, of the elemental abundances of a chemosynthetic community. The communities
have C:N ratios typical of aquatic microorganisms, are low in P, and their metal abundances vary between hot
springs by orders of magnitude.

photosynthesis requires electron transfer intermediates containing Fe (Morel and Price 2003), and the pigment chlorophyll contains Mg. Reactive radicals resulting from photosynthesis are managed by superoxide dismutase enzymes with Mn,
Fe, Ni, Cu, or Zn cofactors (Bowler et al. 1992). Metalloenzymes
with a Zn, Co, or Cd metal center assist microbes with inorganic carbon assimilation (Morel and Price 2003). Organically
bound phosphorus can be made bioavailable by phosphatase
enzymes containing Zn, Mg, or Fe (e.g., Stec et al. 2000).
Because of the importance of these microbial processes, we
are motivated to study major element cycling in situ by measuring the elemental composition of microbial cells extracted
directly from their environment. Carbon, macronutrients
(e.g., nitrogen and phosphorus), and micronutrients (e.g.,
trace metals) need to be taken up from the environment by
microbes. In Earth’s oceans, the availability of metals, P, and N
varies with space and time (Anbar 2008; Falkowski 1997; Papineau et al. 2009) and is known to influence element use by
cells (Morel 2008; Dupont et al. 2010). However, in extreme
environments such as hot springs, the distribution of elements, their use by extremophilic microorganisms, and the

Better understanding of the coupling of major biogeochemical cycles requires knowledge of the cellular elemental composition of microbes (e.g., Williams and Fraústo da Silva 2000;
Haraguchi 2004; Morel 2008). Examples of connections
between trace metals and macronutrient cycling by microbes
abound. Nitrogen is cycled using enzymes that contain Fe, Cu,
and Mo (Morel and Price 2003). Inorganic carbon fixation by

*Corresponding author: E-mail: mneveu@asu.edu

Acknowledgments
Many thanks to Laura Steger, Natalya Zolotova, and Gwyneth
Gordon for their advice on the cell separation protocol and performing
IRMS and ICP-MS analyses, respectively. Elemental analyses were carried
out in the W. M. Keck Foundation Laboratory for Environmental
Biogeochemistry at ASU. Mayra Buenrostro and Nina Popovic separated
most of the Yellowstone samples. Everett Shock helped guide this work
through useful discussions. The Research Office at Yellowstone National
Park granted the permit (YELL-2009-SCI-5434) allowed collection of
these samples. This research was supported by the NASA Astrobiology
Institute grant (NAI5-0018). Finally, we thank two anonymous reviewers
who helped improve this manuscript.
DOI 10.4319/lom.2014.12.519

519

Neveu et al.

Elemental analysis of cells separated from sediment

problems with measurements of cellular elemental composition. We tested the ability of the method of Amalfitano and
Fazi (2008) to produce accurate information about the elemental contents of Escherichia coli cells by comparing elemental abundances in cells that had undergone separation to
those in cells that had not. We then separated cells from sediments and mats collected in hot springs of Yellowstone
National Park (USA) and measured their elemental makeup
using Isotope Ratio Mass Spectrometry (IRMS) for C and N,
and Inductively-Coupled Plasma Mass Spectrometry (ICP-MS)
for the other elements.

influence of extremophiles on element cycling remain largely
unexplored. Hot springs at temperatures above 72°C are also
environments where primary production is supported by
chemosynthetic processes rather than photosynthesis.
Chemosynthesis also involves metals as enzyme cofactors,
such as Fe and Ni in hydrogen oxidation. Determining the elemental contents of microbes in such settings will likely
advance our understanding of modern biogeochemical
processes and potentially those in the past as well.
However, in benthic sediments and microbial mats such as
those found in hot springs, it is difficult to determine the composition of microbes because of the contributions of mineral
grains and particles. It is especially difficult to determine the
microbial content of metals such as Fe, Ni, Cu, Zn, and Mo,
which are also found in sediment. However, these elements
need to be quantified to extend the stoichiometric analysis
beyond the standard “Redfield C:N:P ratio” to a broader suite
of elements of biological importance (Twining et al. 2011;
Nuester et al. 2012). Removing the sediment contribution
from elemental analyses of microbial communities,
extremophilic or otherwise, requires either effective isolation
of cells from complex mineral matrices (Havig et al. 2011;
Steenbergh et al. 2013) or in situ microscopic analyses (Heldal
et al. 2003, Steenbergh et al. 2013).
There is a need to verify that such cell separation procedures do not alter cellular elemental composition before
analysis. Cell separation methods usually involve chemical
(use of detergents and chelating agents) and physical means
(stirring or blending and/or sonication) of disrupting microbemineral bonds, followed by density gradient centrifugation
(Lindahl and Bakken 1995; Amalfitano and Fazi 2008;
Kallmeyer et al. 2008; Amalfitano et al. 2009; Poté et al. 2010,
Morono et al. 2013). Amalfitano and Fazi (2008) compared
combinations of chemical and physical separations. They
found that the most efficient method of detaching cells from
sediment comprised three steps: (1) stirring of the sample
slurry in an extraction mixture containing a detergent and a
chelating agent, (2) gentle sonication of the resulting suspension, and (3) centrifugation through a cushion of the density
gradient medium Nycodenz (Rickwood et al. 1982) to isolate
the cells, which have lower density than sediment. Alternative
methods to that of Amalfitano and Fazi (2008) involve blending, dispersion by rotary pestle, or shaking with glass beads
(e.g., Lindahl and Bakken 1995); all of which could contaminate the separated cells with trace metals. This trace metal
contamination risk from hardware is minimized in the procedure of Amalfitano and Fazi (2008), which can be carried out
entirely in acid-washed plasticware.
For all cell separation methods, there is a trade-off between
efficiently breaking the strong cell-sediment bonds and preserving cell integrity (Lindahl and Bakken 1995). Preserving
cell integrity is essential to conserve cellular elemental contents. The method of Amalfitano and Fazi (2008) can affect
cellular integrity (Amalfitano et al. 2009), which could lead to

Materials and procedures
To evaluate the ability of the cell separation method to preserve the elemental content of cells, wild-type Escherichia coli
in LB broth was cultured to late log or early stationary phase,
then divided into 1-mL aliquots and kept frozen at –80°C
without addition of any cryoprotectant. Freezing and thawing
could result in cell damage. However, when observing thawed
E. coli cells using epifluorescence microscopy for cell counting,
we noticed no physical damage (at least at this limited resolution). Furthermore, cells underwent only one freeze-thaw
cycle, and care was taken to thaw cells slowly (10°C, 20-40
min). Freezing is necessary for field samples because cell separations cannot be performed in situ, and in some cases, the
distance of field sites from laboratory facilities necessitates
sample transport; we therefore applied the same procedure to
E. coli. We measured the abundances of C, N, P, Na, Mg, Al, K,
Ca, V, Cr, Fe, Co, Ni, Cu, Zn, and Mo in E. coli before and after
separation. We did not mix E. coli with sterile sediment before
separation because no significant cell-sediment bonding
would likely occur (Lindahl and Bakken 1995). Cell separations, cell counts, and elemental analyses of E. coli were performed on a minimum of three but typically four replicates.
Cell-sediment separation procedure
The separation procedure is the combined C1+2 + P1+2(N) procedure of Amalfitano and Fazi (2008), followed by washing and
drying of separated cells before elemental analysis. Amalfitano
and Fazi’s C1+2 + P1+2(N) separation procedure consists of two
steps: (a) physical and chemical detachment of cells from sediment by vortexing and sonication of the samples in a buffer containing a detergent and a chelating agent, and (b) isolation of the
cells by centrifugation of the suspension through a gradient of
Nycodenz (Rickwood et al. 1982). Throughout the procedure,
polypropylene containers and spatulas were used to prevent
sample contamination by trace metals from glass or metallic
items. These items were acid-washed and autoclaved before use;
care was taken to avoid any physical contact between the plasticware and metals from the autoclave. All solutions were sterilized by autoclaving except sodium pyrophosphate, which was
filter-sterilized. Although autoclaving is not considered best
practice for trace metal work (Price et al. 1989), the trace metal
content of all solutions was negligible compared with abundances measured in E. coli (Web Appendix A).
520

Neveu et al.

Elemental analysis of cells separated from sediment

450) with the following parameters: 40% output, 35% duty
cycle, 2 min, to not lyse the cells. Amalfitano and Fazi (2008)
sonicated cell suspensions for only 1 min; however, initial
observations using optical and epifluorescence microscopy
showed that more cells were detached from sediment particles
in suspensions sonicated for 2 min than in those sonicated for
1 min.
Each suspension was gently homogenized and carefully layered on top of 5 mL of Nycodenz in 15-mL centrifuge tubes.
The Nycodenz solution was prepared by dissolving 8 g Nycodenz in 10 mL MilliQ water, which yields a density of 1.3 g/mL
(Poté et al. 2010). The density of the Nycodenz solution, which
is intermediate between cells and sediment, allows sediment to
migrate through Nycodenz while cells remain on top. Each
sample was centrifuged at 5000g and 4°C for 30 min using a
swing-out rotor. Centrifugation was repeated for an additional
30 min to 1 h until the Nycodenz layer was clear, and distinct
phases could be seen. The phases of a separated field sediment
sample are shown in Fig. 1B. From bottom to top, these phases
are sediment pellet (absent when the procedure is applied to a
sediment-free E. coli culture); Nycodenz; a thin layer of separated cells (with possible residual detritus of similar density);
and the supernatant. The cell layer was pipetted out, transferred into a clean 2-mL tube, and washed three times to
remove any Nycodenz or cell extraction buffer, using 0.85%
sodium chloride (NaCl) to limit osmotic stress. For each wash,
cells were resuspended in 1 mL NaCl solution, then centrifuged
(10,000g, 5 min, 4°C), and the supernatant was discarded.

Cell extraction buffer was prepared as follows. To prepare
50 mL buffer, we used 0.05 g of the chelating agent sodium
pyrophosphate (0.1% wt./vol. final concentration), 0.25 mL
Tween 20 detergent (0.5% vol./vol. final concentration), and 5
mL phosphate-buffered saline (PBS), brought up to 50 mL
with 18 MOhm-cm MilliQ water. The 5 mL PBS contained 1.3
M sodium chloride (0.13 M final concentration in the extraction buffer), 70 mM sodium phosphate dibasic (7 mM final
concentration), and 30 mM sodium phosphate monobasic (3
mM final concentration); the pH of PBS was adjusted to 7.4
with a few drops of concentrated hydrochloric acid or sodium
hydroxide. Sodium pyrophosphate was purchased from
Sigma-Aldrich (Cat. No. 221368), Tween 20 from Bio-Rad (Cat.
No. 170-6531), sodium chloride from Fisher Scientific (Cat.
No. S271-500), sodium phosphate dibasic from Mallinckrodt
(Cat. No. 7914), sodium phosphate monobasic from EMD Millipore (Cat. No. SX0710-1), and Nycodenz from Axis-Shield
(Cat. No. 1002424).
The procedure is depicted in Fig.!1A. Aliquots of E. coli were
gently thawed in a centrifuge (6000 g, 10°C, 20-40 min). The
supernatant was removed. Samples were weighed and transferred into 20-mL beakers. A different beaker was used for
approximately every 2 g sediment or microbial mat to avoid
overflow during vortexing. In each beaker, extraction buffer (5
mL) was added. The resulting suspensions were then stirred on
a large capacity stirrer for 30 min at 720 rpm using acidwashed, autoclaved stir rods. Each sample subsequently
underwent sonication in an ice-cold bath (Branson Sonifier

Fig. 1. A. Schematic of the cell-sediment separation and preparation for elemental analysis (see text for details). B. Four phases of a separated sediment
sample: top aqueous layer, interface containing cells, Nycodenz layer, and bottom sediment pellet, ideally no longer containing cells. This sample was
collected at an outflow of the hot spring in Yellowstone we unofficially named ‘Bison Pool’.

521

Neveu et al.

Elemental analysis of cells separated from sediment

Cells dried in microcentrifuge tubes were transferred to
acid-cleaned (using concentrated nitric acid followed by concentrated hydrochloric acid) Teflon vessels and digested
overnight in 3 mL concentrated nitric acid on a hot plate at
100 to 150°C. The vessels were then uncapped and heated
until sample volume decreased to approximately 0.5 mL. The
samples were finally diluted to 3.5 mL (the minimum volume
required by the ICP-MS to measure the 14 elements specified)
with 0.32 M nitric acid in 15 mL acid-cleaned centrifuge tubes.
The final dilution volume was weighed to obtain a dilution
factor between initial dry mass and mass of solution analyzed.
A calibration curve was built using a phosphorus-enriched
black shale standard solution of known elemental composition
(characterized using the certified reference material USGS
Devonian black shale SDO-1), serially diluted (1:15, 1:35, 1:150,
1:1000, 1:7500, 1:25000) with nitric acid (0.32 M). This standard was chosen because its elemental composition reasonably
matched that of (sometimes widely varying) Yellowstone sediment samples and E. coli. The standard was not digested along
with the samples to test for element recoveries; however, it is
routinely run in parallel with samples, with generally excellent
recovery, accuracy, and reproducibility. We performed samplefree digestions to check for contamination during digestion,
and subtracted the average of the measured values on these
digestions from the measured value of each E. coli replicate.

Separated cells, resuspended in a minimal amount (200 to
1000 μL) of 18 MOhm-cm MilliQ water to allow pipetting,
were divided into aliquots of 50 μL (IRMS replicates), and 300
μL (ICP-MS replicates) and dried (60°C, overnight) before elemental analysis. Obtaining dry masses of cell fractions was
essential to quantify elemental contents in terms of dry cellular mass. E. coli cells that did not undergo separation were
thawed, washed three times with either 0.85% NaCl or cell
extraction buffer (to assess the capacity of these solutions to
preserve the cellular makeup), resuspended in a minimal volume (200 to 1000 μL) of 18 MOhm-cm MilliQ water, and
dried.
Cell counting
Cell counting using epifluorescence microscopy helped
determine a cell recovery yield. Yields were determined by
comparing the amount of cells present before and after centrifugation through Nycodenz. 45-μL aliquots of cell suspensions were removed immediately after the sonication step and
again at the end of each NaCl wash. Cells were fixed using 5
μL formaldehyde (37%, final concentration 3.7%) and stored
at 4°C until counting. They were stained using 10 μg/mL 4′,6diamidino-2-phenylindole (DAPI), resuspended in 10 mL sterile 18 MOhm-cm MilliQ water, and filtered on a 0.22-μm polycarbonate membrane for enumeration via epifluorescence
microscopy at 1000-fold magnification.
Elemental analysis
Carbon and nitrogen
We used elemental analysis–isotope ratio mass spectrometry (EA-IRMS) to determine the elemental and isotopic abundances of C and N. A Costech EA-IRMS (Costech International
S.p.a.) was used to analyze all samples, calibration standards,
and blanks. Before combustion, samples were loaded into 5 ×
9 mm tin capsules (Costech). Freshly separated cells were
resuspended in as little 18 MOhm-cm MilliQ water as possible
(200 to 1000 μL) and pipetted into tin capsules in 50 μL
aliquots, then dried overnight at 60°C. Serial pipetting-drying
cycles were required for dilute suspensions; cells awaiting drying were stored at 4°C.
A calibration curve for C and N contents was built using a
tomato leaf standard (0.2, 0.5, 1.0, 2.0, 5.0, and 7.0 mg).
Glycine standards (0.7 mg) of known light, medium, and heavy
C and N isotopic composition were analyzed between samples
to determine the accuracy of isotopic determinations. These inhouse standards were characterized using the following standards: USGS 40, USGS 41, IAEA NBS 18, and IAEA NBS 19.
Phosphorus and trace elements
P, Na, Mg, Al, K, Ca, V, Cr, Fe, Co, Ni, Cu, Zn, and Mo abundances were determined by Inductively-Coupled Plasma Mass
Spectrometry (ICP-MS). All samples, appropriate multi-element calibration standards, and blanks were analyzed on an X
Series quadrupole ICP-MS (Thermo Scientific). Sample preparation was carried out in a Class-10 clean laboratory, using 1 ×
distilled nitric and hydrochloric acids. Dilutions were made
with 18 MOhm-cm water.

Assessment
Cell recovery yields
The yield for E. coli aliquots was consistently around 95%.
Therefore, only a small portion of cells was either lysed during
separation or lost to the aqueous or Nycodenz phases.
Amounts of cells washed away in the three washing steps were
about 1%, 0.1%, and 0.01%, respectively.
Preservation of elemental content
The elemental abundances of unseparated and separated E.
coli cells are shown in Web Appendix A, and the relative element enrichments or depletions (i.e., [abundance in separated
cells/abundance in non-separated cells] – 1) in E. coli due to
cell separation are shown in Fig.!2. The measured elemental
content of unseparated E. coli cells did not depend on whether
these cells were washed with 0.85% NaCl or with cell extraction buffer (Web Appendix A).
C and N contents were systematically depleted by about
10% in separated cells, whereas two elements present as
cations in the cells (Na and K, not plotted) were severely
depleted in some separated replicates but highly enriched in
others. P and trace metal (except V, Cr, and Co) contents were
unchanged within 30% or less. V, Cr, and Co contents
appeared to be conserved through the separation just as well
as Ni, Cu, Zn, or Mo but the precision of our ICP-MS measurements did not allow reliable determination of their abundance (Web Appendix A). While V, Cr, and Co were present at
quantifiable levels (>10 × blank level), the spread between
replicates was high.
522

Neveu et al.

Elemental analysis of cells separated from sediment

Furthermore, the Nycodenz molecule contains 3 atoms of
iodine (I), which makes up 46% of its mass. Steenbergh et al.
(2013) noted that measuring I in separated cells can help estimate Nycodenz contamination, and thus its contribution to
the measured C and N. Microorganisms contain 1 to 100 ppm
I (Bowen 1979). Up to 1.4% I by mass (roughly 3% Nycodenz)
was measured in the field samples, equivalent to 0.83% C and
0.15% N enrichments from Nycodenz. This falls well within
measurement uncertainties (1.6% for C and 0.6% for N). Since
Nycodenz contributes the most C and N of the solutions used
in the procedure, we infer that contributions of other chemicals were negligible as well.
The trace element contents of the solutions used in the procedure are shown in Web Appendix A. Most elemental signals
were too noisy (below ten times the instrumental uncertainty
on the analytical blank value) to be accurately quantified. The
only signals clearly above the noise level were for one isotope
of Cu, 63Cu (not confirmed in noisy 65Cu measurements below
or around quantification limit) and for K, which was not conserved in E. coli. We therefore infer that trace metal contamination due to autoclaving of the solutions used in the cell separation procedure is negligible.
Metal contamination could also occur during sample digestion. We carried out sample-free digestions for each ICP-MS
analysis and measured the trace element signal of these procedural blanks. This signal is compared with that of E. coli
digestions in Table!1, where it can be seen that contamination
at this step was negligible.
Loss of elements
Loss of elements from the cells can occur either if cell
integrity is compromised by freeze-thaw, osmotic, or mechanical shocks, or if elements can leave through the cell membrane. Indeed, the cell separation procedure of Amalfitano
and Fazi (2008) used in this study was found to result in membrane damage for 7 to 30% of cells separated from sediment
(Amalfitano et al. 2009).
Separated E. coli had C and N abundances that were systematically lowered by about 10%. This depletion might
reflect loss of some molecular pools in the cells, where compounds such as sugars, organic acids, nitrates, or amino acids
are stored for future use after synthesis or uptake from the
environment. Britten and McClure (1962) showed that E. coli
amino acid pools can be emptied following freeze-thaw or
osmotic shocks, and that the C and N contents of these pools
can account for 10% of organismal abundances. Regardless of
the source of the depletion, the consistent loss of C and N suggests that a correction factor of ~10% should be used to adjust
C and N contents of samples analyzed with this procedure.
Epstein and Schultz (1965) and Rhoads and Epstein (1978)
showed that osmotic shock during cold cell washing steps
deplete the K content of E. coli, because intracellular K concentrations depend on osmolarity, and to a lesser extent, temperature of the external medium. This can explain the depletion in E. coli cellular K that we measured after some

Fig. 2. Relative enrichment or depletion of elements in cultured wildtype E. coli that underwent separation, compared with the same culture
without separation ([abundance after separation in μg/(g dry
mass)/abundance before separation in μg/(g dry mass)] × 100 – 100%).
Error bars show one standard deviation over four replicates (i.e., 2 × the
standard error on the mean over four replicates, approximately a 95%
confidence limit); standard deviations were propagated through the calculation. The dashed lines at +30% and –30% bracket the range of relative enrichments or depletions.

We note that E. coli grown in LB has a high metal content,
up to an order of magnitude more than E. coli grown in minimal medium (Outten and O’Halloran 2001). This could mask
small variations in metal content due to potential contamination, which may thus have been underestimated. Further
investigations may wish to repeat this experiment with E. coli
grown in minimal medium to assess potential contamination
on a lower scale.
Gain of elements (contamination)
Artificially high elemental abundances could have resulted if
the cells were contaminated by autoclaved solutions used during separation (Steenbergh et al. 2013), or during acid digestion.
P- and Na-rich PBS and sodium pyrophosphate, C- rich Tween
20, and C- and N- rich Nycodenz are used in the separation procedure, raising the possibility of contamination by these elements. However, we observed a modest depletion in C and N
after separation (Fig. 2), whereas P abundance remained
unchanged within the limits of analytical uncertainties. In addition, the C and N isotopic compositions of E. coli did not
change during separation (δ13CVPDB = –21.8 ± 0.5‰, δ15NAIR = 4.0
± 0.4‰), and were very different from those of Nycodenz
(δ13CVPDB = –30.3 ± 0.7‰, δ15NAIR = 0.0 ± 0.4‰), further supporting the inference that cell contamination by Nycodenz or
Tween 20 was negligible. The isotope measurement errors were
determined by measuring the isotopic content of glycine standards of known isotopic composition. Constant P abundances
also suggest no contamination from the cell extraction buffer.
523

Neveu et al.

Elemental analysis of cells separated from sediment

Table 1. Comparison between the trace element signal from sample-free digestions and from separated E. coli, in part per billion per
mass of solution analyzed (i.e., digestion diluted to 3.5 mL with 2% HNO3). These values were automatically produced by the mass
spectrometer software, which first subtracted the signal from an analytical blank (nondigested, trace metal-free 2% HNO3 solution) from
the physical mass spectrometer measurement, then averaged over three measurements of the same sample. The uncertainty represents
the standard deviation between these three measurements of a given sample.
Element
(ppb)
23

Na
Mg
27
Al
31
P
39
K
44
Ca
48
Ti
51
V
52
Cr
55
Mn-H2
55
Mn
54
Fe
56
Fe
57
Fe
59
Co
58
Ni
60
Ni
63
Cu
65
Cu
66
Zn
68
Zn
95
Mo

24

16 December 2011
95 ± 25
0.4 ± 0.2
0.3 ± 1.3
68 ± 14
5.3 ± 5.8
119 ± 58
N/A
0.004 ± 0.012
0.042 ± 0.038
N/A
N/A
1.1 ± 0.7
0.25 ± 0.14
0.32 ± 0.24
0.10 ± 0.02
0.27 ± 0.08
0.7 ± 0.2
–0.08 ± 0.02
0.04 ± 0.02
–0.18 ± 0.12
–0.003 ± 0.044
0.11 ± 0.01

Sample-free digestion
21 April 2012
4 June 2012
–0.3 ± 1.1
0.5 ± 0.1
1.9 ± 0.2
41.6 ± 0.9
–7.7 ± 11.2
–6.2 ± 5.8
0.40 ± 0.06
0.006 ± 0.024
0.075 ± 0.034
N/A
N/A
4.7 ± 1.6
2.0 ± 0.2
1.9 ± 1.1
–0.01 ± 0.02
0.03 ± 0.19
–0.8 ± 0.2
0.63 ± 0.03
0.37 ± 0.05
0.38 ± 0.04
N/A
–0.12 ± 0.02

–20.3 ± 9.5
0.6 ± 0.2
7.6 ± 5.1
118 ± 4
–38 ± 22
–63 ± 10
3.5 ± 0.4
0.076 ± 0.024
0.42 ± 0.03
N/A
N/A
137 ± 4
0.25 ± 0.79
–26 ± 4
–0.03 ± 0.01
–0.12 ± 0.01
–0.85 ± 0.05
2.2 ± 0.7
0.9 ± 0.7
0.9 ± 0.3
0.79 ± 0.39
0.07 ± 0.01

20 November 2012

Example-separated
E. coli

14.9 ± 0.9
1.31 ± 0.04
2.23 ± 0.03
42.8 ± 3.6
–7.8 ± 1.0
21 ± 11
0.05 ± 0.04
–0.098 ± 0.014
0.22 ± 0.06
0.032 ± 0.008
0.024 ± 0.034
2.1 ± 0.6
1.4 ± 0.2
1.6 ± 1.3
0.000 ± 0.002
0.007 ± 0.018
0.007 ± 0.016
–0.98 ± 0.04
–0.59 ± 0.04
0.4 ± 0.1
0.35 ± 0.06
0.37 ± 0.01

(2.6 ± 0.009) × 107
(9.4 ± 0.04) × 105
(6.4 ± 0.3) × 103
(1.9 ± 0.02) × 107
(2.5 ± 0.02) × 106
(2.0 ± 0.2) × 105
(1.4 ± 0.2) × 103
43.3 ± 8.3
8.3 ± 9.1
(5.1 ± 0.2) × 103
(4.5 ± 0.02) × 103
(2.1 ± 0.2) × 105
(2.1 ± 0.04) × 105
(2.1 ± 0.2) × 105
(2.6 ± 0.04) × 103
(2.7 ± 0.07) × 103
(3.2 ± 0.04) × 103
(5.3 ± 0.09) × 103
(5.0 ± 0.04) × 103
(4.6 ± 0.05) × 104
(4.2 ± 0.07) × 104
(4.8 ± 0.09) × 103

ples confirmed the applicability of the cell-sediment separation method for hot spring sediment samples (Fig. 1B). Four
phases—aqueous, cell, Nycodenz, and sediment—could be
seen after centrifugation of all samples, although the cell
phase was at times difficult to distinguish in samples with low
bulk biomass content.
For each cell separation of field-collected samples, we also
separated a 1-mL E. coli aliquot, not mixed with any sediment,
from the same culture as that used in the above element conservation studies. This extra E. coli separation was an independent check on the integrity of the method, allowing us to
track anomalous element variations or recovery yields.
Cell recovery yields
Cell recovery yields were estimated for a few samples using
the same method as for E. coli cells. Cell counting also provided
an opportunity to examine the isolated and resuspended cell
fraction to check for the presence of detrital and sedimentary
particles. Cell counts for the field-collected samples indicated
recovery yields between 5 and 40% (Table!2). These rates, lower
than the 95% E. coli recovery rate, likely indicate that a significant fraction of the cells was still bound to sediment. Different
sediment characteristics, such as humic content and grain size,
may affect the cell separation efficiency. Moreover, the 0.85%

separations. When cell separation caused a K enrichment in E.
coli cells, we suspect a contamination by trace K+ from the
extraction buffer: we measured 10-fold differences in K abundances in solutions between the buffer and supernatant
phases, where most buffer should end up. Finally, the mismatch in E. coli Na is not surprising, given its ubiquity in the
buffer and wash solution and the fact that Na+ can pass
through the cell membrane.
Application to field samples
Cell-sediment separation
Due to ongoing research projects at hot springs within Yellowstone National Park (Cox et al. 2011; Havig et al. 2011), we
sought to determine cellular elemental ratios in samples from
the area’s diverse hot spring ecosystems. Field sediment samples were collected in July and August 2009 at hot springs with
temperatures ranging from 20°C to boiling (~93°C). The pH of
the springs ranged from 1 to 9. Four replicate 15-mL tubes
were filled with sediment to 5 mL. Sediment samples without
a cryoprotectant added were kept on ice in the field and then
frozen at –20°C until transport back to Arizona State University, where samples were transferred to a –80°C freezer.
Cells were separated using the same method as for the E.
coli samples. Visual examination of the freshly separated sam524

Neveu et al.

Elemental analysis of cells separated from sediment

Table 2. Cell recovery yields for several separated samples collected at hot springs of Yellowstone National Park, as well as cultured E.
coli. Uncertainties are one standard deviation over three replicates and were propagated for the recovery yields. Note that the names
given to sampling sites are unofficial and correspond only to our own sampling program. All samples were frozen following sampling
and thawed before separation.

Sample

Tspring (°C)

pHspring

Site description

‘Bison Pool’

66.2

8.1

‘Green Cheese’
‘Tinkerbell’

57.4
61.4

6.1
4.7

‘Mr. Clean’

76.4

6.1

E. coli

N/A

N/A

Outflow of siliceous sinter
spring, phototrophic site
Phototrophic site
Red, turbid, Al-rich spring,
chemotrophic site
Black, turbid spring,
chemotrophic site
Cell culture

Cell concentration
in unseparated
sample (per g
sediment)

Cell concentration
in separated
sample (per g
sediment separated)

Cell recovery
yield
(±SD)

(3.6 ± 2.8) 108

(9.8 ± 6.0) 107

28 ± 27%

(2.9 ± 1.2) 108
(7.5 ± 2.4) 108

(1.5 ± 0.1) 107
(2.8 ± 0.9) 107

5 ± 2%
37 ± 16%

(3.7 ± 1.3) 108

(2.8 ± 1.2) 107

8 ± 4%

(2.4 ± 0.03) 1010*

(2.2 ± 0.03) 1010†

95 ± 2%

*

Sum of the numbers of E. coli cells in all four separation phases and all three washes for 1 g E. coli culture separated.
†
Number of E. coli cells in cell layer only (Fig. 1B) for 1 g E. coli culture separated.

reported for microorganisms (see Sterner and Elser [2002] for a
review). The chemosynthetic community sampled had C:N:P
molar ratios of 850:109:1. Thus, its C:N ratio of 7.8 is similar
to that measured by Redfield (1958) for marine phytoplankton
(C:N = 106:16 = 6.6). To our knowledge, these data constitute
the first measurement of the C:N:P ratios of a hot spring
chemosynthetic community.
The abundances of metals in Yellowstone cells collected at
a dozen sites are shown in Fig.!4. Most metal abundances (Mg,
Al, Ca, Ti, Mn, Fe, Cu, and Mo) varied from site to site by
orders of magnitude whereas others (Ni and Zn) were similar
across all the sites. Al, Ca, and Fe tended to be the most abundant metals. The Al content of cells should be on the order of
1 ppm (Bowen 1979; Bruland et al. 1991) but was measured at
0.1 to 1% in the samples. Such artificially high abundances
can result from non-biomass sediment matter (e.g., particles
with similar densities to cells) remaining in the cell fraction.
In the bulk sediment, the amount of Al was 1 to 5% of the dry
mass, ten times higher than in the separated cell fraction. If
100% of the Al measured in the cell fraction was sediment,
then sediment comprised 10% of the dry mass of separated
sample analyzed, the rest being cells. This 10% mineral contribution is roughly represented by the segmented line in Fig.
4, calculated as the geometric average over all field samples of
the sediment contribution, assuming sediment makes up 10%
of the dry mass of separated sample analyzed. Because sediment abundances vary significantly between field samples,
this line should only be taken as a guide to the eye. As shown
on Fig. 4, sediment accounts for most of the measured Ti and
Mn in the separated samples, and part of the measured Mg,
Ca, and Fe. However, sediment contribution is negligible for
C, N, P, Ni, Cu, Zn, and Mo. For all bioessential metals (Mg,
Ca, Mn, Fe, Ni, Cu, Zn, and Mo), the metal:Al ratio is several-

NaCl wash solution may not always closely match the osmolarity of field samples given their considerable compositional
diversity; the resulting osmotic stress may have lysed some
cells. However, the reported yield percentages are in line with
literature data (Lindahl and Bakken 1995; Kallmeyer et al.
2008; Morono et al. 2013).
Elemental composition of cells separated from sediment
When the separated cell phase was too small for quantitative analysis by both IRMS and ICP-MS, determination of C
and N content by IRMS of three to four replicates was given
priority, but we still kept one replicate for ICP-MS analysis.
This was the case for seven Yellowstone samples out of the
eleven separated.
Sample preparation for elemental analysis was carried out
as described previously for E. coli. Samples for ICP-MS analysis
were considered digested if the solution was clear. Turbid solutions were digested again by repeating the addition of 3 mL
nitric acid and heating overnight (usually 2 repeats). If turbidity was still observed, aqua regia (1 mL of concentrated nitric
acid and 3 mL of concentrated hydrochloric acid) was added
and samples heated overnight for up to 10 repeats, until the
turbidity disappeared. Rarely, samples contained particles that
resisted digestion. These particles, considered residual sediment, were weighed and discarded; their mass was subtracted
from those of the samples to calculate final element concentrations but was usually negligible.
Exemplary results are presented in Figs. 3 and 4 and Web
Appendix A, in which all elemental abundances are reported
as a percentage or part-per-million (ppm) of dry mass. The
C:N:P relationships for four communities are shown in Fig.!3.
C and N abundances were corrected to account for the 10%
loss measured in E. coli. The hot spring photosynthetic communities sampled had C:N:P ratios in the range normally
525

Neveu et al.

Elemental analysis of cells separated from sediment

Fig. 3. C:N, C:P, and N:P ratios of cells separated from sediment collected at photosynthetic- and chemosynthetic-dominated Yellowstone hot springs.
Error bars show one standard deviation uncertainties determined from both instrumental uncertainties by comeasurement of standards with the samples, and from one standard deviation over a minimum of three replicates. C and N abundances were corrected for a 10% loss during cell-sediment separation. Elemental abundances are reported as percentages of dry mass.
tion. We minimized this factor by freezing the samples only
once, in the field, since separations cannot be performed
there; we also thawed the samples gently.
Cellular C and N contents can be determined but appear to
require correction of measured abundances for a 10% depletion during separation. Abundances of P, Mg, Al, Ca, Ti, Mn,
Fe, Ni, Cu, Zn, and Mo were conserved during separation and
could be determined by ICP-MS with an accuracy of 30% (1
SD) or better. Whereas V, Cr, and Co contents are likely conserved as well during separation, their cellular abundances
were too low in our samples to obtain an accurate determination by ICP-MS.
Trace element abundances in field-collected samples show
that there remains significant sediment contamination for elements enriched in sediments and depleted in cells, such as Al,
Ti, Mn, and Fe, after the separation procedure (Fig. 4). Provided that the mineralogy of the sediment samples is accurately known, it may be possible to apply an assumed correction factor for sedimentary contribution, for example by using
Al or Ti as a lithogenic tracer (Frew et al. 2006). We attempted
to use Al, assumed nonbiological in origin, to estimate the
extent of mineral contamination of purified cells. Aluminum
abundances by mass range between 1 ppm and 0.1% in
microorganisms, but between 0.9% and 30% in rocks, soil,
and sediment (Bowen 1979; Bruland et al. 1991). E. coli, not
mixed with sediment, had an average Al content of 5 ± 2.5
ppm, both with and without separation. (Uncertainties for
both E. coli and field samples are 1 SD). These amounts are
consistent with Al trapped in the cell membrane of E. coli, as
measured by Coughlin et al. (1983). Al could also have been

fold higher in separated samples compared with bulk sediment; whereas the Ti:Al is several-fold lower. We take this to
indicate removal of at least part of the sediment contribution
to measured elemental abundances in separated samples.

Discussion
These data show that the cell separation method we
applied is successful at separating microbial cells from hot
spring sediments, and based on the E. coli assessments, promising for retaining composition across a broad suite of biologically important elements.
The cell recovery yield was low for some field sites. Because
our original samples were relatively large, these recovery yields
were satisfactory to obtain sufficient material to determine an
elemental composition profile of the sediment community.
However, some microorganisms in a given community might
be more easily separated than others from the sediment
matrix; therefore there is likely to be further heterogeneity in
elemental composition that this method cannot resolve. Such
data can only be obtained by methods such as single-cell
analyses via X-ray fluorescence (Heldal et al. 2003; Steenbergh
et al. 2013).
This technique cannot separate detrital material (biogenic
or otherwise) with a density similar to cell material, but a different element stoichiometry, from intact cells. In some samples, detrital and sedimentary particles were observed along
with cells even after separation; however, the amount of detritus was much smaller compared with unseparated samples.
Another point of uncertainty arises from possible cell damage due to freezing and thawing of the samples before separa526

Neveu et al.

Elemental analysis of cells separated from sediment

sediment constituted up to 50% of the separated fraction). For
these metals, analyses can be interpreted relatively unambiguously in terms of cellular content, provided that supporting
data (such as high Cu:Al, Zn:Al, and Mo:Al compared with
sediment) are reported.
For the purposes of determining the C:N:P stoichiometry of
samples, rejection of separated samples with a C content less
than typical for microbes (35% to 50%, Sterner and Elser
2002) should suffice. This assumes that carbonates are not a
major component of the sediment sample; if they are, a carbonate removal step before separation can be added
(Kallmeyer et al. 2008). In practice, detrital biological material
with a C:N:P stoichiometry different from that of the cells but
with a similar density could also bias the measured C:N:P
ratios of cells; this is a fundamental limitation of this method.
Thus, with further refinement, this cell separation method
enables the determination of elemental contents, in mass per
dry cell mass, of C, N, P, and key trace biological elements in
sediment communities. This procedure is complementary to
laborious electron microscopy methods that investigate the
elemental composition of single cells (Heldal et al. 2003;
Steenbergh et al. 2013). However, using both approaches
would allow comparison of element distributions across broad
components of a microbial community with those of the individual cells comprising the biomass.
This method will help quantify the multi-element stoichiometric envelope within which microbes operate (Sterner and
Elser 2002; Ho et al. 2003; Twining et al. 2011; Cameron et al.
2012; Nuester et al. 2012). It will also allow comparisons of
microbial elemental compositions with those of their environment to determine their mutual influence (Havig et al. 2011).

Fig. 4. Metal abundances in cells extracted from sediment samples collected at a dozen photosynthesis- and chemosynthesis-dominated Yellowstone hot springs. Elemental abundances are reported as ppm of dry
mass. To avoid clutter, indicative error bars show typical one standard
deviation uncertainties for this sample set, determined from both instrumental uncertainty by co-measurement of a standard with the samples,
and from one standard deviation over at least three replicates. If not
enough replicates were available, uncertainties were determined from
one standard deviation over at least four replicates of E. coli separated and
prepared along with the samples. As can be seen in Web Appendix A, the
variability between replicates for E. coli and field samples is comparable
and tends to scale with abundance. Therefore, the error bars provide an
indicative estimate of the variability between replicates anywhere on the
y-axis. The segmented line represents the mineral contribution to elemental abundances if the samples analyzed consist of 10% sediment and
90% cells, as estimated from the Al content measured in separated samples compared to that measured in bulk sediment (see text for details).

Comments and recommendations
Overall, we suggest that this method for separating microbial cells from sediment and analyzing their elemental contents at the community level should be applicable for samples
from a wide variety of environments, such as soils, soil crusts,
or lacustrine or riverbed sediments. Because high temperatures
and pH extremes of hot springs result in low biomass of microbial cells, the abiotic contents of key elements can be unusually high, increasing the risk of elemental contamination of
cells during separation. The possibility of contamination is
likely lower for samples collected in milder conditions. This
method should also be applicable for water samples containing high concentrations of suspended sediments, provided
that enough sample is collected and filtered to obtain sufficient biomass for elemental analyses.
Based on our studies, here is a set of recommendations for
future applications:
(1) Energy parameters for the sonication step should be considered as approximate values. As Lindahl and Bakken
(1995) pointed out, the sonicator output depends on the
age of the electromechanical converter, due to changes in
its crystal structure.

picked up in minute amounts during autoclaving of the LB
solution, although Al is not part of LB medium itself. Al was
present in sample-free digestions at levels of about 0.1% of
that measured in E. coli, which makes the possibility of contamination during digestion unlikely. In half of the field samples, Al abundances were above 0.1%, reaching up to 1 ± 0.5%.
The bulk field sediment samples contained 1 to 5% Al. Therefore, assuming 100% of the Al in separated cell fractions is
from sediment, the separated cell fractions contained about
10% (up to 20%) sediment. However, the widely varying mineralogy of Yellowstone National Park hot springs complicates
the determination of corrections based on metal:Al ratios in
sediment and cells. We consider Al a worst-case contamination proxy, because Al3+ ions are more easily mobilized than
other metal ions by ligands, such as the pyrophosphate used
in the procedure (Williams 1996).
Another useful criterion is to reject data for samples or elements for which the metal:Al ratios closely mimic sedimentary ratios. Cu, Zn, Mo, and (to a lesser extent) Ni are systematically much higher than the sediment level (even if
527

Neveu et al.

Elemental analysis of cells separated from sediment

active influences of bioactive trace metals on biological
production in oceanic waters. Limnol. Oceanogr. 36:15551577 [doi:10.4319/lo.1991.36.8.1555].
Cameron, V., C. H. House, and S. L. Brantley. 2012. A first
analysis of metallome biosignatures of hyperthermophilic
archaea. Archaea 2012:789278 [doi:10.1155/2012/789278].
Coughlin, R. T., S. Tonsager, and E. J. McGroarty. 1983. Quantitation of metal cations bound to membranes and
extracted lipopolysaccharide of Escherichia coli. Biochemistry 22:2002-2007 [doi:10.1021/bi00277a041].
Cox, A., E. L. Shock, and J. R. Havig. 2011. The transition to
microbial photosynthesis in hot spring ecosystems. Chem.
Geol. 280:344-351 [doi:10.1016/j.chemgeo.2010.11.022].
Dupont, C. L., A. Butcher, R. E. Valas, P. E. Bourne, and G. Caetano-Anollés. 2010. History of biological metal utilization
inferred through phylogenomic analysis of protein structures.
PNAS
107:10567-10572
[doi:10.1073/pnas.
0912491107].
Epstein, W., and S. G. Schultz. 1965. Cation transport in E.
coli. V. Regulation of cation content. JGP 49:221-234
[doi:10.1085/jgp.49.2.221].
Falkowski, P. G. 1997. Evolution of the nitrogen cycle and its
influence on the biological sequestration of CO2 in the
ocean. Nature 387:272-275 [doi:10.1038/387272a0].
Frew, R. D., and others. 2006. Particulate iron dynamics during FeCycle in subantarctic waters southeast of New
Zealand. Glob. Biogeochem. Cycles 20:GB1S93 [doi:10.
1029/2005GB002558].
Haraguchi, H. 2004. Metallomics as integrated biometal science. J. Anal. Atom. Spectrom. 19:5-14 [doi:10.1039/
b308213j].
Havig, J. R., J. Raymond, D. R. Meyer-Dombard, N. Zolotova,
and E. L. Shock. 2011. Merging isotopes and community
genomics in a siliceous sinter-depositing hot spring. J. Geophys. Res. 116:G01005 [doi:10.1029/2010JG001415].
Heldal, M., D. J. Scanlan, S. Norland, F. Thingstad, and N. H.
Mann. 2003. Elemental composition of single cells of various strains of marine Prochlorococcus and Synechococcus
using X-ray microanalysis. Limnol. Oceanogr. 48:17321743 [doi:10.4319/lo.2003.48.5.1732].
Ho, T.-Y., A. Quigg, Z. V. Finkel, A. J. Milligan, K. Wyman, P. G.
Falkowski, and F. M. M. Morel. 2003. The elemental composition of some marine phytoplankton. J. Phycol.
39:1145-1159 [doi:10.1111/j.0022-3646.2003.03-090.x].
Jones, G. E., L. G. Royle, and L. Murray. 1979. Cationic composition of 22 species of bacteria grown in seawater
medium. Appl. Environ. Microbiol. 38:800-805. (see Web
Appendix A)
Kallmeyer, J., D. C. Smith, A. J. Spivack, and S. D’Hondt. 2008.
New cell extraction procedure applied to deep subsurface
sediments. Limnol. Oceanogr. Methods 6:236-245 [doi:10.
4319/lom.2008.6.236].
Lindahl, V., and L. R. Bakken. 1995. Evaluation of methods for
extraction of bacteria from soil. FEMS Microbiol. Ecol.

(2) Precise measurement of cellular dry masses is key to the
normalization of elemental abundances to mass and eventual determination of elemental ratios. Mass spectrometers
measure the elemental masses in a sample more precisely
than most balances measure dry masses of a few milligrams. Thus, we recommend to use a balance of microgram precision to measure the dry mass of recovered cell
phases.
(3) For future analyses of samples from a specific location,
standards used to determine the ICP-MS calibration curve
and digestion recovery should be chosen as close to sample composition as possible to yield measurements as accurate as possible.
(4) As a coarse means of screening out contamination, we suggest discarding samples of C content too low to be biological (lower limit of about 35 to 40%, Sterner and Elser
2002). However, even minute amounts of sediment can
significantly alter the metal content of the recovered cell
fraction. Therefore, the use of complementary, more precise contamination indicators is advised. We recommend
developing a correction factor for metal abundances based
on lithogenic tracers such as Al or Ti (Frew et al. 2006) if
the mineralogy of the samples is relatively uniform.
Reporting of metal:Al or metal:Ti ratios is also advised to
assess the reliability of a measurement.
(5) Finally, we recommend that follow-up work focuses on
determining the extent of the selective biases due to overrepresentation of microorganisms that are more easily separated from sediment grains relative to those that contribute the most to the measured community
composition. Energy-dispersive X-ray spectroscopy of single cells that remain in the sediment fraction may help
(Heldal et al. 2003; Steenbergh et al. 2013).

References
Amalfitano, S., and S. Fazi. 2008. Recovery and quantification
of bacterial cells associated with streambed sediments. J.
Microbiol. Methods 75:237-243 [doi:10.1016/j.mimet.
2008.06.004].
———, S. Fazi, and A. Puddu. 2009. Flow cytometric analysis
of benthic prokaryotes attached to sediment particles. J.
Microbiol. Methods 79:246-249 [doi:10.1016/j.mimet.
2009.09.005].
Anbar, A. D. 2008. Oceans. Elements and evolution. Science
322:1481-1483 [doi:10.1126/science.1163100].
Bowen, H. J. M. 1979. Environmental chemistry of the elements. Academic Press.
Bowler, C., M. Van Montagu, and D. Inzé. 1992. Superoxide
dismutase and stress tolerance. Annu. Rev. Plant Physiol.
Plant Mol. Biol. 43:83-116 [doi:10.1146/annurev.pp.43.
060192.000503].
Britten, R. J., and F. T. McClure. 1962. The amino acid pool in
Escherichia coli. Bacteriol. Rev. 26:292-335.
Bruland, K. W., J. R. Donat, and D. A. Hutchins. 1991. Inter528

Neveu et al.

Elemental analysis of cells separated from sediment

factors in the environment. Am. Sci. 46:205-221.
Rhoads, D. B., and W. Epstein. 1978. Cation transport in
Escherichia coli. IX. Regulation of K transport. JGP 72:283295 [doi:10.1085/jgp.72.3.283].
Rickwood, T., T. Ford, and J. Graham. 1982. Nycodenz: A new
nonionic iodinated gradient medium. Anal. Biochem.
123:23-31 [doi:10.1016/0003-2697(82)90618-2].
Stec, B., K. M. Holtz, and E. R. Kantrowitz. A revised mechanism for the alkaline phosphatase reaction involving three
metal ions. J. Molec. Biol. 299:1303-1311 [doi:10.1006/
jmbi.2000.3799].
Steenbergh, A. K., P. L. E. Bodelier, M. Heldal, C. P. Slomp, and
H. J. Laanbroek. 2013. Does microbial stoichiometry modulate eutrophication of aquatic ecosystems? Environ.
Microbiol. 15:1572-1579 [doi:10.1111/1462-2920.12042].
Sterner, R. W., and J. J. Elser. 2002. Ecological stoichiometry.
Princeton Univ. Press.
Twining, B. S., S. B. Baines, J. B. Bozard, S. Vogt, E. A. Walker,
and D. M. Nelson. 2011. Metal quotas of plankton in the
equatorial Pacific Ocean. Deep-Sea Res. II 58:325-341
[doi:10.1016/j.dsr2.2010.08.018].
Wade, H. E. 1952. Variation in the phosphorus content of
Escherichia coli during cultivation. J. Gen. Microbiol. 7:2430 [doi:10.1099/00221287-7-1-2-24].
Williams, R. J. P. 1996. Aluminium and biological systems: an
introduction. Coord. Chem. Rev. 149:1-9 [doi:10.1016/
S0010-8545(96)90004-1].
———, and J. J. R. Fraústo da Silva. 2000. The distribution of
elements in cells. Coord. Chem. Rev. 200-202:247-348
[doi:10.1016/S0010-8545(00)00324-6].

16:135-142 [doi:10.1016/0168-6496(94)00077-A].
Morel, F. M. M. 2008. The co-evolution of phytoplankton and
trace element cycles in the oceans. Geobiology 6:318-324
[doi:10.1111/j.1472-4669.2008.00144.x].
———, and N. M. Price. 2003. The biogeochemical cycles of
trace metals in the oceans. Science 300:944-947 [doi:10.
1126/science.1083545].
Morono, Y., T. Terada, J. Kallmeyer, and F. Inagaki. 2013. An
improved cell separation technique for marine subsurface
sediments: applications for high-throughput analysis using
flow cytometry and cell sorting. Environ. Microbiol.
15:2841-2849 [doi:10.1111/1462-2920.12153].
Nuester, J., S. Vogt, M. Newville, A. B. Kustka, and B. S. Twining. 2012. The Unique Biogeochemical Signature of the
Marine Diazotroph Trichodesmium. Front. Microbiol.
3:150-165 [doi:10.3389/fmicb.2012.00150].
Outten, C. E., and T. V. O’Halloran. 2001. Femtomolar sensitivity of metalloregulatory proteins controlling zinc homeostasis. Science 292:2488-2492 [doi:10.1126/science.106
0331].
Papineau, D., and others. 2009. High primary productivity
and nitrogen cycling after the Paleoproterozoic phosphogenic event in the Aravalli Supergroup, India. Precambr.
Res. 171:37-56 [doi:10.1016/j.precamres.2009.03.005].
Poté, J., A. G. Bravo, P. Mavingui, D. Ariztegui, and W. Wildi.
2010. Evaluation of quantitative recovery of bacterial cells
and DNA from different lake sediments by Nycodenz gradient centrifugation. Ecol. Indic. 10:234-240 [doi:10.1016/j.
ecolind.2009.05.002].
Price, N. M., G. I. Harrison, J. G. Hering, R. J. Hudson, P. M. V.
Nirel, B. Palenik, and F. M. M. Morel. 1989. Preparation and
chemistry of the artificial algal culture medium Aquil. Biol.
Oceanogr. 6:443-461.
Redfield, A. C. 1958. The biological control of the chemical

Submitted 11 November 2013
Revised 5 May 2014
Accepted 6 June 2014

529