Dynamics of the Genetic Diversity of Marine Bacterial
zur Erlangung des Grades eines
Doktors der Naturwissenschaften
– Dr. rer. nat. –
dem Fachbereich Biologie/Chemie der
Die vorliegende Arbeit wurde in der Zeit von Oktober 1997 bis März 2001 am Max-Planck-
Institut für marine Mikrobiologie in Bremen, und am Nederlands Instituut voor Onderzoek
der Zee auf Texel (Niederlande) angefertigt.
Prof. Dr. Friedrich Widdel
Dr. Gerard Muyzer
Tag des öffentlichen Promotionskolloquiums: 26. Oktober 2001
Table of contents
Chapter 2 Denaturing gradient gel electrophoresis in marine microbial
Chapter 3 Successional changes in the genetic diversity of a marine
bacterial assemblage during confinement
Chapter 4 Microbial community dynamics in Mediterranean nutrient-
enriched seawater mesocosms: changes in the genetic diversity of
Chapter 5 Bacterial activity and genetic richness along an estuarine gradient
(Rhône river plume, France).
Chapter 6 Does eutrophication alter bacterioplankton diversity?
A four year experimental study in the Norwegian landlocked bay
Chapter 7Genetic diversity of ‘satellite’ bacteria present in cultures of
Appendix Summary 146
List of abbreviations149
Contribution to the scientific publications presented in this
Further publications 152
The biogeochemical importance of heterotrophic, pelagic, marine bacteria
Classical concepts of trophic interaction in the ocean never ascribed an important role to
marine bacteria. These implied that most of the primary production produced by
phytoplankton is channelled into higher trophic levels, thus to larger organisms through
microzooplankton. Even after estimates of the actual numbers of bacteria present in seawater
had been corrected (Hobbie et al., 1977; Jannasch & Jones, 1959) some doubt remained about
the activity of the cells. The ocean was then regarded as just too dilute in organic carbon
concentrations to support growth and activity of such a large number of cells. However, the
application of radiotracer experiments suggested that the organic matter pool in seawater had
short turn-over times (see references in Azam, 2001). Incorporation of tritiated thymidine was
used to estimate bacterial production (Fuhrman & Azam, 1982) showing that bacteria are an
important component of marine food webs (Azam et al., 1983). There is still considerable
uncertainty about the actual amounts, but accepted estimates attribute 15 to 50% of the
organic carbon produced by phototrophic primary producers to be channelled through bacteria
(see Williams, 2000). However, to date it is still not clear what fraction of bacterial cells can
be regarded as active, dormant or dead, as different methods to quantify the categories yield
different numbers (Choi, 1999; Gasol et al., 1995; Karner & Fuhrman, 1997; Zweifel &
Hagström, 1995). Another matter of debate are bacterial growth efficiencies. There are
indications that these have usually been overestimated significantly (see review of Del
Giorgio & Cole, 1998) and that bacterial growth efficiency may be as low as 15 to 20%. This
finding has important consequences for estimating what fraction of organic matter oxidation
is eventually represented by a bacterial cell. Variations in growth efficiency estimates lead to
great uncertainties in the general balancing of marine production and consumption of organic
matter (Williams, 2000), and make assessment of whole ocean regions as net sources or sinks
of CO2 extremely difficult (Del Giorgio et al., 1997; Williams, 1998). The importance of
bacteria in cycling of organic matter is further underlined by their ability to act as competitors
to phytoplankton in the uptake of inorganic nutrients (Caron, 1994). Thus bacterial nutrient
uptake can affect nutrient pools available for primary production, and bacterioplankton
growth may be limited by the availability of inorganic nutrients itself (Cotner et al., 1997).
Diversity of marine microbial communities
It is clear that bacteria are important, but why should one study bacterial diversity? Studies of
marine bacterioplankton have often applied a so-called “black-box” approach, meaning that
bacteria have mainly been investigated according to the extent of fluxes or transformations
that are mediated by them, without taking into account that only a subset of the community
might actually be contributing to the measured rate. Thus, without knowledge which bacteria
are responsible for the activity and hence relating the performance of the active fraction to the
community as a whole, bacterioplankton remains a black-box. While this may seem to be
ignorant at a first glance, the objective to quantify a bacterially mediated transformation and
at the same time determine the identity of the responsible population(s) is far from trivial.
Microbial ecological studies have only recently begun to explore the links between structure
and function or identity and activity of microbial assemblages by applying new tools.
Especially molecular biological techniques using small subunit ribosomal RNA sequences or
the encoding genes as a molecular marker have been used for assessing the taxonomic
composition of microbial communities (Amann et al., 1995; Giovannoni et al., 1990; Muyzer,
1998; Ward et al., 1990). The application of molecular biological techniques in biological
oceanography has resulted in a revised conception of marine microbial community
composition (see Giovannoni & Rappé, 2000 for a review). The conventional method of
plating bacteria on more or less diverse agar based media typically results in certain culturable
groups of bacteria being detected, while the larger portion of bacteria can not be readily
cultured using this strategy. An inventory made by molecular biological techniques
commonly displays a greater diversity than culture-dependent methods, and moreover gives a
different display of taxonomic composition, with most types retrieved not related to cultured
bacterial isolates. Fluorescent in situ hybridisation (FISH) with taxon-specific gene probes has
confirmed the general problems associated with the culture-dependent methods (Amann, et
al., 1995; Glöckner et al., 1999). The incongruent nature of displays of community
composition given by culture-dependent and independent approaches has also been verified in
studies using the two approaches simultaneously (Eilers et al., 2000; Suzuki et al., 1997).
The consequence of the strong biases associated with the culture-dependent approach
to analysing community composition has been a trend in microbial ecology to replace
cultivation based methods with the new molecular tools. However, up to now, most studies
using molecular biological techniques have only provided snap-shots of community
composition only. Cloning and sequencing of complete 16S rRNA genes to analyse
community composition with subsequent in situ identification of defined populations by FISH
are powerful tools in microbial ecology, but there is always a trade-off between the amount of
sequence information that can be gathered by the cloning approach and the number of
samples that can be analysed. Genetic fingerprinting techniques, e.g. denaturing gradient gel
electrophoresis (DGGE) (Muyzer, 1999; Muyzer & Smalla, 1998), are excellent tools for
studying the changes in bacterial genetic diversity over both, temporal and spatial scales,
because they facilitate the analysis of numerous samples.
Temporal changes in marine bacterial community composition
Few studies have applied molecular methods to study microbial community composition over
longer time-spans, for instance to study seasonal and annual patterns of microbial community
composition. Therefore, studies using cloning and sequencing to address dynamics in
bacterial community composition have usually been limited to few samples resulting in a low
temporal resolution (Kelly & Chistoserdov, 2001; Kerkhof et al., 1999). Using gene probes
and hybridisation techniques a few autecological studies have provided higher temporal
resolution of the distribution of specific bacterial and archaeal phylotypes in the marine
environment (Field et al., 1997; Gordon & Giovannoni, 1996; Massana et al., 2000).
Furthermore, a more extensive application of FISH detection of bacterial populations in
combination with cloning and culturing was done by Eilers and colleagues (Eilers, et al.,
2000). FISH of bacterial populations showed that culturable bacteria form the North Sea made
up only a minor portion of the bacterioplankton assemblage, and at the same time
demonstrated that abundance of probe-defined populations may undergo seasonal fluctuation.
The first report on seasonal changes in marine microbial communities using a genetic
fingerprinting approach, was the study by Murray et al. (Murray et al., 1998) who used PCR-
DGGE to analyse fluctuation in the composition of bacterial assemblages in the waters around
Anvers Island (Antarctica) along period of a nine months. They noted that the genetic
richness of the assemblage increased at the beginning of the productive summer season and
decreased again thereafter.
Bacteria underlie various control mechanisms within the microbial food web.
The goal of microbial ecology is to identify the factors determining the composition of
microbial assemblages and their performance in driving the global biogeochemical cycles.
Only a thorough understanding of how individual populations contribute to community
performance and how they are regulated in their abundance and activity will ultimately add
predictive power to microbial ecology, which is lacking at present. An answer to these
questions can often only be given by correlation of measurements of abiotic and biotic
parameters with fluctuation in microbial community composition. Mechanisms controlling
pelagic bacterial communities are classified as bottom-up and top-down regulation. Bottom-
up regulation refers to the control conferred by availability of nutrients and carbon sources on
the growth of microorganisms. Mortality conferred through grazing by microzooplankton
(heterotrophic nanoflagellates, and ciliates) and viral infection is important in regulating the
abundance of pelagic bacteria (Sanders, 1992; Sherr & Sherr, 1994; Weinbauer & Höfle,
1998) and is referred to as top-down control, and side-in control respectively (Suttle, 1994).
Populations of picoplanktonic microorganisms may vary greatly in their response to these
differing controlling factors. Reaction to bottom-up mechanisms will be determined by
physiological characteristics, such as the substrate spectrum, substrate uptake affinity and
kinetics. Likewise, there is variation in the susceptibility to different mortality mechanisms. It
has been shown, for instance, that the chance of a bacterium being ingested by a grazing
heterotroph is correlated to its size (Monger & Landry, 1991), which in turn is correlated to its
activity (Del Giorgio et al., 1996; Gasol, et al., 1995). Some bacteria have evolved defence
mechanisms that can reduce the mortality of the population, e.g. by developing an inedible
morphotype (Hahn et al., 1999; Jürgens & Güde, 1994; Pernthaler et al., 1997) or outgrowing
predation pressure (Pernthaler, et al., 1997). Thus, it is conceivable that variations in bottom-
up and mortality mechanisms should have the potential to affect the composition of pelagic
bacterial assemblages. Other studies that have addressed temporal changes of bacterial
community composition recently, have provided indications that the development of
phytoplankton blooms may affect bacterial community composition (Kelly & Chistoserdov,
2001; Kerkhof, et al., 1999). Thus, to understand bacterial community composition and its
regulation a variety of factors has to be regarded. In this context it is of interest how grazing
and eutrophication on the one hand and the predominant type of phytoplankton on the other
hand may be able to affect the diversity and activity of bacterial assemblages.
Eutrophication of coastal waters is a problem with world-wide distribution and is of
global importance. Eutrophication not only has economical impact, but also is a risk for the
species diversity of natural ecosystems. Harmful algal blooms have been considered one of
the effects of eutrophication in coastal environments (Paerl, 1998; Vollenweider, 1992).
Eutrophication may disrupt natural pelagic food webs, and might therefore entail potential
consequences for marine bacterial assemblages (Paerl, 1998).
Outline of the thesis
The aim of this thesis was to characterise fluctuations in the genetic diversity of bacterial
communities occurring at a variety of temporal scales (hours to years) and to try to identify
factors that influence the observed dynamics. Special attention was given to the influence of
eutrophication on bacterial diversity.
A good part of the work presented in this dissertation has been carried out in the
framework of the European-Union-funded project CHABADA (‘Changes in bacterial activity
and diversity in Mediterranean coastal waters as effected by eutrophication’, project number
MAS3-CT96-0047), which focussed on the effect of nutrient addition on bacterial activity and
diversity. Analyses of microbial community composition reported on in the different chapters
of this Thesis were mainly performed by denaturing gradient gel electrophoresis. Chapter 2 is
an overview of the application of DGGE genetic fingerprinting in marine microbial ecology
and also provides detailed protocols covering the practical aspects of the technique.
The CHABADA-project started out with mesocosm experiments that were used as
model systems to evaluate short-term changes in Mediterranean bacterial communities as a
consequence of nutrient addition. Generally, marked fluctuation occurred in the genetic
diversity during the incubation, even in control mesocosms. Grazing of heterotrophic
microzooplankton on bacteria was tentatively identified as an important factor affecting the
genetic diversity of bacterial assemblages. The results of the mesocosm experiments are
reported in Chapter 3 and Chapter 4.
Originally, it was planned to complement the insights gained in the mesocosm
experiments with observations from a natural eutrophication gradient such as in the Northern
Adriatic Sea. Unfortunately, it was not possible to go there, so as an alternative the outflow
plume of the Rhone river was studied, where strong gradients in both nutrient and salt
concentrations occur. The results of that study are reported in Chapter 5.
Eventually, the chance to study the potential effects of eutrophication on bacterial
diversity in a natural system arose in a collaboration with Olav Vadstein and Yngvar Olsen,
from the Trondhjem Biological Station of the Norwegian University of Science and
Technology, Trondheim, who were involved in the EU-funded project COMWEB
(‘Comparative Analysis of Food Webs Based on Flow Networks: Effects of Nutrient Supply
on Structure and Function of Coastal Plankton Communities’; project number MAS3-CT96-
0052). Within the COMWEB-project a eutrophication experiment in the landlocked bay
Hopavågen on the coast of central Norway had been performed. Samples from the euphotic
zone of the bay were analysed which were taken during a period of 40 months from 1996
until 1999. Two years, 1996 and 1997, served as control years, and during the summer
seasons of 1998 and 1999 nutrients were added artificially. The samples were analysed for
seasonal distribution patterns in the bacterial community composition and it was assessed
whether emerging patterns were affected by eutrophication. The results are reported in
Part of the rationale behind the final study, a survey of the diversity of so-called
‘satellite’ bacteria from diatom cultures (Chapter 7), was that eutrophication effects on
bacterial diversity might only be indirect, and brought about through interactions that bacteria
have with other components of pelagic food webs. One of the suggested effects of
eutrophication on coastal zones has been the increased incidence of nuisance algal blooms,
which might entail subsequent changes in the bacterial community, if algal-bacterial
interactions (mutualisms and/or antagonisms) exist. Thus, the motivation to gather basic
information by studying the bacterial diversity at the example of diatom laboratory cultures.
The results of the first mesocosm experiment showed that successional changes in the
composition were occurring even in the untreated control mesocosm. Bacterial and protozoan
counts revealed a typical mesocosm succession consisting of a growth phase of bacteria in the
beginning, which was followed by a peak in protozoan counts (mainly heterotrophic
nanoflagellates) and a concomitant reduction of bacterial numbers to initial levels.
Estimations of bacterial mortality due to grazing indicated that grazing was the main factor
responsible for bacterial mortality. Important changes in the composition of the bacterial
community were suggested by marked changes in DGGE fingerprints of the bacterial
assemblage during the incubation. Cloning and sequencing of 16S rRNA genes confirmed
that a major shift in species composition was occurring during incubation and clone libraries
of samples from after the peak in grazing activity were dominated by 16S rRNA genes related
to those of the genus Alteromonas.
In a second mesocosm experiment, replicate tanks were used to evaluate the
reproducibility of treatments (control and nutrient-enrichment with inorganic N and P).
Additionally, samples from the original coastal sampling station were also analysed before
and after the experiment to isolate the effect of manipulation in mesocosms. PCR-DGGE was
also performed on reverse transcribed rRNA, which favours amplification of the most active
bacterial populations due to their high rRNA content. Marked fluctuations in community
composition were evident from DGGE analyses and nutrient enrichment also entailed some
differences in the community composition, but there were also similarities between the
development of tanks from the two treatments, and nutrient addition seemed to affect mainly
the speed and extent of the changes. Interestingly, the phase of intense grazing again marked
important shifts in community fingerprints, and numbers of rRNA-derived DGGE bands
(indicating active populations) that had increased during the initial growth phase, were
reduced during the phase of protozoan grazing. Microbial populations were identified by
sequencing of DGGE bands and contrary to the first mesocosm experiments the post grazing
phase seemed to be dominated by α-Proteobacteria and members of the Cytophaga-
Flavobacterium-Bacteroides group (CFB). Strong changes in community fingerprints of the
bacterial assemblages in mesocosms were not mirrored at the natural site, where only little
change was observed in community fingerprints during two weeks indicating a significant
effect of confinement as suggested in the first mesocosm study.
While the mesocosm experiments had lasted for about two weeks, the temporal scale of the
samplings in the Rhone river plume was much shorter, extending just over a couple of hours.
Riverine and marine bacterial assemblages behaved differently in the mixing zone. Bacterial
abundances and activities showed a more drastic decrease in the low salinity range of the
gradient than expected from dilution models, indicating that an important fraction of
freshwater bacteria disappeared in the mixing area. The plume zone had a high total bacterial
genetic richness (estimated by the number of DNA-derived DGGE bands), 13-55 bands
compared to that reported in other aquatic ecosystems, which was the consequence of the
mixing of riverine and marine assemblages. The proportion of active populations was
estimated using the ratio of DGGE bands derived from RNA and DNA. This ratio was lower
in Rhone water than in marine water indicating that only a part of the constitutive populations
were active, while the activity was distributed within a larger fraction of populations in the
marine assemblage. This fitted well with the observation of higher specific leucine
incorporation rates in the marine assemblages. The marine community appeared to be
strongly affected by decreasing salinity, which probably was a consequence of the mixing of
marine assemblage with a much more abundant riverine community. No marked modification
of the marine community by inflowing nutrient rich river water was observed. This lack of a
response was probably the consequence of a very short residence time of water in the studied
mixing area which did not allow for a growth response of marine populations to display in
Seasonal and annual patterns in the bacterial community composition of the pristine,
landlocked bay Hopavågen on the Norwegian coast were analysed in years with and without
sustained experimental nutrient addition. Hopavågen was used as an experimental field for
testing the consequences of nutrient addition in the summer seasons of 1998 and 1999, while
1996 and 1997 were investigated as control years without artificial eutrophication. Denaturing
gradient gel electrophoresis analysis of bacterial 16S rRNA gene fragments showed seasonal
variations in community composition that seemed to be recurrent annually with some
phylotypes appearing at similar times during the years. Synechococcus–related cyanobacteria
dominated the late summer community, α-Proteobacteria of the Roseobacter group seemed
permanently present during the phase of phytoplankton production. The study showed that
recurrent seasonal patterns exist in bacterial assemblages, with especially marked changes
during the transition from winter to the productive seasons. Doubled amount of nutrient
addition in 1999 as compared to 1998 had an effect on phytoplankton primary production and
also some effect on phytoplankton community structure. An effect of eutrophication on the
genetic diversity of the bacterial assemblage, was, however, not obvious, suggesting that the
level of artificial eutrophication, was not yet sufficient to alter the microbial food web
structure and bacterial diversity of Hopavågen.
‘Satellite’ bacteria in cultures of marine diatoms
The motivation to analyse the genetic diversity of ‘satellite-bacteria’, bacteria accompanying
uni-algal cultures of marine phytoplankton was to assess whether the different algal cultures
may constitute niches for specific bacterial species. The six diatom cultures that were
analysed were accompanied by distinct satellite assemblages, as the majority of the
phylotypes detected in the six cultures was unique, only some phylotypes were common to
more than one culture. Only minor variations of satellite assemblage genetic fingerprints was
observed suggesting that the bacterial-algal associations were stable. An experimental
approach to find evidence for specific algae-bacteria interactions by challenging algae
cultures with heterologous satellite assemblages was unsuccessful. It was not possible to
avoid carry-over of algae. Most satellite populations were identified by sequencing of DGGE
bands as typical marine phylotypes of the α-Proteobacteria (related to the genera Ruegeria,
Sulfitobacter, Roseobacter, and Erythrobacter), or of different genera of the CFB phylum.
Surprisingly, β-Proteobacteria were also found in two of the cultures. A unifying theme in
satellite bacterial assemblage composition was the presence of at least one representative of
the α-Proteobacteria and of the CFB phylum, both of which have been identified as important
representatives of the marine picoplankton. The results indicate that algae diversity may be an
important factor for explaining some of the enormous bacterial diversity in marine
assemblages, and vice versa.
The present study was successful in analysing dynamics of marine bacterioplankton
assemblages in artificial as well natural systems at a variety of temporal scales. Simultaneous
measurements of a variety of other biological parameters related to the microbial food web in
the context of the European projects allowed to correlate fluctuations in bacterial diversity
with co-varying parameters such as bacterial production, grazing conferred mortality, and
phytoplankton production and composition.
Results presented in this dissertation have for instance demonstrated that the typical
tri-phasic course often observed in incubation experiments (i.e. a growth-, a grazing, and a
post-grazing phase; (Jürgens & Güde, 1994)) is accompanied by strong fluctuations in
bacterial community composition. It was suggested that grazing by heterotrophic
microzooplankton was an important factor in such incubations and was likely to be
responsible for shifts in bacterial community composition between peak in bacterial biomass
and subsequent reduction of bacterial numbers by grazing. Grazing has also been identified in
other studies as a structuring force of bacterioplankton diversity (Jürgens et al., 1999;
Pernthaler, et al., 1997; Suzuki, 1999; van Hannen et al., 1999) and it is now widely accepted
that grazing by bacterivorous microzooplankton may affect community composition. Besides
confirming that grazing can affect community composition in mesocosm experiments,
DGGE-fingerprinting performed on DNA as well as on RNA (after reverse transcription) in
the second mesocosm experiment also suggested that it most strongly affected active bacterial
populations, matching previous suggestions and observations (Del Giorgio, et al., 1996;
Gasol, et al., 1995). There was circumstantial evidence for removal of active bacterial
populations in the fluctuation of the numbers of rRNA-derived bands representing active
bacterial populations, which decreased after the grazing phase in mesocosm experiments. It
has to be pointed out, however, that an increase in activity and growth rate may also be a
bacterial strategy to compensate for grazing inflicted losses and hence to allow to co-exist
with their predators (Pernthaler, et al., 1997).
Another important aspect of the present study is related to the question how mesocosm
experiments, or incubation experiments in general, can help to answer questions in aquatic
microbial ecology. Despite their wide application in the field (Duarte et al., 1997) only
recently we and other authors have begun to analyse community composition of incubated
bacterial assemblages by molecular methods, e.g. (Eilers et al., 2000; Jürgens, et al., 1999;
Riemann et al., 2000). The present study extends previous observations on the effects of
confinement (Ferguson et al., 1984) by giving examples of the way community composition
can change significantly as analysed by molecular biological methods. Furthermore, the
results presented here contrast sharply to results of previous studies suggesting only limited
shifts in confined samples (Lee & Fuhrman, 1991), possibly because of different handling and
incubation times and conditions. Additionally, the different methods to evaluate shifts in
community composition, DGGE and cloning of 16S rRNA genes on the one hand, and
community DNA hybridisation on the other hand, may have contributed to drawing the
On the one hand the mesocosm experiments were useful to map changes in
community composition in phases of varying gross activity, and hence to tentatively identify
factors that structure the community, such as grazing by heterotrophic protozoa. Thus,
mesocosm experiments may be of use for gaining information about potential short-term
microbial community dynamics in marine environments, such as exemplified by episodic
up-welling events and build-up and decay of phytoplankton blooms (Kerkhof, et al., 1999).
On the other hand, the results show that mesocosm experiments may not be useful for
experimental perturbation studies, despite the interesting successions of bacterial populations
that are usually occurring. Although artificial nutrient addition to Mediterranean coastal water
had a strong effect on global parameters such as bacterial production and biomass (Lebaron et
al., 2001; Lebaron et al., 1999), the impact on bacterial diversity was less pronounced than
might be expected. The aspect of food web manipulation in combination with confinement
had much greater impact on bacterial diversity than had nutrient addition. This was evident
from the marked deviation of DGGE patterns of microbial assemblages in control and nutrient
enriched mesocosms from those in the natural situation (see Chapter 4). Apparently, a true
negative control can hardly be achieved due to effects of sample handling (e.g. enrichment
with organic carbon from damaged cells) and it seems obvious, therefore, that untreated
mesocosms cannot be considered real controls against which to test the effect of experimental
nutrient addition. Paerl noted (Paerl, 1998) that “microbial taxa (and their interactions) may
respond opportunistically to anthropogenic alterations”, a description that also fits which the
microbial assemblages of the mesocosms. In the control mesocosm of the first experiment the
development of Alteromonas related bacteria was observed, which are often referred to as
typical opportunistic bacteria. Thus, to some extent all mesocosms seemed to behave as if
they were eutrophied, regarding their deviation from the natural state as even in control tanks
strong fluctuations in parameters related to microbial activity and diversity were recorded.
Eilers and colleagues (Eilers, et al., 2000) found that strong shifts in community composition
such as those observed in our mesocosm experiments were due to selective activation of
culturable groups of bacterioplankton, and, similar to the first mesocosm experiment (see
Chapter 3), they observed an increase in the abundance of bacteria affiliated with
Alteromonas during confinement without substrate addition, confirming our observations.
Studies in the field were conducted in the Rhone outflow plume, and in a landlocked
bay on the coast of Norway. The short residence time of water in the Rhone outflow plume
however, did not allow to follow the true response of bacterial populations to the changing
environmental conditions in terms of changes in genetic diversity. The situation was further
complicated by the significant differences in bacterial abundance in river and seawater, which
biased the detection on DGGE of bacterial populations in favour of freshwater populations.
Nevertheless, there were interesting differences regarding the fraction of active bacterial
populations as estimated by specific leucine incorporation rates on the one hand and the ratio
of DNA and RNA derived DGGE bands on the other hand. Interestingly, both indicated that
the activity was distributed amongst a larger fraction in the marine end part of the gradient.
To date there have been only few reports on seasonal changes in marine microbial
assemblages. The study of the landlocked bay Hopavågen over a period of 40 month therefore
certainly provides new information in showing that there were seasonal and annually
recurring distribution of specific phylotypes. A marked change in community composition
was concomitant with the beginning of the period of phytoplankton production in spring. At
that time α-proteobacterial phylotypes related to Roseobacter became predominant in DGGE
banding patterns and persisted throughout summer, emphasising that these bacteria are tightly
coupled to phytoplankton production as has been suggested previously (González et al.,
2000). Although, eutrophication affected primary production and algal biomass in 1999
(when nutrient loads were doubled with respect to those in 1998) and some marked peaks in
diatom biomass during summer, there was no evidence of marked changes in the seasonal
distribution of dominant bacterial phylotypes. However, this observation does not exclude a
potential of eutrophication to affect bacterial community composition and dynamics, but may
be a consequence of a still too low level of artificial eutrophication.
Eutrophication has also been implied in changing patterns of phytoplankton
community composition (Paerl, 1998). While the attempt to experimentally check specificity
of bacterial-algal associations in laboratory cultures was not completely successful, the results
of the study demonstrate that there is potential for co-occurrence of algae and certain groups
of bacteria. Especially, bacteria from the CFB and the α-Proteobacteria formed mini-consortia
with the algae that may resemble the functional roles of the constituents in natural
ecosystems. Furthermore, the observation that β-proteobacterial phylotypes are propagating in
seawater media brought indirect evidence for existence of salt-water tolerant β-Proteobacteria.
Influence of eutrophication?
In principle a number of environmental conditions is potentially determining the composition
of bacterial assemblages. The factors emphasised above, i.e. grazing, phytoplankton activity
and production and composition, are those that have been implicated a pivotal role in shaping
bacterial community composition in the present study, but what about the influence of
A direct influence of nutrient concentrations on physiological state of bacteria may be
possible as bacteria differ in their uptake kinetics of inorganic nutrients. Furthermore, gross
differences in the quality and availability of nitrogen in either inorganic or and organic form
might potentially favour growth of adapted bacterial populations. In the mesocosms, however,
it seemed that increases in nutrient concentrations only had limited effect on the composition
of the community, although some predominant phylotypes in mesocosms (i.e. the dominant
bands), were similar in controls and nutrient-enriched mesocosms. Eutrophication mainly
affected the speed and magnitude of changes in biomass and thymidine incorporation
(compare Lebaron, et al., 2001). Divergence between control and nutrient-enriched
mesocosms was more profound at the end of the experiment when many other processes had
had the chance to modify the composition of the bacterial community.
In the landlocked bay experiment an influence on bacterial diversity could neither be
seen. However, this does not exclude the possibility that eutrophication affected some
community members. On the one hand such populations might have been below the detection
threshold of the DGGE method, on the other hand similar bands between years might
represent physiologically distinct populations with regards to nutrient acquisition.
Additionally, the level of nutrient addition may still have been to low in order to induce
changes in food web structure sufficient to affect bacterial community dynamics and
Methodology applied and its constraints
Molecular biological techniques have provided new insights in aquatic microbial ecology.
They have helped substantially to change our view of the organisation and complexity of
natural microbial assemblages. All molecular biological techniques have their strong and
weak points, and therefore not a single technique is ideal for every purpose. Cloning and
sequencing of complete 16S rRNA genes certainly allows for phylogenetic inferences with a
superior confidence than is offered by partial 16S rRNA genes, such as those recovered from
DGGE bands. However, DGGE allows processing and analysis of a greater number of
samples than the cloning approach and hence, for good reason, was the preferred method in
the work described here.
While 16S rRNA based techniques are still (and will be for some time) widely used
tools for analysing microbial community composition, or fluctuation of composition, they do
not yield information about the function of individual community members. Combination of
molecular biological methods with microsensor measurements have resolved some of these
problems in studies of biofilms, aggregates and sediments, by making it possible to sample
the chemical microenvironment of the bacterial populations, and thus allowing to study fluxes
and quantify substrate turnover in situ (Amann & Kühl, 1998; Santegoeds et al., 1998;
Schramm et al., 1998). Unfortunately, no comparable tool is available currently for studying
pelagic bacterial populations. Non-destructive sampling of the presumably gel-like matrix of
the ocean (Azam, 1998) is not possible. Substrate addition experiments for probing the
physiological status of pelagic bacterial communities rely on incubation for periods of a few
hours and may fail to provide information on the populations predominating in situ, due to
growth of opportunistic bacterial populations (Eilers, et al., 2000; Sherr et al., 1999).
Combination of microautoradiography with FISH is one possibility to study uptake of defined
substrates by natural prokaryotic populations that yields new information on potential
activities of microbes (Cottrell & Kirchman, 2000; Lee et al., 1999; Ouverney & Fuhrman,
1999), yet the method does not discriminate between mere uptake and ability to oxidise a
Another new technique holding promise for linking activity and identity of microbial
populations is stable isotope labelling. Boschker and colleagues used stable isotope probing to
label microbial populations involved in acetate and methane metabolism in estuary sediments.
Subsequently the isotopic composition of lipid biomarkers was compared to known
phospholipid fatty acid (PLFA) profiles as a means of identification of microbial populations
(Boschker et al., 1998). A limitation for reliable identification may reside in a relatively small
PLFA database, and the possibility that populations lack diagnostic biomarkers molecules.
A variation of stable isotope probing allows identification of microorganisms involved
in processing of specific substrates on the basis of 13C-enriched genomic DNA (Radajewski et
al., 2000). Radajewski and colleagues applied this technique to study methanol-consuming
methylotroph populations in forest soil microcosms. DNA of populations assimilated 13C-
labelled methanol became relatively heavy and could be separated by density gradient
centrifugation from the 12C-DNA of microorganisms that had not incorporated the heavy
carbon. Analysis of diversity in the heavy DNA-fraction literally identifies organisms that are
processing the added substrate. These may also be involved in substrate turnover in situ, but
the importance of the identified organisms in the environment should subsequently be verified
by hybridisation of RNA with taxon-specific oligonucleotide probes. A drawback of the
method is that it requires relatively long incubation times. Furthermore, dilution of substrates
and/or crossfeeding of metabolites between different populations may render the isotopic
enrichment of DNA ineffective (Radajewski, et al., 2000).
Application of environmental genomics as well as proteomics are exciting new
possibilities to learn more about uncultured microbial taxa, and begin to make their mark in
marine microbial ecology (DeLong, 2001). More genome sequences have become available
recently, and those of strains of the same species may differ markedly in gene content
(Boucher et al., 2001). That adds to the concerns about a priori linking of 16S phylotype to
phenotype and suggests a severe limit of 16S rRNA as a marker in ecological studies to
discriminate between potentially phenotypic diverse members of clades such as the
Thus, special emphasis should therefore be put on the functional aspects of bacterial
populations in future studies. This can, for instance, be achieved by targeting of functional
genes which are associated with biogeochemical cycling of specific compounds. Sequencing
of the so-called ‘metagenome’ and transcriptome of environmental samples may be an
additional possibility to identify functional roles of microbes in the environment. Recently,
Beja and colleagues identified by cloning and sequencing of genomic environmental DNA
that uncultivated bacteria from the SAR86 group contain genes encoding bacteriorhodopsin,
so far only been known to be expressed by halophilic archaea. Subsequently, they were able
to show the presence of rhodopsin in membrane preparations from native marine picoplankton
and also found evidence for widespread distribution of related proteorhodopsin genes in the
surface ocean (Beja et al., 2001). Such genome sequencing studies may help to elucidate
metabolic features of as yet uncultivated groups of bacterioplankton and might prove helpful
in formulation of more promising culture conditions for retrieval of these groups as cellular
On the other hand, the technological advance especially in genomic sequencing
technology has been immense over the past few years, and only few research centres can keep
up with the costs imposed by instrumentation and consumables required for genomic studies.
Furthermore, it is debatable whether the possibilities for obtaining new cultures representative
of dominant bacterioplankton have really been exhaustively exploited. A more extensive and
maybe automated approach to finding more suitable culturing conditions for as yet
uncultivated bacterioplankton might also have potential to contribute to progress in the field.
At present and probably for some time in the future, isolated strains will be required for an
understanding of the biochemical pathways that drive the biogeochemical cycling.
Suggestions that microbial ecology is a field that no longer requires input from cultivation
attempts but that can be brought forward simply by sequencing (of 16S rRNA genes) (Pace,
1996) are therefore clearly inappropriate.
16S rRNA based molecular ecological techniques have given microbial ecology a new
direction and have yielded invaluable information on the composition and dynamics of
microbial communities. Further progress in the field will rely on a combination of methods,
such as the cloning techniques with stable isotope probing, or the use of functional genes as
molecular markers. Genomic studies will probably provide more exciting findings in the
future, but culturing of environmentally important bacteria will be as important for furthering
our understanding of how bacterial populations are involved in biogeochemical cycling.
Amann, R. & Kühl, M. (1998). In situ methods for assessment of microorganisms and their
activities. Curr. Opin. Microbiol. 1, 352-358.
Amann, R. I., Ludwig, W. & Schleifer, K. H. (1995). Phylogenetic identification and in situ
detection of individual microbial cells without cultivation. Microbiol. Rev. 59, 143-169.
Azam, F. (1998). Microbial control of oceanic carbon flux: the plot thickens. Science 280,
Azam, F. (2001). Introduction, history, and overview: the 'methods' to our madness. In
Marine Microbiology. Edited by J. H. Paul. London: Academic Press.
Azam, F., Fenchel, T., Field, J. G., Meyer-Reil, R. A. & Thingstad, F. (1983). The
ecological role of water column microbes in the sea. Mar. Ecol. Prog. Ser. 10, 257-263.
Beja, O., Spudich, E. N., Spudich, J. L., Leclerc, M. & DeLong, E. F. (2001).
Proteorhodopsin phototrophy in the ocean. Nature 411, 786-789.
Boschker, H. T. S., Nold, S. C., Wellsbury, P., Bos, D., Degraaf, W., Pel, R., Parkes, R. J.
& Cappenberg, T. E. (1998). Direct linking of microbial populations to specific
biogeochemical processes by C-13-labelling of biomarkers. Nature 392, 801-805.
Boucher, Y., Nesbø, C. L. & Doolittle, W. F. (2001). Microbial genomes: dealing with
diversity. Curr. Opin. Microbiol. 4, 285-289.
Caron, D. A. (1994). Inorganic nutrients, bacteria, and the microbial loop. Microb. Ecol. 28,
Choi, J. W., Sherr, B. F. & Sherr, E. B. (1999). Dead or alive? A large fraction of ETS-
inactive marine bacterioplankton cells, as assessed by reduction of CTC, can become ETS-
active with incubation and substrate addition. Aquat. Microb. Ecol. 18, 105-115.
Cotner, J. B., Ammerman, J. W., Peele, E. R. & Bentzen, E. (1997). Phosphorus-limited
bacterioplankton growth in the Sargasso Sea. Aquat. Microb. Ecol. 13, 141-149.
Cottrell, M. T. & Kirchman, D. L. (2000). Natural assemblages of marine proteobacteria
and members of the Cytophaga-Flavobacter cluster consuming low- and high-molecular-
weight dissolved organic matter. Appl. Environ. Microbiol. 66, 1692-1697.
Del Giorgio, P. A. & Cole, J. J. (1998). Bacterial growth efficiency in natural aquatic
systems. Ann. Rev. Ecol. Syst. 29, 503-541.
Del Giorgio, P. A., Cole, J. J. & Cimbleris, A. (1997). Respiration rates in bacteria exceed
phytoplankton production in unproductive aquatic systems. Nature 385, 148-151.
Del Giorgio, P. A., Gasol, J. M., Vaque, D., Mura, P., Agusti, S. & Duarte, C. M. (1996).
Bacterioplankton community structure: protists control net production and the proportion of
active bacteria in a coastal marine community. Limnol. Oceanogr. 41, 1169-1179.
DeLong, E. F. (2001). Microbial seascapes revisited. Curr. Opin. Microbiol. 4, 290-295.
Duarte, C. M., Gasol, J. M. & Vaque, D. (1997). Role of experimental approaches in
marine microbial ecology. Aquat. Microb. Ecol. 13, 101-111.
Eilers, H., Pernthaler, J. & Amann, R. (2000). Succession of pelagic marine bacteria during
enrichment: a close look at cultivation-induced shifts. Appl. Environ. Microbiol. 66, 4634-
Eilers, H., Pernthaler, J., Glöckner, F. O. & Amann, R. (2000). Culturability and in situ
abundance of pelagic bacteria from the North Sea. Appl. Environ. Microbiol. 66, 3044-3051.
Ferguson, R. L., Buckley, E. N. & Palumbo, A. V. (1984). Response of marine
bacterioplankton to differential filtration and confinement. Appl. Environ. Microbiol. 47, 49-
Field, K. G., Gordon, D., Wright, T., Rappe, M., Urbach, E., Vergin, K. & Giovannoni,
S. J. (1997). Diversity and depth-specific distribution of SAR11 cluster rRNA genes from
marine planktonic bacteria. Appl. Environ. Microbiol. 63, 63-70.
Fuhrman, J. A. & Azam, F. (1982). Thymidine incorporation as a measure of heterotrophic
bacterioplankton production in marine surface waters: evaluation and field results. Mar. Biol.
Gasol, J. M., Del Giorgio, P. A., Massana, R. & Duarte, C. M. (1995). Active versus
inactive bacteria: size-dependence in a coastal marine plankton community. Mar. Ecol. Prog.
Ser. 128, 91-97.
Giovannoni, S. & Rappé, M. (2000). Evolution, diversity, and molecular ecology of marine
prokaryotes. In Microbial ecology of the oceans, pp. 47-84. Edited by D. L. Kirchman. New
York: Wiley-Liss Inc.
Giovannoni, S. J., Britschgi, T. B., Moyer, C. L. & Field, K. G. (1990). Genetic diversity
in Sargasso Sea bacterioplankton. Nature 345, 60-63.
Glöckner, F. O., Fuchs, B. M. & Amann, R. (1999). Bacterioplankton compositions of
lakes and oceans: A first comparison based on fluorescence in situ hybridization. Appl.
Environ. Microbiol. 65, 3721-3726.
González, J. M., Simó, R., Massana, R., Covert, J. S., Casamayor, E. O., Pedrós-Alió, C.
& Moran, M. A. (2000). Bacterial community structure associated with a
dimethylsulfoniopropionate-producing North Atlantic algal bloom. Appl. Environ. Microbiol.
Gordon, D. A. & Giovannoni, S. J. (1996). Detection of stratified microbial populations
related to Chlorobium and Fibrobacter species in the Atlantic and Pacific oceans. Appl.
Environ. Microbiol. 62, 1171-1177.
Hahn, M. W., Moore, E. R. B. & Höfle, M. G. (1999). Bacterial filament formation, a
defense mechanism against flagellate grazing, is growth rate controlled in bacteria of different
phyla. Appl. Environ. Microbiol. 65, 25-35.
Hobbie, J. E., Daley, R. J. & Jasper, S. (1977). Use of Nuclepore filters for counting
bacteria by fluorescence microscopy. Appl. Environ. Microbiol. 33, 1225-1228.
Jannasch, H. W. & Jones, G. E. (1959). Bacterial populations in seawater as determined by
different methods of enumeration. Limnol. Oceanogr. 4, 128-139.
Jürgens, K. & Güde, H. (1994). The potential importance of grazing-resistant bacteria in
planktonic systems. Mar. Ecol. Prog. Ser. 112, 169-188.
Jürgens, K., Pernthaler, J., Schalla, S. & Amann, R. (1999). Morphological and
compositional changes in a planktonic bacterial community in response to enhanced
protozoan grazing. Appl. Environ. Microbiol. 65, 1241-1250.
Karner, M. & Fuhrman, J. A. (1997). Determination of active marine bacterioplankton: A
comparison of universal 16S rRNA probes, autoradiography, and nucleoid staining. Appl.
Environ. Microbiol. 63, 1208-1213.
Kelly, K. M. & Chistoserdov, A. Y. (2001). Phylogenetic analysis of the succession of
bacterial communities in the Great South Bay (Long Island). FEMS Microbiol. Ecol. 35, 85-
Kerkhof, L. J., Voytek, M. A., Sherrell, R. M., Millie, D. & Schofield, O. (1999).
Variability in bacterial community structure during upwelling in the coastal ocean.
Hydrobiologia 401, 139-148.
Lebaron, P., Servais, P., Troussellier, M., Courties, C., Muyzer, G., Bernard, L.,
Schäfer, H., Pukall, R., Stackebrandt, E., Guindulain, T. & Vives-Rego, J. (2001).
Microbial community dynamics in Mediterranean nutrient-enriched mesocosms: changes in
abundances, activity, and composition. FEMS Microbiol. Ecol. 34, 255-266.
Lebaron, P., Servais, P., Troussellier, M., Courties, C., Vives-Rego, J., Muyzer, G.,
Bernard, L., Guindulain, T., Schäfer, H. & Stackebrandt, E. (1999). Changes in bacterial
community structure in seawater mesocosms differing in their nutrient status. Aquat. Microb.
Ecol. 19, 255-267.
Lee, N., Nielsen, P. H., Andreasen, K. H., Juretschko, S., Nielsen, J. L., Schleifer, K. H.
& Wagner, M. (1999). Combination of fluorescent in situ hybridization and
microautoradiography - a new tool for structure-function analyses in microbial ecology. Appl.
Environ. Microbiol. 65, 1289-1297.
Lee, S. & Fuhrman, J. A. (1991). Species composition shift of confined bacterioplankton
studied at the level of community DNA. Mar. Ecol. Prog. Ser. 79, 195-201.
Massana, R., DeLong, E. F. & Pedrós-Alió, C. (2000). A few cosmopolitan phylotypes
dominate planktonic archaeal assemblages in widely different oceanic provinces. Appl.
Environ. Microbiol. 66, 1777-1787.
Monger, B. C. & Landry, M. R. (1991). Prey-size dependency of grazing by free-living
marine flagellates. Mar. Ecol. Prog. Ser. 74, 239-248.
Murray, A. E., Preston, C. M., Massana, R., Taylor, L. T., Blakis, A., Wu, K. & Delong,
E. F. (1998). Seasonal and spatial variability of bacterial and archaeal assemblages in the
coastal waters near Anvers island, Antarctica. Appl. Environ. Microbiol. 64, 2585-2595.
Muyzer, G. (1998). Structure, function and dynamics of microbial communities: the
molecular biological approach. In Advances in molecular ecology, pp. 87-117. Edited by G.
R. Carvalho: NATO Science Series.
Muyzer, G. (1999). DGGE/TGGE a method for identifying genes from natural ecosystems.
Curr. Opin. Microbiol. 28, 317-322.
Muyzer, G. & Smalla, K. (1998). Application of denaturing gradient gel electrophoresis
(DGGE) and temperature gradient gel electrophoresis (TGGE) in microbial ecology. Antonie
Van Leeuwenhoek 73, 127-141.
Ouverney, C. C. & Fuhrman, J. A. (1999). Combined microautoradiography-16S rRNA
probe technique for determination of radioisotope uptake by specific microbial cell types in
situ. Appl. Environ. Microbiol. 65, 1746-1752.
Pace, N. R. (1996). New perspective on the natural microbial world: molecular microbial
ecology. ASM News 62, 463-470.
Paerl, H. W. (1998). Structure and function of anthropogenically altered microbial
communities in coastal waters. Curr. Opin. Microbiol. 1, 296-302.
Pernthaler, J., Posch, T., Simek, K., Vrba, J., Amann, R. & Psenner, R. (1997).
Contrasting bacterial strategies to coexist with a flagellate predator in an experimental
microbial assemblage. Appl. Environ. Microbiol. 63, 596-601.
Radajewski, S., Ineson, P., Parekh, N. R. & Murrell, J. C. (2000). Stable isotope probing
as a tool in microbial ecology. Nature 403, 646-649.
Riemann, L., Steward, G. F. & Azam, F. (2000). Dynamics of bacterial community
composition and activity during a mesocosm diatom bloom. Appl. Environ. Microbiol. 66,
Sanders, R. W., Caron, D. A., Berninger, U. G. (1992). Relationships between bacteria and
heterotrophic nanoplankton in marine and fresh waters: an inter-ecosystem comparison. Mar.
Ecol. Prog. Ser. 86, 1-14.
Santegoeds, C. M., Ferdelman, T. G., Muyzer, G. & De Beer, D. (1998). Structural and
functional dynamics of sulfate-reducing populations in bacterial biofilms. Appl. Environ.
Microbiol. 64, 3731-3739.
Schramm, A., De Beer, D., Wagner, M. & Amann, R. (1998). Identification and activities
in situ of Nitrosospira and Nitrospira spp. as dominant populations in a nitrifying fluidized
bed reactor. Appl. Environ. Microbiol. 64, 3480-3485.
Sherr, E. B. & Sherr, B. F. (1994). Bacterivory and herbivory: Key roles of phagotrophic
protists in pelagic food webs. Microb. Ecol. 28, 223-235.
Sherr, E. B., Sherr, B. F. & Sigmon, C. T. (1999). Activity of marine bacteria under
incubated and in situ conditions. Aquat. Microb. Ecol. 20, 213-223.
Suttle, C. A. (1994). The significance of viruses to mortality in aquatic microbial
communities. Microb. Ecol. 28, 237-243.
Suzuki, M., Rappé, M. S., Haimberger, Z. W., Winfield, H., Adair, N., Strobel, J. &
Giovannoni, S. J. (1997). Bacterial diversity among small-subunit rRNA gene clones and
cellular isolates from the same seawater sample. Appl. Environ. Microbiol. 63, 983-989.
Suzuki, M. T. (1999). Effect of protistan bacterivory on coastal bacterioplankton diversity.
Aquat. Microb. Ecol. 20, 261-272.
van Hannen, E. J., Veninga, M., Bloem, J., Gons, H. J. & Laanbroek, H. J. (1999).
Genetic changes in the bacterial community structure associated with protistan grazers. Arch.
für Hydrobiol. 145, 25-38.
Vollenweider, R. A. (1992). Coastal marine eutrophication: principles and control. In Marine
coastal eutrophication. Edited by R. A. Vollenweider, R. Marchetti & R. Viviani: Elsevier.
Ward, D. M., Weller, R. & Bateson, M. M. (1990). 16S ribosomal RNA sequences reveal
numerous uncultured microorganisms in a natural community. Nature 345, 63-65.
Weinbauer, M. G. & Höfle, M. G. (1998). Significance of viral lysis and flagellate grazing
as factors controlling bacterioplankton production in a eutrophic lake. Appl. Environ.
Microbiol. 64, 431-438.
Williams, P. J. le B. (1998). The balance of plankton respiration and photosynthesis in the
open ocean. Nature 394, 55-57.
Williams, P. J. le B. (2000). Heterotrophic bacteria and the dynamics of dissolved organic
material. In Microbial ecology of the oceans, pp. 153-200. Edited by D. L. Kirchman. New
York: Wiley-Liss Inc.
Zweifel, U. L. & Hagström, Å. (1995). Total counts of marine bacteria include a large
fraction of non-nucleoid-containing bacteria (ghosts). Appl. Environ. Microbiol. 61, 2180-
Denaturing Gradient Gel Electrophoresis in Marine Microbial
Hendrik Schäfer and Gerard Muyzer
In: Methods in Microbiology, John Paul (Ed), volume 30, pp 425-468,
Academic Press, London (2001)
During the past decade the approach of microbial community composition analysis has
changed considerably. Classical techniques such as cultivation and microscopic identification
are not sufficient to assess the diversity of bacteria in natural samples. On the one hand, lack
of conspicuous morphology and small cell size do not allow microscopic identification of the
majority of naturally occurring bacteria. On the other hand, media used for the cultivation of
microbial strains are selective and hence give a biased view of the community composition.
Furthermore, the isolation of the vast majority of naturally occurring bacteria in pure culture
is hindered by our lack of knowledge of the specific culture conditions they need and by the
potential for synergy between different organisms. Comparisons of culturable and total
microscopic cell counts from diverse habitats have demonstrated the inadequacy of the
culture-dependent approach to analyse microbial community composition (summarised in
Amann et al., 1995). Therefore, other tools are required to supplement the conventional
microbiological techniques. The introduction of molecular techniques in microbial ecology
including those that use the gene sequences of the small subunit ribosomal RNA as a
molecular marker for identification of microorganisms has changed our perception of the
diversity of microbial communities. The genes encoding small subunit ribosomal RNAs
reflect the evolutionary relationship of microorganisms (Woese, 1987) and the sequences of
these genes allow to group and identify microorganisms. Despite some uncertainties about the
phylogeny inferred from rRNA (e.g. the rooting of the different domains) which have
emerged as a result of whole-genome sequencing studies and the use of alternative molecular
markers (see e.g., Pennisi, 1998; Doolittle and Logsdon, 1998), the 16S rRNA approach
remains the standard marker (see Ludwig and Schleifer, 1999). Giovannoni and co-workers
(1990) for instance used a cultivation-independent approach consisting of PCR amplification,
cloning and sequencing of 16S rRNA gene fragments to characterise the composition of
Sargasso Sea bacterioplankton. The sequences obtained represented unknown 16S rRNA
genes of heretofore uncultivated bacteria, and confirmed the limitations of cultivation-
dependent approaches. Similar differences between culture-dependent and molecular
approaches were observed by Ward and colleagues for a hot spring cyanobacterial mat
community (Ward et al., 1990) and have been reported from microbial ecology studies
repeatedly (for a review see Muyzer, 1998).
To get a better insight into the temporal dynamics or spatial variation of microbial
communities, microbial ecosystems need to be studied over longer periods of time (e.g., days
to years) or samples from many different locations have to be analysed. Although successful,
the application of cloning and sequencing of 16S rRNA genes is too laborious and time
consuming to analyse a large number of samples, even with the recent progress in sequencing
technology. Genetic fingerprinting techniques, however, are excellently suited for comparison
of large numbers of samples. Genetic fingerprinting of microbial communities provides
banding patterns or profiles that reflect the genetic diversity of the community. Denaturing
gradient gel electrophoresis (DGGE) of PCR-amplified gene fragments is one of the genetic
fingerprinting techniques used in microbial ecology (Muyzer, 2000). In DGGE similar-sized
DNA fragments are separated in a gradient of DNA denaturants according to differences in
sequence. A variant of DGGE, temperature gradient gel electrophoresis (TGGE) makes use of
a temperature gradient to separate gene fragments. DGGE is relatively easy to perform and is
especially suited for the analysis of multiple samples. Since its introduction into microbial
ecology by Muyzer et al. (1993) it has been adapted in many laboratories as a convenient tool
for assessment of microbial diversity in natural samples. A general overview of PCR-DGGE
fingerprinting of microbial communities is shown in Figure 1.
DGGE in marine microbial ecology
16S rRNA gene sequences
Extraction of nucleic acids
16S rRNA gene fragments
DNA & RNA
Sequencing of bands
F lectobacillus glomeratus
marine aggregate clone agg58
Prionitis lanceolata gall symbiont
Chlorella mirabilis plastid
clone OM20 - Eukaryote plastid
C ytophaga &
Flow diagram of PCR-DGGE analysis of microbial communities. The different steps are discussed in detail in
this chapter. Briefly, bacteria are collected on filters, their nucleic acids are extracted and used as template in the
PCR. The mixture of PCR products is analysed by DGGE. Community profiles can be further analysed with
statistical methods, such as UPGMA and MDS (see Figure 3 for an example). To identify the community
members, bands are excised from the denaturing gradient gels, re-amplified and sequenced. The sequence data
are used for phylogenetic analysis, or can be used for the design of specific probes to detect bacterial cells in situ
(see chapter by Amann). The gel shows temporal shifts in the bacterial diversity of mesocosm samples which are
reflected in different community profiles. The time interval between the samplings were: 2 days between
samples run on lane 1 and 2, and 3 days between samples of lane 2 and 3 (total time between sample 1 to 3: 5
days). Lane M shows a marker composed of PCR-products from 5 different DNAs (see section on DGGE
standards). Sequences determined from the DGGE bands are shown in bold-type in the tree. The phylogenetic
tree has been created with the special parsimony tool implemented in the software program ARB (Ludwig et al.,
1998, Strunk and Ludwig, 1998), which allows the reliable positioning of partial sequence data in a tree derived
from complete sequences, without affecting the topology of the tree.
Principle of DGGE separation
Amplification of DNA extracted from mixed microbial communities with primers specific for
16S rRNA gene fragments of bacteria result in mixtures of PCR products. Because these
products all have the same size, they can not be separated from each other by agarose gel
electrophoresis. However, sequence variations between different bacterial rRNAs bring about
different melting properties of these DNA molecules, and separation can be achieved in
polyacrylamide gels containing a gradient of DNA denaturants, such as a mixture of urea and
formamide. PCR products enter the gel as double-stranded molecules; as they proceed
through the gel, the denaturing conditions gradually become stronger. PCR products with
different sequences therefore start melting at different positions (i.e. at different denaturant
concentrations) in the gel. Melting proceeds in so-called ‘melting domains’. Once a domain
with the lowest melting temperature reaches its melting temperature at a particular position in
the denaturant gradient, a transition from a double stranded to a partially melted molecule
occurs. The protruding single strands practically cause a halt of the molecule at that position.
To prevent the complete dissociation of the two DNA strands, a 40-nucleotide GC-rich
sequence (‘GC-clamp’) is attached at the 5’-end of one of the PCR primers.
Applications of PCR-DGGE in marine microbial ecology
PCR-DGGE fingerprinting is a tool for monitoring variations in microbial genetic diversity,
providing a minimum estimate of the richness of predominant community members.
Furthermore, DGGE facilitates the identification of individual populations by hybridisation
analysis of patterns with specific probes, or by sequence analysis of individual bands.
PCR-DGGE has been used to investigate the diversity of microbial communities, to determine
the spatial and temporal variability of bacterial populations, and to monitor community
behaviour after natural or induced environmental perturbations. It has been used to study
communities in various habitats, such as soil, sediments, water column, hydrothermal vents,
microbial mats, mesocosms, or sewage treatment plants. Here we will only give some
examples of the application of PCR-DGGE in marine ecosystems. For a more comprehensive
overview of the use of PCR-DGGE in microbial ecology the reader is referred to Muyzer
(1998, 1999) and Muyzer and Smalla (1998).
DGGE to study spatial and temporal variability of bacterial populations
The distribution of microbial populations in the marine water column depends on numerous
factors and variables. Especially in stratified systems exhibiting strong physicochemical
gradients, DGGE fingerprinting can reveal a concomitant stratification of resident microbial
assemblages. Teske et al. (1996) used PCR-DGGE to study the distribution of sulphate-
reducing bacteria (SRB) in a stratified Danish fjord. PCR-DGGE combined with hybridisation
analysis showed that the presence of SRB increased at and below the chemocline. Most-
probable number (MPN) counts of SRB were done in parallel and showed a similar trend for
the distribution of SRB. Interestingly, DGGE patterns of PCR-products obtained from cDNA
after reverse transcription of RNA, representing the active populations, were different from
those obtained after amplification of genomic DNA. Despite the agreement between MPN
and DGGE, the hybridisation of DGGE patterns with oligonucleotide probes and sequencing
analysis of DGGE bands revealed that the SRB enriched in the MPN-tubes had a different
phylogenetic affiliation than the SRB detected in the natural samples. The finding that SRB
obtained from the MPN cultures belonged to the genera Desulfovibrio, Desulfobulbus, and
Desulfobacter, but those in the DGGE patterns of natural samples represented an independent
DGGE in marine microbial ecology
lineage of the δ-Proteobacteria, verified the potential disagreement between culture-dependent
and molecular methods due to selection of culturable types of SRB.
The potential of PCR-DGGE for the analysis of large sets of samples was recognised
by Ferrari and Hollibaugh (1999). They processed 100 samples from different stations in the
Arctic Ocean to analyse the spatial variation in the diversity of bacterioplankton assemblages.
DGGE fingerprints of the samples were subjected to image analysis and the spatial variation
of the bacterioplankton assemblage was inferred by regression analysis of the similarity of
densitometric curves derived from the DGGE patterns. The resulting dendrogram separated
all DGGE patterns into five major clusters with minimally 80% similarity. While clustering of
some samples corresponded to samples taken in a specific region of the Arctic Ocean, there
was no correlation of geography and clustering of other samples. The authors noted that
clustering of the majority of samples rather seemed to reflect different phases of the cruise
and might therefore be confounded with temporal variation over the 44 day period of the
cruise (Ferrari and Hollibaugh, 1999).
The bacterioplankton assemblages of two estuaries in California, San Francisco Bay
and Tomales Bay, differing markedly in a number of physical and biological factors, had been
shown to differ in metabolic properties. The analysis of samples from both estuaries by PCR-
DGGE supported the hypothesis that metabolic differences were reflected in a different
bacterioplankton composition (Murray et al., 1996). Yet, a few bands were common in all
samples, and a number of bands were detectable at different times in both estuaries, raising
the question as to what extent factors, such as the relative activity of the detected populations
or metabolic plasticity, might influence the differences in metabolic profiles (Murray et al.,
In another study, Murray and colleagues (1998) addressed spatial as well as temporal
variations in bacterial community composition in the waters around Anvers Island
(Antarctica). No obvious variation was detected between samples taken within one month
from different points in an area of about three square nautical miles (3 and 50 m depth).
However, samples retrieved from several depths up to 1,200 m on two occasions within seven
weeks showed variations in DGGE-patterns especially at depths of 500 and 1,200 m
indicating compositional changes of the bacterial community. The authors argued that due to
the low bacterial activity (estimated by leucine incorporation) advective mixing processes
rather than bacterial growth might have caused most of the variation. Seasonal variation in
bacterial community composition of the surface waters was inferred from changes in DGGE-
fingerprints over a period of almost 9 months at one station. Interestingly, the number of
phylotypes decreased during the transition from spring to summer and increased from summer
Riemann et al. (1999) used PCR-DGGE to map the genetic diversity of
bacterioplankton in the surface-, mid- and deep water of the Arabian Sea during two
consecutive monsoon periods and concluded that there was a high horizontal homogeneity of
the microbial assemblages. Moreover, the dominant bands in DGGE profiles of the bacterial
communities sampled eleven months apart, were remarkably similar, suggesting that if there
was a seasonal variation in the bacterioplankton assemblage, it might be a predictable one.
Predominant phylotypes were identified by cloning and sequencing of DGGE bands and were
members of groups common in oceanic waters, e.g. members of the SAR11-cluster and the
cyanobacteria. However, it was remarkable that none of the bands corresponded to γ-
Proteobacteria or to members of the Cytophaga-Flavobacterium-Bacteroides phylum (CFB),
and that 16S rRNA gene fragments similar to those of magnetotactic bacteria were retrieved.
West and Scanlan (1999) investigated the genetic structure of Prochlorococcus
communities by molecular methods in two depth profiles from the surface to around 100 m
water depth in the Eastern North Atlantic, to assess the distribution of high-light (HL) and
low-light (LL) adapted populations. Cloning and sequencing, as well as hybridisation with HL
and LL specific gene probes of DNA amplified from different depths were performed. PCR
products amplified with a cyanobacteria specific primer (Nübel et al., 1997) and a
Prochlorococcus specific primer were separated on DGGE. All three methods indicated a
niche-partitioning of Prochlorococcus genotypes HL and LL in the water column and provide
a genetic support for flow cytometric observations of dim and bright Prochlorococcus
DGGE to monitor population shifts after environmental perturbation
As pointed out above, PCR-DGGE analyses can be performed with DNA as well as with
RNA. While DNA-derived PCR amplified 16S rRNA gene fragments are related to the
presence of different bacterial populations, analyses of rRNA-derived PCR products can give
an indication of which bacterial populations contribute to the RNA pool. As the cellular
concentration of ribosomal RNA is related to the (recent) activity of cells it helps in surveying
changes in the activity of bacterial populations. An example of potential differences between
DNA- and RNA-derived DGGE fingerprints is shown in Figure 2. Similarly, the analysis of
the genetic diversity and expression of functional genes can be performed using either DNA
or mRNA. Here, PCR-DGGE analysis of DNA-derived PCR products show the genetic
diversity (presence) of certain functional genes, while PCR-products obtained after DNase
digest and reverse transcription of mRNA show the diversity of expressed genes (Wawer et
Rossello-Mora et al. (1999) investigated the response of the microbial community of
marine sediments to amendment with cyanobacterial biomass under anaerobic conditions.
Fluorescence in situ hybridisation (FISH), DGGE of PCR products obtained from DNA as
well as from cDNA after reverse transcription of RNA, and sequencing of 16S rDNA PCR
products were used to assess changes in the microbial community composition. Concomitant
changes in the activity of the community were followed by measurements of carbon
mineralisation, sulphate reduction, and ammonium production rates. Addition of
cyanobacterial biomass resulted in marked changes in the composition. Dominant bands from
RNA-derived banding patterns were affiliated with members of the CFB. FISH with probes
specific for these CFB-populations showed that, although sulphate reduction was the main
mineralisation process, members of the CFB, but not SRB showed the highest increase in
abundance as detected by FISH. The authors concluded that these CFB played an important
role in the anaerobic decomposition of complex organic matter and suggested that CFB might
be responsible for hydrolysis of macromolecules and fermentation.
Mesocosm experiments were performed by Lebaron et al. (2001) and Schäfer et al.
(2001) to study changes in the activity and diversity of bacterial assemblages from the
Mediterranean Sea after addition of nutrients. Fluctuations in activity were recorded in
parallel to variation in community composition, which was assessed by PCR-DGGE.
Different phases were observed during the incubation corresponding to an initial increase of
bacterial numbers, followed by an increase of heterotrophic protozoa cropping the bacterial
production and a new increase of bacterial production after the peak in grazing activity
(growth-, grazing, and post-grazing phase, respectively). These phases were reflected by
concomitant changes in DGGE-fingerprints of the bacterial assemblage. Both, nutrient
addition as well as grazing of protozoa seemed to effect changes in the bacterial genetic
diversity. Multidimensional scaling analysis of DGGE patterns showed that differences in the
development of the bacterial communities occurred between nutrient-enriched and control
mesocosms and indicated that duplicate treatments were reproducible. Sequencing of DGGE
DGGE in marine microbial ecology
bands was used to identify several microbial populations. DGGE-bands of some populations
disappeared from the DGGE patterns during the grazing phase, while members of the
Cytophaga-Flavobacterium-Bacteroides phylum and Ruegeria-like bacteria became
especially important after the peak in grazing activity. The latter populations also dominated
the RNA-derived DGGE-fingerprints and hence it was suggested that these populations
escaped the grazing pressure and were important contributors to bacterial production and
activity in the post-grazing phase of the experiment.
DNA RNARNA DNA
DGGE patterns of PCR-amplified 16S rRNA gene fragments obtained from
bacterial DNA, indicating the presence of bacteria, or from bacterial ribosomal
RNA, showing the most active populations within the assemblage. Water samples
from two different locations were analysed: Sample A (lanes 1 and 2) is from
surface water taken off the coast of Banyuls-sur-mer (France) in May 1997;
Sample B (lanes 3 and 4) is from coastal water taken near the mouth of the river
Rhône (France) in April 1998. Fingerprints of the natural bacterial assemblages
were obtained from DNA (lanes 1 and 3) or after DNA digestion and reverse
transcription of RNA (lanes 2 and 4). Note the differences in DNA and RNA
derived patterns in the upper part of lane 1 and 2, where some of the bands seen in
the DNA derived pattern are weaker or not represented at all in the RNA derived
patterns. This indicates that corresponding populations have a relatively low RNA
content, and hence are probably less active than others. The marked difference in
intensity of the band at the bottom of the profiles shown in lanes 3 and 4 indicates
that a population contributes relatively less to the DNA pool, but relatively much
to the RNA pool indicator of recent cellular activity. Hence, cells of this
populations probably have a high rRNA content, which might indicate that they
In a multidisciplinary approach McCaig et al. (1999) studied the impact of fish
farming in cages on N-cycling and community structure of the underlying sediment. Organic
carbon content and ammonium concentration of the sediment measured along a transect from
the fish cage to a distance of 40 m from the cage. Carbon content and ammonium
concentration were much higher under the fish cage than at the other sampling sites along the
transect. Furthermore, nitrification and denitrification were strongly inhibited beneath the fish
cage. DGGE was used to profile the diversity of beta-ammonium oxidising bacteria (β-AOB),
along a transect from underneath the fish cages to 40 m from the cage, after PCR
amplification of 16S rRNA gene fragments with primers specific for β-AOB (Kowalchuk et
al., 1997). DGGE-profiles were blotted onto a membrane and hybridised with probes specific
for subclusters of the β-AOB; to reveal the identity of β-AOB populations. DGGE-profiles of
highly polluted sediments under the fish cage showed two prominent bands that were only
faintly visible in DGGE profiles of samples from 20 and 40 m from the cage.
DGGE to study archaea, eukaryotes, and viruses
The examples described above show that PCR-DGGE is increasingly being used in ecological
contexts to get a better understanding of the factors regulating bacterial community
composition. Bacterial community composition, however, is not only influenced by
physicochemical factors, but may be effected by biotic factors such as primary production,
grazing (Jürgens et al., 1999; Pernthaler et al., 1997; Schäfer et al., 2000; Suzuki, 1999; van
Hannen et al., 1999a) and viral infection (Fuhrman 1999; van Hannen et al., 1999b).
Therefore, it is of great interest to extend the molecular approach to study the genetic
diversity of primary producers, grazers and viruses. Furthermore, Archaea might also play an
important role in the marine system (DeLong 1992). Other than analysing bacterial
communities by PCR-DGGE, specific PCR-DGGE assays allow to study eukaryotic
microorganisms (van Hannen et al., 1998) and Archaea. Furthermore, PCR-DGGE
applications for viral communities have been introduced (Scanlan and Wilson, 1999, Short
and Suttle, 1999). These assays will be valuable for a more integrated study of the microbial
Reports of unusual crenarchaeal 16S rRNA gene sequences retrieved from marine
waters (DeLong, 1992, Fuhrman et al., 1992) have triggered further studies into the
importance of Archaea in the marine environment. It has been shown by oligonucleotide
hybridisation that archaeal rRNA may amount to a high percentage of total rRNA extracted
from concentrated marine picoplankton (Murray et al., 1998), suggesting that Archaea have
been largely ignored as potentially important members of marine microbial communities.
Different PCR-DGGE assays have been used for the analysis of archaeal 16S rRNA gene
fragments (Casamayor et al., 2000; Øvreås et al., 1997; Rölleke et al., 1998; Vetriani et al.,
1999). PCR-DGGE analyses of microbial communities from meromictic lakes by Øvreås and
colleagues (1997) and by Casamayor et al. (2000) have also reported crenarchaeal sequence
types related to those reported from coastal surface waters (DeLong, 1992), marine
archaeoplankton communities however, have not yet been analysed by PCR-DGGE.
DGGE-fingerprinting is not limited to the use of 16S rRNA gene fragments, but can
also be performed with functional genes (e.g. Fesefeldt and Gliesche, 1997; Wawer and
Muyzer, 1995). Using primers targeting the gene encoding [NiFe] hydrogenase of
Desulfovibrio species Wawer et al. (1997) were able to analyse expression of these genes in
complex microbial communities by DGGE.
Eukaryotic microbial communities
Although small eukaryotes such as protozoa can be identified much easier by microscopy
than bacteria because of their discriminative morphological features, their identification is
time-consuming and can often be done by experts only. Identification of eukaryotic microbes
by molecular methods can be achieved with primers developed by van Hannen et al. (1998),
which amplify a 210 bp 18S rRNA gene fragment that can be separated by DGGE. Due to the
limited size of the fragment, sequencing of gel bands may make identification possible at the
phylum level only (van Hannen et al., 1998). However, the authors demonstrate, that using
species specific oligonucleotide probes for hybridisation analysis of DGGE gels,
identification at the species level is possible. Van Hannen et al. (1998) used the eukaryote-
specific PCR-DGGE assay to compare the diversity of five Dutch lakes of a lagoon system.
Analysis of DGGE-fingerprints and environmental variables of these lakes by UPGMA
resulted in similar clustering of lakes and the respective genetic fingerprints of their
DGGE in marine microbial ecology
Viruses are a numerically important part of the microbial food web. Virus-induced mortality
may contribute significantly to overall mortality of natural microbial populations (Suttle,
1994) and has therefore the potential to affect the diversity of bacterioplankton and primary
producers (van Hannen et al., 1999b). Unfortunately, viruses do not contain ribosomal RNA,
hence, for the study of natural virus communities other molecular markers are needed.
Essential virus genes present in a large number of certain virus groups are candidates for
molecular assays, and applications of PCR-DGGE assays to study diversity in virus
communities have been described by Short and Suttle (1999) and Scanlan and Wilson (1999).
The assay by Short and Suttle is based on primers that specifically amplify gene
fragments of the DNA polymerase genes (pol) of viruses infecting microalgae
(Phycodnaviridae) (Chen and Suttle, 1995; Short and Suttle, 1999). Short and Suttle
suggested that similarly, the development of primers specific for DNA polymerases of
cyanophages and bacteriophages should be possible. Separation of virus PCR-amplified pol-
gene fragments derived from cultures was accomplished by DGGE. A preliminary analysis of
natural marine Phycodnaviridae communities demonstrated that they may undergo seasonal
changes, and that community composition may vary over relatively small spatial scales (Short
and Suttle, 1999).
Scanlan and Wilson (1999) have applied a cyanophage-specific PCR-DGGE assay
based on primers described by Fuller et al. (1998) which target genes encoding capsid-
proteins. DGGE was used to separate fragments of PCR-amplified virus-capsid protein genes
obtained from a variety of virus strains. Application of cyanophage DGGE fingerprinting
holds promise for gaining more insight into the influence of cyanophages on the diversity of
cyanobacterial populations. Furthermore, they might facilitate elucidation of environmental
stimuli, e.g. phosphate-limitation (P-limitation) that might decide whether a lytic,
pseudolysogenic or lysogenic infection is established by cyanophages. Wilson et al. (1998)
induced P-limitation in a mesocosm experiment by addition of excess nitrogen (N) at an N:P
ratio of 75:1. Nutrient addition lead to a large Synechococcus bloom, which was shown to
become P-limited by using a immunological marker for P-limitation in Synechococcus. Virus
concentrations increased at the same time, just before the bloom collapsed after re-addition of
P. The authors suggested that the P status of the Synechococcus population had important
implications for the result of the host/phage interaction. They pointed out, however, that there
was not sufficient data and too much variation in virus abundance estimates to establish
whether or not the collapse of the Synechococcus bloom was due to lysogenic or
pseudolysogenic host/phage interaction.
In the above section we have shown that PCR-DGGE is a useful tool for the analysis of
complex microbial communities, and is contributing to change our perception of how the
diversity of microbial communities is controlled. The examples discussed above emphasise
that DGGE is not a stand-alone technique, but should rather be used in combination with
other measurements related to the physicochemical and biotic factors that regulate the activity
and diversity of microbial communities.
The different steps of the PCR-DGGE approach are described below. The method can
also be used to study microbial communities from other ecosystems, yet some modifications
in sample preparation and processing may be necessary.
Practical aspects of PCR-DGGE
Apart from standard laboratory facilities, such as a refrigerator, freezer (-20°C and –80°C),
and fume hood, the basic equipment needed for the analysis of bacterial communities by
PCR-DGGE consists of:
? bench-top centrifuge with refrigeration for 15 ml tubes
? bench-top centrifuge for 0.5 and 1.5 ml tubes
? water bath
? power supplies for electrophoresis systems
? agarose gel electrophoresis system
? denaturing gradient gel electrophoresis system, e.g., DCode system (Bio-Rad 170-9080;
includes gradient former) or PhorU system (Ingeny)
? UV-transilluminator and Polaroid camera or a fluorescence imaging system, e.g., Fluor-S
Multiimager (Bio-Rad 170-7701, Macintosh version)
? peristaltic pump (Model EP-1, Bio-Rad 731-8142) and gradient former (Model 385, Bio-
Rad 165-2000) for casting gradient gels
? automatic DNA sequencer, e.g. ABI 310 Genetic Analyzer (Perkin Elmer)
Sampling of bacteria
About 109 bacterial cells are collected from water samples by gentle filtration on hydrophilic
Durapore filters (Millipore, GVWP04700, polyvinylidene fluoride membrane, 0.22 µm, 47
mm diameter). After filtration, the filters are transferred into cryovials and immediately
frozen in liquid nitrogen or otherwise. Upon return to the laboratory the filters can be stored at
Extraction of nucleic acids
Several protocols have been published in the literature for extraction of nucleic acids from
marine microorganisms. We routinely use the protocol described here; it represents a
combined extraction of DNA and RNA from bacteria collected on Millipore GVWP filters.
The protocol is also suited for extraction of DNA and RNA from Gram-positive bacteria
(Rossello-Mora et al., 1999), and Archaea (Casamayor et al., 2000).
Reagents and disposables
? safety glasses, lab coat and gloves
? 15 ml sterile disposable centrifuge tubes
? 1.5 ml sterile disposable microcentrifuge tubes
? tube racks
DGGE in marine microbial ecology
? AE-buffer (20 mM sodium acetate, 1 mM EDTA, pH 5.5)
? phenol:chloroform:isoamylalcohol (25:24:1), pH 7 (PCI)
? 10% (w/v) sodium dodecyl sulphate (SDS)
? 3 M sodium acetate, pH 5.2
? 100% (v/v) ethanol
? 70% (v/v) ethanol
? water (Sigma W4502) or TE-buffer (10 mM Tris, 1 mM EDTA, pH 8.0)
1. Pre-warm the PCI to 60°C in a water bath. Pre-cool the AE-buffer on ice. Set the
temperature of the centrifuge at 4°C.
2. Transfer the filter from the cryovial to a 15 ml tube with clean forceps. Keep the tube on
3. Rinse the filter with 2 ml of ice-cold AE-buffer. Vortex briefly, and put the tube back on
4. Add 5 ml of hot PCI and 150 µl of SDS. Incubate for 5 minutes at 60°C. Vortex briefly
5. Cool the tube on ice.
6. Centrifuge at 4,000g for 5 minutes at 4°C to separate the aqueous and organic phase.
7. Transfer the aqueous phase to a clean 15 ml tube, and add 1/10 of a volume (ca. 200 µl) of
8. Add 5 ml of PCI. Vortex briefly, and separate the two phases as described in step 6.
9. If necessary, repeat steps 7 and 8, until no proteineous material is visible at the interface
between the aqueous and organic phase.
10. Transfer the aqueous phase to a clean 15 ml tube. Add 2.5 volumes of ice-cold 100%
ethanol, and incubate for at least 3 hours at -20°C.
11. Centrifuge at 4,000g for 60 minutes at 4°C to pellet the precipitated nucleic acids.
Remove the supernatant by gentle aspiration (use a fresh sterile pipette tip for each
12. Rinse the DNA pellet with 1 ml of ice-cold 70% ethanol.
13. Centrifuge at 4,000g for 5 minutes at 4°C. Remove the supernatant.
14. Dry the pellet under vacuum.
15. Add 100 µl water (or TE-buffer) and incubate overnight at 4°C.
16. Redissolve the pellet by gentle pipetting. Aliquot the nucleic acid solution into sterile
reaction tubes and store at –80°C. Inspect 5 µl of the solution by electrophoresis in a 1%
(w/v) agarose gel together with an appropriate molecular weight standard, e.g., lambda
Hind III-digest (Stratagene 201109).
? All steps involving the handling of phenol or phenol-containing solutions should be
performed in a fume hood, wearing safety glasses, gloves, and a lab coat.
Purification of RNA
Reagents and disposables
? 10x DNase buffer (400 mM Tris, 60 mM MgCl2, pH 7.5)
? DNase I, RNase-free (10 u/µl, Pharmacia 27-0514-01)
? 3M sodium acetate pH 5.2
? phenol:chloroform:isoamylalcohol (25:24:1), pH 7 (PCI)
? 1.5 ml sterile disposable microcentrifuge tubes
? tube racks
? safety glasses, lab coat and gloves
1. Add the following reagents to a 1.5 ml tube:
7 µl water
2 µl 10x DNase buffer
15 µl nucleic acid extract
2. Add 1 µl DNase, and incubate for 30 minutes at 37°C in a water bath or thermocycler.
3. Add 280 µl water and 30 µl sodium acetate. Vortex briefly. Remove the DNase enzyme
by extracting with 300 µl of PCI.
4. Centrifuge at 4,000g for 15 minutes at 4°C to separate the aqueous and organic phase.
5. Transfer the aqueous phase to a clean 1.5 ml tube
6. Add 2.5 volumes 100% ethanol and incubate 2 hours at -20°C.
7. Centrifuge at 4,000g for 1 hour at 4°C.
8. Remove the supernatant by gentle aspiration. Rinse the RNA pellet with ice-cold 70%
9. Dry the pellet under vacuum and redissolve in 15 µl water.
10. Use the solution directly for first strand synthesis and store the remainder at –80°C.
? All steps involving the handling of phenol or phenol-containing solutions should be
performed in a fume hood, wearing safety glasses, gloves and a lab coat.
? To avoid contamination with RNase enzymes all solutions should be prepared with RNase
free water and chemicals should be molecular biology reagent grade (“RNase – none
detected”). All steps should be done wearing gloves, additionally all lab bench surfaces
and pipettes should be wiped with 70% ethanol.
? Using the above pipetting scheme the maximal amount of nucleic acids solution that can
be used for preparation of RNA is 17 µl. In our experience 10-15 µl of nucleic acid extract
usually is a sufficient amount to perform RT-PCR analyses. In case of nucleic acid
extracts that are low in RNA content the above scheme must be scaled up (e.g. to 100 µl
volumes or more).
DGGE in marine microbial ecology
Preparation of 1st strand cDNA
Reagents and disposables
? sterile PCR reaction tubes (0.2 or 0.5 ml)
? random hexanucleotides (1:50 [v/v] dilution of Boehringer 1277081 in water (Sigma
W4502) ca. 40 ng/µl)
? dNTP solution (2.5 mM each dNTP; prepared from ultrapure dNTP-set [100 mM each
dNTP; Pharmacia 27-2035-01] and PCR water [Sigma W4502]
? 5x RT buffer (250 mM Tris-HCl, pH 8.3 at 25°C, 375 mM KCl, 15 mM MgCl2, 50 mM
? MMLV-reverse-transcriptase (200 u/µl; Promega M1701)
1. Add 1 µl hexanucleotides (ca. 40 ng/µl) to 10 µl of RNA preparation in a microcentrifuge
tube. Incubate for 10 minutes at 70°C in a water bath or thermocycler to denature the
RNA. Cool on ice.
2. Centrifuge briefly to collect the liquid in the bottom of the tube and add 4 µl of 5x RT-
buffer and 4 µl of dNTP-solution. Incubate for 2 minutes at 37°C.
3. Add 1 µl (200 u) of MMLV reverse transcriptase and incubate for 1 hour at 37°C
4. Incubate the tube at 95°C for 5 minutes. Cool on ice.
5. Use 1 to 5 µl as of the solution as template in the PCR (or make dilutions if too much
PCR product is obtained). Store the remainder at –20°C.
? We do not determine the RNA concentration, but by following this protocol 10 µl of RNA
preparation is usually sufficient for obtaining PCR products.
? It is important to save some of the RNA preparation for doing PCR controls to check the
completeness of the DNA digestion.
? According to the above scheme, 1 µl of cDNA preparation corresponds to an initial input
of 0.5 µl nucleic acid extract containing DNA and RNA. This will be important to
consider for PCR controls.
PCR and RT-PCR for DGGE
Reagents and disposables
? sterile PCR reaction tubes
? 10x PCR reaction buffer (100 mM Tris-HCl, pH 9.0, 15 mM MgCl2, 500 mM KCl)
? Taq DNA-polymerase (5 u/µl; Pharmacia 27-0799-02)
? dNTP solution (2.5 mM each dNTP; prepared as above)
? primers (50 µM) (see Table 1)
? water (Sigma W4502)
Make a 10-fold serial dilution of the extracted DNA and test several dilutions in the PCR to
find the best concentration of template DNA that gives a good specific product. Note that at
very high dilutions, some less abundant templates in the mixture may be lost and hence not be
amplified sufficiently to form a band in the DGGE analysis.
1. Prepare a master-mix for the PCR reactions by adding for each reaction the following
10x PCR reaction buffer 10 µl
dNTPs (2.5 mM each) 10 µl
forward primer (50 µM) 1 µl
reverse primer (50 µM)1 µl
water 76.8 µl
Taq DNA polymerase (5 u/µl)0.2 µl
2. Vortex, and spin briefly to collect the reagents in the bottom of the tube. Dispense 99 µl to
each of the PCR-reaction tubes.
3. Add 1 µl of DNA (or cDNA) solution to each tube. Close the lid, and mark the tubes.
Note which template and dilution was added to which tube.
4. Spin briefly to collect all the fluid in the bottom of the tube. If your thermocycler is not
equipped with a heated bonnet, overlay the reactions with a drop of mineral oil (e.g.
5. Insert the tubes into the thermocycler and start the appropriate PCR-program (see Table
6. When the run has completed store the reactions at 4°C or at –20°C until further use
7. Inspect 5 µl of the PCR products by electrophoresis on 1.5% (w/v) agarose gels together
with appropriate size and mass standards.
? Always perform the following PCR controls: (i) without addition of DNA template
(negative control), and (ii) with addition of known DNA (positive control).
? We prefer to use single PCR-tubes with attached lids instead of strips of tubes with strips
of caps. To avoid the risk of cross-contamination of samples during pipetting of template
DNAs, the tubes are opened one at a time to add template DNA and immediately closed
before proceeding to the next tube.
? For RT-PCR, an extra control reaction is necessary to check for the completeness of DNA
digestion. A volume of RNA preparation corresponding to the amount of cDNA used as
template in the PCR is added to a separate PCR reaction. If this control gives a product,
then the DNA digestion was not complete, and traces of genomic DNA were present.
DGGE in marine microbial ecology
? Addition of bovine serum albumin (BSA; Sigma B6917) to the PCR reactions may help to
overcome the effects of inhibitory substances present in the nucleic acid preparations,
such as humic acids. Use a final concentration of up to 3 mg/ml in PCR reactions.
Quantification of PCR products
An important point to consider in comparative DGGE analysis of multiple samples is that
similar amounts of PCR products should be loaded onto the denaturing gels. Faint bands
visible in one lane, might not be detectable in another lane. If the total amount of PCR
product applied in these two lanes differ markedly, a fair comparison between samples is not
possible. We use a mass standard (Precision Molecular Mass Standard; Bio-Rad 170-8207)
for quantification of PCR products by agarose gel electrophoresis. At least three lanes are
loaded with undiluted, 2-fold diluted and 4-fold diluted mass standard. Using the software
MULTIANALYST (Bio-Rad) the pixel density values of marker bands with varying DNA-
amounts (100, 70, 50, 35, 25, 20, 17.5, 12.5, 10 and 5 ng) can be used for regression analysis.
Pixel densities are also read of unknown PCR-product bands, as well as of the background
staining which is determined at least at one representative point of the gel. The regression
curve and formula derived from the marker fragments is then used to estimate the DNA-
concentration of the unknown samples.
? Samples that are out of the range of the standard curve should be avoided. Instead, these
samples should be analysed again with more or less sample applied
? Avoid oversaturation of portions of the gel-pictures (by overexposure)
? Thoroughly rinse the ethidium bromide stained gels in Milli-Q water to increase the
Troubleshooting – PCR
No PCR product If no product was obtained for positive control either, probably due to
accidental omission of a vital ingredient, try again.
If positive control worked well, this may be related to presence of substances
inhibiting Taq polymerase in nucleic acid extracts
Use more template, or pool replicate reactions and concentrate by
Use less template
Poor product yield
Too much product or
Product in negative
Contamination of solutions and/or plastic ware with DNA. UV-resistant
plastics can be decontaminated by exposure to UV-light source (e.g. cross
linker, clean bench). Use fresh aliquots of reagents, if problems persist.
Prepare new stock solutions with nucleic acid-free water
DNA digestion was not complete. Repeat DNA digestion with more DNase
and/or longer incubation time and/or less nucleic acid extract
Substances, such as humic acids are co-extracted with the nucleic acids and
inhibit Taq-polymerase To overcome this problem (i) further purify the
nucleic acid extracts or (ii) dilute nucleic acids which will also dilute
inhibitory compounds and/or (iii) add BSA to PCR reactions (see notes above)
Product from RNA
Inhibition of Taq
Table 1. Target sites, sequences and specificity of primers targeting small subunit ribosomal RNA used for DGGE analysis
Sequence (5' to 3') Specificityc
341-357 CCT ACG GGA GGC AGC AGBacteria Muyzer et al., 1993
907-926 CCG TCA ATT CMT TTG AGT TTBacteria Muyzer et al., 1998
518-534 ATT ACC GCG GCT GCT GGuniversal Muyzer et al., 1993
1055F 1055-1070ATG GCT GTC GTC AGC TBacteria Ferris et al., 1996
1392R-GC 1392-1406 ACG GGC GGT GTG TACuniversal Ferris et al., 1996
968F-GC 968-984AAC GCG AAG AAC CTT AC
Nübel et al., 1996
1330R 1330-1346 TAG CGA TTC CGA CTT CA BacteriaNübel et al., 1996
1385R 1385-1401CGG TGT GTA CAA GAC CCBacteria Nübel et al., 1996
PRBA338F-GC338-357 ACT CCT ACG GGA GGC AGC AGBacteria Øvreås et al., 1997
PARCH340FGC 340-357 CCC TAC GGG CYG CAS CAGArchaea Øvreås et al., 1997
PARCH519R519-533 TTA CCG CGG CKG CTG ArchaeaØvreås et al., 1997
ARC344 344-363 ACG GGG AGC AGC AGG CGC GA
CCC CGT CAA TTC ATT TGA GTT TTg
Archaea Rölleke et al., 1998
universal Rölleke et al., 1998
344-363 ACG GGG CGC AGC AGG CGC GAArchaea Vetriani et al., 1999
ARC344F-GC 344-363 ACG GGG YGC AGC AGG CGC GAArchaeaCasamayor et al., 2000
915R 915-934 GTG CTC CCC CGC CAA TTC CTArchaea Casamayor et al., 2000
359-378GGG GAA TYT TCC GCA ATG GGCyanobacteria Nübel et al., 1997
781-805 GAC TAC WGG GGT ATC TAA TCC CWT TCyanobacteriaNübel et al., 1997
PRO1017R1017-1035 TCC CGA AGG CAC CCT CWA Amost
West and Scanlan, 1999
190-208GGA GRA AAG YAG GGG ATC G
Kowalchuk et al., 1997
CTO654R 633-654CTA GCY TTG TAG TTT CAA ACG C
Kowalchuk et al., 1997
Eukaryotic forward1427-1453TCT GTG ATG CCC TTA GAT GTT CTG GG van Hannen et al., 1998
Eukaryotic reverse1616-1637 GCG GTG TGT ACA AAG GGC AGG G Eukaryavan Hannen et al., 1998
a F (forward) and R (reverse) indicate the orientation of the primers in relation to the rRNA sequence. Primers with identical designation but
different sequence are reported individually.
b E. coli numbering according to Brosius et al. (1981) for primer sequences targeting 16S rRNA of Prokaryotes, and according to the numbering
of Saccharomyces cerevisiae 18S rRNA for eukaryote specific primers (van Hannen et al., 1998)
c The specificity denoted is quoted from the original citation and may in some cases be applicable to the PCR assay (Table 2) with the appropriate
second primer and cycling conditions. Furthermore, due to the increasing amount of sequence information in public databases, the stated
specificities of primer sequences may not hold true upon re-assessment against the most recent database. It is therefore strongly recommended
to re-assess the applicability of a certain primer sequence for specific applications.
d GC denotes that a GC-rich sequence (GC-clamp) is attached to the 5'-end of a PCR primer. For primers 341F-GC, ARC344F-GC used by
Casamayor et al. (2000) and CYA359F-GC it comprises 40 nucleotides (sequence: 5'-CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG
CCC CCG CCC G-3'). See other references for particular GC-clamp sequence used. For a compilation of different GC-clamps and calculated
free energies of hairpin and primer-dimer formation see Muyzer et al. (1998).
e Component primers of 907R, i.e. 907RC (5'- CCG TCA ATT CCT TTG AGT TT-3') and 907RA (5'-CCG TCA ATT CAT TTG AGT TT-3')
can be used individually [Schäfer, 2001). Component primer 907RC may exhibit a one nucleotide mismatch with certain bacterial 16S rRNA
genes (e.g. members of γ-proteobacterial lineages Acinetobacter, Pseudomonas, Beggiatoa, Enterobacteriacaea, Shewanella,
Pseudoalteromonas, Alteromonas, and many ε-proteobacterial 16S rRNA genes, as well as in members of Gemmata, Verrucomicrobiales,
Chlamydia. This mismatch is avoided in most cases with primer 907RA.
f Primer 518R can be used in combination with 341F-GC for amplification of bacterial 16S rRNA genes (Muyzer et al., 1993). Furthermore,
Vetriani et al. (1999) used it in combination with primer 344F-GC for amplification of archaeal 16S rRNA gene fragments.
g The primers has up to 4 mismatches with most Archaea.
h The reverse primer CYA781R used in PCR is an equimolar mixture of CYA781R(a) (5’-GAC TAC TGG GGT ATC TAA TCC CAT T-3’) and
CYA781R(b) (5’-GAC TAC AGG GGT ATC TAA TCC CTT T-3’) (Nübel et al., 1997)
i See Kowalchuk et al. (1997) for details of primer synthesis. Note that in original paper component primers CTO189A-GC and CTO189B-GC
had a typing error in the GC-clamp sequence (first 3 bases missing). The GC-clamp sequence should be the same as for component primer
CTO189C-GC (Kowalchuk, personal communication). The same primers have been used by McCaig et al. (1999) with designations
CTO178fGC and CTO637r.
Table 2. PCR-cycling conditions for various PCR-DGGE assays
Primer combinationSpecificityPCR programa
5 min at 94°C, followed by 20 cycles of 1 min at 94°C, 1 min at
65°..55°C (touchdown –0.5°C cycle-1), and 3 min at 72°C, followed by 15
cycles of 1 min at 94°C, 1 min at 55°C and 3 min at 72°C, followed by 7
min final extension at 72°
Muyzer et al.,
338F-GC/518RBacteria 2 min 92°C, followed by 30 cycles of 1 min at 92°C, 30 s at 55°C, 1 min
at 72°C, followed by 6 min final extension at 72°C
Øvreås et al.,
1055F/1392R-GC Bacteria5 min 94°C, followed by 11 cycles 1 min at 94°C, 1 min at 53..43
(touchdown –1°C cycle-1), and 3 min at 72°C, followed by 20 cycles of
1 min at 94°C, 1 min at 43°C, and 3 min at 72°C, final extension phase of
Ferris et al., 1996
Bacteria 5 min at 94°C, followed by 35 cycles of 1 min at 94°C, 1 min at 63°C,
and 1 min at 72°C, followed by 5 min final extension at 72°C
Nübel et al., 1996
5 min at 94°C, followed by 35 cycles of 1 min at 94°, 1 min at 60°C, and
1 min at 72°
Nübel et al., 1997
5 min at 95°C, followed by 30 cycles of 1 min at 95°C, 1 min at 55°C,
and 1 min at 72°C
1 min at 93°C, followed by 35 cycles of 30 s at 92°C, 60 s at 57°C, and 45
s at 68°C (+1 s cycle-1), followed by 5 min final extension at 68°C
Kowalchuk et al.,
Archaea2 min at 92°C, followed by 30 cycles of 1 min at 92°C, 30 s at 53.5°C, 1
min at 72°C, followed by 6 min final extension at 72°C
Øvreås et al.,
ARC344/907R-GC Archaea 5 min at 94°C, followed by 35 cycles of 1 min at 94°C, 1 min at 65°C, 3
min at 72°C, then 1 µl of the PCR reaction is transferred to fresh reaction
mixture and another 30 cycles are donec.
Rölleke et al.,
ARC344F-GC/915R Archaea 5 min at 94°C, followed by 20 cycles of 1 min at 94°C, 1 min 71-61°C
(touchdown: -0.5°C cycle-1), and 3 min at 72°C, followed by 15 cycles of
1 min at 94°C, 1 min at 61°C, and 3 min at 72°C, followed by 10 min
final extension at 72°C
Casamayor et al.,
Archaea 40 cycles of 30 s at 94°, 30 s at 48°C, and 30 s at 72°C Vetriani et al.,
EukaryoticF/EukaryoticR Eukarya 5 min at 94°C, followed by 25 cycles of 0.5 min at 94°C, 1 min at 52°C,
and 1.5 min at 68°C, followed by 10 min final extension at 68°C
van Hannen et al.,
a potential hot start phases at 80°C have been omitted; touchdown denotes that annealing temperature is decreased in consecutive cycles
b second PCR of a nested PCR approach, see details in original study
c the two-step PCR was performed since only very limited amounts of original sample (Medieval wall paintings) were at disposition.
Sometimes PCR products of the first round were gel-purified first, to avoid carry-over of by-products.
Casting and running of denaturing gradient gels
To achieve the maximum resolution in DGGE patterns of unknown samples it is
recommended to find out the best gradient conditions. This requires running perpendicular
denaturing gels with the unknown sample to define the range of denaturant concentrations that
allow the best separation possible. In our experience gradients ranging from ca. 10-20% to
70-80% denaturant concentration (urea and formamide; UF) result in a good separation of
fragments obtained by PCR with primers 341f-GC and 907R and provide a security margin
for fragments melting at unexpectedly high denaturant concentrations at the same time. It is
strongly recommended to run time-travel experiments when starting DGGE analysis to check
for optimal separation. For a description of casting and running perpendicular denaturing
gradient gels and time travel experiments the reader is referred to Muyzer et al.(1996), or to
the manual coming with the DGGE system.
Preparation of reagents
Add 10 g of mixed bed resin (e.g. Sigma M8032) to 100 ml formamide in an Erlenmeyer and
stir for 30-60 minutes. Decant or filter (e.g., Schleicher & Schuell folded filter 595 1/2, order
no. 311647) the formamide to separate it from the resin beads. Store the de-ionised
formamide in volumes of 35 ml at –20°C for the preparation of the 80% denaturing gel
Acrylamide/bis-acrylamide stock solution (37.5:1; 40% w/v)
Acrylamide is a powerful neurotoxin and should be handled with extreme care. To avoid
exposure to acrylamide dust, we recommend to buy ready-made acrylamide/bis-acrylamide
stock solution (e.g., Bio-Rad 161-0149). If you prepare the solution from acrylamide powder,
wear safety glasses, gloves, a lab coat, and a dust mask.
Filter the solution (e.g., through a e.g., Schleicher & Schuell folded filter 595 1/2) and store at
4°C in a dark bottle.
to 100 ml
DGGE acrylamide/bis-acrylamide solutions
Prepare 6% (w/v) acrylamide/bis-acrylamide gradient solutions according to the amounts of
reagents shown below. We use 6% acrylamide/bis-acrylamide solutions for PCR products
obtained with primers 341F-GC / 907R as well as for CYA359F-GC / CYA781R (see Table
1). Higher concentrations of acrylamide/bis-acrylamide may be necessary for DGGE analysis
of other 16S rRNA gene fragments (check original citations for details). The use of an 80%
denaturing gel solution as high denaturing solution is usually sufficient for preparation of
denaturing gradient gels. However, care has to be taken that bands are not lost from the
analysis due to migration to higher denaturant concentrations than 80%. In this case a 100%
denaturant acrylamide solution should be used.
DGGE in marine microbial ecology
15 ml acrylamide/bis-acrylamide
50x TAE (pH 8.3)
formamide (deionised) (F)
to 100 ml to 100 ml
Filter through 0.45 µm filter or a Schleicher & Schuell folded filter 595 1/2 (Schleicher &
Schuell 311647). Degas the acrylamide/bis-acrylamide solution for 15 minutes under vacuum,
and store at 4°C in a dark bottle.
10% Ammonium persulphate solution
Aliquot into single use portions and store at –20°C.
to 10 ml
TEMED is bought as a ready-to-use solution (e.g., from Fluka or Bio-Rad)
50x TAE buffer (2M Tris, 2M Acetic acid, 50 mM EDTA; pH 8.3)
0.5 M EDTA, pH 8.0
Acetic acid (glacial)
to 1000 ml
Autoclave the buffer solution for 20 minutes and store at room temperature.
1X TAE running buffer
Dilute 1 volume of 50x TAE-buffer with 49 volumes of Milli-Q water.
To visually inspect proper gradient formation after casting, a dye solution can be added to the
high denaturant solution.
Bromophenolblue (0.5% w/v final)
Xylenecyanole (0.5% w/v final)
1X TAE buffer
10x gel loading solution
Glycerol (100% v/v)
Bromophenolblue (0.25% w/v final)
Xylenecyanole (0.25% w/v final)
Mix and store in small aliquots at room temperature.
Assembly and casting of parallel denaturing gradient gels
1. Clean the glass plates and spacers with water and soap. Rinse them with de-mineralised
2. Wipe the glass plates first with 70% ethanol and then with acetone. Use a dust-free cloth
(e.g. Kimwipes). Do not wipe any plastics (e.g., spacers, combs, etc) with acetone.
3. Wipe the spacers (1 mm thickness) with ethanol and let them dry, then sparingly smear
grease (High vacuum grease; Dow Corning, Auburn, MI, USA) along one of the long
edges, such that around 2 mm are covered with a thin grease-film on each face of the
4. Place the large glass plate on a clean surface, and put the spacers onto the left and right
margins, such that the greased edges face the outside edge of the glass plate.
5. Put the small glass plate on top of the spacers, to form a ‘sandwich’.
6. Align the spacers and the glass plates in such a way that they are flush at the bottom of the
sandwich (this can be done on an even surface, or in the aligning slot of the casting stand).
7. Attach the clamps to the sandwich, tighten the clamp screws (finger-tight) and put the
sandwich in the casting stand, fix in casting slot by turning the levers.
8. Before proceeding to the next step make sure that the device used for casting the gradient
gel is ready installed and you are familiar with the procedure described below. The work
has to proceed quickly otherwise you run the risk that gel solutions will polymerise before
casting is finished. A gradient former comes with the DCode system from Bio-Rad. We
use a combination of peristaltic pump (Model EP-1, Bio-Rad 731-8142) and gradient
former (Model 385, Bio-Rad 165-2000) to cast gels. For detailed instruction on set-up and
operation of these refer to the technical instructions of Bio-Rad. Connect the tubing of the
pump with the outflow chamber of the gradient chamber. Attach an injection needle to the
Luer-lock of the outlet tubing of the pump and insert the needle between the glass plates
in the middle of the gel sandwich.
9. Prepare the high and low denaturant solutions for the gradient as required in disposable
plastic tubes. Using 1 mm thick spacers, 12 ml each are recommended. The gradient gel
will finally be overlaid with a 0% denaturant acrylamide solution (prepare 5 ml), as
otherwise the presence of the denaturants hinders the formation of good sample wells.
10. Add ammonium persulphate and TEMED to the gradient solutions. We add 60 µl of
ammonium persulphate and 8 µl of TEMED to each solution, directly pipett these into the
solutions. Close tubes and mix thoroughly by inverting several times.
11. To inspect the gradient, add 120 µl of gradient-dye-solution to the high denaturant
solution. Close the tube and mix by inverting several times.
DGGE in marine microbial ecology
12. Close the connection pipe between the two chambers of the gradient chamber, make sure
pump is not running. Pour the high denaturant gel solution into the outflow chamber of the
13. To remove air bubbles in the connection pipe, slowly open the pipe by turning the lever
aside until the air has been expelled from the pipe and a drop of high denaturant gel
solution is visible on the bottom of the second chamber. Then close connection pipe and
pipette back all high denaturant gel solution back to the outflow chamber using a clean
14. Carefully add the low denaturant gel solution into the second chamber.
15. Turn on the magnetic stirrer at 250 rpm, then turn on the pump and slowly open the
connection pipe, such that no extra high denaturant gel solution enters the second
chamber. Cast the gel with ca. 4 ml/min. The last ca. 1 ml of the gradient gel will not be
mixed properly (due to remains of high denaturant gel solution in the connecting pipe of
the gradient chamber), hence avoid delivery of that last bit of gradient solution, as it will
disturb the top of the gradient gel.
16. Remove needle from gel sandwich. Rinse gradient chamber and pump tubing with water
to remove residual gel solution.
17. Clean a comb (1 mm thick) with ethanol and let it dry.
18. Add 25 µl of ammonium persulphate and 5 µl of TEMED to 5 ml of 0% denaturant gel
solution. Close the tube and mix.
19. Carefully overlay the gradient gel with about half of the 0% denaturant gel solution using
a 1000 µl pipette.
20. Insert the comb at an angle to avoid the formation of air-bubbles. Completely fill the gel
sandwich with the remainder of the 0% denaturant gel solution.
21. Let the solution polymerise for at least 2 hours.
? There are two different kinds of spacers, those for casting perpendicular gels, which have
grooves on the inside side of the gel sandwich and normal spacers without grooves. For
normal parallel DGGE analysis we recommend to use spacers without grooves, as they are
easier to grease and better safeguard against current leakage, which may cause
considerable smiling of the gel.
? To facilitate mixing of the gradient solutions, the gradient chamber should be placed on a
magnetic stirrer and small magnetic stirring bars should be added to each chamber.
Furthermore the gradient chamber should be placed at a higher level than the peristaltic
pump to improve gradient formation.
? To avoid too fast polymerisation of acrylamide solutions these can be kept on ice before
casting. This may be especially important when temperatures in the lab rise to high levels
during summer months.
? Polymerised gels can be stored overnight. To avoid drying out, the comb is removed, the
wells filled with water and the gel covered with cellophane.
Troubleshooting – DGGE gel casting
Acrylamide solution leaves outflow chamber of
gradient former, but gel solution from second
chamber is not flowing into and mixing with solution
in outflow chamber
During casting, bubbles appear in the tubing between
pump and needle
During casting many air bubbles are formed at the
needle that get into the gradient gel between the glass
Acrylamide solution leaks at the bottom of the glass
Gel does not polymerise
Gel polymerises, but remains viscous
Mostly due to air bubbles in the connection pipe
between the two chambers of the gradient
Mostly due to defect tubing, replace tubing
Mostly due to old needle, replace needle
Spacers and glass plates are not flush at the
No or not enough TEMED and / or APS added
Make sure that proper mixture or percentage of
acrylamide and bis-acrylamide was used
Running parallel DGGE gels
After quantification of PCR-products, the samples are mixed with 10x gel loading solution.
The total volume of PCR product to be loaded may vary between 15 µl and 60 µl. Using a
1 mm-thick, 16-well comb of the DCode system it is possible to load volumes up to
approximately 70 µl. Apply the sample very slowly into the sample wells to avoid mixing
with the electrophoresis buffer and to avoid overflow into the neighbouring wells. As bands
tend to focus in DGGE there is no need to apply equal sample volumes. Alternatively, PCR
products of low concentration can be precipitated and be re-dissolved in smaller volumes.
Sometimes more than 20 samples are to be compared on denaturing gradient gels, exceeding
the number of wells formed with the 20-well comb, hence multiple gels are needed.
Denaturing gradient gels, however, show some degree of gel-to-gel variation, caused by
differences in the gradient. Therefore, it is recommended to use a marker standard on the gels
that is composed of fragments halting at a range of denaturant concentrations. Such a marker
facilitates gel-to-gel comparison, the marker we use routinely (see lane M of gel shown in
Figure 1) is composed of five different fragments derived from chloroplast 16S rDNA of a
Nitzschia sp., two cloned 16S rRNA genes obtained from an earlier study (Schäfer et al.,
2000), and two commercially available genomic DNAs of Clostridium perfringens (Sigma
D1760) and Micrococcus lysodeikticus (Sigma D8259). Bands halting at high denaturant
concentrations can be used to normalise the migration length of individual bands which may
vary between gels (Ferrari and Hollibaugh, 1999).
DNA amounts to load
There is no general rule for the amount of PCR product to apply on denaturing gels, since the
optimal amount will depend on the number of different sequence types (i.e. bands) in a given
sample, as well as the relative contribution of the bands to the total PCR-product (i.e. the
relative intensity of particular bands). For instance, loading 500 ng of PCR product in a
DGGE in marine microbial ecology
situation where the fluorescence intensity is equally distributed over 5 different bands will be
different from samples showing 30 to 40 different bands. The absolute DNA amount to be
loaded should therefore be tested empirically. Typically, we use about 500 ng (range 300-
600 ng) PCR product for the analysis of marine bacterioplankton communities obtained by
amplification with primers 341F-GC/907R. In our experience, using around 1 µg often leads
to high background and overloading of individual dominant bands, potentially obscuring
some other, fainter bands. Ferrari and Hollibaugh (1999) reported that about 1 µg was the
optimal amount to use, however they often observed multiple bands for single organism
templates, which may have been an effect of overloading DGGE lanes rather than
representing sequence heterogeneity of multiple rrn operons. For analysis of oxygenic
phototrophic communities Nübel and colleagues (1999) used around 500 ng.
1. Fill the electrophoresis tank with approximately 7 litres of 1x TAE buffer.
2. Insert the core. Two gels can be attached to the core and run at the same time. If only one
gel is run, attach a buffer dam at the other site. The buffer dam can be made of a large and
small glass plate without spacers and held together by the sandwich clamps.
3. Carefully place the lid (i.e. the electrophoresis/temperature control module) on the
electrophoresis tank. Take care that the end of the stirring bar comes in its proper position.
4. Switch on the DCode system with the on/off button on the electrophoresis / temperature
control module. Switch on the buffer recirculation pump and the heating element. Set the
temperature to 60°C and set the ramp rate to 0. The buffer will reach the temperature in
about 1 hour.
5. Prepare the samples by adding between 5 and 10 µl of gel loading solution. Mix the
samples and spin briefly.
6. Remove the comb slowly, when the acrylamide gel is polymerised.
7. When the buffer has reached 60°C, switch off the electrophoresis unit, wait at least 15
seconds before removing the lid, and place the lid on the lid stand.
8. Take out the core, pre-wet the sandwich clamps of the gel sandwich and attach to core.
Replace the core in the electrophoresis tank.
9. Take a 25 ml syringe, pull up the buffer from the electrophoresis tank, attach a needle and
rinse the wells of the denaturing gel to remove traces of non-polymerised acrylamide.
10. Load the samples into the wells with a 50 µl Hamilton syringe. Thoroughly rinse the
syringe with electrophoresis buffer between the different samples.
11. Put the lid on the buffer tank turn on electrophoresis unit and connect the cords to the
12. Run the gel at constant voltage of 10 volts for 10 minutes while the temperature is brought
back to 60°C.
13. If some samples could not be loaded completely due to a too large sample volume, switch
off power unit and electrophoresis unit, and repeat steps 10 and 11.
14. Run the gel at a constant voltage of 100 V for 18 hours. The amperage should be around
15. After 18 hours, turn off the power supply and the electrophoresis unit. Wait at least 15 sec
before removing the lid. Take out the core and detach the gel sandwich.
16. Remove carefully one of the glass plates as well as the spacers. Stain the gel on the glass
plate with ethidium bromide solution for 30 minutes (ethidium bromide 0.5 µg/ml in
17. Rinse the gel for 20 to 30 minutes in distilled water.
18. Transfer the gel to an UV-transilluminator and photograph with a Polaroid camera or
preferably use a gel documentation system equipped with a CCD camera and coupled to a
computer (e.g. Fluor-S Multiimager, Bio-Rad). Take several photos of the gel with
varying exposure times (optimal, underexposed, overexposed). Underexposed
photographs may help to define very intense bands, while overexposed photographs may
help to identify additional faint bands.
? Avoid powdered gloves as they may leave a background on the gel.
? DGGE gels can also be stained with Sybr Green (Muyzer et al., 1998) or Gelstar
(Moeseneder et al., 1999). Specific filters might be necessary to optimise the acquisition
of gel images. Denaturing gels can also be stained with silver. However, this might be
disadvantageous for further re-amplification and sequencing of excised bands.
? Gels can be easily transferred into the Fluor-S Multiimager (Bio-Rad) using a large gels-
coop (Sigma G7152). Avoid scratches in the scoop as this will show in gel images.
? In most cases, DGGE gels are 1 mm thin and therefore difficult to handle. However, gels
can be transferred easily from UV-tables back to glass plates or moved to a blotting device
or another UV-table using Whatman filter paper. Cut a piece to match the size of the gel
and carefully put it on top of the gel, avoiding bubbles. Carefully lift the filter paper, make
sure the gel remains attached, and put down on a glass plate/UV-table/blotting stack. Soak
the filter paper completely with water (or buffer when moving to a blotting stack) and the
filter paper will come off easily.
Analysis of DGGE patterns
DGGE patterns from mixed microbial communities may be very complex. Different kinds of
information can be extracted from DGGE patterns, i.e. the number, position (absence or
presence of particular bands) and relative intensity of bands. Furthermore, the nucleotide
sequence of bands can be determined. Information extracted from DGGE-patterns can be
subject to numerical analysis to determine the extent of variation between DGGE patterns of
different samples and thus help in the interpretation of DGGE analyses. A prerequisite for
comparative analysis of DGGE patterns is that similar amounts of PCR-products were applied
on the gel. Figure 3 schematically shows the steps in numerical analysis of DGGE patterns.
Deciding which features of gels represent bands and which do not is of pivotal
importance. DGGE patterns can be analysed with band-finding algorithms after digitisation of
gel photographs. Ferrari and Hollibaugh (1999), however, noted that visual inspection of gel
patterns provides the most sensitive way. This agrees with our experience, although subjective
assessment can not be ruled out with visual inspection, and analysis may vary between
persons. Fragments of the 16S rDNA from different microorganisms may show varying
degrees of sharpness as DGGE bands, some may focus very well, whereas others remain
somewhat fuzzy. These are probably intrinsic features of the melting behaviour of different
nucleic acid sequences. To remain as objective as possible, all features that look like a band
should be scored as such. The basic assumption in DGGE analysis is that each band in a
DGGE fingerprint corresponds to a unique type of 16S rRNA gene. Yet, there are some
circumstances that prompt to think of this in relative terms (see section, Limitations of PCR-
DGGE in marine microbial ecology
A first step in the analysis of DGGE patterns by statistical methods, such as unweighted pair-
wise grouping with mathematical averages (UPGMA) and multidimensional scaling (MDS) is
to set up a binary matrix that is representative of the bands occurring in a set of DGGE
patterns. The presence or absence of DGGE bands in a sample are scored as present (1) or
absent (0), relative to the DGGE bands detectable in all samples of a set of DGGE patterns.
Unweighted pair-wise grouping with mathematical averages (UPGMA)
UPGMA is a clustering method for binary data whereby pair-wise similarities of DGGE
patterns are used to infer a dendrogram that depicts these distances in graphical form. For
UPGMA analysis of DGGE patterns a binary matrix is translated into a distance matrix
representing the similarities of the DGGE-patterns using a similarity coefficient. Different
similarity coefficients have been used by several authors. The Dice coefficient used for cluster
analysis of data from restriction fragment length polymorphism (RFLP) of 16S rRNA genes
(Heyndrickx et al., 1996) and ribopatterns of bacterial strains (Vachee et al., 1997) is identical
to the Sorensen coefficients used by Murray et al. (1998) for calculation of pair-wise
similarities and the Nei & Li coefficient used by van Hannen et al. (1998) and Lebaron et al.
(1999) for cluster analysis of DGGE patterns. Other authors (Curtis and Craine, 1998; Ferrari
and Hollibaugh, 1999; Liu et al., 1997) have used the Jaccard coefficient (Jaccard, 1908) for
clustering of fingerprint patterns (T-RFLP and DGGE). This coefficient has also been used in
the schematic example depicted in Figure 3. Both, the Jaccard and the Dice coefficients seem
to be appropriate since they do not consider double absence of bands in the calculation of the
pairwise similarity, and thereby avoid spuriously high similarity values in pairwise
similarities of samples (i.e. DGGE patterns of two lanes in a DGGE gel) with high numbers of
The Jaccard-similarity is calculated according to the formula:
SJaccard = NAB / ( NA + NB – NAB)
Where NAB is the number of bands common in both samples (patterns), NA and NB represent
the total number of bands in sample A and B, respectively.
The formula for the Dice coefficient as shown in (Heyndrickx et al., 1996) is:
SDice = 2 NAB / ( NA + NB)
the designations of the terms are the same as given for Jaccard coefficient.
The distance matrix is further analysed by UPGMA (for examples see Lebaron et al., 1999;
van Hannen et al., 1998).
Multidimensional scaling analysis (MDS)
MDS is a powerful data-reduction technique that may aid in the interpretation of large sets of
complex DGGE patterns. Van Hannen et al. (1999a) were the first to use this statistical
method in conjunction with DGGE fingerprinting in their study on the influence of predation
on the genetic diversity of a microbial community. Schäfer and colleagues (2001) analysed by
DGGE the development of Mediterranean bacterioplankton in nutrient-enriched mesocosms.
Here, MDS not only served to show deviations between control and treatment mesocosms,
but also confirmed the reproducibility of duplicate mesocosms. For MDS analyses the
information of the DGGE patterns is again represented as a 0/1 binary matrix, which is used
to derive a distance matrix, using the Dice or Jaccard coefficient (the Jaccard coefficient is for
instance implemented in the statistics software SYSTAT 7.0).
0.310.40 0.600.80 1.00
distance matrix (Jaccard coefficient)
Schematic example of statistical analysis of DGGE patterns. Briefly, the presence (1) and absence (0) of DGGE
bands in different samples are scored in a binary matrix. The binary matrix is translated into a distance matrix
using a similarity coefficient (e.g. Jaccard coefficient) that is used for UPGMA or MDS.
DGGE in marine microbial ecology
MDS reduces a complex DGGE pattern to a point in a two-dimensional space (when
restricted to two dimensions). When, for instance, the development of a microbial community
is studied during time by DGGE, the patterns can be analysed by MDS. Connecting the dots
representing consecutive samples by lines, the development of the banding patterns can be
visualised (for an example see van Hannen et al., 1999a).
Densitometric analysis – relative fluorescence of DGGE bands
DGGE data may also be amenable to quantitative analysis. For this, the relative fluorescence
(staining intensity) of DGGE bands has to be measured. This can usually be achieved using
software such as NIH-image (available at http://rsb.info.nih.gov/nih-image/) by plotting the
pixel density along the DGGE profile. This results in a peak pattern of which individual peaks
and the baseline have to be defined. Subsequently relative fluorescence values can be
obtained for individual bands.
DGGE derived values of genetic richness and abundance (defined as relative fluorescence of
DGGE bands) can be used to calculate diversity indices. In a study of hyper-saline microbial
mat communities Nübel and colleagues compared the diversity of oxygenic phototrophic
microorganisms in mat samples from different sites (Nübel et al., 1999). Using a specific PCR
(Nübel et al., 1997) they amplified 16S rRNA gene fragments of oxygenic phototrophs and
separated them by DGGE. Different samples were compared according to the number of
DGGE bands detectable (i.e. genetic richness), and their relative staining intensity (i.e.
evenness). Using these PCR-DGGE-defined richness and evenness values, a Shannon-Weaver
diversity index could be calculated which was compared to two other cultivation
independently derived diversity estimates. It is important to note that the PCR conditions have
to be adjusted such that the PCR does not reach the plateau-phase. Furthermore, using
bacterial/universal primers with complex communities might not result in valid diversity
estimates due to complex DGGE patterns.
Identification of community members
Apart from facilitating the comparison of larger numbers of samples, DGGE-fingerprinting
also makes possible the identification of predominant community members. Two approaches
have been applied successfully. The first is hybridisation analysis of blotted denaturing
gradient gels with oligonucleotide (e.g., Brinkhoff and Muyzer, 1997, Muyzer et al., 1998) or
polynucleotide probes (Heuer et al., 1999). The second is sequencing of excised denaturing
bands. The latter approach is, however, more straightforward than hybridisation analysis and
also more universal, because only few of the “group-specific” target sites (Snaidr et al., 1997)
lie within the fragment of the 16S rRNA encoding gene used for DGGE analysis of mixed
Excision of bands and re-amplification
After documentation of the denaturing gel, make a printout of the gel and mark all bands that
are to be excised and sequenced. Assign each band a number and label a corresponding
number of 0.5 ml reaction tubes, accordingly.
1. Transfer the gel to a UV-table and set the UV-table to “preparative” (or “low”) instead of
“analytical” (or “high”) mode.
Chapter 2 Download full-text
2. Wipe a scalpel blade with ethanol and switch on UV-table, cut out band of interest and
pick it up with the blade or with forceps.
3. Immediately switch off the UV-source to minimise the damage to the DNA bands in the
4. Transfer the gel piece to the labelled tube.
5. Continue excising bands as described in step 2-4, until all bands have been excised.
6. Rinse the bands by adding 200 µl of nucleic acid free water, spin down contents of tubes
and incubate at room temperature for 1-2 hours.
7. Remove the water by gentle aspiration (use a clean sterile tip for each band).
8. Add 25-50 µl of nucleic acid free water, spin down, and incubate at 4°C overnight.
9. Use water from the supernatant as template for re-amplification with the same primers as
for the PCR for DGGE, store the remainder at –20 °C
10. Check the PCR product from the re-amplification alongside the original DGGE pattern to
make sure it is the proper band and to see if it is a single band (see Figure 4)
DGGE of re-amplified bands that were excised from a
denaturing gel for sequence analysis (lanes 1 and 2,
and lanes 4 to 6). The re-amps were run side by side
with PCR products of the original samples (lanes 3 and
7) to verify that (i) the re-amplified products were
single bands, and (ii) correspond to the excised band in
the original pattern. Sometimes re-amplified products
might consist of more than one band (e.g. lanes 5 and
6). In such cases the band should be excised and re-
amplified again. Alternatively, such PCR products can
be cloned to isolate the band of interest.
? Ethidium bromide is a powerful mutagen. Wear always at least one pair of protective
? Protect yourself against exposure of UV radiation by wearing a UV-filtering face-shield.
Shield your arm wrists by taping the ends of your lab coat sleeves tight around the wrists
? UV-light will also damage the DNA that you want to re-amplify. Therefore, excision
should proceed as quickly as possible and UV-exposure has to be kept as short as
possible. This can be achieved by switching off the UV-source as soon as a band has been
excised and only turning it on when you are ready for excision of the next band