Quantum dot conjugates for SEM of bacterial communities

Quantum dot conjugates for SEM of bacterial communities

Jay Nadeaua, Randall Mielkeb, and Samuel Clarkea a Department of Biomedical Engineering, McGill University, 3775 Rue University, Montréal, QC b Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109 ABSTRACT Biologically compatible quantum dot (QD) nanoparticles are hybrid inorganic-organic materials with increasing popularity as fluorescent probes for studying biological specimens. QDs have several advantageous optical features compared to fluorescent dyes and they are electron-dense, allowing for correlated fluorescence and electron microscopic imaging. Despite these features, widespread use of QDs as biological probes has generally been limited by the complex chemistry required for their synthesis and the conjugation. In this work, we show that easily prepared quantum dot (QD) probes provide excellent contrast for fluorescent confocal and environmental scanning electron microscopy (ESEM) analysis of pure microbial cultures and microbial communities. Two conjugation strategies were employed in order to specifically target the QDs to bacterial cell surfaces. The first was biotinylation of the bacteria followed by labeling with commercially available QDs incorporating the high-affinity partner for biotin (QD-streptavidin). Second, we designed a novel QD probe for Gram negative bacteria: QD-polymyxin B (PMB), which binds to lipopolysaccharide (LPS) in the Gram negative cell wall. Pure cultures of Gram positive and Gram negative strains were used to illustrate that QDs impart electron density and irradiation stability to the cells, and so no other preparation apart from QD labeling is required. The techniques were then extended to a set of recently characterized microbial communities of perennial cold springs in the Canadian High Arctic, which live in close association with unusual sulfur crystals. Using correlated confocal and and ESEM, we were able to image these organisms in living samples and illustrate their relationship to the minerals. Keywords: Quantum dots, biotin, polymyxin B, ESEM, fluorescence

1. INTRODUCTION Biofilms are organized multicellular systems that allow bacteria to conserve nutrients, resist predation and toxins, and colonize extreme environments in nature and in higher organisms [1]. They are responsible for antibiotic-resistant infections, especially those of implantable medical devices, and for biofouling. Controlling their initiation and growth requires studies of biofilm formation and structure to elucidate the physiological and chemical interactions between the bacteria and their substrates. Such studies will also enable insight into microbial evolution and its implications for adaptation to extreme conditions. Although advanced microscopic techniques have allowed for significant recent advances in this field, imaging of intact, particularly live biofilms remains a significant challenge. Labeling of cells with fluorescent dyes followed by laser-scanning confocal microscopy has been shown in living [2] and fixed [3] biofilms, but resolution and depth of field are limited. Scanning electron microscopy (SEM) has the advantage of increased resolution and the possibility of elemental analysis using energy-dispersive X-ray spectroscopy (EDAX), but standard protocols are difficult to perform, involving complex freezing and drying protocols that often destroy the smallest cells [4]. Environmental SEM (ESEM), performed under 8 mbar of water vapor, allows electron microscopic analysis of live cells but its resolution has typically been limited due to the fragility of the cells which are rapidly destroyed by the electron beam [5]. Metals can improve contrast, but usually require fixation of the cells [6]. In this work we show that quantum dot (QD) conjugates provide excellent labeling for fluorescent confocal and ESEM analysis in pure cultures and biofilms. No other preparation apart from QD staining was required, as the QDs imparted electron density and irradiation stability to the cells, although co-labeling with a fluorescent nucleic acid dye confirmed labeling of bacteria rather than debris.

Scanning Microscopy 2009, edited by Michael T. Postek, Dale E. Newbury, S. Frank Platek, David C. Joy, Proc. of SPIE Vol. 7378, 73781Y · © 2009 SPIE · CCC code: 0277-786X/09/$18 · doi: 10.1117/12.824201

Proc. of SPIE Vol. 7378 73781Y-1

2. MATERIALS AND METHODS 2.1 Chemicals All chemicals were of the highest grade available and were purchased from Sigma-Aldrich Canada unless otherwise noted. Solvents were purchased from Fisher Scientific. Sodium phosphate buffer was used at a final reaction concentration of 50 mM at pH 7.2. Borate buffer was used at a final concentration of 50 mM at pH 7.8. 2.2 Organometallic Synthesis and Surface Functionalization of CdSe/ZnS CdSe QDs were synthesized using a method adapted from Asokan et al. [7], and based on the noncoordinating solvent 1octadecene (ODE). Briefly, 0.05 g of cadmium oxide (CdO) and 2 mL oleic acid (OA) were added to a three-neck flask containing 20 mL of ODE. This mixed was degassed and heated under an atmosphere of nitrogen to 250°C. The mixture turned colorless around 150°C. The selenium (Se) precursor was prepared by mixing 0.016 g of Se with 0.75mL trioctylphoshine (TOP), under inert atmosphere, followed by sonication until the solution became transparent. Once the CdO/OA/ODE mixture reached 250 °C, the heat was turned off, and the Se precursor was injected rapidly using a needled syringe. To make CdSe/ZnS core/shell QDs, a zinc sulfide (ZnS) precursor was prepared as follows: 0.75 mL of TOP was combined with 0.2 mL hexamethyldisilathiane [(TMS)2S] and 0.3 mL dimethylzinc [Zn(CH3)2] under inert atmosphere and diluted to 5 mL with ODE. This precursor was injected over a time course of 5 min during the desired stage of QD growth. Afterwards, the temperature was allowed to drop to 100°C and it was maintained at this temperature for several hours. To purify core or core/shell quantum dots, repeated extractions with equivolume hexanes/methanol were performed. The purified QDs were finally stored in the ODE/hexanes phase until further use. Mercaptopropionic acid (MPA) was used to replace the OA surfactant by a thiol-exchange reaction. 1 mL of concentrated QDs (OD > 5) in ODE/hexanes was diluted to 5 mL using methanol. MPA was added to a final concentration of 1 M and the pH was adjusted to 10 using tetramethylammonium hydroxide pentahydrate (TMAH). This solution was stirred at room temperature for 72 hours. The thiol-modified QDs were separated from excess MPA ligand by precipitation and several washings with ethyl acetate. The QDs were dried at room temperature for several hours under air and resuspended in Milli-Q water (Millipore). 2.3 Conjugation and characterization of QDs QDs were conjugated to streptavidin (SA) using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC)-mediated coupling with a SA:QD molar ratio of 5: 1. Due to the negative charge of the QDs and the positive charge of polymyxin B, QD-PMB was prepared by mixing 30-100 molar equivalents of polymyxin B with carboxylate-terminated QDs in H2O and allowing them to associate electrostatically. Conjugates were characterized by UV-Vis spectroscopy on a Spectra Max Plus plate reader (Molecular Devices) and by gel electrophoresis on 0.5% agarose gels in 0.5X TAE buffer (20mM Tris, 20mM Acetate, 0.5mM EDTA, pH 8.3). QD samples were diluted to 100 nM in 3% glycerol loading buffer immediately before use. A 20 V/cm electric field was applied for 20 seconds followed by 10 V/cm electric field for 15 minutes. Gels were imaged by fluorescence detection using a Model 120 Electrophoresis Analysis System (Kodak, Rochester, NY). 2.4 Growth and labeling of bacterial test strains To determine strain affinity of QD-PMB, we used fifteen test strains (8 Gram negative, 7 Gram positive). Strains were grown overnight in LB medium and seeded 20:1 – 100: 1 the morning of the experiment, then allowed to grow to an optical density at 600 nm of 0.4 to 0.6. Then 0.5 mL of each strain in LB was added to 0.5 mL of at 1 μM QD-PMB and incubated with shaking at room temperature for 3-5 min. Bacteria were sedimented by centrifugation at 5000 x g for 5 min and all of the supernatant removed with a micropipette. The centrifugation speed does not remove QDs. The samples were then resuspended in 0.9% NaCl and the fluorescence read in a SpectraMax Gemini plate reader with excitation at 400 nm and emission at 450-700 nm in 5 nm steps. All samples were also examined by epifluorescence microscopy to ensure that the fluorescent signal arose from QDs bound to cells and not from QD aggregates remaining in the solution. If aggregates were present, the data point was abandoned. Each experiment was repeated 3-5 times per strain. Bacteria-only controls showed no fluorescence when excited at this wavelength. 2.5 Arctic biofilm samples Filaments were collected in July 2007 or July 2008 using sterile tweezers and were subsequently stored in spring water inside sterile polypropylene tubes (~ 1 g biofilm/ 50 mL water). Tubes were transported to McGill University, in

Proc. of SPIE Vol. 7378 73781Y-2

Montreal, Canada, at subzero temperatures (unfrozen) and stored at 4°C under sterile conditions; images shown were taken after < 4 mo of storage. Viability was assessed periodically by observation of motility. For staining with QDs or dyes, a 1-5 mm strip of biofilm was removed from the sample with sterile tweezers and placed inside a 1.5 mL Eppendorf tube, and the tube was filled with spring water. To label with QD-streptavidin, the samples were first biotinylated by adding 50 μg of Fluo Reporter cell surface biotinylation reagent (biotin-XX sulfosuccinimidyl ester, Invitrogen) in DMSO to a 1-5 mm strip of biofilm in a 1.5 mL Eppendorf tube containing spring water. The sample was incubated on ice for 30 min, then rinsed with cold spring water. 10 nM final concentration of QD-streptavidin was added, and the sample was rocked for 10-30 min before imaging. For labeling with QD-PMB, the conjugate was simply added to a final concentration of 500 nM, and the sample was rocked gently in the dark at room temperature for 30 min. When used, the DNA stain SYTO 9 was added to a final concentration of 5 μm. Controls for nonspecific binding were prepared by adding 500 nM of carboxylate-terminated QDs to identical samples. 2.6 Confocal microscopy and ESEM Confocal imaging was performed on live samples on a Zeiss LSM 510 with a 100x PlanApo objective and excitation at 488 nm; emission was in two channels: green (505-535 nm) and red (560 nm longpass). ESEM was performed as described [8]at 20 kV with pressures of 1.4-3.9 torr.

3. RESULTS 3.1 Characterization of QD-PMB Because the highly charged polymyxin B (Fig. 1A) had not been previously conjugated to QDs, we quantified its effects on the QD fluorescence and optimized the conjugation ratios for ease of handling. Greater than about 20-30 PMB molecules per QD led to particle aggregation and loss of water solubility (Fig. 1B). PMB also led to a brightening and slight blue-shift of the QD emission peak, consistent with a positive charge density around the particle (Fig. 1 C, D). PMB itself is not fluorescent when excited at 400 nm (not shown).

A

02.5510204080 160

Ratio PMB:QD 2.8 W

2.62.42.2-

o 6

2-

.t2

1.81.61.41.2-

E

W

.

a

U

1-

-602

20000

-600

18000

-598-. E

_596.E.

.

>, 14000

-

-594 ' 4

-

-592 -590

A

A

A

-588

A

0.8 10

100

Number of Equivalent PMB

586 1000

U

16000 12000

10000

8000 a

0

6000

E

4000

uJ

2000 0

450

500

550

600

650

700

750

Wavelength (nm)

Figure 1. Characterization of QD-PMB. (A) Structure of polymyxin B sulfate. (B) Gel electrophoresis of QD-PMB for PMB:QD ratios of 0 – 160 equivalents. Polarity is indicated and the horizontal line denotes the loading position. (C) Relationship between the concentration of PMB and the relative QD emission intensity (squares) and peak wavelength location (triangles). Data are an average of four experiments and error bars are the standard deviation. (D) Example emission spectra of QDs before (solid line) and after (dashed line) the addition of 160 equivalents of PMB. Note the increase in fluorescent emission and red-shift of the peak wavelength. Black arrows denote samples used for bacteria labeling experiments.

Proc. of SPIE Vol. 7378 73781Y-3

3.2 Labeling of pure cultures Before proceeding to labeling of biofilm samples, QD-PMB was tested on pure cultures for its labeling intensity and specificity. Labeling with QDs imparted both fluorescence and electron density to the cultures. Unlabeled bacteria showed no fluorescence and ESEM imaging at 20 kV showed virtually no contrast (Fig. 2 A). QD-PMB labeled samples showed excellent contrast and stability permitting high-resolution images of strains that labeled strongly, such as E. coli (Fig. 2 B). Other strains, such as B. subtilis, showed virtually no fluorescence or EM contrast after this labeling procedure (Fig. 2 C). To quantify binding, we measured total fluorescent counts from cultures of fifteen test strains, allowing for a semiquantitative estimate of binding. QD-PMB was estimated to label most Gram negative strains 2-3 times more strongly than Gram positive strains, with the exception of very strong binding to S. aureus (Table 1). Unconjugated QDs also showed nonspecific binding to S. aureus; there was no nonspecific binding of unconjugated QDs to any Gram negative strains (Table 1).

A

2 im

10 im Figure 2. QD-PMB as a fluorescent and electron micrographic stain. Left panels, fluorescence images; right panels, ESEM images. For fluorescence, samples were excited by a mercury lamp through a 400-450 nm bandpass filter and imaged through a 500 nm long pass filter over the spectral range 500-750 nm. (A) Pure culture of E. coli, unstained. (B) E. coli incubated with 500 nM QD-PMB. (C) B. subtilis with QD-PMB. Note the lack of contrast in the ESEM image, making the bacteria appear dark.

Proc. of SPIE Vol. 7378 73781Y-4

Gram positive strains

QDOnly

QD-PMB

Gram negative strains

QDOnly

QD-PMB

Rhodococcus MD-2

sp.

ND

30±10

Halomonas sp. NP35

ND

70±20

Athrobacter Eur3 AL25

sp.

ND

30±10

Psychrobacter sp. NP42

ND

80±10

Paenibacillus Eur3 1.8

sp.

ND

20±6

Stenotrophomonas sp. Eur3 AL.5

ND

50±6

Staphylococcus aureus

50

100±10

Pseudomonas sp. Eur3 AL.16

ND

60±8

Staphylococcus epidermidis

10

30±6

Pseudomonas aeruginosa

ND

80±10

Bacillus subtilis

50

50±8

E. coli

ND

60±10

Serratia marcescens

ND

60±10

Proteus vulgaris

ND

80±6

Enterobacter aerogenes

ND

90±10

Table 1. Binding of QD-PMB and unconjugated QDs to environmental and laboratory strains. All values are fluorescent counts, means ± standard error for at least 3 experiments, and values were taken after washing. ND = not detectable.

3.3 Confocal imaging of labeled streamer biofilms Recent studies have identified nearly monogenetic biofilms in cold saline springs at Gypsum Hill, in the Canadian High Arctic. The location and setting [9], community DNA analyses [10] [11], and results of staining with conventional fluorescent dyes [12] have all been reported. The biofilms consist of tightly packed sulfur minerals and a high density (~1011/g wet weight) of Gram negative sulfur-oxidizing bacteria of the genus Thiomicrospira. The function and origin of the sulfur crystals is unknown, as is the possible relationship between the microorganisms and the other minerals in the springs, notably gypsum. Confocal imaging of these biofilms presented some difficulties for interpretation. The sulfur minerals present in the biofilm showed significant green autofluorescence in a variety of shapes and sizes (Fig. 3 A). After labeling with QDs, patterns suggestive of bacteria on minerals emerged but were not conclusive. Biotinylation with QD-streptavidin showed discrete areas of intense labeling (Fig. 3 B). This was similar to what was seen with SYTO9 staining, although it was difficult to tell whether the bacteria adhered to the minerals or whether they mostly floated free (Fig. 3 C). QD-PMB labeling was more intense than that of QD-SA; large areas of labeled minerals were seen, but again it was difficult to ascertain whether this was because of adherent cells (Fig. 3 D). Polymyxin B-induced toxicity could be observed as rounding of the cells (Fig. 3 D, right panel).

Proc. of SPIE Vol. 7378 73781Y-5

'k 4

S

'

B

01

F.

0

Figure 3. Confocal images of labeled and unlabeled streamer biofilms. Red and green channels defined in text; yellow indicates overlap. (A) Unlabeled samples, showing different morphologies of autofluorescent sulfur minerals. The structures on the right were particularly abundant and embedded throughout the biofilm. (B) Samples labeled with QDstreptavidin, showing discrete areas of labeling which may be individual cells or clusters of cells. (C) Traditional SYTO 9 staining of the biofilm showing abundant organisms over the mineral background. (D) Staining with SYTO 9 and QD-PMB. >99.9% of organisms stained with SYTO 9 also stained with QD-PMB (> 105 cells studied). Note the rounding of the cells due to the antibiotic effect of the polymyxin B, which disrupts cell permeability.

Proc. of SPIE Vol. 7378 73781Y-6

3.4 ESEM imaging of labeled and unlabeled biofilms ESEM was found to be an excellent technique for imaging the biofilm samples, as it allowed for identification of mineral elemental composition and structure (Fig. 4). Most of the 10-20 μm long, cigar-shaped structures as seen in Fig. 3 A were found to be elemental sulfur crystals. Also seen in abundance were gypsum and calcium carbonate. At high magnification, clefts or indentations suggestive of bacterial cells were seen on some of the minerals, although they could not be identified with certainty (Fig. 4, right panel).

Ar 'W.JI

2m

r 10 urn

kPeW I

I- \ Figure 4. Unlabeled samples of streamer biofilms imaged by ESEM. The left, middle, and right panels show increased levels of magnification of the same field. “G” indicates gypsum, “C” calcium carbonate, “S” elemental sulfur, and “Al” aluminum silicate, identified from morphology and EDAX. On the highest magnification panel, clefts or outlines suggestive of bacteria can be seen on three different types of minerals (arrows indicate possible cells).

QD labeling of the cell surface allowed the bacteria to be imaged in ESEM mode using higher voltages (20 kV) and exposure times than is typically possible without destroying cells. Using QD-streptavidin resulted in patchy outlining of clusters of bacteria on the sulfur minerals (Fig. 5 A). Bacteria could also be identified on the smooth surfaces of elemental sulfur crystals (Fig. 5 B). Samples of spring water without macroscopic crystal formations, which appeared entirely dark when unlabeled, showed clusters of Thiomicrospira cells (Fig. 5 C).

Proc. of SPIE Vol. 7378 73781Y-7

I

ii __J._I

5 im

r

Figure 5. Streamer biofilms biotinylated and labeled with QD-streptavidin. (A) Labeled bacteria on elemental sulfur. (B) Sulfur crystals showing outlines of bacterial cells. (C) Bacterial cells observed in the liquid surrounding the biofilm. The background features are the aluminum stub used to mount the ESEM samples.

Using QD-PMB allowed for even greater resolution of the QDs on sulfur crystals (Fig. 6). Large populations of cells were found only associated with sulfur, not gypsum or calcium carbonate.

Proc. of SPIE Vol. 7378 73781Y-8

10 im

-pI11II_

L. . \ 0

C

,r.

1 VF

A

-

Figure 6. QD-PMB labeling on sulfur crystals. The left, middle, and right panels are increasing magnifications of the same view. Three distinct crystal shapes are shown in (A), (B), and (C), indicating the relationship between the weathered surfaces of the mineral and the microbial colonization.

4. DISCUSSION Semiconductor QDs greatly increase the contrast and stability of bacteria under the electron beam of ESEM. In this work we introduced a new, easily-prepared QD conjugate, QD-PMB, for labeling of Gram negative organisms. If the morphological changes in the bacteria due to polymyxin B exposure are undesirable, a peptide with lipopolysaccharidebinding ability but no toxicity could be used, such as the stereoisomer of polymyxin[13]. However, osmotic disruption did not concern us here because of the low levels of water used in ESEM; rounding was only seen under confocal microscopy.

5. CONCLUSIONS Although the electron density of CdSe is less than that of Au, it is sufficient for SEM imaging and has the added bonus of bright, multicolor fluorescence. This work provides a general framework for the creation of QD-based probes for fluorescence and electron microscopy.

Proc. of SPIE Vol. 7378 73781Y-9

REFERENCES

[1] Costerton, J. W., Cheng, K. J., Geesey, G. G., Ladd, T. I., Nickel, J. C., et al., "Bacterial biofilms in nature and disease," Annu Rev Microbiol 41, 435-464 (1987). [2] Lawrence, J. R., Korber, D. R., Hoyle, B. D., Costerton, J. W.&Caldwell, D. E., "Optical sectioning of microbial biofilms," J Bacteriol 173, 6558-6567 (1991). [3] Decho, A. W.&Kawaguchi, T., "Confocal imaging of in situ natural microbial communities and their extracellular polymeric secretions using Nanoplast resin," Biotechniques 27, 1246-1252 (1999). [4] Douglas, S.&Douglas, D. D., "Environmental scanning electron microscopy studies of colloidal sulfur deposition in a natural microbial community from a cold sulfide spring near Ancaster, Ontario, Canada," Geomicrobiology Journal 17, 275-289 (2000). [5] Walker, J. T., Verran, J., Boyd, R. D.&Percival, S., "Microscopy methods to investigate structure of potable water biofilms," Methods Enzymol 337, 243-255 (2001). [6] Priester, J. H., Horst, A. M., Van de Werfhorst, L. C., Saleta, J. L., Mertes, L. A., et al., "Enhanced visualization of microbial biofilms by staining and environmental scanning electron microscopy," J Microbiol Methods 68, 577-587 (2007). [7] Asokan, S., Krueger, K. M., Alkhawaldeh, A., Carreon, A. R., Mu, Z. Z., et al., "The use of heat transfer fluids in the synthesis of high-quality CdSe quantum dots, core/shell quantum dots, and quantum rods," Nanotechnology 16, 2000-2011 (2005). [8] Douglas, S., Abbey, W., Mielke, R. E., Conrad, P.&Kanik, I., "Textural and mineralogical biosignatures in an unusual microbialite from Death Valley, California," Icarus 193, 620-636 (2008). [9] Omelon, C. R., Pollard, W. H.&Andersen, D. T., "A geochemical evaluation of perennial spring activity and associated mineral precipitates at Expedition Fjord, Axel Heiberg Island, Canadian High Arctic," Applied Geochemistry 21, 1-15 (2006). [10] Perreault, N. N., Andersen, D. T., Pollard, W. H., Greer, C. W.&Whyte, L. G., "Characterization of the prokaryotic diversity in cold saline perennial springs of the Canadian high Arctic," Applied and Environmental Microbiology 73, 1532-1543 (2007). [11] Niederberger, T. D., Perreault, N. N., Lawrence, J. R., Nadeau, J. L., Mielke, R. E., et al., "Novel sulfuroxidizing streamers thriving in perennial cold saline springs of the Canadian high Arctic," Environ Microbiol. 11, 616629 (2009). [12] Nadeau, J. L., Perreault, N. N., Niederberger, T. D., Whyte, L. G., Sun, H. J., et al., "Fluorescence microscopy as a tool for in situ life detection," Astrobiology 8, 859-874 (2008). [13] Tsubery, H., Ofek, I., Cohen, S.&Fridkin, M., "The functional association of polymyxin B with bacterial lipopolysaccharide is stereospecific: studies on polymyxin B nonapeptide," Biochemistry 39, 11837-11844 (2000).

Proc. of SPIE Vol. 7378 73781Y-10