Insight into the composition of the intercellular matrix
of Streptococcus pneumoniae biofilms
Mirian Domenech,1,3 Ernesto García,1,3 Alicia Prieto2
and Miriam Moscoso1,3*
1Departamento de Microbiología Molecular y Biología de
las Infecciones, and 2Biología Medioambiental, Centro
de Investigaciones Biológicas (CSIC), Ramiro de
Maeztu 9, 28040 Madrid, Spain.
3CIBER de Enfermedades Respiratorias (CIBERES),
Madrid, Spain.
Summary
Biofilm matrices consist of a mixture of extracellular
polymeric substances synthesized in large part by
the biofilm-producing microorganisms themselves.
These matrices are responsible for the cohesion
and three-dimensional architecture of biofilms. The
present study demonstrates the existence of a matrix
composed of extracellular DNA, proteins and
polysaccharides in the biofilm formed by the human
pathogen Streptococcus pneumoniae. Extracellular
DNA, visualized by fluorescent labelling, was an
important component of this matrix. The existence of
DNA–protein complexes associated with bacterial
aggregates and other polymers was hypothesized
based on the unexpected DNA binding activity of lys-
ozyme LytC, a novel moonlighting protein. Actually, a
25-amino-acid-long peptide derived from LytC (posi-
tions 408 and 432 of the mature LytC) was also
capable of efficiently binding to DNA. Moreover, the
presence of intercellular DNA–LytC protein com-
plexes in pneumococcal biofilms was demonstrated
by confocal laser scanning microscopy. Evidence
of extracellular polysaccharide different from the
capsule was obtained by staining with Calcofluor dye
and four types of lectin conjugated to Alexa fluoro-
phores, and by incubation with glycoside hydrolases.
The presence of residues of Glcp(1,4) and Glc-
NAc(1,4) (in its deacetylated form) in the pneumo-
coccal biofilm was confirmed by GC-MS techniques.
Introduction
The growth and dispersal of microbes, whether patho-
genic or environmental, commonly involves the produc-
tion of biofilms; indeed, they are thought to be present in
the vast majority of bacterial infections (Wolcott and
Ehrlich, 2008). A biofilm is defined as a thin layer of
microorganisms adhered to the surface of an organic or
inorganic substrate embedded in an extracellular matrix
(Costerton et al., 1995). This matrix consists of a mixture
of biopolymers (extracellular polymeric substances or
EPS) synthesized largely by the biofilm-producing micro-
organisms themselves. Although EPSs such as proteins,
DNA and polysaccharides are common to most biofilms,
matrix composition can vary greatly depending on the
bacterial species involved and environmental conditions
(Flemming and Wingender, 2010).
The human pathogen Streptococcus pneumoniae (or
pneumococcus) is a leading cause of pneumonia, menin-
gitis and bloodstream infections in the elderly, and one of
the main pathogens responsible for middle ear infections
in children. It is carried asymptomatically in the nasophar-
ynx of many healthy adults, and in as many as 20–40% of
healthy children (colonization begins shortly after birth)
(Weiser, 2010). Recent reports have shown the in vivo
formation of S. pneumoniae biofilms on adenoid and
mucosal epithelial tissues in children with recurrent or
chronic ear infections (for a recent review, see Domenech
et al., 2012). Using low-temperature scanning electron
microscopy (LTSEM) techniques, our group detected a
biofilm matrix containing an intercellular, fibre-like material
that linked the pneumococcal cells to one another and to
the glass substrate on which they were grown (Moscoso
et al., 2006). The presence of extracellular proteins in this
matrix was inferred from the biofilm-disaggregating activ-
ity of proteolytic enzymes (Moscoso et al., 2006).
Several authors have reported extracellular DNA
(eDNA) to be an important EPS in pneumococcal biofilms,
based on the dramatic disappearance and inhibited for-
mation of these films following treatment with DNase
(Moscoso et al., 2006; Hall-Stoodley et al., 2008; Carrolo
et al., 2010; Trappetti et al., 2011). However, other
authors have been unable to confirm this (Muñoz-Elías
et al., 2008). As well as being a structural element of the
matrix, other roles have been ascribed to eDNA in differ-
ent biofilms (Fuxman Bass et al., 2011), including: (i)
Received 7 May, 2012; accepted 21 July, 2012. *For correspondence.
E-mail mmoscoso@cib.csic.es; Tel. (+34) 918 373 112; Fax
(+34) 915 360 432.
bs_bs_banner
Environmental Microbiology (2013) 15(2), 502–516 doi:10.1111/j.1462-2920.2012.02853.x
© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd
acting as an adhesin in the initial phase of biofilm forma-
tion (attachment to the substrate) and contributing to
increased mucus viscosity; (ii) being involved in horizontal
gene transfer, thus contributing to the spread of antibiotic
resistance; and (iii) attenuating inflammatory mediator
production and macrophage invasion. Extracellular RNA
seems not be a critical biofilm element (Domenech et al.,
2012).
Controversy exists over whether extracellular polysac-
charide is a component of the S. pneumoniae biofilm
matrix. To our knowledge, only two studies have sug-
gested this to be the case. In one, wheat germ agglutinin
(WGA) was found to stain the biofilm formed by a clinical
pneumococcal isolate of an undefined capsular type
(Donlan et al., 2004). In the other, a cocktail of five lectins
of different specificity (including WGA) stained the bio-
films formed by strains of six different capsular types
(Hall-Stoodley et al., 2008). Using encapsulated S. pneu-
moniae to demonstrate the existence of a polysaccharide
among the EPS is, however, problematic since lectins
also bind the capsular polysaccharide, as demonstrated
more than 30 years ago (Ebisu et al., 1977). However,
Hall-Stoodley and colleagues (2008) reported the biofilm
formed by the non-encapsulated R6 strain to bind lectins,
suggesting the presence of a non-capsular carbohydrate
EPS. Unfortunately, the use of a lectin mixture precluded
any further identification of that putative polysaccharide.
Glycosylated proteins may also be represented among
the EPSs. Protein glycosylation is a common post-
translational modification in eukaryotes, but recently it has
become clear that it also occurs in archaea and bacteria
(Larkin and Imperiali, 2011). Many glycoproteins act as
virulence factors in medically important bacterial patho-
gens, such as the giant (4776 amino acid residues) pneu-
mococcal serine-rich repeat protein (PsrP; SP_1772),
which is synthesized by accessory genes (including
several novel glycosyltransferases) present only in certain
clonal lineages of S. pneumoniae (e.g. the TIGR4 strain
but not the R6 strain) (Tettelin et al., 2001; Löfling et al.,
2011). The PsrP protein, which is encoded by the psrP
gene located in an accessory region, acts as an adhesin
and is involved in biofilm formation, apparently by promot-
ing the formation of large bacterial aggregates (Sanchez
et al., 2010). A number of genes putatively encoding gly-
cosyltransferases lie close to psrP. Recent studies have
shown the first step in the O-linked glycosylation of PsrP
to be the transfer of N-acetylglucosamine (GlcNAc) resi-
dues to the polypeptide, a reaction catalysed by the
GtfA/B complex (Wu et al., 2010); the second might be the
addition of glucose (Glc) to the GlcNAc-modified PsrP in a
reaction catalysed by SP_1768, as recently demonstrated
for the protein Nss, an SP_1768 orthologue of Strepto-
coccus parasanguinis (Zhou et al., 2010). Taken together,
these data indicate that, when working with biofilms syn-
thesized by strains encoding PsrP, such as the common
TIGR4 strain, the glycosylation of this protein may con-
found attempts to identify possible polysaccharides
among the EPS.
The present work provides substantial evidence that
the biofilm formed by S. pneumoniae R6 contains a
matrix made up of eDNA, protein and polysaccharide
components.
Results
Visualization of the pneumococcal biofilm matrix
Despite possible problems of sample shrinkage, LTSEM
has a number of advantages compared with conventional
scanning electron microscopy. For this type of analysis
the latter would require the preparation of samples by
either critical point drying or freeze-drying. These tech-
niques would cause the almost complete disappearance
of EPS from the matrix. In contrast, LTSEM preparations
involve the use of techniques (see Experimental proce-
dures) that allow the structural water of biofilms to be
retained, preserving the natural fabric of the matrix,
including the EPS. Moreover, no chemical fixation is
necessary in LTSEM, eliminating this source of arte-
facts (Sutton et al., 1994). LTSEM examination of the
ultrastructure of S. pneumoniae biofilms previously
revealed the participating cells to be interconnected by
small, thin filaments (Moscoso et al., 2006). The present,
improved electron micrographs of pneumococcal biofilms
growing on polystyrene microtitre dishes confirm the pres-
ence of these intercellular filaments; indeed, they
revealed a conspicuous net-like matrix linking the cells
(Fig. 1).
DNA as a component of EPS
The detection of eDNA (using specific, fluorescent,
double-stranded DNA staining) in S. pneumoniae biofilms
grown for 1 to 8 days in a complex medium has been
previously reported (Hall-Stoodley et al., 2008). However,
the use of stains such as propidium iodide, ethidium
bromide, DAPI (4′,6-diamidino-2-phenylindole) and SYTO
for visualizing this eDNA requires the sensitivity of CLSM
be strongly increased. Consequently, good quality images
are difficult to obtain (Allesen-Holm et al., 2006). Recently,
DDAO [7-hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-
one)] was shown suitable for selectively targeting the
eDNA of biofilms due to its molecular size (which does not
allow the stains to penetrate intact cell membranes) and
because of its enhanced fluorescent properties, which
make it easy to detect by CLSM (Dominiak et al., 2011).
Figure 2 shows the presence of abundant eDNA in a
14-hour-old biofilm formed by S. pneumoniae strain R6
in C medium, detected by in situ staining with DDAO
Composition of the pneumococcal biofilm matrix 503
© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 15, 502–516
and SYTO 9. Although most of the eDNA seems to be
cell-associated, discrete areas of the biofilm that stained
with DDAO but not with SYTO 9 (and vice versa) were
found when the preparations were observed at higher
magnification (Fig. 2D). A DDAO-stained planktonic R6
control culture showed no fluorescence (data not shown).
Evidence of protein–DNA interactions
Earlier work showed that pneumococcal biofilm integrity
is severely impaired by treatment with proteases during
or after biofilm formation (Moscoso et al., 2006; Muñoz-
Elías et al., 2008). It is thus conceivable that extracel-
lular and/or surface-attached proteins form part of the EPS
matrix. In fact, several surface-attached proteins belonging
to the family of the choline-binding proteins (CBPs) (López
and García, 2004), namely LytA N-acetylmuramoyl-L-
alanine amidase (EC 3.5.1.28; NAM-amidase), LytB glu-
cosaminidase, LytC lysozyme (EC 3.2.1.17; muramidase),
CbpA (PspC) adhesin, PcpA putative adhesin and PspA
(pneumococcal surface protein A), are somehow involved
in biofilm formation since mutants that cannot produce just
one of these proteins are less able to produce biofilms
(Moscoso et al., 2006). No such reduction is observed,
however, in Pce phosphocholinesterase or CbpD putative
NAM-amidase/endopeptidase mutants. Reduced biofilm-
formation capacity has also been confirmed for lytC
mutants in two independent studies (Muñoz-Elías et al.,
2008; Domenech, 2012). Although no experimental evi-
dence is available, it is conceivable that one or more of
these (and/or other) surface proteins may interact with
some of the EPS (e.g. eDNA). This hypothesis is here
confirmed. When different CBPs were incubated with
DNAs of different origin, strong binding between DNA and
LytC was shown (by electrophoresis). When purified LytC
was incubated with chromosomal DNAs obtained from
bacteria with different G + C contents [from 28% (Clostrid-
ium acetobutylicum) to 73% (Micrococcus roseus)], the
protein was capable of binding to all DNAs irrespective of
their base composition (Fig. 3A). Moreover, the LytC–DNA
binding reaction took place independently of factors such
as DNA size, strandedness or topological state. However,
binding was completely abolished by the addition of 10 mM
EDTA (but not with 10 mM EGTA). Furthermore, the addi-
Fig. 1. LTSEM image of a S. pneumoniae R6
biofilm formed on the surface of a glass
coverslip. Filamentous material (indicated by
arrows) links the pneumococcal cells to one
another. On the right, a magnification of the
area indicated by the white square is shown.
The reticular nature of the intercellular matrix
can be observed.
Fig. 2. CLSM evidence of eDNA in pneumococcal biofilms. A biofilm of the S. pneumoniae strain R6 was stained with a combination of SYTO
9 (A, green) and DDAO (B, blue). Image (C) is a merger of the two channels. Image (D) is an enlarged vision of the area marked with a
square in (C). Arrows indicate labelling with DDAO alone. Scale bars = 25 mm.
504 M. Domenech, E. García, A. Prieto and M. Moscoso
© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 15, 502–516
tion of 10 mM MgCl2 to an EDTA-inhibited complex
restored complex formation. This indicates the require-
ment of Mg2+ ions for LytC–DNA complex formation
(Fig. 3B). Interestingly, LytC-bound DNA appears to
remain intact since it migrates to its expected position in
gels after its removal from the protein–DNA complex with
proteinase K. In addition, although the binding reaction
was inhibited at NaCl concentrations above 50 mM, and
the LytC–DNA interaction was fully susceptible to SDS
attack (final concentration 1%), the complex could not be
broken by the addition of 1.5 M NaCl, 2% choline chloride,
or both (results not shown). Control experiments per-
formed under varying experimental conditions showed the
optimal conditions for the formation of LytC–DNA com-
plexes to involve a pH of < 9.5 (optimal pH 6–7.5), tem-
peratures of above 16°C (preferably at 37°C), an
LytC : DNA (w/w) ratio of �3 (optimum when ª25), and a
reaction time of at least 1 min (optimal incubation time 1 h)
(Fig. S1). It should be noted that formation of LytC–DNA
led to no detectable inhibition of the catalytic activity of the
muramidase when tested against radioactively labelled
pneumococcal cell walls (results not shown).
Fig. 3. Formation of LytC–DNA complexes.
A. Agarose gel (0.7%) electrophoresis of LytC bound to various chromosomal DNAs (0.1 mg). The G + C content (%) of each DNA is indicated
in parentheses. The binding reactions were performed over 1 h at 37°C in 10 mM Tris-HCl buffer (pH 7.0). A_CIB, Azoarcus sp. CIB; Eco,
E. coli; Aca, A. calcoaceticus; Spn, S. pneumoniae; Cac, Clostridium acetobutylicum; Mro, Micrococcus roseus. The symbols ‘+’ and ‘-’
indicate, respectively, that the incubation mixture included, or did not include, LytC (2.6 mg). S, BstEII-digested l DNA.
B. Binding of LytC to different kinds of DNA, i.e. fragments of HindIII-digested DNA (0.3 mg); the covalently closed circular form of pGL30
(40 ng); pGL30 (40 ng) and pUC19 (40 ng) linearized with SalI or EcoRI respectively; small size fragments of the replicative form (RF) of
fX174 digested with HaeIII (160 ng); single-stranded (ss) M13mp18 DNA (40 ng). The symbols ‘+’ and ‘-’ indicate, respectively, that the
incubation mixture included, or did not include, LytC (2.6 mg). In some cases (PK), the binding mixture was digested with proteinase K
(100 mg ml–1; 15 min, 37°C) after the protein–DNA complex was formed and then analysed by agarose gel electrophoresis. In one experiment
(E), the binding reaction (1 h, 37°C, in Tris-HCl buffer, pH 7.0) was performed in the presence of 10 mM EDTA.
C. Various bacterial-encoded (LytA, LytB, LytC, Pce and CbpF) and phage-encoded (Cpl-1, Skl and Pal) purified CBPs (López and García,
2004) (1 mg each) were incubated with pGL30 (40 ng) and analysed in 0.7% agarose gels. Cpl-7 is a muramidase of phage origin that does
not belong to the CBP family of proteins (García et al., 1990). BSA, bovine serum albumin; HEWL, hen-egg white lysozyme.
D. Inhibitory effect of CbpF on LytC–DNA binding. Plasmid pGL30 (40 ng) was incubated under standard conditions with LytC (1 mg) or LytC
plus CbpF (0.5–2 mg). C, control pGL30.
E. Electron micrographs of pGL30 (left panel) or pGL30-LytC complex (right panel). Scale bars = 200 nm.
F. LytC (1 mg) and the indicated peptides (1 and 5 mg) were incubated with pGL30 (40 ng) and analysed in 0.7% agarose gels.
Composition of the pneumococcal biofilm matrix 505
© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 15, 502–516
Although LytC appears to form the strongest DNA–
protein complexes, LytB and CbpF were also associated
with perceptible delays in DNA migration in agarose gels
(Fig. 3C). These three proteins may therefore be
included in the currently growing list of moonlighting pro-
teins (Henderson and Martin, 2011). Whereas LytB, a
chain-dispersing enzyme involved in cell separation (De
las Rivas et al., 2002), is known to be involved in biofilm
formation (see above), CbpF, an inhibitor of the LytC
autolytic activity (Molina et al., 2009), had not been pre-
viously examined in this respect. Interestingly, when
CbpF was present in sufficient quantity, the LytC–DNA
binding reaction was remarkably inhibited (Fig. 3D).
Electron micrographs of the LytC–DNA complexes con-
firmed the existence of protein–DNA aggregates
(Fig. 3E), which explains their abnormal migration in
agarose gels.
It has been shown that LytC is released to the super-
natant of planktonic pneumococcal cultures in relatively
large amounts (Eldholm et al., 2009). This appears to be
also true for S. pneumoniae biofilms. When biofilms were
incubated with anti-LytC serum, labelling of both the cell
surfaces and the intercellular matrix could be revealed
(Fig. 4B). In addition, colocalization of LytC and eDNA
(stained with DDAO) was observed in the intercellular
spaces (Fig. 4B and D). The formation of LytC–DNA com-
plexes has an ionic basis (in agreement with the depend-
ence of optimal complex formation at a neutral or acidic
pH), and it is strong enough to withstand high ionic
strength environments (i.e. 1.5 M NaCl; see above).
Fig. 4. CLSM evidence of LytC–eDNA complexes in pneumococcal biofilms. A biofilm of S. pneumoniae R6 was stained with a combination of
SYTO 9 (A, green), anti-LytC serum followed by Alexa Fluor 647-labelled goat anti-rabbit IgG (B, red) and DDAO (C, blue). Image (D) is a
merger of the three channels. Image (E) is an enlarged vision of the area marked with a rectangle in (D). Arrows indicate intercellular areas
labelled with anti-LytC and DDAO. Scale bars = 30 mm.
506 M. Domenech, E. García, A. Prieto and M. Moscoso
© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 15, 502–516
However, the mature form of LytC (468 amino acid resi-
dues) has a net negative charge (61 Glu + Asp versus 53
Arg + Lys; predicted pI 6.1) at physiological pH, which
contrasts with that shown by other cationic polymers such
as polylysine and polyarginine, or indeed positively
charged proteins such as avidin and human and hen
egg-white (HEW) lysozymes (pI 10.7), all of which are well
known to interact with DNA (Morpurgo et al., 2004).
Surface plasmon resonance and circular dichroism spec-
troscopy were used to study the interactions between
plasmid pUC19 and positively charged peptides derived
from the human lysozyme. The dissociation constants of
the whole protein and peptides derived from it fell in a
r