Seasonal and spatial community dynamics in the meromictic Lake Cadagno

Abstract

The seasonal and spatial variations in the community structure of bacterioplankton in the meromictic alpine Lake Cadagno were examined by temporal temperature gradient gel electrophoresis (TTGE) of PCR-amplified 16S rDNA fragments. Two different amplifications were performed, one specific for the domain Bacteria (Escherichia coli positions 8-536) and another specific for the family Chromatiaceae (E. coli positions 8-1005). The latter was followed by semi-nested reamplification with the bacterial primer set, allowing comparison of the two PCR approaches by TTGE. The TTGE patterns of samples from the chemocline and the anoxic monimolimnion were essentially identical, whereas the oxic mixolimnion displayed distinctively different banding patterns. For samples from the chemocline and the monimolimnion, dominant bands in the Bacteria-specific TTGE profiles comigrated with bands obtained by the semi-nested PCR approach specific for Chromatiaceae. This observation suggested that Chromatiaceae are in high abundance in the anoxic water layer. All dominant bands were excised and sequenced. Changes in the community structure, as indicated by changes in the TTGE profiles, were observed in samples taken at different times of the year. In the chemocline, Chomatium okenii was dominant in the summer months, whereas Amoebobacter purpureus populations dominated in autumn and winter. This change was confirmed by fluorescent in situ hybridization.

Full text

Abstract The seasonal and spatial variations in the com-
munity structure of bacterioplankton in the meromictic
alpine Lake Cadagno were examined by temporal temper-
ature gradient gel electrophoresis (TTGE) of PCR-ampli-
fied 16S rDNA fragments. Two different amplifications
were performed, one specific for the domain Bacteria
(Escherichia coli positions 8–536) and another specific
for the family Chromatiaceae (E. coli positions 8–1005).
The latter was followed by semi-nested reamplification
with the bacterial primer set, allowing comparison of the
two PCR approaches by TTGE. The TTGE patterns of
samples from the chemocline and the anoxic moni-
molimnion were essentially identical, whereas the oxic
mixolimnion displayed distinctively different banding
patterns. For samples from the chemocline and the moni-
molimnion, dominant bands in the Bacteria-specific
TTGE profiles comigrated with bands obtained by the
semi-nested PCR approach specific for Chromatiaceae.
This observation suggested that Chromatiaceae are in
high abundance in the anoxic water layer. All dominant
bands were excised and sequenced. Changes in the com-
munity structure, as indicated by changes in the TTGE
profiles, were observed in samples taken at different times
of the year. In the chemocline, Chromatium okenii was
dominant in the summer months, whereas Amoebobacter
purpureus populations dominated in autumn and winter.
This change was confirmed by fluorescent in situ hybrid-
ization.
Key words 16S rDNA · Temporal temperature gradient
gel electrophoresis · Fluorescent in situ hybridization ·
Population dynamics · Meromixis · Bacterioplankton ·
Chemocline · Bacterial diversity · Chromatiaceae ·
Purple sulfur bacteria
Abbreviations TTGE Temporal temperature gradient
gel electrophoresis · DGGE Denaturing gradient gel
electrophoresis · FISH Fluorescent in situ hybridization
Introduction
The introduction of molecular biological techniques in
microbial ecology has fundamentally changed our view of
microbial diversity. It is nowadays accepted that only a
minor part of the naturally occurring microbial species
has been cultivated so far. Through the use of molecular
techniques, many new phylotypes and new phylogenetic
divisions have been described (Hugenholtz et al. 1998).
Although the microbial diversity of various ecosystems
has been investigated, relatively little is known about the
temporal population dynamics of the species inhabiting
these ecosystems. Recently developed techniques such as
denaturing gradient gel electrophoresis (DGGE) and fluo-
rescence in situ hybridization (FISH) are new tools to
study biodiversity, community dynamics, and successions
on the species level (Casamayor et al. 2000; Duineveld et
al. 1998; Ferris and Ward 1997; Øvreås et al. 1997; Pern-
thaler et al. 1998).
Lake Cadagno is a small meromictic lake at an altitude
of 1921 m in the southern Swiss Alps. Its water chemistry
is influenced by the dolomite gypsum geology that leads
to a sulfate-rich input to the monimolimnion. The oxic
mixolimnion reaches a depth of about 11 m; below this,
the concentration of hydrogen sulfide formed by sulfate
reducers increases up to 1 mM at the sediment surface.
The bacterial layer, easily detected by its high turbidity, is
situated at the upper part of the anoxic zone, which is still
reached by light and where hydrogen sulfide is available
for anoxic photosynthesis. The lake is covered by ice and
snow from November until May.
In recent years, Lake Cadagno has been the subject
of several limnological investigations (Birch et al. 1996;
Fischer et al. 1996; Joss et al. 1994; Lehmann and Bach-
ofen 1999; Schanz et al. 1998; Tonolla et al. 1999). These
have mainly focused on the physiology of phototrophs,
Philipp P. Bosshard · Rolf Stettler ·
Reinhard Bachofen
Seasonal and spatial community dynamics
in the meromictic Lake Cadagno
Received: 21 October 1999 / Revised: 20 June 2000 / Accepted: 20 June 2000 / Published online: 2 August 2000
ORIGINAL PAPER
P. P. Bosshard (✉) · R. Stettler · R. Bachofen
Institute of Plant Biology, Department of Microbiology,
University of Zürich, Zollikerstrasse 107,
8008 Zürich, Switzerland
e-mail: bossi@botinst.unizh.ch,
Tel.: +41-1-6348286, Fax: +41-1-6348204
Arch Microbiol (2000) 174:168–174
Digital Object Identifier (DOI) 10.1007/s002030000191
© Springer-Verlag 2000

elemental cycles in the lake, and characterization of envi-
ronmental factors. From these studies it is known that
species of Chromatiaceae, presumably Chromatium okenii
and Amoebobacter purpureus [recently reclassified and
proposed as Pfennigia purpurea (Tindall 1999)], are very
abundant in the chemocline of Lake Cadagno. Recently, we
analyzed the bacterial diversity of the redox transition zone
using molecular methods based on a single discrete mea-
surement (Bosshard et al. 2000). We found that the com-
munity was dominated by Chromatiaceae, especially
Amoebobacter species, but, in contrast to the findings of
earlier studies, C. okenii was not numerically abundant.
To obtain a better understanding of bacterial popula-
tion dynamics, we analyzed the spatial and seasonal
changes of the bacterial composition in Lake Cadagno by
temporal temperature gradient gel electrophoresis (TTGE)
over a 1-year sampling period and by FISH at selected
sampling dates. Knowing that species of the family Chro-
matiaceae constitute a major part of the bacterial commu-
nity, we used a Chromatiaceae-specific primer set besides
primers specific for Bacteria. The TTGE technique, simi-
lar to DGGE, allows the separation of equally sized PCR
products in a polyacrylamide matrix based on their differ-
ent melting behaviors due to sequence variations (Muyzer
et al. 1993). In contrast to DGGE, in which the denaturing
environment is formed by a chemical gradient, the TTGE
method requires a linear temperature gradient over the
length of the electrophoresis run.
Materials and methods
Sample collection
Water samples from three different depths (mixolimnion, 4 m;
chemocline, 11–13 m; monimolimnion, 16 m) were collected on
4 September 1997, 26 March 1998 (lake was ice-covered), and bi-
weekly from 22 June 1998 to 8 October 1998. The exact position
of the bacterial layer was assessed by its turbidity. Subsurface wa-
ter samples were collected with a 2-l sampler. Water (120 ml) was
filtered (pore-size 0.22 µm, Durapore, Millipore, Bedford, Mass.)
using a syringe filtration device. The filters were placed in sterile
2-ml centrifuge tubes and covered with 1.5 ml lysis buffer (50 mM
Tris, pH 8.0, 20 mM EDTA, 50 mM sucrose). After processing,
the tubes were immediately frozen and stored at –20°C until DNA
extraction.
For FISH, 5–20 ml of sample water were concentrated on
47-mm diameter polycarbonate filters (pore size 0.2 µm, type
GTTP; Millipore, Volketswil, Switzerland). The filters were fixed
with paraformaldehyde as previously described (Glöckner et al.
1996)
and stored at –20°C until further processing.
Nucleic acid extraction
For total nucleic acid extraction, microorganisms were washed off
the thawed filter by rinsing the membrane with buffer. The sus-
pension was centrifuged and the pellet was resuspended in 0.5 ml
lysis buffer. Lysozyme (10 mg ml
–1
) was added before incubation
at room temperature for 10 min. After adding 1% (v/v) SDS and
100 µg ml
–1
proteinase K, the mixture was incubated at 37°C for
30 min and at 55°C for 10 min. DNA was obtained from the
lysates by using standard phenol-chloroform extraction and
ethanol precipitation procedures (Sambrook et al. 1989). RNA was
removed by incubating the aqueous solution with 5 U of DNase-
free RNase for 15 min at 37°C. The effectiveness of the cell lysis
procedure was examined by microscopy of samples taken after ly-
sis treatment.
PCR amplification of rDNA fragments
The 16S rRNA genes from total bacterial DNA were amplified by
PCR with two different primer sets (Table 1). The first primer set
amplified Chromatiaceae 16S rRNA gene sequences. The reverse
primer complements a region conserved among members of Chro-
matiaceae and is slightly modified from one described by Coolen
and Overmann (1998). The second primer set, specific for the do-
main Bacteria, amplified a 530-bp fragment suitable for subse-
quent TTGE analysis. It was used for semi-nested PCR of ampli-
cons generated with the first primer set or for direct PCR of total
genomic DNA. The GC-clamp of the forward primer prevents
strand dissociation at high temperatures during separation in the
gel (Muyzer et al. 1993). Touchdown PCR was carried out with a
Techne thermocycler (Witec, Luzern, Switzerland) with the fol-
lowing conditions: First, the samples were heated to 94°C for 5 min
to denature template DNA. Subsequently, 20 cycles were performed
beginning with an annealing temperature of 68 (primer set 1) or
65°C (primer set 2) for 30 s, lowering the temperature 0.5°C every
cycle, followed by 20 cycles with annealing at 58 (primer set 1) or
55°C (primer set 2). Primer extension was carried out at 72°C for
90 s. The hot start technique was used, adding the enzymes after
the initial denaturing step at 75°C. Amplification was carried out
in 25-µl reaction solution containing 1 U of polymerase (Expand
High Fidelity PCR System; Roche Diagnostics, Rotkreuz, Switzer-
land), reaction buffer, 1.5 mM MgCl
2
, 2.5 µg bovine serum albu-
min, 200 µM deoxynucleotide triphosphate, 5 pmol of each for-
ward and reverse primer, and 100 ng template DNA.
Temporal temperature gradient gel electrophoresis analysis
TTGE analysis was performed with a DCode system (Bio-Rad
Laboratories, Glattbrugg, Switzerland). PCR samples were applied
directly onto 6% (w/v) polyacrylamide gels (acrylamide:N,N′-
methylene bisacrylamide, 37.5:1 (w/w); 7 M urea; 1×TAE). The
temperature range was 54°C to 64°C, temperature ramp rates
1.7°C h
–1
, and voltage 150 V. After completion of electrophoresis,
the gels were stained in ethidium bromide and photographed under
UV transillumination.
169
Table 1 Set of PCR primers used for Chromatiaceae- and Bacteria-specific amplification
Primer Set Target Sequence Position
S-D-Bact-0008-a-S-20 1 Bacteria 5′AGAGTTTGATCCTGGCTCAG 8–27
S-F-Chrom-986-A-20 1 Chromatiaceae 5′TTCCRAGGATGTCAAGGGCT 1005–986
S-D-Bact-0008-a-S-20–GC 2 Bacteria 5′AGAGTTTGATCCTGGCTCAG 8–27
S-
*
-Univ-0536-a-A-18 2 Universal 5′GWATTACCGCGGCKGCTG 536–519
GC-clamp of forward primer 2 5′CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGG

Sequencing of TTGE bands
For sequence determination of TTGE bands, small pieces were ex-
cised from the acrylamide gel and placed into sterilized vials. The
samples were subjected to passive diffusion (12 h at 4°C) after
adding 20 µl sterilized water. Supernatant (10 µl) was then used as
template for a reamplification using the same primer and reaction
conditions as before. Purity of the PCR reamplification product
was checked by rerunning an aliquot on a TTGE gel. If there were
more bands visible on the gel than just the PCR-reamplified TTGE
band, the target band was reexcised, reamplified, and controlled
twice before sequencing. The PCR products (30–90 ng) were then
bidirectionally sequenced with an automated DNA sequencer (Ap-
plied Biosystems, model 310) using the Taq DyeDeoxy termina-
tor sequencing kit as described by the manufacturer, with the
primer
S-D-Bact-0008-a-S-20 or S-*Univ-0536-a-A-18 as described
above.
The FASTA search option for the EMBL database was used to
search for closest phylogenetic neighbors. Sequences were submit-
ted to the CHECK_CHIMERA program of the ribosomal database
project (RDP) to detect possible chimeric artifacts (Kopczynski et
al. 1994; Maidak et al. 1999).
Fluorescence in situ hybridization
with a Chromatiaceae-specific probe
FISH was performed as previously described (Glöckner et al. 1996).
Filter sections were hybridized with 150 ng of CY3-labeled probe
and stained with 4′,6′-diamidino-2-phenylindole (DAPI). Cells
were viewed with an Olympus BX50 epifluorescence microscope
(Olympus Optical, Volketswil, Switzerland). The following group-
specific probes (Manz et al. 1992) were used: EUB338 (domain
Bacteria), ALF1b (α-Proteobacteria), BET42a (β-Proteobacteria),
GAM42a (γ-Proteobacteria), and Arch915 (domain Archaea).
Along with these known probes, a Chromatiaceae-specific oligo-
nucleotide slightly modified from the primer described by Coolen
and Overmann (1998) was used. The sequence of this probe
S-F-
Chrom-986-b-A-20 is 5′-TTCCRAGGATGTCAAGGGCT-3′
(Escherichia
coli positions 986–1005).
To estimate biovolumes, the size of at least 30 cells of the large
phototrophs Chromatium okenii and Amoebobacter purpureus
were measured. Of all other bacteria more than 500 cells were
classified into three morphotypes: (1) small coccoids, ≤0.5×0.5 µm;
(2) rods, approximately 1×2 µm; (3) larger irregular cells. Cell vol-
umes were calculated on the basis of the best fitting ellipsoid ac-
cording to V=(4π/3)ab
2
, where Vis cell volume and aand bare the
major and minor radii, respectively. Due to the fact that C. okenii
cell forms are between ellipsoids and cylinders, their cell volumes
were calculated as the mean of an ellipsoid and a cylinder.
Nucleotide sequence accession numbers
Nucleotide sequences have been deposited in the GenBank database
under the accession numbers AJ250029 (band C) and AJ250030
(band G).
Results
Sample collection and nucleic acid extraction
The bacterial layer identified by maximum turbidity was
located at a depth of 11–13 m. From earlier findings (Egli
et al. 1998) it is known that the layer oscillates by a verti-
cal distance of 1.5 m several times a day. No tendency
was observed for the mean depth of the layer to increase
or decrease seasonally (data not shown). The turbidity of
the layer was very low during the winter ice cover.
The recovery of predominantly intact high-molecular-
weight DNA was demonstrated by agarose gel elec-
trophoresis (data not shown). The total amount of nucleic
acids extracted from 120-ml water samples was between
0.2 and 2.0 µg DNA. The extracted DNA was used di-
rectly as PCR template; no additional step was necessary
to remove substances that might inhibit PCR amplifica-
tion.
PCR amplification of rDNA fragments
Direct PCR of total community DNA with the bacterial
primer set yielded amplification products for all samples.
This resulted in fragment lengths of about 530 bp, suitable
for TTGE analysis.
Since members of the phototrophic γ-Proteobacteria
play an important role in the lake, a Chromatiaceae spe-
cific primer was employed. Bosshard et al. (2000) re-
cently described a PCR primer to selectively recover 16S
rDNA fragments from organisms belonging to this phylo-
genetic group. The sequence of the primer is slightly
modified from the primer described by Coolen and Over-
mann (1998). We found that by changing the base at posi-
tion 1001 (E. coli numbering) from C to Y (C or T) the
primer recognizes six additional species of this family
without becoming more unspecific. This modification
was introduced on the basis of sequence comparisons of
all available Chromatiaceae sequences in current data-
bases.
PCR with the Chromatiaceae-specific primer set yielded
amplification products only for the anoxic samples from
the chemocline and the monimolimnion (data not shown).
To analyze these fragments by TTGE and to compare the
results with the patterns obtained with the bacterial primer
set, semi-nested PCR was done using the Bacteria-spe-
cific primer pair (data not shown). Negative controls with
water instead of template DNA showed no amplification
in all experiments.
Variation of bacterial community structure with depth
The 16S rDNA amplification products from different
depths in Lake Cadagno were analyzed by TTGE. In Fig.1
(lanes 1–3) the TTGE patterns of the bacterial populations
at different depths from one sampling date (9 September
1998) are shown. The patterns of the chemocline and the
monimolimnion were highly similar (Fig.1, lanes 2, 3),
whereas the mixolimnion showed a distinctly different
pattern (Fig.1, lane 1). Muyzer et al. (1993) found that on
a DGGE gel each band corresponds to a different 16S
rDNA sequence and thus reflects distinct microbial popu-
lations in the community. By sequencing we found that
the same fragment can be separated by TTGE into two
distinct bands (see below). In the figures, bands with same
sequences are indicated by the same letter, e.g. A and A′.
The number of detectable bands in the oxic zone did not
differ significantly from that in the anoxic zone. The
170

TTGE was highly reproducible concerning the number
and order of the bands and showed only slight variation in
the separation distance between successive fragments.
TTGE of Chromatiaceae-specific PCR products
Comparing direct Bacteria-specific and semi-nested Chro-
matiaceae-specific PCR products of equal length by
TTGE (Fig.1, lanes 2–5), it was remarkable that most
bands of Chromatiaceae corresponded to dominant bands
of Bacteria. Moreover, the resolution patterns of these two
amplification products were highly similar, but the Chro-
matiaceae-specific pattern had less background.
Seasonal variations in bacterial community structure
To investigate the temporal community dynamics, several
samples were collected from September 1997 until Octo-
ber 1998 at three different depths (mixolimnion, 4 m;
chemocline, 11–13 m; monimolimnion, 16 m). The com-
munity structure of these three zones varied with time
(Fig2A, B). In the mixolimnion, the community composi-
tion changed greatly during the sampling period. Several
bands occurred only at one or two sampling dates, espe-
cially at the beginning of the summer season, from the
end of ice cover to midsummer (dates 2, 3, and 5) when
the surface water was not yet stratified. During summer
stagnation (dates 5, 7, and 9), the banding patterns re-
mained similar for 3 months.
Most remarkable for the chemocline was the commu-
nity shift from 26 March (date 2, ice cover) to 22 June
(date 3) and again from 9 September (date 7) to 21 Sep-
tember (date 8). Bands D and D′took the place of bands
A, A′, B, B′, and C from date 3 to date 7. This shift was
demonstrated both with the Bacteria- and the Chromati-
aceae-specific primer pairs. Bands E and E′of the bacter-
ial fingerprint were present throughout the year in both
anoxic sample depths. With the bacterial primers, an addi-
tional band (G) was observed from 22 June (date 3) to 19
August (date 6). In the monimolimnion this shift in bacte-
rial community (D and D′to A, A′, B, B′, and C) was less
distinctive, only the intensities of the corresponding bands
varied with time. Overall, the number of bands, and thus
171
Fig.1 Temporal temperature gradient gel electrophoresis (TTGE)
analysis of 16S rDNA fragments of three different depths. Lanes
1–3 Bacterial fingerprints of mixolimnion, chemocline, and moni-
molimnion, respectively; lanes 4 and 5 Chromatiaceae-specific
fingerprint of chemocline and monimolimnion, respectively. All
16S rDNA fragments correspond to Escherichia coli position
8–536. The Chromatiaceae fingerprints were generated using
semi-nested PCR, first using a Chromatiaceae-specific primer pair
and second using the same primers as for the bacterial fingerprints.
Letters and arrows indicate excised and sequenced bands of
chemocline and monimolimnion
Fig.2A,B Seasonal TTGE analysis of 16S rDNA fragments of
different sampling depths. Lanes 1–9 correspond to sampling
dates; September 1997 (1), 26 March 1998 (2), 22 June 1998 (3),
16 July 1998 (4), 4 August 1998 (5), 19 August 1998 (6), 9 Sep-
tember 1998 (7), 21 September 1998 (8), and 8 October 1998 (9).
Letters indicate excised and sequenced bands. ABacterial finger-
prints. BChromatiaceae-specific fingerprints. All 16S rDNA frag-
ments correspond to E. coli position 8–536. The Chromatiaceae
fingerprints were generated using semi-nested PCR, first using a
Chromatiaceae-specific primer pair and second using the same
primers as for the bacterial fingerprints

the number of dominant populations, did not alter signifi-
cantly during the sampling period.
Sequencing of dominant bands
Since this investigation focuses on the anoxic water layer,
dominant bands of TTGE patterns from the chemocline
and monimolimnion were excised, reamplified, and se-
quenced. Bands A, A′, B, B′, C, E, E′, and F were excised
and sequenced from lane 1 (September 1997) of the bac-
terial fingerprints, band G from lane 4 (16 July 1998) ) of
the bacterial fingerprints, and bands D and D′from lane 5
(4 August 1998) both of the Bacteria and Chromatiaceae
fingerprints, respectively. To confirm that correct frag-
ments were sequenced, the reamplification products were
run on a control TTGE (data not shown). From this gel it
became apparent that bands A and A′, B and B′, and C and
C′were double bands representing the same fragment (i.e.,
after reamplification of the excised fragment A two bands,
A and A′, were obtained). Sequencing both bands separately
confirmed this observation. The nearest neighbor and
phylogenetic group of each sequence is listed in Table 2.
Except for bands C and G, all other fragments sequenced
were identical to known sequences in public databases.
This result clearly showed that Chromatium okenii and
Amoebobacter purpureus are the organisms that account
for the community shifts observed by TTGE. Bands H
and I could not be purified enough for proper sequencing.
Fluorescence in situ hybridization
To further validate the population shift from A. purpureus
to C. okenii and vice versa (from 26 March to 22 June and
from 9 September to 21 September), FISH was per-
formed. Samples of the chemocline from 9 September and
21 September were compared. Cell densities, determined
by DAPI staining, were 9.4×10
6
and 9.2×10
6
cells ml
–1
,
respectively. Between 57% and 55% of all DAPI-stained
cells could be visualized with bacterial probe EUB338,
whereas only 11% and 9% counted for Archaea. The per-
centages of the proteobacterial subclasses for the two
sampling dates detected by the corresponding probes were
as follows: α-Proteobacteria 2% and 0%, β-Proteobacte-
ria 9% and 15%, γ-Proteobacteria 9% and 12%, respec-
tively. With the probe specific for Chromatiaceae, 4% and
6% of the DAPI stained cells were detected.
Due to the conspicuously large cell forms and unam-
biguous morphology of these phototrophs, the detected
organisms could be assigned to either C. okenii or A. pur-
pureus. On 9 September, 20% of the cells reacting with
the Chromatiaceae probe were identified as C. okenii,
whereas 80% were found to be A. purpureus and formed
aggregates of up to a few hundred cells. On 21 September,
only 0.4% of the cells were C. okenii, whereas 99.6% rep-
resented A. purpureus. During these 2 weeks, the cell den-
sity of C. okenii dropped from an absolute abundance of
7.0×10
4
to 2.3×10
3
cells ml
–1
, whereas the A. purpureus
cell number increased from 2.8×10
5
to 5.8×10
5
cells ml
–1
.
The average cell volumes of C. okenii and A. purpureus
were 247±98 and 9±4 µm
3
per cell, respectively. Thus, on
9 September, the biovolumes of 17.3±6.9×10
3
and
2.59±1.0×10
3
mm
3
per m
3
of lake water of C. okenii and
A. purpureus accounted for approximately 72% and 11%
of the total microbial biovolume. In contrast, on 21 Sep-
tember, the respective biovolumes were 0.6±0.2×10
3
and
5.4±2.1×10
3
mm
3
m
–3
, or about 6% and 54% of the mi-
crobial biovolume.
172
Table 2 Sequence similarity to nearest neighbors and phylogenetic affiliation of 16S rDNA retrieved from Lake Cadagno
Band Nearest neighbor Phylogenetic group
Similarity Species Accession number
100% Uncultured freshwater bacterium LCK-22
a
AF107320
A, A′99.8% Phototrophic bacteria
b
AJ006058 Chromatiaceae, γ-Proteobacteria
99.0% Lamprocystis roseopersicina, strain DSM 229 AJ006063
B, B′100% Amoebobacter purpureus, strain DSM 4197 AJ223235 Chromatiaceae, γ-Proteobacteria
C99.6% phototrophic bacteria
b
AJ006061 Chromatiaceae, γ-Proteobacteria
94.1% Lamprocystis roseopersicina, strain DSM 229 AJ006063
D, D′100% Uncultured freshwater bacterium LCK-30
a
AF107324 Chromatiaceae, γ-Proteobacteria
99.4% Chromatium okenii AJ223234
E, E′100% Uncultured freshwater bacterium LCK-20
a
AF109141 δ-Proteobacteria
96.4% Desulfocapsa thiozymogenes X95181
F100% Uncultured freshwater bacterium LCK-32
a
AF107326 Chlorobiaceae
94.5% Chlorobium phaeovibrioides Y10654
G 94.3% Geothrix fermentans U41563 Holophaga group
a
Bosshard et al. 2000. Sequences of a 16S rDNA clone library of the chemocline of Lake Cadagno
b
Tonolla et al. 1999. Sequences of a 16S rDNA clone library of the chemocline of Lake Cadagno

Discussion
We have used TTGE analysis of PCR-amplified 16S rDNA
fragments to infer an overall picture of the seasonal dy-
namics and spatial distribution of the bacteria in Lake
Cadagno. Although a large amount of physical, chemical,
and biological background data are available for this lake,
no previous study has focused on the community dynam-
ics of lake bacteria. In the study presented here, we report
results for samples collected over a 1-year period. Be-
cause TTGE allows comparison of a large number of en-
vironmental samples, it is a useful technique for a general
assessment of the spatial distribution of bacterial popula-
tions and their changes over time.
However, as with most of techniques, TTGE also has
limitations. First, it is subject to the usual biases intro-
duced by the DNA extraction and PCR amplification pro-
cedures. Second, only numerically dominant populations
will be detected by TTGE. In addition to these concerns,
we found that some of the bands with different running
distances had identical sequences. That means that differ-
ent bands do not a priori belong to different populations
in every case. A possible explanation for the formation
of double bands could be the use of degenerate primers
(S-
*
-Univ-0536-a-A-18), which produce a mixture of
PCR products in the primer site. Degenerate primers are
often used in PCR analysis of environmental 16S rDNA
fragments to recover all members of a target group. How-
ever, the variable positions could lead to different melting
behaviors of the PCR products. An earlier study with
cloned 16S rDNA fragments (Bosshard et al. 2000) showed
that, indeed, some of the sequences amplified with degen-
erate primers produced double bands. Double bands could
also be due to hairpin formation within the GC-clamp of
the forward primer. Despite the drawbacks outlined here,
our results indicate that TTGE can be effectively used to
monitor spatial and temporal dynamics of different phylo-
genetic groups in natural samples and to detect commu-
nity shifts.
In a first step, we determined the genetic diversity with
primers specific for the domain Bacteria. In meromictic
lakes, the chemocline separates spatial zones with differ-
ent physical and chemical parameters. Indeed, between
the oxic and anoxic parts of the lake a distinct variance in
TTGE pattern was observed (Fig.1). These results indi-
cate that the natural communities from the mixolimnion
and the monimolimnion are significantly different. By
contrast, only very few differences were observed in pro-
files between chemocline and monimolimnion, suggesting
a very low population variability. The presence of an iden-
tical community structure in both anoxic zones was unex-
pected. However, TTGE analysis of 16S rDNA fragments
does not distinguish whether the template DNA originates
from living or dead organisms. Therefore, it cannot be ex-
cluded that the bacteria at a depth of 16 m are dead or in a
physiologically inactive state during sedimentation.
In addition to these spatial studies, we investigated the
seasonal dynamics of the bacterial community structure
over a 1-year sampling period (1998). Figure2A shows
the changes in the composition of the bacterial assem-
blage of the mixolimnion. From winter ice cover, over the
period after icebreak and until midsummer, the dominant
fraction of the community changed totally. This result was
not unexpected, since environmental factors such as tem-
perature, exposition to sunlight and organic loading can
affect the composition of a microbial community. A tem-
poral variation in the upper layer among the communities
is also in accordance with results from other studies in
which molecular methods were used (Pernthaler et al.
1998; Höfle et al. 1999). However, in contrast to these
studies the assemblage remained nearly stable during
summer stagnation in Lake Cadagno (Fig.2A), whereas in
the other lakes the microbial assemblages continuously al-
tered throughout the year. It is notable that dominant
bands on TTGE gels from a water sample of September
1997 occurred again in samples collected in summer and
autumn 1998.
During the seasonal study we also observed changes in
the community structure in the anoxic zone of the lake.
Because sequencing of the dominant bands revealed that
the nucleotide sequence of most bands had a high similar-
ity to the DNA sequence of members of Chromatiaceae,
particular attention was paid to the changes in abundance
and taxonomic composition of this group. Therefore, in a
second approach a group-specific probe was used to re-
cover sequences from members of this group (Fig.2B). To
compare the group-specific pattern from one sample di-
rectly to the bacterial community pattern of the same sam-
ple, we used nested PCR with Chromatiaceae-specific
primers followed by a second PCR with bacterial primers.
From earlier investigations (Birch et al. 1996; Fischer et
al. 1996; Joss et al. 1994; Lehmann and Bachofen 1999;
Schanz et al. 1998; Tonolla et al. 1999; Bosshard et al.
2000), it is known that species of purple sulfur bacteria
(Chromatiaceae) make up a considerable fraction of the
bacteria in the chemocline of Lake Cadagno. Our sequenc-
ing results of both intensely stained TTGE bands and in
situ analysis corresponded well to the results determined
in previous years. Both in terms of biomass and number
Chromatiaceae were the most dominant group in the
chemocline. This indicates a significant role of these
populations. For instance, on 9 September, 83% of the
microbial biovolume was formed by A. purpureus and
C. okenii.
Potentially, the most interesting finding in this study is
that the structure of the Chromatiaceae community
changed significantly with time. The DNA-based finger-
prints of the chemocline indicated that a predominance of
C. okenii was replaced by A. purpureus in late summer
(Fig.2A, B). This shift occurred against a background
of relative stability in the other bacterial populations, as
determined by PCR-TTGE. Immediately after icebreak
C. okenii formed the dominant fraction until September,
when A. purpureus populations were more abundant. This
population shift was confirmed using FISH. In the anoxic
monimolimnion a similar observation was made. Although
the bands never disappeared completely, C. okenii bands
173

174
gained intensity in summer while A. purpureus bands
dominated in winter. It is not clear at the moment whether
these organisms at a depth of 16 m are dead or in a phys-
iologically inactive state during sedimentation. Further-
more, it would be interesting to know whether they are
able to move up the water column again to reach the light
intensities needed for photosynthesis. There are indica-
tions for a directed upwards movement of C. okenii to the
chemocline in spring, as these purple bacteria are found
after the melt of the ice cover on the underside of sedi-
mentation traps exposed during winter (C. Lehmann, Uni-
versity of Zürich, Switzerland, personal communication).
These observations lead to the suggestion that the
environmental lake conditions in late summer favor
A. purpureus. Further investigations will have to be made
to understand the environmental factors leading to the ob-
served population shift and to learn more about the verti-
cal movement of C. okenii.
Acknowledgements We are grateful to Dominik Grüter for tech-
nical help. This research was supported by grant 31–50950.97 of
the Swiss National Foundation for Scientific Research.
References
Birch L, Hanselmann KW, Bachofen R (1996) Heavy metal con-
servation in Lake Cadagno sediments: historical records of an-
thropogenic emissions in a meromictic alpine lake. Water Res
30:679–687
Bosshard PP, Santini Y, Grüter D, Stettler R, Bachofen R (2000)
Bacterial diversity and community composition in the chemo-
cline of the meromictic alpine Lake Cadagno as revealed by
16S rDNA analysis. FEMS Microbiol Ecol 31:173–182
Casamayor EO, Schäfer H, Bañeras L, Pedrós-Alió C, Muyzer G
(2000) Identification of and spatio-temporal differences be-
tween microbial assemblages from two neighboring sulfurous
lakes: comparison by microscopy and denaturing gradient gel
electrophoresis. Appl Environ Microbiol 66:499–508
Coolen MJL, Overmann J (1998) Analysis of subfossil molecular
remains of purple sulfur bacteria in a lake sediment. Appl En-
viron Microbiol 64:4513–4521
Duineveld BM, Rosado AS, van Elsas JD, van Veen JA (1998)
Analysis of the dynamics of bacterial communities in the rhi-
zosphere of the chrysanthemum via denaturing gradient gel
electrophoresis and substrate utilization patterns. Appl Environ
Microbiol 64:4950–4957
Egli K, Wiggli M, Klug J, Bachofen R (1998) Spatial and tempo-
ral dynamics of the cell density in a plume of phototrophic mi-
croorganisms in their natural environment. Doc Ist Ital Idrobiol
63:121–126
Ferris MJ, Ward DM (1997) Seasonal distributions of dominant
16S rRNA-defined populations in a hot spring microbial mat
examined by denaturing gradient gel electrophoresis. Appl En-
viron Microbiol 63:1375–1381
Fischer C, Wiggli M, Schanz F, Hanselmann KW, Bachofen R
(1996) Light environment and synthesis of bacteriochlorophyll
by populations of Chromatium okenii under natural environ-
mental conditions. FEMS Microbiol Ecol 21:1–9
Glöckner FO, Amann R, Alfreider A, Pernthaler J, Psenner R,
Trebesius K, Schleifer K-H (1996) An in situ hybridization
protocol for detection and identification of planktonic bacteria.
Syst Appl Microbiol 19:403–406
Höfle MG, Haas H, Dominik K (1999) Seasonal dynamics of bac-
terioplankton community structure in a eutrophic lake as deter-
mined by 5S rRNA analysis. Appl Environ Microbiol 65:3164–
3174
Hugenholtz P, Pitulle C, Hershberger KL, Pace NR (1998) Novel
division level bacterial diversity in a Yellowstone hot spring.
J Bacteriol 180:366–376
Joss A, Mez K, Känel B, Hanselmann KW, Bachofen R (1994)
Measurement of fluorescence kinetics of phototrophic bacteria
in the natural environment. J Plant Physiol 144:333–338
Kopczynski ED, Bateson MM, Ward DM (1994) Recognition of
chimeric small-subunit ribosomal DNAs composed of genes
from uncultivated microorganisms. Appl Environ Microbiol
60:746–748
Lehmann C, Bachofen R (1999) Images of concentrations of dis-
solved sulphide in the sediment of a lake and implications for
internal sulphur cycling. Sedimentology 46:537–544
Maidak BL, Cole JR, Parker CT, Garrity GM, Larsen N, Li B, Lil-
burn TG, McCaughey MJ, Olsen GJ, Overbeek R, Pramanik S,
Schmidt TM, Tiedje JM, Woese CR (1999) A new version of
the RDP (Ribosomal Database Project). Nucleic Acids Res 27:
171-173
Manz W, Amann R, Ludwig W, Wagner M, Schleifer K-H (1992)
Phylogenetic oligodeoxynucleotide probes for the major sub-
classes of proteobacteria: problems and solutions. Syst Appl
Microbiol 15:593–600
Muyzer G, De Waal EC, Uitterlinden AG (1993) Profiling of com-
plex microbial populations by denaturing gradient gel elec-
trophoresis analysis of polymerase chain reaction-amplified
genes coding for 16S rRNA. Appl Environ Microbiol 59:695–
700
Øvreås L, Forney L, Daae FL, Torsvik V (1997) Distribution of
bacterioplankton in meromictic lake Sælenvannet, as deter-
mined by denaturing gradient gel electrophoresis of PCR-am-
plified gene fragments coding for 16S rRNA. Appl Environ
Microbiol 63:3367–3373
Pernthaler J, Glöckner FO, Unterholzner S, Alfreider A, Psenner
R, Amann R (1998) Seasonal community and population dy-
namics of pelagic bacteria and archaea in a high mountain lake.
Appl Environ Microbiol 64:4299–4306
Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: a
laboratory manual, 2nd edn. Cold Spring Harbor Laboratory,
Cold Spring Harbor, N.Y.
Schanz F, Fischer-Romero C, Bachofen R (1998) Photosynthetic
production and photoadaption of phototrophic sulfur bacteria in
Lake Cadagno (Switzerland). Limnol Oceanogr 43:1262–1269
Tindall BJ (1999) Taxonomic note: transfer of Amoebobacter pur-
pureus to the genus Pfennigia gen. nov., as Pfennigia purpurea
comb. nov., on the basis of the illegitimate proposal to make
Amoebobacter purpureus the type species of the genus Amoe-
bobacter. Int J Syst Bacteriol 49:1307–1308
Tonolla M, Demarta A, Peduzzi R, Hahn D (1999) In situ analysis
of phototrophic sulfur bacteria in the chemocline of meromic-
tic Lake Cadagno (Switzerland). Appl Environ Microbiol 65:
1325–1330