The ISME Journal (2010) 4, 509–519

& 2010 International Society for Microbial Ecology All rights reserved 1751-7362/10 $32.00
www.nature.com/ismej

ORIGINAL ARTICLE
Microtoming coupled to microarray analysis to
evaluate the spatial metabolic status of Geobacter
sulfurreducens biofilms
Ashley E Franks, Kelly P Nevin, Richard H Glaven1 and Derek R Lovley
Department of Microbiology, University of Massachusetts, Amherst, MA, USA

Further insight into the metabolic status of cells within anode biofilms is essential for understanding
the functioning of microbial fuel cells and developing strategies to optimize their power output. Cells
throughout anode biofilms of Geobacter sulfurreducens reduced the metabolic stains: 5-cyano-2,3-
ditolyl tetrazolium chloride and Redox Green, suggesting metabolic activity throughout the biofilm.
To compare the metabolic status of cells growing close to the anode versus cells in the outer portion
of the anode biofilm, anode biofilms were encased in resin and sectioned into inner (0–20 lm from
anode surface) and outer (30–60 lm) fractions. Transcriptional analysis revealed that, at a twofold
threshold, 146 genes had significant (Po0.05) differences in transcript abundance between the inner
and outer biofilm sections. Only 1 gene, GSU0093, a hypothetical ATP-binding cassette transporter,
had significantly higher transcript abundances in the outer biofilm. Genes with lower transcript
abundance in the outer biofilm included genes for ribosomal proteins and NADH dehydrogenase,
suggesting lower metabolic rates. However, differences in transcript abundance were relatively low
(othreefold) and the expression of genes for the tricarboxylic acid cycle enzymes was not
significantly lower. Lower expression of genes involved in stress responses in the outer biofilm may
reflect the development of low pH near the surface of the anode. The results of this study suggest
that cells throughout the biofilm are metabolically active and can potentially contribute to current
production. The microtoming/microarray strategy described here may be useful for evaluating gene
expression with depth in a diversity of microbial biofilms.
The ISME Journal (2010) 4, 509–519; doi:10.1038/ismej.2009.137; published online 24 December 2009
Subject Category: integrated genomics and post-genomics approaches in microbial ecology
Keywords: Geobacter; microtoming; microbial fuel cell; transcription; biofilm

Introduction and are still a matter of considerable debate (Lovley,

2006, 2008; Rittmann et al., 2008; Logan, 2009).
One of the most surprising claims in the study of A key assumption in this model for current
microbial respiration in past five years is the production is that the cells at a distance from the
suggestion that microorganisms are capable of anode are metabolically active. Indirect evidence for
long-range electron transfer through the biofilms this in G. sulfurreducens anode biofilms (Reguera
that form on the anodes of microbial fuel cells. et al., 2006) was the finding that cells throughout the
The first evidence for this was the finding that anode biofilm fluoresced green when treated with
the current output of microbial fuel cells of the LIVE/DEAD BacLight bacterial viability kit from
Geobacter sulfurreducens increased in direct propor- Molecular Probes (Eugene, OR, USA), designed to
tion to the accumulation of biomass on the anode stain metabolically active cells green and metaboli-
surface and with the increasing height of the anode cally inactive cells red (Boulos et al., 1999).
biofilm (Reguera et al., 2006). This suggested that Studies that have attempted to model the dis-
cells at a substantial distance from the anode surface tribution of microbial activity in anode biofilms
(that is, 50–100 mm) contributed to current produc- have suggested that different layers of the biofilm
tion. The mechanisms for this potential long-range may have substantially different activities, but
electron transfer have been intensively investigated which layers are predicted to be most active
depends on the assumptions incorporated into the
Correspondence: AE Franks, Department of Microbiology, modeling. For example, inital models predicted
University of Massachusetts, Morrill IVN, 639 North Pleasant large variation in biofilm thickness and activity
Street, Amherst, MA, USA. depending on the conductivity of the biofilm. In low
E-mail: [email protected] conductive biofilms, it was estimated that almost all
1
Current address: Center for Biomolecular Science and Engi-
neering, Naval Research Laboratories, Washington DC, USA. the active biomass would be located within the first
Received 24 August 2009; revised 5 November 2009; accepted 5 10 mm from the anode surface, with inert biomass
November 2009; published online 24 December 2009 dominating the biofilm at a distance greater then
 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
510
3 mm (Marcus et al., 2007). If a higher degree of likely to have application for the study of gene
conductance was assumed then thicker (ca. 50 mm) expression in other types of biofilms as well.
biofilms were predicted, but it was predicted that
dual limitations of substrate concentration and
localized potential differences within the biofilms Materials and methods
would create a zone of 20–30 mm from the anode
surface in which metabolic activity was the highest Staining for metabolic activity
(Marcus et al., 2007). When the release of protons Microbial reduction of metabolic stains in current-
within the biofilm was considered, it was predicted producing anode biofilms was imaged in the
that the generation of low pH near the anode surface previously described microbial fuel cells that permit
would limit metabolism in this zone, and that real-time imaging of the current producing biofilm
members of the biofilm at a distance from the anode (Franks et al., 2009). These microbial fuel cells
would have higher rates of metabolism and contain a solid graphite (grade G10; Graphite
contribute more to current production (Torres Engineering and Sales, Greenville, MI, USA) anode
et al., 2008). (0.81 cm2) in a flow-through chamber (0.5 mm deep
Experimental evidence also suggests that the and 6.35 mm wide), which allows microscopic
environment within anode biofilms is likely to be observation through a cover slip. The system was
far from homogenous. For example, protons accu- inoculated with G. sulfurreducens pRG5Mc that
mulate within anode biofilms of G. sulfurreducens, constitutively produces the fluorescent protein,
particularly near the anode surface, lowering the pH mcherry, allowing fluorescent detection of the cells
to levels that may inhibit metabolism (Franks et al., throughout the biofilm (Franks et al., 2009). As
2009), as previously predicted in modeling studies previously described (Franks et al., 2009), acetate
(Torres et al., 2008). Gradients in electron donor (10 mM) was the electron donor for current produc-
availability (Logan and Regan, 2006; Marcus et al., tion and fresh medium was continuously supplied
2007) and possibly other important environmental at a flow rate of 0.1 ml min1.
parameters are also likely. The impacts of these Metabolic activity in the anode biofilms was
gradients on the metabolic status of cells throughout evaluated using two different dyes. When in the
the anode biofilm are yet unknown. oxidized state, 5-cyano-2,3-ditolyl tetrazolium
Previous studies have demonstrated that it is chloride (CTC; BacLight RedoxSensor CTC Vitality
possible to obtain significant insights into the Kit; Molecular Probes, Eugene, OR, USA) is soluble,
functioning of anode biofilms through an analysis colorless, and exhibits no fluorescence. Components
of gene transcript abundance. Genome-scale analy- of microbial electron transfer chains can reduce
sis of transcript abundance in cells growing in anode CTC to form a fluorescent intracellular, insoluble
biofilms compared with transcript levels either in formazan (Rodriguez et al., 1992). The CTC staining
planktonic cells (Holmes et al., 2005) or biofilm cells has been found to be useful in evaluating the
using fumarate as the electron acceptor (Nevin et al., metabolic state of a diversity of microorganisms,
2009), has revealed genes whose expression is including Geobacter species, (Bhupathiraju et al.,
specifically upregulated in cells producing current 1999; Gruden et al., 2003) and for localizing
and has helped identify components, such outer- metabolic activity within biofilms (Schaule et al.,
surface c-type cytochromes and pili, that seem to be 1993; Huang et al., 1995; Zheng and Stewart, 2004).
important in electron transfer to the anode. Quanti- Redox Sensor Green reagent (Redox green;
fying the level of transcripts of the gene for citrate BacLight RedoxSensor Green Vitality Kit; Molecular
synthase, a key gene in the tricarboxylic acid (TCA) Probes) yields green fluorescence when modified by
cycle, demonstrated that citrate synthase transcript bacterial reductases (Gray et al., 2005). This dye has
abundance within the anode biofilm was directly been used to detect metabolism of methylotrophs in
related to rates of current production (Holmes et al., lake sediments and may be less likely to inhibit
2005). However, these previous studies evaluated microbial metabolism than CTC (Kalyuzhnaya et al.,
transcript abundance within the complete anode 2008).
biofilm, averaging any differences with depth in the Fuel cells were grown to maximum power for
biofilm, and could not account for potential differ- staining. Medium was amended with CTC (5 mM) or
ences in gene expression within different zones of Redox green (0.1 mM) and 10 ml of these solutions
the anode biofilm. This is a well-recognized limita- where injected into the anode chamber. The micro-
tion of transcriptional profiling in biofilms in bial fuel cells were incubated without flow in the
general (An and Parsek, 2007). dark for 30 min and then washed with fresh water
The purpose of this study was to evaluate the acetate medium under a flow rate of 0.1 ml/min1 for
hypothesis that there are major differences in 30 min in the dark. The MFCs were immediately
the metabolic states between cells close to the anode imaged with a Leica TCS SP5 microscope (Leica
surface and those at a greater distance from Microsystems GmbH, Wetzlar, Germany) with a HCX
the anode. The new technique for evaluating gene APO  63 (numerical aperture: 0.9) objective.
expression at different depths in the anode biofilm, Images where processed and analyzed with Leica
which was developed to address this question, is LAS AF software (Leica). Consecutive line scanning

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
511
was used to detect CTC (excitation 488 nm/emission placed on ice and prepared by wiping the inside
630 nm), redox green (488 nm/emission 520) with a sterile cotton wool bud with a single drop of
and mcherry red fluorescent protein that the accelerator (#02648-AB, SPI Supplies, West Chester,
cells constitutively produced (excitation 588/ PA, USA). Excess accelerator was removed before
emission 600). the graphite anode sections were place into the
mold. A volume of 10 ml of London Resin White
resin was mixed with one drop of accelerator and
Transcriptional profiling immediately used to fill the space in the mold.
Studies on gene expression in G. sulfurreducens Samples were allowed to cure overnight at 4 1C. This
anode biofilms were conducted with wild-type procedure produced a solid block of the intact
G. sulfurreducens strain PCA (ATCC 51573, DSMZ biofilm associated with the graphite anode for
12127; Caccavo et al., 1994). Cells were grown in sectioning (Figure 1).
‘H-cells’ with a continuous flow of medium that The blocks were cut into 100-nm shavings with a
contained acetate (10 mM) as the electron donor, as 451 diamond cyro knife (Diatome, Hatfield, PA,
previously described (Reguera et al., 2006), with the USA) in a Leica Ultracut UCT ultramicrotome (Leica
exception that the anode was modified to permit Microsystems, Bannockburn, IL, USA). Embedded
sub-sampling for microtoming. The usual ‘stick’ sections were trimmed and three-dimensionally
graphite anode, consisting of a 2.5  7.5  1.25 cm positioned to take 100-nm shavings parallel to the
piece of unpolished graphite, was replaced with two anode surface at a cutting speed of 1 mm s1.
polished graphite electrodes of 2.5  7.5  0.1 cm Inspection of the sample and knife position using
with grooves cut from the base to form five graphite a  20 magnification microscope of the ultramicro-
‘fingers’ of 0.2  0.1 cm. These dual electrodes tome helped to ensure correct positioning. Shavings
provided a surface area of 60 cm2, which is compar- were pooled every 50 slices, representing 5-mm
able to the 67.5 cm2 of the usual electrodes and the sections, and stored on ice. The block and knife
fingers could be easily cut from the main body of the were cleaned with a sterile brush and compressed
electrode for microtoming studies. air after each of these 50 slices allocates to reduce
Electrodes with mature anode biofilms which cross contamination of sections. The microtoming
had reached maximum power production levels continued until the graphite surface was reached, as
(ca. 16 mA), corresponding to maximum biofilm thick- evidenced by black graphite visually apparent in
ness, were removed and 2 ml of RNAprotect Bacteria the shavings. Sections were separated into inner
Reagent (Qiagen, Alameda, CA, USA) was immedi- (0–20 mm from the surface) and outer (30–60 mm
ately added to the side of the anode to be analyzed from the surface) sections, flash frozen in an ethanol
for gene expression. After 5 min, the graphite finger dry ice bath and stored at 80 1C until processed for
portions of the anode were cut into 5-mm sections RNA extraction.
without disturbing the anode biofilm. The RNApro- The RNA was extracted from the biofilm sections
tect Bacterial Reagent was removed by touching the with an acetone extraction method, as previously
other side of the electrode with a clean sterile Kim described (Holmes et al., 2004). Samples were
Wipe tissue. The sections were then covered in the crushed with the end of sterile glass hockey sticks
low viscous hydrophilic London Resin White for 2 h in 2-ml tubes and suspended in 800 ml of 4 1C TPE
at 4 1C. Excess London Resin White was then buffer (100 mM Tris–HCl, 100 mM KH2PO4, 10 mM
removed with a Kim Wipe tissue as before. EDTA (pH 8)). Plant RNA Isolation Aid (100 ml;
Disposable Flat Embedding molds (#70906, Electron Ambion, Austin, TX, USA) and 1 ml of 20 1C
Microscope Services, Columbia, MD, USA) were acetone was added. The mixture was mixed by

Biofilm

Biofilm

Anode Anode

Figure 1 Current producing biofilm embedded in LR (London Resin) White resin. For these images under light (a) or phase-contrast (b)
microscopy the resin was cut perpendicular to the biofilm surface in contrast to the microtoming in parallel used for sectioning, to reveal
the full biofilm and its association with the graphite. The embedded biofilm is indicated by the arrows. Remains of the graphite anode,
some of which was lost during the perpendicular cutting, is labeled as anode.

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
512
inverting ca. 40 times and then centrifuged at Quantitative real-time PCR (qRT–PCR) analysis
16 000  g for 10 min. The supernatant was Gene expression patterns for select genes were
discarded and 2 ml of Superase-In RNase inhibitor further evaluated using quantitative RT–PCR. For-
(Ambion) was added to the pellet, which was then ward and reverse primers were designed with
resuspended in 1 ml of diethyl pyrocarbonate- Primer3 software (Rozen and Skaletsky, 2000) and
treated water (Ambion). A volume of 10 ml of are listed in Supplementary Table 1. Each reaction
lysozyme (50 mg ml1), 5 mg of proteinase K was performed in triplicate for each biological
(20 mg ml1), and 30 ml of 10% sodium dodecyl replicate for each gene tested.
sulfate solution were added and the mixture was Single strand cDNA was created through reverse
heated to 37 1C for 10 min. The supernatant was transcription of 2 mg cRNA in a 100-ml reaction
separated by centrifugation at 16 000  g for 15 min. volume of TaqMan Reverse Transcription Reagents
Plant RNA Isolation Aid (50 ml), 10 ml of tRNA (Applied Biosystems, Foster City, CA, USA) as a
(10 mg ml1; Ambion), 600 ml of heated acidic phenol template for real-time PCR. Forward and reverse
(pH 4.5, 70 1C, Ambion), and 400 ml of chloroform: primers were added at a final concentration of
isoamyl alcohol (24:1, Sigma, St Louis, MO, USA) 200 nM to SYBR Green PCR Master Mix (Applied
were added. The supernatant was mixed on a rotary Biosystems) and 1 ml of 1:10 diluted template cDNA.
shaker for 5 min and then centrifuged at 16 000  g The incorporation of SYBR Green dye into the PCR
for 5 min. The aqueous layer was removed and products was detected in real time on an ABI Prism
mixed with 100 ml 5 N NH4OAc (Ambion), 20 ml 7900HT Sequence Detection System (Applied Bio-
glycogen (Ambion) and 1 ml of 20 1C isopropanol systems). Normalization of non-PCR-related fluor-
(Sigma). The RNA was precipitated for 1 h at 20 1C, escent signal variation was performed with a ROX
centrifuged at 16 000  g for 30 min and washed with (6-carboxyl-X-rhodamine) passive reference dye.
70% EtOH at 20 1C. Pellets where then dried at The SYBR Green incorporation was used to deter-
room temperature and resuspended in 50 ml of mine the threshold cycle (Ct), which identifies the
diethyl pyrocarbonate-treated water. PCR cycle at which exponential production of the
Resuspended pellets were cleaned with PCR amplicon begins. Standard curves were deter-
RNeasy RNA clean up kit (Qiagen) and mined for each cDNA sample analyzed for each
DNA-free DNase Kit (Ambion) following the primer set. Expression was normalized to the house
manufactures’ instructions and tested for genomic keeping gene proC (Holmes et al., 2005). To verify
DNA contamination as previously described (Postier amplification and correct amplicon size, aliquots
et al., 2008). from qRT–PCR were examined on an ethidium
The RNA extracts had A260/280 ratios, as bromide stained 2% agarose gel.
determined with a Nanodrop ND-1000 spectro-
photometer (NanoDrop Technologies, Wilmington,
DE, USA), of 1.8–2.0, indicating high purity Results and discussion
(Ausubel et al., 1997). Total RNA (0.5 mg) was
amplified and biotin-labeled using the MessageAmp Evaluation of metabolic activity with redox stains
II-Bacteria Kit (Ambion) as previously described Current production and biofilm formation in
(Postier et al., 2008) following the manufacturer’s the flow-through cells designed for real-time
instructions. imaging were the same as previously reported
Synthesis of cDNA, array hybridization and (Franks et al., 2009). A maximum current density
imaging were performed at the Genomic Core of 3.5 A m2 was achieved within 300 h of inocula-
Facility at the University of Massachusetts Medical tion and this current level could be maintained
Center. A total of 10 mg of amplified cRNA was used indefinitely. As previously reported (Franks et al.,
as template to synthesize labeled cDNAs using 2009), the biofilm covered the entire available
Affymetrix (Santa Clara, CA, USA) GeneChip DNA graphite anode surface and formed pillar-structured
Labeling Reagent Kits. Labeled cDNA samples were biofilms greater than 50 mm thick (Figures 2a and
hybridized to Affymtrix GeneChip G. sulfurreducens 3a). Although a low shear and turbulence forces due
arrays according to Affymetrix guidelines. After to a low flow rate may have favored a more complex
hybridization, the arrays were scanned with a biofilm structure, this biofilm structure was
GeneChip 3000 Scanner and normalized gene similar to those observed in a diversity of other
expression data were analyzed by ArrayStar microbial fuel cell designs (Reguera et al., 2006;
(DNASTAR, Madison, WI, USA) using the Robust Nevin et al., 2009), suggesting that the flow of
Multichip Average algorithm. A Student t-test with a medium through the fuel cell did not alter typical
cutoff P-value of 0.05 was used to compare the mean biofilm morphology.
gene expression value and differentially regulated The addition of CTC resulted in a 15–20%
genes were detected using a minimum twofold decrease in current, consistent with the CTC serving
change from this group. Microarray data have been as an electron acceptor for electrons derived from
deposited with the NCBI Gene Expression Omnibus acetate oxidation. When the CTC was removed
(http://www.ncbi.nlm.nih.gov/geo) under accession current production returned to the level observed
numbers GSE17591. before the addition of CTC.

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
513

Figure 2 Confocal scanning laser micrographs of anode biofilms treated with the metabolic stain 5-cyano-2,3-ditolyl tetrazolium
chloride (CTC). Geobacter sulfurreducens pRG5Mc, which constitutively produces the fluorescent protein mcherry, was imaged here as
green. Top down three-dimensional images of the cellular biomass (a), the purple insoluble formazan produced from the microbial
reduction of CTC (b) and both images superimposed (c). Biomass (d) and formazan (e) in perpendicular cross-sections. The circled areas
illustrate rare zones within (b and c) or on the outer surface (d and e) of the biofilm in which the cells did not reduce CTC. White
bar ¼ 75 mm.

After staining the current-producing biofilm with Transcriptional analysis of outer versus inner members
the CTC, the red fluorescent formazan that is a of current-producing biofilms
product of microbial reduction of CTC could be To obtain more detailed information on metabolic
detected, but only were cells were present status of cells within the biofilm, gene transcript
(Figure 2c). The CTC level was reduced through abundance in cells growing near the anode was
out most of the biofilm (Figures 2b and c), including compared with those at the outer surface with
the outer pillared structures that reached up to whole-genome microarray analysis. Current produc-
50 mm from the anode surface. There were a few tion (Figure 4a) in the flow-through ‘H-cell’ design
isolated zones within the biofilms that did not with the ‘fingered’ graphite anodes that permitted
reduce CTC and were apparently metabolically portions of the anode to be cut for microtoming was
inactive (Figures 2b and c). In rare instances, the comparable to the power production previously
cells in the very top of a pillar (B50 mm from the reported for the same type of H-cells with the
anode surface) did not reduce CTC, indicating that, previously used solid graphite stick anodes (Nevin
sporadically, there was a lack of metabolic activity at et al., 2009). The biofilm was comprised of a layer
this outer fringe of the biofilm (Figures 2d and f). ca. 30-mm thick, which completely covered the
These results suggested that the vast majority of the entire surface with differentiated pillar structures
cells within the cells had the potential for metabolic up to 55-mm thick (Figures 4b and c).
activity. A total 146 genes were identified as being
This conclusion was supported in additional differentially expressed (Po0.05) between the inner
studies in which the capacity for electron transfer (0–20 mm) and outer (30–60 mm) portions of the
was evaluated with Redox Green (Figure 3). The biofilm using a twofold cut off in expression
spatial pattern of Redox Green reduction was similar (Supplementary Table 2). GSU0093, the only gene
to that for CTC reduction (Figures 3a and b). more highly expressed (2.1-fold) in the cells in the
Reductase activity was observed throughout the outer portion of the biofilm compared with the inner
entire biofilm with only rare spots within the portion, encodes a putative ATP-binding cassette
biofilm (Figure 3c), or at outer surface (Figures 3d transporter, ATP-binding/membrane protein. There
and e), in which the metabolic stain was not are homologs of GSU0093 in other Geobacter
reduced. species, including G. metallireducens, G. daltonii

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
514

Figure 3 Confocal scanning laser micrographs of anode biofilms of Geobacter sulfurreducens pRG5Mc treated with Redox Green
reagent. Biomass imaged in green as in Figure 1. Top down three-dimensional images of the cellular biomass (a), the blue reduced redox
green (b) and both images superimposed (c). Biomass (d) and reduced redox green (e) in perpendicular cross-sections. The circled areas
illustrate rare zones within (b and c) or on the outer surface (d and e) in which the cells did not reduce CTC. White bar ¼ 75 mm.

18.00

16.00

14.00

12.00
mAmps

10.00

8.00

6.00

4.00

2.00

0.00
0 50 100 150 200 250 300 350
hours

Figure 4 Power production (a) and confocal scanning laser micrograph of biofilm formation on graphite anodes used for microarray
analysis represented as three-dimensional image (b) and in perpendicular cross sections (c). Cells were imaged with Live/Dead stain.
White bar ¼ 250 mm.

(formerly strain FRC-32), G. uraniireducens, (Supplementary Table 3). Greater ribosome produc-
G. bemidjiensis, Geobacter strain M21 and G. lovleyi. tion is associated with faster growth rates in many
The function of this gene, and hence the signi- microorganisms (Wagner, 1994; Hua et al., 2004;
ficance of its increased expression in the outer Beste et al., 2005; Boccazzi et al., 2005) and there
portion of the anode biofilm, is unknown. was more ribosomal protein in faster growing
The transcript levels of a number of ribosomal cultures of G. sulfurreducens (Ding et al., 2006). In
protein-encoding genes had a ca. 2–3-fold lower the closely related G. uraniireducnes, transcript
abundance in cells in the outer layer of the biofilm abundance for ribosomal protein genes was related

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
515
to growth rate (Holmes et al., 2009). Thus, the outer biofilm sections. The lack of significant
slightly lower abundance of ribosomal protein gene change in the expression of any of these genes
transcripts in the outer layer of the biofilm suggests between the inner and outer portions of the biofilm
that those cells might be growing slower. suggests that the cells in both portions of the biofilm
Further evidence for slightly slower metabolic are experiencing similar requirements for extra-
rates in the cells in the outer portion of the biofilm cellular electron transfer.
was the finding that genes in the nuo operon Transcript levels in the outer biofilm of genes for
(GSU0338–GSU0351) had lower (ca. 1.6–2.8-fold) several cytochromes that do not seem to have a direct
transcript levels in the outer biofilm. Genes in role in extracellular electron transfer were slightly (ca.
the nuo cluster encode components of a large 2.05–2.85-fold) decreased (Supplementary Table 4).
membrane-associated NADH dehydrogenase that These included two putative outer-surface c-type
transfers reducing equivalents generated in the cytochromes, OmcX and OmcQ, putative cytochromes
TCA cycle to the menaquinone pool (Izallalen encoded by GSU0593 and GSU2743, as well as the
et al., 2008). periplasmic cytochrome PpcA (Lloyd et al., 2003).
However, differences in the rates of metabolism However, there seemed to be no significant differ-
between the inner and outer layer are unlikely ence in the expression levels of the vast majority of
to have been substantial because there was not the ca. 100 c-type cytochrome genes (Methé et al.,
a significant difference in the abundance of 2003) in G. sulfurreducens.
transcripts for genes encoding proteins of the If there were substantial disparities in electron-
TCA cycle. For example, transcript levels for gltA, acceptor availability between the inner and outer
which encodes a subunit of citrate synthase in sections of the biofilm it might be expected that this
G. sulfurreducens (Bond et al., 2005), is directly would result in some other differences in metabo-
related to metabolic rates, including electron lism that were not reflected in changes in gene
transfer to electrodes (Holmes et al., 2005). A similar transcript abundance. For example, G. sulfurreducens
response is expected for other TCA cycle enzyme will reduce protons to hydrogen if alternative electron
genes (Holmes et al., 2008, 2009). In general, values acceptors are not available (Cord-Ruwisch et al.,
for transcript abundance of TCA cycle genes seemed 1998). Therefore, if cells at the outer surface of the
to be lower in the samples from the outer layer of the biofilm were unable to transfer electrons derived
biofilm, but the differences were small (otwo-fold) from acetate metabolism to the anode, an alternative
and not statistically significant. Quantitative RT– response would be to increase hydrogen production
PCR analysis of gltA transcript levels indicated that with a corresponding increase in transcription of
the difference between the inner and outer layers hydrogenase genes. However, there was no difference
was only 1.3-fold. in expression of hydrogenase genes between the inner
Transcripts for outer-surface electron-transfer and outer sections of the biofilm, suggesting that the
components, considered important for electron cells in the outer section did not have an electron
transfer at the anode, were similar in the inner and acceptor limitation.
outer sections of the biofilms. For example, a A potential limitation on the metabolism for cells
previous comparison between gene expression in closer to the anode surface is lack of electron donor
G. sulfurreducens biofilms producing current versus or nutrients due to cells in the outer section
biofilms growing on the same surface material, but consuming these components (He et al., 2005;
using fumarate as the electron acceptor (Nevin et al., Marcus et al., 2007; Rabaey et al., 2007; Torres
2009), revealed increased expression of the gene for et al., 2008). G. sulfurreducens genes encoding
PilA, the structural protein for the pili that seem to acetate transporters that are more highly expressed
be electrically conductive pili (Reguera et al., 2005) when acetate is limiting have been identified (Risso
and may be involved in electron transfer through et al., 2008). However, transcript abundance for
anode biofilms (Reguera et al., 2006; Nevin et al., these genes was comparable in the inner and outer
2009). Cells in the inner and outer portions of the sections of the biofilms, suggesting that acetate
biofilm did not have a significant difference in pilA limitation is not an important consideration in
expression, but the adjacent gene, GSU1497, was current production. In a similar manner, genes that
less expressed (2.6-fold) in the outer biofilm (Nevin are more highly expressed under nitrogen- (Holmes
et al., 2009). et al., 2004; Methé et al., 2005; Mouser et al., 2009b),
OmcZ is an outer-surface, c-type cytochrome that iron- (O’Neil et al., 2008) or phosphate-limiting
is essential for high-density current production and (N’Guessan et al., 2009) conditions had similar
the OmcZ gene is more highly expressed in current- transcript abundances in the outer and inner
producing cells (Nevin et al., 2009; Richter et al., portions of the biofilm.
2009). Expression of omcZ was similar in the inner Cells in the outer section of the biofilm had lower
and outer biofilm. Two other outer-surface c-type expression of a number of genes expression of which
cytochromes, OmcB and OmcE, which are also more is expected to increase in response to environmental
highly expressed in current-producing cells versus stress (Supplementary Table 5). These included
cells reducing fumarate (Nevin et al., 2009), were genes encoding putative cold shock proteins, sodA,
also expressed at similar levels in the inner and mscL and universal stress response proteins. The

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
516
4.0

3.0

2.0

1.0

0.0
GSU1994 GSU0093 omcX ppcA GSU0207 omcQ GSU0872 rpsD atpE gtlA
-1.0

-2.0

-3.0

-4.0

-5.0

-6.0

-7.0

Figure 5 Fold change in transcript abundance of a diversity of genes between inner and outer biofilms as determined with quantitative
RT–PCR. Error bars illustrated the s.d. values of the triplicate determinations.

greatest increase in abundance in transcripts of section of the biofilm includes transcript level
stress-reponse genes was for sodA. Although this differences for some ribosomal proteins and a
gene is known to be involved in the oxidative stress NADH dehydrogenase. However, transcript abun-
response in some microorganisms, its role in dances for key metabolic genes, such as citrate
Geobacter species seems to be different (Mouser synthase, which previous studies have shown have
et al., 2009a), with highest expression associated significantly changes in expression levels when
with the use of Fe(III) as an electron acceptor (Methé metabolic rates change, were not significantly
et al., 2005; Mouser et al., 2009b). Increased different between the inner and outer fractions of
expression of mscL in Escherichia coli is associated the biofilm.
with entry into stationary phase or hyperosmotic There also was little indication from the gene
shock (Booth et al., 2007; Kloda et al., 2008). Under expression results that cells throughout the anode
acidic stress the regulation of numerous genes, biofilms were electron donor or nutrient-limited or
which overlap with oxidative stress, heat shock had significant differences in the expression of
and envelop stress responses have been identified proteins thought to be important in electron transfer
(Maurer et al., 2005). The increase in transcript to the anode. Increased transcript abundance for
abundance of these stress response genes in cells several genes associated with stress responses in
closer to the anode may be in response to the lower cells within the inner section of the biofilm do
pH that is expected closer to the anode surface suggest that these cells may be under some environ-
(Franks et al., 2009). mental stress, possibly associated with the expected
As a check on the microarray results, transcript lower pH in that environment, but this apparent
abundance was also evaluated with quantitative stress does not seem to significantly impact on
RT–PCR for a few select genes (Figure 5). These metabolism.
included: the only upregulated gene, GSU093; the The finding that cells throughout the bulk of the
most highly downregulated gene GSU1994 and biofilm are metabolically active and likely to be
genes for several other hypothetical proteins; genes contributing to current production is consistent
encoding representative ribosomal, electron transfer with the previous finding of a linear increase in
and energy conservation proteins; and, as noted current production with increasing anode biomass
above, the citrate synthase gene, gtlA. Results from and biofilm height (Reguera et al., 2006). This
the quantitative RT–PCR were comparable to the requires that cells not in direct contact with the
microarray results. anode can still transfer electrons to the anode and it
was suggested that electrically conductive pili may
mediate this long-range electron transfer (Reguera
Implications et al., 2006). Modeling studies have predicted that
The results of the staining for metabolic activity, as the conductivity of the biofilm would have to be
well as the analysis of differential gene expression 104 mS cm 1 to avoid electron transfer limitation
between the inner and outer sections of the anode (Marcus et al., 2007). Although biofilms are
biofilm suggest that cells throughout the biofilm are generally considered to act as insulators rather than
metabolically active and likely to be contributing to conductors (Herbert-Guillou et al., 1999; Muñoz-
current production. Evidence for slightly lower rates Berbel et al., 2006; Dheilly et al., 2008), the finding
of growth and metabolism in cells in the outer that cells at substantial distance from the anode are

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
517
metabolically active is consisent with the concept Beste DJV, Peters J, Hooper T, Avignone-Rossa C, Bushell
that anode biofilms may be conductive. ME, McFadden J. (2005). Compiling a molecular
Clearly, the more cells that can release electrons inventory for Mycobacterium bovis BCG at two growth
from metabolic activity, the greater the potential rates: evidence for growth rate-mediated regulation of
ribosome biosynthesis and lipid metabolism. J Bacteriol
current output. However, there do seem to be limits
187: 1677–1684.
on how thick anode biofilms can grow and meta- Bhupathiraju VK, Hernandez M, Landfear D, Alvarez-
bolic staining provided evidence that some cells in Cohen L. (1999). Application of a tetrazolium dye
the most outer surface of the biofilm may not as an indicator of viability in anaerobic bacteria.
metabolically active. The factors limiting metabo- J Microbiol Methods 37: 231–243.
lism in the small areas at the outer reaches of the Boccazzi P, Zanzotto A, Szita A, Bhattacharya S, Jensen
biofilm are yet not understood, but overcoming KF, Sinskey AJ. (2005). Gene expression analysis of
these limitations could be a useful strategy for Escherichia coli grown in miniaturized bioreactor
increasing the power output of microbial fuel cells. platforms for high-throughput analysis of growth and
The microtoming–microarray approach for depth genomic data. App Microbiol Biotechnol 68: 518–532.
Bond DR, Mester T, Nesbø CL, Izquierdo-Lopez AV, Collart
resolution of gene expression in biofilms described FL, Lovley DR. (2005). Characterization of citrate
here may be applicable to the study of gene synthase from Geobacter sulfurreducens and evidence
expression in other types of biofilms. Genes speci- for a family of citrate synthases similar to those of
fically expressed during biofilm formation in eukaryotes throughout the Geobacteraceae. Appl
response to environmental conditions or specific Environ Microbiol 71: 3858–3865.
mutations have been identified (Whiteley et al., Booth IR, Edwards MD, Black S, Schumann U, Miller S.
2001; Schembri et al., 2003; Stanley et al., 2003; Zhu (2007). Mechanosensitive channels in bacteria: signs
and Mekalanos, 2003), but these studies on whole of closure? Nat Rev Microbiol 5: 431–440.
biofilms lacked the spatial resolution that may be Boulos L, Prévost M, Barbeau B, Coallier J, Desjardins R.
necessary to account for the heterogeneity in (1999). LIVE/DEAD BacLight: application of a new
rapid staining method for direct enumeration of viable
metabolic states that can be expected in biofilms and total bacteria in drinking water. J Microbiol
(Stewart and Franklin, 2008). Introducing reporters Methods 37: 77–86.
that express unstable green fluorescent proteins can Caccavo F, Lonergan DJ, Lovely DR, Davis M, Stolz JF,
provide insight into the expression of individual Mcinerney MJ. (1994). Geobacter sulfurreducens sp.
genes throughout biofilms (Sternberg et al., 1999; nov., a hydrogen- and acetate-oxidizing dissimilatory
Teal et al., 2006; Stewart and Franklin, 2008). metal-reducing microorganism. Appl Environ Micro-
However, this strategy is limited to evaluating the biol 60: 3752–3759.
expression of a small number of genes and cannot be Cord-Ruwisch R, Lovley DR, Schink B. (1998). Growth of
applied to microorganisms for which systems for Geobacter sulfurreducens with acetate in syntrophic
genetic manipulation have not been developed. The cooperation with hydrogen-oxidizing anaerobic part-
ners. Appl Environ Microbiol 64: 2232–2236.
microtoming–microarray approach described here Dheilly A, Linossier I, Darchen A, Hadjiev D,
can be applied to any organism for which a genome Corbel C, Alonso V. (2008). Monitoring of microbial
sequence is available. Furthermore, it could readily adhesion and biofilm growth using electro-
be adapted for the study of multi-species biofilms or chemical impedancemetry. Appl Microbiol Biotechnol
natural biofilm communities with the appropriate 79: 157–164.
metagenomic/metatranscriptomic approaches. Ding YH, Hixson KK, Giometti CS, Stanley A, Esteve-
Núñez A, Khare T et al. (2006). The proteome of
dissimilatory metal-reducing microorganism Geobacter
sulfurreducens under various growth conditions.
Acknowledgements Biochim Biophys Acta 1764: 1198–1206.
We thank L Raboin, Polymer Science, University of Franks AE, Nevin KP, Jia H, Izallalen M, Woodard TL,
Massachusetts, Amherst, for his help and suggestions Lovley DR. (2009). Novel strategy for three-dimen-
with the microtoming procedure and Dr H Bagdadi for sional real-time imaging of microbial fuel cell com-
discussions leading to the development of this approach. munities: monitoring the inhibitory effects of proton
This study was supported by the Office of Science (BER), accumulation within the anode biofilm. Energy Enviro
U.S. Department of Energy, Cooperative Agreement No. Sci 2: 113–119.
DE-FC02–02ER63446 and Office of Naval Research Award Gray D, Yue RS, Chueng CY, Godfrey W. (2005). Bacterial
No. N00014-07-1-0966. vitality detected by a novel fluorogenic redox dye
using flow cytometry. In: Abstracts of the American
Society of Microbiology Meeting American Society for
References Microbiology: Washington, DC, USA.
Gruden CL, Fevig S, Abu-Dalo M, Hernandez M. (2003).
An D, Parsek MR. (2007). The promise and peril of 5-Cyano-2,3-ditolyl tetrazolium chloride (CTC) reduc-
transcriptional profiling in biofilm communities. Curr tion in a mesophilic anaerobic digester: Measuring
Opin Microbiol 10: 292–296. redox behavior, differentiating abiotic reduction, and
Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman comparing FISH response as an activity indicator.
JG, Smith JA et al. (1997). Current Protocols J Microbiol Methods 52: 59–68.
in Molecular Biology. John Wiley and Sons, Inc.: He Z, Minteer SD, Angenent LT. (2005). Electricity
New York. generation from artificial wastewater using an

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
518
upflow microbial fuel cell. Environ Sci Technol 39: motility, catabolism, and oxidative stress in
5262–5267. Escherichia coli K-12. J Bacteriol 187: 304–319.
Herbert-Guillou D, Tribollet B, Festy D, Kiéné L. (1999). Methé BA, Nelson KE, Eisen JA, Paulsen IT, Nelson W,
In situ detection and characterization of biofilm in Heidelberg JF et al. (2003). The genome of Geobacter
waters by electrochemical methods. Electrochimica sulfurreducens: insights into metal reduction in
Acta 45: 1067–1075. subsurface environments. Science 302: 1967–1969.
Holmes DE, Mester T, O’Neil RA, Perpetua LA, Larrahon- Methé BA, Webster J, Nevin K, Butler J, Lovley DR. (2005).
do MJ, Glaven R et al. (2008). Genes for two multi- DNA microarray analysis of nitrogen fixation and
copper proteins required for Fe(III) oxide reduction in Fe(III) reduction in Geobacter sulfurreducens. Appl
Geobacter sulfurreducens have different expression Environ Microbiol 71: 2530–2538.
patterns both in the subsurface and on energy-harvest- Mouser PJ, Holmes DE, Perpetua LA, DiDonato R, Postier
ing electrodes. Microbiol 154: 1422–1435. B, Liu A et al. (2009a). Quantifying expression of
Holmes DE, Nevin KP, Lovley DR. (2004). In situ Geobacter spp. oxidative stress genes in pure culture
expression of Geobacteraceae nifD in sub- and during in situ uranium bioremediation. ISME J 3:
surface sediments. Appl Environ Microbiol 70: 454–465.
7251–7259. Mouser PJ, N’Guessan AL, Elifantz H, Holmes DE,
Holmes DE, Nevin KP, O’Neil RA, Ward JE, Adams LA, Williams KH, Wilkins MJ et al. (2009b). Influence of
Woodard TL et al. (2005). Potential for quantifying heterogeneous ammonium availability on bacterial
expression of the Geobacteraceae citrate synthase gene community structure and the expression of
to assess the activity of Geobacteraceae in the subsur- nitrogen fixation and ammonium transporter genes
face and on current-harvesting electrodes. Appl during in situ bioremediation of uranium-contami-
Environ Microbiol 71: 6870–6877. nated groundwater. Environ Sci Technol 43:
Holmes DE, O’Neil RA, Chavan MA, N’Guessan LA, Vrionis 4386–4392.
HA, Perpetua LA et al. (2009). Transcriptome of Muñoz-Berbel X, Muñoz FJ, Vigués N, Mas J. (2006). On-
Geobacter uraniireducens growing in uranium-contami- chip impedance measurements to monitor biofilm
nated subsurface sediments. ISME J 3: 216–230. formation in the drinking water distribution network.
Hua Q, Yang C, Oshima T, Mori H, Shimizu K. (2004). Sens Actuators: B Chem 118: 129–134.
Analysis of gene expression in escherichia coli in Nevin KP, Kim BC, Glaven RH, Johnson JP, Woodard TL,
response to changes of growth-limiting nutrient Methé BA et al. (2009). Anode biofilm transcriptomics
in chemostat cultures. Appl Environ Microbiol 70: reveals outer surface components essential for high
2354–2366. density current production in Geobacter sulfurredu-
Huang CT, Yu FP, McFeters GA, Stewart PS. (1995). cens fuel cells. PLoS ONE 4: e5628.
Nonuniform spatial patterns of respiratory activity N’Guessan AL, Elifantz H, Nevin KP, Mouser PJ et al.
within biofilms during disinfection. Appl Environ (2009). Molecular analysis of phosphate limitation in
Microbiol 61: 2252–2256. Geobacteraceae during the bioremediation of a ura-
Izallalen M, Mahadevan R, Burgard A, Postier B, Didonato nium-contaminated aquifer. ISME J (in press).
Jr R, Sun J et al. (2008). Geobacter sulfurreducens O’Neil RA, Holmes DE, Coppi MV, Adams LA, Larrahondo
strain engineered for increased rates of respiration. MJ, Ward JE et al. (2008). Gene transcript analysis of
Metab Eng 10: 267–275. assimilatory iron limitation in Geobacteraceae during
Kalyuzhnaya MG, Lidstrom ME, Chistoserdova L. (2008). groundwater bioremediation. Environ Microbiol 10:
Real-time detection of actively metabolizing microbes 1218–1230.
by redox sensing as applied to methylotroph popula- Postier B, DiDonato R, Nevin KP, Liu A, Frank B,
tions in Lake Washington. ISME J 2: 696–706. Lovley DR et al. (2008). Benefits of electrochemically
Kloda A, Petrov E, Meyer GR, Nguyen T, Hurst AC, synthesized oligonucleotide microarrays for analysis
Hool L et al. (2008). Mechanosensitive channel of gene expression in understudied microorganisms.
of large conductance. Int J Biochem Cell Biol 40: J Microbiol Methods 74: 26–32.
164–169. Rabaey K, Rodrı́guez J, Blackall LL, Keller J, Gross P,
Lloyd JR, Leang C, Hodges Myerson AL, Coppi MV, Cuifo Batstone D et al. (2007). Microbial ecology meets
S, Methe B et al. (2003). Biochemical and genetic electrochemistry: electricity-driven and driving
characterization of PpcA, a periplasmic c-type cyto- communities. ISME J 1: 9–18.
chrome in Geobacter sulfurreducens. Biochem J 369: Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen
153–161. MT, Lovely DR. (2005). Extracellular electron
Logan BE. (2009). Exoelectrogenic bacteria that power transfer via microbial nanowires. Nature 435:
microbial fuel cells. Nat Rev Microbiol 7: 375–381. 1098–1101.
Logan BE, Regan JM. (2006). Electricity-producing Reguera G, Nevin KP, Nicoll JS, Covalla SF, Woodard TL,
bacterial communities in microbial fuel cells. Trends Lovley DR et al. (2006). Biofilm and nanowire
Microbiol 14: 512–518. production leads to increased current in Geobacter
Lovley DR. (2006). Bug juice: harvesting electricity with sulfurreducens fuel cells. Appl Environ Microbiol 72:
microorganisms. Nat Rev Microbiol 4: 497–508. 7345–7348.
Lovley DR. (2008). The microbe electric: conversion of Richter H, Nevin KP, Jia H, Lowy DA, Lovley DR, Tender
organic matter to electricity. Curr Opin Biotechnol 19: LM. (2009). Cyclic voltammetry of biofilms of wild
564–571. type and mutant Geobacter sulfurreducens on fuel cell
Marcus AK, Torres CI, Rittmann BE. (2007). Conduction- anodes indicates possible roles of OmcB, OmcZ, type
based modeling of the biofilm anode of a microbial IV pili, and protons in extracellular electron transfer.
fuel cell. Biotechnol Bioeng 98: 1171–1182. Energy Environ Sci 2: 506–516.
Maurer LM, Yohannes E, Bondurant SS, Radmacher M, Risso C, Methé BA, Elifantz H, Holmes DE, Lovley DR.
Slonczewski JL. (2005). pH regulates genes for flagellar (2008). Highly conserved genes in Geobacter species

The ISME Journal

 Spatial metabolic status of Geobacter sulfurreducens biofilms
AE Franks et al
519
with expression patterns indicative of acetate limita- Bacillus subtilis by use of DNA microarrays. J Bacteriol
tion. Microbiol 154: 2589–2599. 185: 1951–1957.
Rittmann BE, Torres CI, Marcus AK. (2008). Understand- Stewart PS, Franklin MJ. (2008). Physiological heteroge-
ing the distinguishing features of a microbial neity in biofilms. Nat Rev Microbiol 6: 199–210.
fuel cell as a biomass-based renewable energy tech- Sternberg C, Christensen BB, Johansen T, Toftgaard Nielsen
nology. In: Shah V (ed). Emerging Environmental A, Andersen JB, Givskov M et al. (1999). Distribution of
Technologies. Springer Netherlands: Netherlands, pp bacterial growth activity in flow-chamber biofilms.
1–28. Appl Environ Microbiol 65: 4108–4117.
Rodriguez GG, Phipps D, Ishiguro K, Ridgway HF. (1992). Teal TK, Lies DP, Wold BJ, Newman DK. (2006).
Use of a fluorescent redox probe for direct visualiza- Spatiometabolic stratification of Shewanella oneiden-
tion of actively respiring bacteria. Appl Environ sis biofilms. Appl Environ Microbiol 72: 7324–7330.
Microbiol 58: 1801–1808. Torres CI, Marcus AK, Rittmann BE. (2008). Proton
Rozen S, Skaletsky H. (2000). Primer3 on the WWW for transport inside the biofilm limits electrical current
general users and for biologist programmers. In: generation by anode-respiring bacteria. Biotechnol
Krawetz S, Misener S (eds). Bioinformatics Methods Bioeng 100: 872–881.
and Protocols: Methods in Molecular Biology. Humana Wagner R. (1994). The regulation of ribosomal RNA
Press: Totowa, NJ, 365–386. synthesis and bacterial cell growth. Arch Microbiol
Schaule G, Flemming HC, Ridgway HF. (1993). Use of 161: 100–109.
5-cyano-2,3-ditolyl tetrazolium chloride for quantifying Whiteley M, Bangera MG, Bumgarner RE, Parsek MR, Teitzel
planktonic and sessile respiring bacteria in drinking GM, Lory S et al. (2001). Gene expression in Pseudo-
water. Appl Environ Microbiol 59: 3850–3857. monas aeruginosa biofilms. Nature 413: 860–864.
Schembri MA, Kjaergaard K, Klemm P. (2003). Global gene Zheng Z, Stewart PS. (2004). Growth limitation of
expression in Escherichia coli biofilms. Mol Microbiol Staphylococcus epidermidis in biofilms contributes
48: 253–267. to rifampin tolerance. Biofilms 1: 31–35.
Stanley NR, Britton RA, Grossman AD, Lazazzera BA. Zhu J, Mekalanos JJ. (2003). Quorum sensing-dependent
(2003). Identification of catabolite repression as biofilms enhance colonization in Vibrio cholerae. Dev
a physiological regulator of biofilm formation by Cell 5: 647–656.

Supplementary Information accompanies the paper on The ISME Journal website (http://www.nature.com/ismej)

The ISME Journal