Intercellular Nanotubes Mediate Bacterial Communication
Gyanendra P.Dubey1and Sigal Ben-Yehuda1,*
1Department of Microbiology and Molecular Genetics,Institute for Medical Research Israel-Canada(IMRIC),The Hebrew University-Hadassah Medical School,POB12272,The Hebrew University of Jerusalem,91120Jerusalem,Israel
*Correspondence:sigalb@ekmd.huji.ac.il
DOI10.ll.2011.01.015
SUMMARY
Bacteria are known to communicate primarily via
secreted extracellular factors.Here we identify a
previously uncharacterized type of bacterial commu-
nication mediated by nanotubes that bridge neigh-
boring cells.Using Bacillus subtilis as a model
organism,we visualized transfer of cytoplasmic
fluorescent molecules between adjacent cells.Addi-
tionally,by coculturing strains harboring different
antibiotic resistance genes,we demonstrated that
molecular exchange enables cells to transiently
acquire nonhereditary resistance.Furthermore,non-
conjugative plasmids could be transferred from
one cell to another,thereby conferring hereditary
features to recipient cells.Electron microscopy re-
vealed the existence of variously sized tubular
extensions bridging neighboring cells,serving as
a route for exchange of intracellular molecules.
These nanotubes also formed in an interspecies
manner,between B.subtilis and Staphylococcus aureus,and even between B.subtilis and the evolu-tionary distant bacterium Escherichia coli.We
propose that nanotubes represent a major form of
bacterial communication in nature,providing a
network for exchange of cellular molecules within
and between species.
INTRODUCTION
Bacteria in nature display complex multicellular behaviors that enable them to execute sophisticated tasks such as antibiotic production,secretion of virulence factors,bioluminescence, sporulation,and c
ompetence for DNA uptake(Bassler and Losick,2006;Lazazzera,2001;Nealson et al.,1970;Ng and Bassler,2009;Tomasz,1965).Such social activities ultimately benefit the population and are unproductive if performed by a single bacterium.Furthermore,nearly all bacteria are capable of forming a resilient multicellular structure,termed biofilm,comprising cells with different functionalities.Natural biofilms are typically composed of several bacterial species and therefore demand a coordinated gene expression of the various inhabitants(Kolter and Greenberg,2006;Kuchma and O’Toole,2000;Lemon et al.,2008;Straight and Kolter, 2009).
Multicellular activity is achieved by the ability of group members to exchange information in order to synchronize their behavior.Importantly,bacteria are not limited to communicate within their own species but are also capable of sending and receiving messages in an interspecies manner.In both Gram-positive and-negative bacteria,cell-to-cell exchange of informa-tion is mediated primarily by signaling molecules belonging to the general classes of low molecular weight autoinducers and signaling oligopeptides(Bassler and Losick,2006;Fuqua and Greenberg,2002;Lazazzera,2001;Ng and Bassler,2009).In a process known as quorum sensing(QS),the production and detection of these signaling molecules is employed by bacteria to monitor population density and modulate gene expression accordingly(Bassler and Losick,2006;Fuqua and Greenberg, 2002;Lazazzera,2001;Nealson et al.,1970;Ng and Bassler, 2009;Tomasz,1965).
Secretion and detection of small extracellular molecules to the surrounding environment is not the only form of molecular exchange between bacteria.Many Gram-negative bacteria trade information by packaging molecules into extracellular membrane vesicles(MVs).These MVs can travel and fuse with distal cells,thus providing a secure mode for delivering various cellular moieties,including QS molecules,antimicrobial factors, toxins,and DNA(Mashburn-Warren and Whiteley,2006). Furthermore,in some cases neighboring daughter cells have been found to exchange molecular information by establishing intimate cytoplasmic connections.In cyanobacteria,for exam-ple,the movement of small molecules(e.g;sugars and amino acids)within afilament was shown to be mediated by intercel-lular channels.This cytoplasmic sharing enables vital coopera-tive behavior between nitrogenfixing heterocysts and photosyn-thetic nurturing cells(Giddings and Staehelin,1981;Golden and Yoon,2003;Mullineaux et al.,2008).
An additional type of molecular exchange that involves phys-ical interactions between neighboring bacterial cells is conjuga-tion(Lederberg and Tatum,1946).During this process DNA is transferred from a donor to a recipient through a pilus,a tube-like structure that physically connects the participating cells (Madigan et al.,2003).Notably,conjugation represents a key mechanism of horizontal gene transfer in nature(Juhas et al., 2009),whereby hereditary genetic information,rather than nonhereditary molecular signal,is delivered.
590Cell144,590–600,February18,2011ª2011Elsevier Inc.
Tubular conduits between cells that allow exchange of cellular content are typical of multicellular organisms.In plants,neighboring cells are connected by cytoplasmic tubes called plasmodesmata,which provide multiple routes for intercellular transfer of nutrients,signals,proteins and transcripts (Heinlein and Epel,2004;Lucas et al.,2009).In mammalian cells,intercel-lular communication is mediated locally through gap junctions and synapses;however,recent reports demonstrate the exis-tence of a network of intercellular membrane nanotubes enabling long-distance communication.These tunneling nano-tubes have been shown to facilitate intercellular transfer of cyto-plasmic molecules and even organelles and viruses (Belting and Wittrup,2008;Hurtig et al.,2010;Schara et al.,2009).Here we report the identification of analogous nanotubular channels formed among bacterial cells grown on solid surface.We demonstrate that nanotubes connect bacteria of the same and different species,thereby providing an effective conduit for exchange of intracellular content.
RESULTS
Neighboring B.subtilis Cells Exchange Cytoplasmic Constituents
Given the complex intercellular communication required within natural bacterial communities,we reasoned that bacterial cells grown on a solid surface can physically interact in order to estab-lish an effective route for exchange of molecular information.Initially,we examined whether adjacent cells exchange cyto-plasmic GFP molecules.Bacillus subtilis cells (SB444)harboring a chromosomally encoded gfp reporter gene (gfp+)were spotted on solid medium alongside B.subtilis cells (PY79)lacking gfp (gfp À).Cells were allowed to grow for 15hr and then visualized by fluorescence microscopy (Figure 1A).Remarkably,a green fluorescence gradient was observed to emanate from the gfp+cells toward the gfp Àcells,covering a distance of approximately 40m m (Figures 1Ab and 1B).Superimposing the green fluores-cence and the phase contrast image,which demarcates the
cells
Figure 1.Visualizing a Molecular Gradient
between Neighboring B.subtilis Cells
(A)PY79(gfp À)and SB444(gfp +)cells were grown side by side on an LB agar plate at 37 C and visualized by fluo-rescence microscopy 15hr after plating,when small colonies were visible.The dashed line indicates the border between the two populations.(a)Phase contrast image (blue).(b)GFP fluorescence image (green).(c)Overlay of phase and GFP fluorescence images.The scale bar represents 10m m.
(B)Average fluorescence intensity of the gfp Àpopulation (as indicated in Aa)as a function of the distance from the gfp+population.The gfp Àregion was divided into iden-tical sub-regions and the average fluorescence signal was defined in arbitrary units (AU).
(C)Exponentially growing PY79(gfp À)and SB444(gfp+)cells were mixed,plated on an LB agarose pad,and incubated in a temperature controlled chamber at 37 C.Cells were visualized by time-lapse fluorescence micros-copy and phase contrast (blue)and fluorescence (green)images collected at 10
min intervals.Select overlay images are shown from the following time points:(a)t0min,(b)t30min,and (c)t60min.Each pair of colored arrows (red and yellow)indicates different locations where transfer of fluorescent molecules between neighboring cells is increasing over time.Larger fields of the same region are shown in Figure S1.The scale bar represents 1m m.
(D)Average fluorescence intensity of the gfp Àcells as a function of their distance from the gfp+cells at t0min (light blue bars)and at t60min (dark blue bars)of the co-incubation experiment as described in (C)(see Extended Experimental Procedures ).No detectible signal was measured when cells were located beyond 1m m at t0min.Average fluorescence signal is expressed in arbitrary units (AU).Error bars represent standard deviation (SD)of the mean fluorescence signal calculated from at least 40cells located at the indicated distance.Shown is a representa-tive experiment out of three independent biological repeats.
See also Figure S1and Figure S2.
Cell 144,590–600,February 18,2011ª2011Elsevier Inc.591
boundary,demonstrated that thisfluorescence gradient was associated exclusively with the presence of cells(Figure1Ac). The observed cell-associated gradient of the GFP signal con-cords with our pre
mise that cytoplasmic molecular exchange occurs between neighboring cells.However,it remained possi-ble that the gradient was due to migrating gfp+cells.
To further explore this
phenomenon,time-lapse microscopy was carried out to follow the formation of the GFP gradient at a single-cell level.gfp+and gfpÀcells were mixed,applied to an agarose pad,and their growth andfluorescence were moni-tored.Immediately after mixing(t0min),thefluorescence signal was confined to the gfp+cells and no detectablefluorescence was seen in adjacent gfpÀcells(Figure1Ca).However,after 30min,gfpÀcells lying in proximity to gfp+cells acquired a weakfluorescence signal(Figure1Cb).Thefluorescence inten-sity of gfpÀcells increased over time in a manner inversely proportional to their distance from the gfp+:cells residing closer to the gfp+cells acquired morefluorescence than distant ones(Figure1D).Conversely,thefluorescence dis-played by gfp+cells decreased over time(Figure1Cc),suggest-ing that they distribute theirfluorescence among proximal cells. Observing a largerfield highlights that as time progresses,gfpÀcells not directly contacting gfp+cells,also gained afluores-cence signal(Figure S1A available online).To rule out the possi-bility that these observations are a consequence of multiplefluo-rescence exposures,we imaged unexposed regions of the growing cells at thefinal time point,and a similarfluorescence pattern was detected(Figure S1B).Further,when gfp+and gfpÀcells were residing apart from one other,neither the gfpÀcells gained nor the gfp+cells lostfluorescence(Figure S2). The contact-dependent nature of thefluorescence gradient excludes the possibility that the signal derives from cell migration and corroborates that cytoplasmic GFP molecule
s(27kDa)can be transferred from one cell to another in a temporal and spatial manner.Of note,we cannot exclude the possibility that to some extent gfp transcripts are also being traded among the cells.
In a complementary approach,cytoplasmic exchange was examined with calcein,a nongenetically encoded cytoplasmic fluorophore.Calcein is a small nonfluorescent acetoxymethyles-ter(AM)derivative that is sufficiently hydrophobic to traverse cell
membranes.After passage into the cytoplasm,hydrolysis of cal-cein by endogenous esterases gives rise to afluorescent hydro-philic product(623Da)unable to traverse membranes and thus caged within the cytoplasm(Haugland,2005).When B.subtilis cells(PY79)were incubated with calcein-AM(see Extended Experimental Procedures),they rapidly acquired a strongfluo-rescence signal indicating calcein hydrolysis.Next,labeled cells were washed,mixed with nonlabeled cells,and the mixture was placed on an agarose pad and tracked by time-lapse micros-copy.At t0min,only labeled cells exhibited a detectablefluores-cence signal(Figures2A and2A0).After15min,an apparent fluorescence signal was monitored from nonlabeled cells located in the vicinity of labeled ones(Figures2B and2B0). Remarkably,however,by t30min almost all the nonlabeled cells displayed significantfluorescence while thefluorescence from labeled cells decreased(Figures2C and2C0and Fi
gure S3Ba). When unexposed regions of the growing cells were photo-graphed at the latest time point,a similarfluorescence pattern was observed(data not shown).Consistently,labeled cells, located apart from nonlabeled ones,largely maintained their fluorescence signal(Figure S3A).Thus,same as GFP,calcein can be transferred from one cell to another;yet it appears to be delivered more rapidly,suggesting that the speed of transfer inversely correlates with the size of the traversed molecule. Taken together,our results establish that adjacent B.subtilis cells are able to exchange cytoplasmic molecules in a spatially ordered manner.To the best of our knowledge,this is thefirst report of cytoplasmic sharing between neighboring B.subtilis cells.
Intercellular Nanotubes Connect Neighboring
modulateB.subtilis Cells
The exchange of cytoplasmic molecules between adjacent cells raised the notion that intercellular connections,facilitating this process,exist.To examine this possibility,we grew B.subtilis cells(PY79)on solid Luria Bertani(LB)medium and visualized Figure2.Transfer of Calcein between Neighboring B.subtilis Cells Exponentially growing PY79cells were labeled with calcein(see Extended Experimental Procedures).Labeled cells were washed and mixed with non-labeled cells,plated on an
LB agarose pad,and incubated in a temperature controlled chamber at37 C.Cells were visualized by time-lapsefluorescence microscopy and images of phase contrast(blue)andfluorescence(green) were collected at5min intervals.Selectfluorescence(A–C)and corresponding overlay images(A0–C0)are shown from the following time points:(A and A0)t0 min,(B and B0)t15min,and(C and C0)t30min.The scale bar represents1m m. See also Figure S3.
592Cell144,590–600,February18,2011ª2011Elsevier Inc.
them with high-resolution scanning electron microscopy (HR-SEM;see Experimental Procedures ).Surprisingly,tubular protrusions (nanotubes)bridging neighboring cells were plainly visible (Figure 3A).The nanotubes seem to project from the cell surface at different positions in a nonspecific manner.Higher-magnification micrographs clearly evidence a network of inter-cellular connecting nanotubes whereby cells frequently attach to more than one partner simultaneously (Figures 3B and 3C and Figure S4A).Occasionally,we observed the occurrence of branched nanotubes linking together several cells at once (Fig-ure S4B).Notably,these tubes were structurally distinguishable from classical conjugative pili (Figure S4C).Examining cells of an undomesticated B.subtilis strain (3610)with the same procedure revealed a similar or an even enhanced ability to form nanotubes (Figure S4D).The existence of nanotubes was also
detected
Figure 3.Intercellular Nanotubes Form between Neighboring B.subtilis Cells
(A–D)PY79cells were grown to midexponential phase,plated on LB agar,incubated for 6hr at 37 C,and visualized by HR-SEM (see Experi-mental Procedures ).(A)A typical field of B.subtilis cells (315,000).Green arrows indicate intercellular nanotubes connecting neighboring cells.The scale bar represents 5m m.(B)A higher-magnifi-cation image (340,000)of the boxed region in (A).Membrane bulging is indicated by an asterisk (*).The scale bar represents 500nm.(C)An additional field of cells demonstrating the occurrence of a network of intercellular nanotubes (350,000).The scale bar represents 1m m.(D)A field of cells where a cluster of smaller nanotubes (highlighted by a dashed circle)as well as a more pronounced larger tube (indicated by an arrow)are apparent (3100,000).The scale bar represents 500nm.(E)An immuno-EM section of cocultured PY79(gfp À)and SB444(gfp+)cells,stained with anti-GFP and secondary gold-conjugated antibodies (see Extended Experimental Procedures ).Black dots indicate the expression and localization of GFP molecules.The scale bar represents 200nm.(F)A magnification of the dashed square in (E).The arrow highlights the flow of GFP molecules within a tube.The scale bar represents 200nm.
(G)An additional example of an immuno-EM section,showing the localization of a GFP mole-cule within a tube,as indicated by an arrow.The scale bar represents 200nm.
See also Figures S4and Figure S5.
when cells were incubated on minimal medium yet at a lower frequency (Fig-ure S4G).However,nanotubes seem to be absent when cells were grown in liquid medium (data not shown),suggesting that growth on solid medium induces their formation.
Tube dimension appears to vary with the distance between connected cells.Generally,tube length ranged up to 1m m,whereas width ranged approxi-mately from 30to 130nm (e.g.,Figures S4E and S4F).The rela-tively large size of the tubes concords with our assumption that they could easily accommodate the passage of proteins such as
GFP (approximately 40A
˚;[Yang et al.,1996])and even larger cytoplasmic molecules.Closer investigation of the HR-SEM images revealed that beside the large nanotubes,an additional type of smaller nanotubes was visible,though more challenging to detect (Figure 3D).These smaller tubes tended to be clustered co
nnecting nearby cells intimately,appearing to actually ‘‘stitch’’one cell to another.We speculate that these smaller nanotubes are more ubiquitous than the larger ones and are capable of traversing small molecules.
In an alternative approach,intercellular connections were visualized with transmission electron microscopy (TEM),where cells were imaged without employing any contrasting agent
Cell 144,590–600,February 18,2011ª2011Elsevier Inc.593
(see Extended Experimental Procedures ).Consistent with the HR-SEM images,a network of pronounced nanotubes tying one cell to another was readily visible (Figure S5A).Interestingly,higher-magnification analysis of a typical tube appears to indi-cate a structure comprising outer and inner layers,hinting at a multilayered structure (Figures S5B–S5D).Moreover,thin section analysis suggests that the tubes contain cell wall mate-rial,membrane and cytoplasmic content (Figures S5E and S5F).To demonstrate that nanotubes indeed serve as a route for trading cytoplasmic molecules,we carried out immunoelectron microscopy (immuno-EM).gfp+and gfp Àcells were mixed and grown on solid medium.Next,cells were gently fixed,sectioned,incubated with anti-GFP antibodies and then immunostained with gold-conjugated secondary antibodies (see Extended Experimental Procedures )
.Remarkably,the gold particles could be visualized within nanotubes connecting neighboring cells (Figures 3E–3G),corroborating that indeed intercellular nano-tubes serve as a path for molecular exchange.In many images,a GFP gradient was observed whereby a GFP-producing cell containing multiple gold particles was connected to an adjacent cell containing few gold particles (Figure 3E),resembling the phenomenon observed by time-lapse microscopy (Figure 1C).Importantly,when only gfp Àcells were similarly processed,no significant gold signal was detected (data not shown).
Thus,intercellular nanotubes bridge adjacent B.subtilis cells,thereby generating a network of tubular conduits that enable the exchange of cytoplasmic content.
Transient Nonhereditary Resistance to Antibiotics Can Be Acquired from Adjacent Cells
Having established the existence of intercellular nanotube networks,we sought to explore their capability to generate new phenotypes.We anticipated that when two strains,each harboring a different antibiotic resistance gene,are grown together,the exchange of cytoplasmic molecules (proteins and possibly transcripts)through the tubes could yield a population of cells temporarily resistant to both antibiotics in a nonhereditary fashion (Figure 4A).
To test this prediction,we examined the exchange of chloram-phenicol acetyltransferase (Cat)and ery
thromycin resistance methylase (Erm)between two different B.subtilis strains.The Cat protein confers resistance to chloramphenicol (Cm)and the Erm protein confers resistance to lincomycin (Lin).Strains harboring chromosomally encoded resistance to Cm (P1:Cm R )or Lin (P2:Lin R )were spotted separately or in a mixture onto LB agar plate and incubated for 4hr in the absence of any anti-biotic selection.Next,the ability of the strains to grow on selec-tive plates containing Cm,Lin,or both was examined by replica plating (Figure 4B)in order to maintain the spatial arrangement of the cells.Strikingly,the mixed population of P1and P2cells was able to survive on the antibiotic plate containing both Cm and Lin (Figure 4B).To explore the genotype of the survivors,cells growing on the Cm+Lin plate were streaked onto a nonselective LB plate.Then individual colonies from the streak were picked,grown as stripes on LB plates,and their genotype
was
Figure 4.Transient Nonhereditary Pheno-types Can Be Acquired from Neighboring Cells
(A)A schematic model for the transient gain of nonhereditary phenotypes via intercellular nano-tubes.Shown on the left are two B.subtilis cells,each harboring a different antibiotic resistance gene,providing Cm R or Lin R .Genes (colored stripes)are depicted on the chromosomes (olive lines)with colored circles and colored combs indicating their respective proteins and transcripts.Shown on the right is the gain of antibiotic resis-tance by proteins and transcripts passing through intercellular nanotubes in a mixed population.Molecular transfer through the connecting tubes yields a population of cells temporarily resistant to both antibiotics in a nonhereditary fashion.
(B)An antibiotic assay examining the exchange of Cat and Erm proteins (and possibly transcripts)between two different B.subtilis strains.Left:Equal numbers of cells from PY79(WT),SB463(amyE::P hyper-spank -cat-spec )(P1:Cm R ),and GD57(amyE::P hyper-spank -erm-spec )(P2:Lin R )strains were spotted separately on LB agar.In parallel,equal numbers of mixed P1and P2cells (1:1)were spotted similarly.Cells were grown for 4hr at 37 C.Right:Grown cells were replica plated onto the indicated selective plates and finally onto LB.Plates were incubated O/N at 37 C.
(C)An antibiotic assay examining the exchange of Cat and Kan resistance proteins (and possibly transcripts)between two different B.subtilis strains.Left:Equal numbers of cells from PY79(WT),SB463(amyE::P hyper-spank -cat-spec )(P1:Cm R ),and SB513(amyE::P hyper-spank -gfp-kan )(P3:Kan R )strains were spotted separately on LB agar.In parallel,equal numbers of mixed P1and P3cells (1:1)were spotted similarly.Spotted cells were grown for 4hr at 37 C.Right:Grown cells were replica plated onto the indicated selective plates and finally onto LB.Plates were incubated O/N at 37 C.See also Figure S6and Figure S7.
594Cell 144,590–600,February 18,2011ª2011Elsevier Inc.
版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。
发表评论