Home
Search
Collections
Journals
About
Contact us
My IOPscience
In situ probing the interior of single bacterial cells at nanometer scale
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Nanotechnology 25 415101 (http://iopscience.iop.org/0957-4484/25/41/415101) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 155.97.178.73 This content was downloaded on 15/10/2014 at 13:36
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 25 (2014) 415101 (13pp)
doi:10.1088/0957-4484/25/41/415101
In situ probing the interior of single bacterial cells at nanometer scale Boyin Liu1, Md Hemayet Uddin2, Tuck Wah Ng1, David L Paterson3, Tony Velkov4, Jian Li4 and Jing Fu1 1
Department of Mechanical and Aerospace Engineering, Monash University, Clayton, VIC 3800, Australia Melbourne Centre for Nanofabrication, Clayton, VIC 3800, Australia 3 Centre for Clinical Research, University of Queensland, Brisbane, QLD 4072, Australia 4 Monash Institute of Pharmaceutical Sciences, Monash University, Parkville, VIC 3052, Australia 2
E-mail:
[email protected] and
[email protected] Received 6 May 2014, revised 4 August 2014 Accepted for publication 15 August 2014 Published 26 September 2014 Abstract
We report a novel approach to probe the interior of single bacterial cells at nanometre resolution by combining focused ion beam (FIB) and atomic force microscopy (AFM). After removing layers of pre-defined thickness in the order of 100 nm on the target bacterial cells with FIB milling, AFM of different modes can be employed to probe the cellular interior under both ambient and aqueous environments. Our initial investigations focused on the surface topology induced by FIB milling and the hydration effects on AFM measurements, followed by assessment of the sample protocols. With fine-tuning of the process parameters, in situ AFM probing beneath the bacterial cell wall was achieved for the first time. We further demonstrate the proposed method by performing a spatial mapping of intracellular elasticity and chemistry of the multi-drug resistant strain Klebsiella pneumoniae cells prior to and after it was exposed to the ‘last-line’ antibiotic polymyxin B. Our results revealed increased stiffness occurring in both surface and interior regions of the treated cells, suggesting loss of integrity of the outer membrane from polymyxin treatments. In addition, the hydrophobicity measurement using a functionalized AFM tip was able to highlight the evident hydrophobic portion of the cell such as the regions containing cell membrane. We expect that the proposed FIB–AFM platform will help in gaining deeper insights of bacteria–drug interactions to develop potential strategies for combating multi-drug resistance. S Online supplementary data available from stacks.iop.org/NANO/25/415101/mmedia Keywords: single cell analysis, focused ion beam, atomic force microscopy, antibiotic resistance, elastic modulus (Some figures may appear in colour only in the online journal) 1. Introduction
recently achieved resolution is in the order of 100 nm [4]. By overcoming the light diffraction limit, electron microscopy (EM) including transmission electron microscopy/scanning electron microscope (TEM/SEM) and scanning probe microscopy (e.g. atomic force microscopy (AFM)) have become the popular approaches to investigate bacterial architectures [5–7]. For TEM, however, the sample thickness has to be less than several hundred nanometres to allow sufficient electrons to penetrate and to be collected by the detectors [8]. Hence,
It is of fundamental interest in life sciences to understand the hierarchical organization of molecules and organelles at the subcellular level, as well as their interactions to extracellular stimuli [1]. Various microscopic and spectroscopic methods have been developed to investigate cell structural and compositional architectures with ever-increasing resolutions [2, 3]. High-resolution imaging of single bacterial cells has been a challenge for light microscopy, even though the 0957-4484/14/415101+13$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 1. (a) Schematic diagram of the proposed slice-and-probe approach by combining AFM and FIB for probing the surface and interior of single bacterial cells. (b) SEM image of a single bacterial cell (Klebsiella pneumoniae) sliced by FIB (Ga+, 30 keV). (c) The corresponding height map of interior obtained by AFM.
confined to the sample surface so far [16, 17]. Since the cell interior has not been interrogated yet, there is a wealth of information can be retrieved through AFM measurements carried out at cell interior. Focused ion beam (FIB), originally developed for the semiconductor sector, has recently become an important tool for precisely milling single cells for imaging [18–20]. A recent study has demonstrated that FIB-based tomography can also be applied to reconstruct the 3D structures of single bacterial cells [21]. Maintained in cryogenic environment, bacterial cells can be sliced by cryo-FIB in situ with minimal damage, directly followed by cryo-TEM imaging [22]. In addition to structural studies, FIB milling also enables cellular interior to be accessible for various chemical imaging approaches, including secondary ion mass spectrometry
resin embedded or cryo-fixed cells are typically sectioned by microtome or cryo-microtome prior to imaging [9]. It relies on highly sophisticated machinery, and the thin sectioning of samples in a frozen state may result in artefacts such as compression. AFM offers a unique opportunity to probe living cells at nanometre resolution and to acquire information such as surface topology, mechanical moduli and chemistry through functionalized tips [10–15]. The most notable characteristic of AFM is its ability to probe samples under aqueous conditions, hence allowing for the interrogation of biological samples in their near native state, providing a new dimension in the characterization of the biomechanical properties of cells and tissues, sharp cantilever tips are employed to indent into the sample surface to elucidate its stiffness or elastic modulus, and hence the investigations are 2
Nanotechnology 25 (2014) 415101
B Liu et al
(SIMS) [23], atom probe tomography [24] and synchrotron xray [25]. In the current study, a novel strategy combining FIB and AFM was developed to investigate the surface, and more importantly, the interior of single bacterial cells as illustrated in figure 1(a). After removing layers of pre-defined thickness on the target bacterial cells by FIB milling (figure 1(b)), the target cells were subject to AFM probing under both ambient and aqueous environments (figure 1(c)). Based on the developed protocols, we studied the multi-drug resistant strain Klebsiella pneumoniae (K. pneumoniae) prior to and after it was exposed to the ‘last-line’ lipopeptide antibiotic polymyxin B, and successfully acquired the spatial mapping of intracellular elasticity and chemistry.
Both sample alignment and milling processes were monitored with SEM imaging at accelerating voltage of 2–3 kV. Cells were milled at three different ion beam incidence angles to compare the result of the sectioned surface (figure 2(a)). For milling at both 0° and 30°, a rectangular pattern was applied to the target cell to remove an approximately 100 nm thick layer from the top surface of the target cell. The depth of milling was calibrated and adjusted, and yield was approximately five times of milling Si3N4. For milling at 60° incident angle, the cells were sliced by half with the cleansing pattern of the ion beam, and the milling depth was set to 1 μm to ensure ion beam could section the whole cell surface. Milling depth was controlled at 100–200 nm to expose the interior of bacterial cells. This allowed the exposure of the cytoplasm and nucleoid regions for probing. The bacterial cells for chemical mapping were milled at a grazing angle where the incidence ions were parallel to the substrate surface with a customized pre-tilted sample holder. Prior to the cell sectioning, 1 μm deep triangular markers with the side length ranging between 20 and 50 μm were also milled by high ion beam current of 6.5 nA to mark the positions of sliced target cells. The size of the triangular groove also allowed locating the approximate position of the processed cells under optical microscope in the subsequent AFM studies. All milling tasks on bacterial cells were performed with 30 keV accelerating voltage with a low ion current of 9.7 pA.
2. Materials and methods 2.1. Cell growth and harvesting
Bacterial sample K. pneumoniae ATCC 13883 was subcultured from frozen stock onto nutrient agar plates and incubated overnight at 37 °C. One colony was inoculated into cation-adjusted Mueller-Hinton broth (CAMHB) and incubated in a shaking waterbath at 37 °C for 16–20 h. Log phase cultures were prepared for each treatment by inoculating 200 μL overnight culture into 19.8 mL CAMHB and left in a 37 °C shaking waterbath for 2–3 h until it reached ∼108 CFU mL−1. For experiments on the antibiotic-treated cells, the samples were treated with 2 mg L−1 polymyxin B for 24 h. Then both the treated and untreated samples underwent three washes with Milli-Q water and centrifuged at 3220g for 10 min at room temperature. After treatment with 1 mL of 2.5% glutaraldehyde in phosphate buffered saline for 20 min, the samples were washed two times and re-suspended in Milli-Q water. A recent approach for improved cell collection at specific regions was applied in this process [26].
2.3. AFM topography imaging and surface roughness
The sectioned cells were then transferred to an AFM instrument (Dimension Icon, Bruker, Billerica, USA), and a triangular cantilever with nominal frequency 70 kHz and nominal spring constant 0.4 N m−1 (SCANASYST-AIR, Bruker, Billerica, USA) was applied for scanning the cells. The sectioned cells were firstly estimated under the optical microscope equipped on AFM by locating the triangular markers. The first scan was then performed using the ScanAsyst mode on a 30 × 30 μm2 area to locate the single cells followed by zooming in to a 2 × 2–5 × 5 μm2 area for detailed scanning with a typical scan rate 1 Hz and 512 scanning lines. During scanning, the parameters such as integral gain and deflection set point were adjusted to obtain a better fit between trace and retrace signals. Inbuilt nanoscope analysis (v1.40r, Bruker, Billerica, USA) software was employed for calculating average surface roughness values after a first order flattening procedure. For each FIB milled surface, two 200 × 200 nm2 regions were randomly selected for roughness measurements, and at least three different cells were chosen in each case for studying the angular effects. In addition, regions containing random large porous structures of several hundred nanometres in size were excluded. The roughness data were represented by mean values (Ra) and root mean square values (Rq).
2.2. FIB setup for micromachining
A 20 μL bacterial suspension was placed on a freshly cleaved mica disk of 10 mm diameter. Dehydration was performed either by air drying in a fume hood for 30 min, or vacuum freeze drying after plunge freezing in liquid nitrogen. The mica disk surface was rinsed with deionized water to remove the loosely attached bacteria. Before FIB milling, the mica disk was firstly inspected under optical microscope to ensure a single layer of bacterial cell was smeared on. An approximately 10 nm thin layer of gold was coated with a sputter coater (Emitech K550X, Quorum Technologies Ltd, Lewes, UK) on the mica disk to reduce the charging effect in FIB/ SEM. The samples were then mounted on an aluminium SEM stub and transferred to an FIB/SEM instrument (Helios Nanolab 600, FEI Company, Hilboro, USA) equipped with a gallium liquid metal ion source. The vacuum chamber was pumped down below 10−3 Pa, and the sample stage was tilted and maintained at the eucentric point for milling and imaging. 3
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 2. (a) Bacterial cells micromachining by FIB at different FIB incidence angles from surface normal: 0° and 60°, and the corresponding AFM height images. (b) Quantitative analysis of surface roughness of bacterial interior machined at different beam incident angles (0°, 30° and 60°; standard deviations are shown in error bars). Ra: mean roughness value, Rq: root mean square value.
∼35°. Cantilevers were cleaned placing them in an UV/ozone chamber for about 15 min before use. The detector sensitivity (S) was determined by pressing the cantilever on a clean mica surface, which is considered as an infinite stiff sample in respect of a low spring constant cantilever. The actual spring constant for each cantilever was measured from thermal fluctuations [27]. Cantilever tip was engaged on the top of the bacterial surface using an optical microscope associated with the AFM. A few first scans (2 Hz) were conducted on a 2 × 2–5 × 5 μm2 area and 256 × 256 pixels to locate the region of interest on the sample surface. Once the centre region was located, the surface topography of a 200 × 200–500 × 500 nm2 area was imaged by contact mode AFM at a scanning rate of ∼1 Hz and 512 × 512 pixels. A cyclic loading–unloading force-mapping was carried out on the same site and area in a 16 × 16 curves grids yielding 256 force–displacement curves with at a constant
2.4. AFM force mapping
As the surface roughness results indicated that the ion beam sectioning at higher incident angle (60°) produced a smoother surface (lower surface roughness values) as presented in figure 2(b), the subsequent force mapping experiments were conducted only on the cells sectioned by ion beam milling at 60° incident angle. To avoid artefacts from metal coating, force curves of target cells without sputter coating were also measured and compared. All force-mapping experiments were carried out using a second AFM instrument (JPK NanoWizard II, JPK Instruments AG, Berlin, Germany) both in air and Milli-Q water at the room temperature of 25 °C. V-shaped cantilevers (SNL10-D, Bruker, Billerica, USA) with nominal spring constant 0.06 N m−1 and quadratic Si3N4 pyramid shaped tip were employed. The nominal radius of the silicon nitride pyramidal tip is ∼10 nm and half-open angle of the pyramidal face is 4
Nanotechnology 25 (2014) 415101
B Liu et al
maximum applied load (∼10 nN) and at a vertical frequency of 2 Hz. Force-mapping measurements were carried out at three different sites at least several hundreds of nanometers apart from each other for each control or treated bacteria in both dehydrated and hydrated conditions. After imaging and force mapping in air, the cantilever was withdrawn from the sample surface. Then the sample surface was covered with added Milli-Q water. The AFM head was lowered to immerse the cantilever into water, followed by 30 min waiting time for the system to reach equilibrium. The detector sensitivity (S) was determined again in water before conducting force mapping on bacterial surface by pressing the cantilever on a clean mica surface immersed in water. The force mapping in water was carried out in a similar manner as described for force mapping in air.
assuming that the sample has infinite thickness compared to indentation depth, the deformation is elastic and there is no adhesion between tip and sample. In this study, the adhesion forces when cantilever approaching the sample surface were always negligible or minimal, and the measured indentation depth is in the order of tens of nanometres in comparison to sample thickness of microns. Also considering the high displacement rate at 2 Hz, it was reasonable to choose Hertz model for model fitting in this study. For a square pyramidal tip, Young’s modulus (E) can be calculated as suggested by [37, 38] F=
AFM force-mapping raw data was processed by JPK data processing software (spm-4.2.59, JPK Instruments AG, Berlin, Germany). All the curve fitting operations were conducted on approach curves as the adhesion effect in the retract curve might affect the determination of the contact point. Curve smoothing was first performed in order to optimize the procedure to find the contact point. Baseline correction was then performed to remove cantilever offset and tilt by matching the baseline to the state of zero of cantilever deflection. As AFM does not measure absolute distances, the contact point where the cantilever tip reached the sample surface was derived from the position of the piezoelectric scanner by built-in algorithms [28, 29]. AFM force-mapping records the cantilever deflection (detector photodiode voltage) as a function of z-piezo displacement (z, in nm). The cantilever deflection (dV, in volts) was converted to force (F) by multiplying the change in cantilever deflection (ΔdV, V), the AFM photodiode detector sensitivity (S, nm V−1), and the measured cantilever spring constant (k, nN nm−1):
)
δ 2,
(3)
2.6. AFM chemical force spectroscopy
For chemical force spectroscopy, the tip–surface interactions were first calibrated using chemically functionalized substrate surface. Silicon wafers (3 × 3 mm2) were firstly washed by absolute ethanol then cleaned in a UV–ozone cleaning system (UV-1, SAMCO, Kyoto, Japan) for 10 min. The silicon substrates were further cleaned by argon plasma in an electron beam evaporator, followed by coating with 5 nm chromium adhesive layer and 30 nm gold layer (Nanochrome II, Intlvac, Niagara Falls, USA). The coated silicon substrates were immersed in 1 mM octadecanethiol in ethanol solution for 18 h to assemble monolayers of octadecanethiol molecules on the gold surface [41]. The AFM cantilever tips were chemically functionalized in the same manner as described above and used to probe the functionalized substrate. A similar procedure with non-functionalized cantilever tip and nonfunctionalized silicon substrate was done in parallel to compare the chemical forces. The interior of bacterial cells in the control group was probed by functionalized AFM cantilever tips with 2 nN vertical force in contact mode. All the experiments on chemical force spectroscopy were conducted in aqueous condition.
(1)
The tip–sample separation distance (D) can be calculated by subtracting cantilever deflection in nm (SΔdV) from the change in z-piezo displacement (Δz): D (nm) = Δz − SΔd v.
(
4 1 − ν2
where α is the half-angle of the pyramidal face (∼35°) of the tip and ν is Poisson’s ratio of the elastic material [39, 40]. Bacterial cells were assumed as an incompressible material with ν = 0.5. As shown in equation (3), Young’s modulus is inversely proportional to the square of indentation depth (δ2) at a given force, and the modulus values can be calculated from the slopes of the F verses δ2 curves. The AFM forcemapping raw data were analyzed using JPK data processing software, and modulus histograms were presented. Statistical analyses including probability distribution fitting and ANOVA were performed on Minitab (v16, Minitab, State College, USA).
2.5. Data acquisition, processing and statistical analysis
F (nN) = kSΔd v.
3E tan α
(2)
The indentation depth, δ was determined by searching the tip–bacteria cell surface contact point on the F verses D curve, where the force begins to level off from the horizontal line representing the average force (F ∼ 0) when the cantilever tip is off contact. Several mechanical models have been used in the analysis of the biomechanical properties, including Oliver–Pharr method [30], Hertizian contact mechanics [31], Johnson– Kendall–Roberts (JKR) [32] and Derjaguin–Muller–Toporov (DMT) [33]. Among them, Hertizian contact mechanics method has been extensively used to analyze biomechanical properties measured using AFM [34–36]. It has been applied on the approaching curve to obtain Young’s modulus
3. Results and discussions 3.1. Cell sectioned at different ion beam incidence angles
During FIB milling, the energetic ions, typically of gallium, bombard the sample surface and sputter away the surface 5
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 3. (a) A bacterial cell (K. pneumoniae) sliced by FIB at 60° incidence angle was examined by AFM indentation with increasing
vertical forces. The corresponding Young’s modulus maps were presented on the left. (b) The trends of median modulus and average indentation depth of the measured region (250 nm × 250 nm2) on different cantilver loads. (c) Hysteresis feature observed in liquid medium to demonstrate the viscoelastic property of a bacterial cell.
figure 2(b), and the Ra and Rq values of the cell surface milled at 60° incident angle were less than 50% compared to those of the surface milled at 0° and 30°. The following Young’s modulus measurements on cell interior were then performed based on FIB sectioning with incident angle at least 60°, to minimize the effects of surface topology during AFM measurements.
atoms and molecules. The interactions between the energetic gallium ions and the sample surface may produce various artefacts such as ion implantation [42, 43]. From the AFM images (figure 2(a)) of cells interior, it is evident that the morphologies of the resultant surface varied with milling at different incidence angles. The cell interior surface processed with 0° and 30° incident ion beam was relatively rough, with large pores observed. Some sputtered cellular materials were possibly trapped with the milled geometry known as redeposition, such that both redeposition and ion damage may contribute to the observed surface morphology. In contrast, the sectioned surface with cleansing pattern (grazing angle close to 90°) or 60° incident angle milling showed significantly smoother patterns. Due to the nature of the bacterial cell surface, the roughness values obtained from unsectioned cell surface were the highest. Detailed numerical comparisons are presented in
3.2. Effect of cantilever load and selection of force threshold
Due to the heterogeneity of biological materials, Young’s modulus is expected to vary depending on the indentation depth. Gram-negative bacterial cells such as K. pneumoniae studied here contain an outer membrane envelope and primarily cytoplasmic region. Preliminary data of Young’s modulus were obtained from a 250 × 250 nm2 area with varied loads (2, 4, 6, 8 and 10 nN) under ambient environments. The results of 16 × 16 force curves were collected for each 6
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 4. Force mapping experiments were performed on three arbitrary regions of each target cell (a)–(h). The results were corrected and
fitted by the Hertz model to obtain the elastic modulus of each contact point, followed by conversion to a elastic modulus map (scale bar on the right). AFM height maps of each cell were presented as background.
cantilever load. Based on the results in figure 3(a), the softer region became larger as the cantilever load increased, implying the nature of heterogeneity in the bacterial cell. Figure 3(b) displays the median values of calculated Young’s modulus along with the corresponding indentation depth. It is evident that the median values decreased significantly with deeper indentation, suggesting variation of Young’s modulus at different depth of intracellular regions. In addition, hysteresis was also observed in the data points under liquid operation, which indicated some viscoelastic behaviour of the hydrated sectioned cell surface (figure 3(c)). This result is due to the fact that the sectioned cell surface is more viscous in liquid with a long relaxation time during deformation. As the Hertz model is only valid at small indentation, it was important to limit the indentation depth within 10% of cell height [44]. Moreover, increased indentations could also affect the accuracy in determining the elastic modulus [45, 46]. A cantilever deflection force should also be determined to retract the cantilever from the surface when the deflection threshold was reached. In practice, the load of cantilevers had to reach at least 2 nN to generate distinctive signal in contact mode. By collecting all the pilot force curves acquired from preliminary studies described in figure 3, the maximum indentation depths for each cantilever loading were limited to 20 nm. With the same scanner extend speed (2 μm s−1) and same sample rate (2048 data points per force ramp), a higher deflection threshold was considered to provide more data points after tip–sample contact point for curve fitting, as well as to mitigate the issue of soft cantilever drifting. In the current study, 10 nN was eventually chosen as the optimal force threshold for the following elastic modulus measurements to increase the data size without compromising the measurement accuracy.
3.3. Modulus map on bacterial surface and interior
Fitted Young’s moduli corresponding to the extended curve associated with each pixel were presented as maps in figures 4(a)–(h), and overlayed with the corresponding AFM height images (grayscale) to show the actual locations of the moduli measurements. The range of values was set to 0–10 MPa (white color representing moduli equal to or larger than 10 MPa) in order to provide an optimal contrast for comparison. It has been found that, overall, larger moduli were present in the treated cells (figures 4(c) and (d) for dehydrated and g–h for hydrated) compared to the control cells (figures 4(a), (b) and (e), (f)). Changes in morphology and stiffness of bacterial membrane are a common indicator of effectiveness in antibiotic treatments [47–49], and results presented in this study imply stiffness increases after the treatment of polymyxin B. In the control cells that were not subject to polymyxin B exposure, a significant portion of the measured regions appeared to be softer (darker as lower modulus) both in ambient environment (figures 4(a) and (b)) and in hydrated state (figures 4(e) and (f)). For the same control cells, the average modulus measured in ambient environment was higher than those obtained in the hydrated state. Despite the greater stiffness of the polymyxin B treated cells, no measurable difference was found with regard to the hydration status. This is possibly due to the loss of soft cytoplasmic materials during polymyxin B exposure, and the modulus of the remaining cellular materials are less likely affected by the presence of extracellular fluids. An interesting finding here is that the changes of stiffness after treatment were also found in the bacteria’s interior when interrogated with AFM after FIB sectioning. In a recent EM study using the same polymyxin-bacteria treatment regime, significant morphological changes were revealed in the cytoplasmic 7
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 5. (a)–(d) The modulus data was fit to lognormal distribution with signficant confidence level (p > 0.95). (e) Summary of the log scaled data into box plots for comparisons.
regions [21]. This observation is in line with the result of interior modulus presented in this study (figure 4), and all the evidence suggested significant compositional changes of the cells induced by polymyxin B exposure.
Further analyses were also conducted to explore statistical information. All the datasets of Young’s modulus fit lognormal distributions for comparisons (figures 5(a)–(d)), with goodness-of-fit values (also referred to as the p-value) larger than 0.95. Comparison of the distributions between 8
Nanotechnology 25 (2014) 415101
B Liu et al
the control (figures 5(a) and (b)) and treated (figures 5(c) and (d)) groups also confirmed that the cells became stiffer after polymyxin B treatment. Figure 5(e) summaries the log scaled moduli data as box plots. Clearly the moduli in the untreated control cells contained a number of outliers and some of them reached a fairly high value, implying active metabolism in native physiological cycles. For the treated samples as shown in figure 5(e), patterns of higher modulus but minimal outliers were observed, suggesting more homogenous and stiffer cellular materials resulted from drug treatments. In addition, the majority of Young’s moduli of control cell measured in liquid ranged from 0 to 2 MPa, and shifted to higher values (1–4 MPa) if dehydrated (figures 5(a)–(d)). The moduli measured in liquid appeared to be reduced than those measured in air for all the groups except for the cell interior of treated bacterial cells (figure 5(d)). After polymyxin B treatment, both the interior and surface of cells became stiffer, however, the extent of the change was possibly greater in the interior regions. These observations can be explained by our contemporary understanding of the antibacterial mode of action of polymyxins. Polymyxins are believed to permeabilize the Gram-negative bacterial outer membrane via a direct interaction with the lipid A component of the lipopolysaccharide (LPS). The disruptive effect of polymyxins on the outer membrane is thought to involve a two-stage interaction mechanism. In stage one, the binding of polymyxin to LPS involves an initial electrostatic interaction of the cationic diaminobutyric acid side chains with the anionic phosphate groups of the lipid A component of LPS, displacing divalent cations (Ca2+ and Mg2+) that bridge adjacent LPS molecules [50–53]. This initial electrostatic interaction allows the Nterminal fatty acyl chain of the polymyxin molecule to insert into the fatty acyl chain layer of the lipid A molecules (stage 2). This two-stage ‘self-promoted’ mechanism is believed to produce lesions in the outer membrane structure, which leads to bacterial cell death. Upon loss of cytoplasm material, the interior space of the treated cell is irreversibly dominated by organelles and other large proteins and polysaccharides, which contribute to the higher modulus values and reduced variations as shown in this study (figure 5(e)). These effects were less pronounced in the hydrated state (figure 5(d)) possibly due to the amelioration effect of liquid filling in the interstitial spaces.
scanning was done at ∼20 nm intervals in the ambient environment (figure 6(d)). The boundary between the cell and biofilm was more distinct, with a number of soft regions present on the surface surrounded by stiffer features which are to be possibly multi-protein complexes. While the use of probe tips with higher aspect ratio may provide better resolution and thus clearer insights into the morphology of this important boundary (e.g. using nanotubes) [54], yet it should be noted that there are potential errors arising from tip buckling. Quantitative analyses based on two-way ANOVA were performed to examine the effects of polymyxin B treatment on Young’s moduli. The logarithmic scale data were rearranged in such a way that the treatment and environment (ambient or aqueous) were considered as two factors to test. The significance of the two factors was analyzed using the two-way ANOVA with 95% confidence interval. The p-value of the two factors was both lower than 0.001, suggesting that both polymyxin B treatment and the measuring media had significant impacts on Young’s modulus measurements. It should be noted that microbial cells are generally considered as soft materials and Young’s moduli measured by previous studies ranges widely from 1 kPa to 10 MPa [7, 45, 48, 55, 56]. In this study, modulus measurements of the control cells in hydrated status are close to previous reported data in different microbial species such as Escherichia coli [57], Pseudomonas aeruginosa [58], Aspergillus nidulans [59] and Saccharomyces cerevisiae [60]. Whereas the higher moduli obtained with the nanoindentation measurements on the K. pneumoniae may be caused by the effectiveness of polymyxin B. As discussed in previous sections, increased stiffness on treated cell surface was expected and confirmed after loss of integrity of the outer membrane, which consequently caused loss of water containing cytoplasmic materials and/or reduction of nucleoid regions. The other two possible contributing factors are fixation and indenter shape. Previous studies reported that the cells fixed by glutaraldehyde were harder compared to the same cells prior to fixation due to protein crosslinking [46, 58, 61, 62]. Although spherical indenter was common in cell measurements, the sharp quadratic pyramid shape was chosen in this study in order to yield a higher resolution and to detect the effect of cellular composition or structural changes [63].
3.4. Multiple imaging modes in the hydrated state
3.5. Chemical spectroscopy of bacterial interior
To further demonstrate the possibility of the proposed approach for interrogating single cells with multiple AFM modes, a sectioned cell after polymyxin B treatment was scanned in both ambient and liquid environments and with multiple resolutions (figure 6). Consistently, the measured Young’s moduli of the hydrated interior were one magnitude lower than those measured in dehydrated condition (figures 6(a) and (c)). In both modulus maps, a large soft region was identified in the central region, which was surrounded by stiffer materials. A higher resolution stiffness map on the surface of the sectioned cell was obtained when
As gold has a high affinity to sulfur, the octadecanethiol molecules can self-assemble on the gold surface with sulfur attached to the gold with the long carbon chain extended. After CH3 functionalization, the CH3/CH3 interaction in aqueous condition should result in a significant increase in the adhesion force during tip pull off from the substrate. This is because CH3 is non-polar, making it typically hydrophobic. If CH3 is attached to another non-polar molecule, it forms a tertiary hydrophobic interaction in the protein structure [64]. The calibration of adhesion force showed an average adhesion value of 568 pN on both the functionalized cantilever tips and 9
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 6. Multi-mode AFM mapping of a single cell interior. (a) Modulus map of a sliced bacterial cell (K. pneumoniae), and (b) the
corresponding height map acquired in air. (c) After rehydration, a modulus map of the same cell was obtained showing significant lower values, and (d) a follow-up scan of higher magnification revealed the heterogeneous modulus distribution close to the cell envelop.
In this study, the bacterial cells for surface measurements were not gold coated to avoid artefacts. Although an approximately 10 nm thick gold layer was coated on the cell surface to minimize the charging effect during FIB milling, the coating was completely removed by the FIB milling prior to the measurements on cell interior. In addition, FIB milling at a grazing angle is also well-recognized as the standard micromachining method to minimize material damage and ion implantation [69–72]. An additional test was performed on PDMS to investigate the moduli change after grazing angle FIB milling (supplementary materials). Limited modulus increase was reported on the PDMS surface after grazing angle milling and AFM measurements (figure S1(b)). This confirmed that FIB milling offers a ‘surgical’ removal of soft material, with structural and chemical information well-maintained for tomographic reconstructions [20, 22] and future chemical/modulus probing of cells. One bottleneck for obtaining high-throughput data with the proposed FIB–AFM approach is the acquisition speed of AFM particularly for larger sections such as with mammalian cells. Both modulus and chemical mapping require repetitive probing on each target spot to retrieve reliable measurements for model fitting. In the present study, a typical 2D modulus map of a bacterial cell with a resolution of 10 nm required approximately 30 min for an AFM image. Recently, several new quantitative nanomechnical mapping techniques have been developed to achieve higher resolution and faster data
substrate, while they were reduced to 157 pN on the nonfunctionalized tips and substrates. The map of adhesion forces from a K. pneumoniae cell (control) is presented in figure 7(d), which showed significant higher adhesion forces over 1 nN at the close proximity of the cell envelope. This is likely due to the interactions between the functionalized tips and the large hydrocarbon core in the phospholipid bilayers of bacterial cell membrane there. The central region containing cytoplasm and nucleoid is supposed to be less hydrophobic, and this is confirmed by the obtained adhesion force map. All these results suggest a promising prospect of being able to investigate specific molecular interactions inside cells via proper tip functionalization [13].
3.6. Limitations and future applications
Vacuum freeze drying is employed for the FIB–AFM approach, and the artefacts are expected to be minimal. For chemical mapping of cells, freeze drying is well-known for its ability to preserve the majority of the macromolecular structures and biochemical compositions, and has been employed in numerous previous studies including SIMS, Synchrotron imaging, x-ray photoelectron spectroscopy [23, 65–68]. The chemical signals obtained are usually assumed as near native. In terms of AFM modulus mapping, glutaldehyde fixation introduces higher modulus, yet in some occasions it had to be conducted as per safety protocol of the AFM laboratories. 10
Nanotechnology 25 (2014) 415101
B Liu et al
Figure 7. Adhesion force calibration and chemical mapping of cell interior with octadecanethiol functionalized AFM tips. (a) The adhesion results on a 2 × 2 μm2 area of Au/Cr coated silicon wafer surface probed by non-functionalized (control) AFM tips. (b) The adhesion results on a 2 × 2 μm2 area of Au/Cr coated silicon wafer surface with both substrate surface and AFM tip functionalized, with average adhesion force significantly higher compared with (a) due to the interactions of CH3 molecules. (c) Height map of a sliced K. pneumoniae cell measured in liquid (left), and the corresponding hydrophobicity measurements of cell interior based on adhesion forces obtained by CH3 functionalized AFM tips (right).
4. Conclusion
acquisition, which are particularly useful for studying biological cells [73–75]. With improved integration such as the installation of AFM in FIB/SEM, three dimensional modulus and chemical mapping of a single cell will be obtained in a reasonable timeframe to reveal the biochemical nature currently shielded by the cellular membrane.
The proposed strategy in this study allows removing a controlled layer on bacterial surface and subsequently investigating the mechanical and chemical characteristics of the inaccessible interior of cells. Young’s modulus results 11
Nanotechnology 25 (2014) 415101
B Liu et al
obtained by AFM nanoindentation on cell surface with carefully chosen AFM cantilever tips revealed the intracellular mechanical property change between the control cells and the polymyxin treated cells. In addition, the hydrophobicity measurement using a functionalized AFM tip was able to highlight the evident hydrophobic portion of the cell such as regions containing cell membrane. In a previous study, adhesion force was applied to quantity antibiotic and peptidoglycan interaction [14], and combining the same tip functionalization protocols with the current study has great potential for intracellular drug localization studies. Together with recently developed approaches for improved bacterial cell collection within specific regions [26, 76] and correlative imaging [77–79], we expect that the complete FIB–AFM platform will help in gaining deeper insights of bacteria–drug interactions to develop potential strategies for combating multi-drug resistance.
[17] Arnoldi M, Fritz M, Bäuerlein E, Radmacher M, Sackmann E and Boulbitch A 2000 Phys. Rev. E 62 1034–44 [18] Heymann J A, Hayles M, Gestmann I, Giannuzzi L A, Lich B and Subramaniam S 2006 J. Struct. Biol. 155 63–73 [19] Heymann J A, Shi D, Kim S, Bliss D, Milne J L and Subramaniam S 2009 J. Struct. Biol. 166 1–7 [20] Rigort A, Bäuerlein F J B, Villa E, Eibauer M, Laugks T, Baumeister W and Plitzko J M 2012 Proc. Natl. Acad. Sci. 109 4449–54 [21] Liu B, Yu H H, Ng T W, Paterson D L, Velkov T, Li J and Fu J 2014 Microsc. Microanal. 20 537–47 [22] Marko M, Hsieh C, Schalek R, Frank J and Mannella C 2007 Nat. Methods 4 215–7 [23] Szakal C, Narayan K, Fu J, Lefman J and Subramaniam S 2011 Anal. Chem. 83 1207–13 [24] Narayan K, Prosa T J, Fu J, Kelly T F and Subramaniam S 2012 J. Struct. Biol. 178 98–107 [25] James S A, Feltis B N, De Jonge M D, Sridhar M, Kimpton J A, Altissimo M, Mayo S, Zheng C, Hastings A and Howard D L 2013 ACS Nano 7 10621–35 [26] Liu B, Fu J, Somers A, Muradoglu M and Ng T W 2014 RSC Adv. 4 2611–9 [27] Hutter J L and Bechhoefer J 1993 Rev. Sci. Instrum. 64 1868–73 [28] Melzak K A, Moreno-Flores S, Yu K, Kizhakkedathu J and Toca-Herrera J L 2010 Microsc. Res. Tech. 73 959–64 [29] Benítez R, Moreno-Flores S, Bolós V J and Toca-Herrera J L 2013 Microsc. Res. Tech. 76 870–6 [30] Oliver W C and Pharr G M 1992 J. Mater. Res. 7 1564–83 [31] Hertz H 1882 J. Reine Angew. Math. 92 156–71 [32] Johnson K, Kendall K and Roberts A 1971 Proc. R. Soc. A 324 301–13 [33] Derjaguin B, Muller V and Toporov Y P 1975 J. Colloid Interface Sci. 53 314–26 [34] Grant C A, Thomson N H, Savage M D, Woon H W and Greig D 2011 J. Mech. Behav. Biomed. Mater. 4 535–40 [35] Darling E M, Wilusz R E, Bolognesi M P, Zauscher S and Guilak F 2010 Biophys. J. 98 2848–56 [36] Tracqui P, Broisat A, Toczek J, Mesnier N, Ohayon J and Riou L 2011 J. Struct. Biol. 174 115–23 [37] Bilodeau G 1992 J. Appl. Mech. 59 519–23 [38] Rosenbluth M J, Lam W A and Fletcher D A 2006 Biophys. J. 90 2994–3003 [39] Coles J M, Zhang L, Blum J J, Warman M L, Jay G D, Guilak F and Zauscher S 2010 Arthritis Rheum. 62 1666–74 [40] Ng L, Hung H-H, Sprunt A, Chubinskaya S, Ortiz C and Grodzinsky A 2007 J. Biomech. 40 1011–23 [41] Dufrêne Y F 2008 Nat. Protoc. 3 1132–8 [42] Drobne D, Milani M, Lešer V and Tatti F 2007 Microsc. Res. Tech. 70 895–903 [43] Fu J, Sanjay B J and Jeffrey M C 2008 J. Micromech. Microeng. 18 095010 [44] Mathur A B, Collinsworth A M, Reichert W M, Kraus W E and Truskey G A 2001 J. Biomech. 34 1545–53 [45] Kuznetsova T G, Starodubtseva M N, Yegorenkov N I, Chizhik S A and Zhdanov R I 2007 Micron 38 824–33 [46] Vadillo-Rodriguez V, Beveridge T J and Dutcher J R 2008 J. Bacteriol. 190 4225–32 [47] Formosa C, Grare M, Duval R E and Dague E 2012 Nanomed. Nanotechnol. Biol. Med. 8 12–6 [48] Longo G, Rio L M, Trampuz A, Dietler G, Bizzini A and Kasas S 2013 J. Microbiol. Methods 93 80–4 [49] Soon R L, Nation R L, Harper M, Adler B, Boyce J D, Tan C-H, Li J and Larson I 2011 Int. J. Antimicrob. Agents 38 493–501 [50] Wiese A, Münstermann M, Gutsmann T, Lindner B, Kawahara K, Zähringer U and Seydel U 1998 J. Membr. Biol. 162 127–38
Acknowledgements Funding for this study was provided by the Australian National Health & Medical Research Council (NHMRC, APP1046561) and Seed Grant from the Faculty of Engineering, Monash University. This work was performed in part at the Melbourne Centre for Nanofabrication (MCN) in the Victorian Node of the Australian National Fabrication Facility (ANFF). TV is an NHMRC CDF Industrial Fellow and JL is an NHMRC Senior Research Fellow.
References [1] Yamada K M 1983 Annu. Rev. Biochem. 52 761–99 [2] Peddie C J, Blight K, Wilson E, Melia C, Marrison J, Carzaniga R, Domart M-C, Larijani B and Collinson L M 2014 Ultramicroscopy 143 3–14 [3] Yang X, Gu C, Qian F, Li Y and Zhang J Z 2011 Anal. Chem. 83 5888–94 [4] Stephens D J and Allan V J 2003 Science 300 82–6 [5] Ubbink J and Schär-Zammaretti P 2005 Micron 36 293–320 [6] Touhami A, Jericho M H and Beveridge T J 2004 J. Bacteriol. 186 3286–95 [7] Touhami A, Nysten B and Dufrêne Y F 2003 Langmuir 19 4539–43 [8] Hayat M A 2000 Principles and Techniques of Electron Microscopy: Biological Applications 4th edn (Cambridge: Cambridge University Press) [9] Wilson S M and Bacic A 2012 Nat. Protoc. 7 1716–27 [10] Dufrêne Y 2001 Micron 32 153–65 [11] Dague E, Alsteens D, Latgé J-P, Verbelen C, Raze D, Baulard A R and Dufrêne Y F 2007 Nano Lett. 7 3026–30 [12] Dufrêne Y F 2000 Biophys. J. 78 3286–91 [13] Dufrêne Y F, Martínez-Martín D, Medalsy I, Alsteens D and Müller D J 2013 Nat. Methods 10 847–54 [14] Gilbert Y, Deghorain M, Wang L, Xu B, Pollheimer P D, Gruber H J, Errington J, Hallet B, Haulot X and Verbelen C 2007 Nano Lett. 7 796–801 [15] Verbelen C, Dupres V, Raze D, Bompard C, Locht C and Dufrêne Y F 2008 J. Bacteriol. 190 7614–20 [16] Dufrêne Y F 2002 J. Bacteriol. 184 5205–13 12
Nanotechnology 25 (2014) 415101
B Liu et al
[67] Dengis P B and Rouxhet P G 1996 J. Microbiol. Methods 26 171–83 [68] Nygren H and Malmberg P 2004 J. Microsc. 215 156–61 [69] Kiener D, Motz C, Rester M, Jenko M and Dehm G 2007 Mater. Sci. Eng. A 459 262–72 [70] Ishitani T, Umemura K, Ohnishi T, Yaguchi T and Kamino T 2004 J. Electron Microsc. 53 443–9 [71] Langford R and Petford-Long A 2001 J. Vac. Sci. Technol. A 19 2186–93 [72] Kato N, Kohno Y and Saka H 1999 J. Vac. Sci. Technol. A 17 1201–4 [73] Zhang S, Andreasen M, Nielsen J T, Liu L, Nielsen E H, Song J, Ji G, Sun F, Skrydstrup T and Besenbacher F 2013 Proc. Natl. Acad. Sci. 110 2798–803 [74] Dong M, Hovgaard M B, Mamdouh W, Xu S, Otzen D E and Besenbacher F 2008 Nanotechnology 19 384013 [75] Dong M and Sahin O 2011 Nat. Commun. 2 247 [76] Connell J L, Wessel A K, Parsek M R, Ellington A D, Whiteley M and Shear J B 2010 mBio. 1 e00202–10 [77] Yang Y, Heine R, Xu F, Suhonen H, Helfen L, Rosenhahn A, Gorniak T, Kirchen S, Schwartz T and Baumbach T 2013 J. Phys.: Conf. Ser. 463 1–4 [78] Stukalov O, Korenevsky A, Beveridge T J and Dutcher J R 2008 Appl. Environ. Microbiol. 74 5457–65 [79] Murphy G E, Narayan K, Lowekamp B C, Hartnell L M, Heymann J A, Fu J and Subramaniam S 2011 J. Struct. Biol. 176 268–78
[51] Zavascki A P, Goldani L Z, Li J and Nation R L 2007 J. Antimicrob. Chemother. 60 1206–15 [52] Hermsen E D, Sullivan C J and Rotschafer J C 2003 Infect. Dis. Clin. North Am. 17 545–62 [53] Cardoso L, Araujo M, Goes A, Pacifico L, Oliveira R and Oliveira S 2007 Microb. Cell Fact. 6 1–6 [54] Jin K, Feng X, Ng T W and Xu Z 2012 Nanotechnology 23 415–502 [55] Alonso J L and Goldmann W H 2003 Life Sci. 72 2553–60 [56] Radmacher M 1997 IEEE Eng. Med. Biol. Mag. 16 47–57 [57] Boulbitch A, Quinn B and Pink D 2000 Phys. Rev. Lett. 85 5246 [58] Yao X, Jericho M, Pink D and Beveridge T 1999 J. Bacteriol. 181 6865–75 www.ncbi.nlm.nih.gov/pubmed/10559150 [59] Zhao L, Schaefer D, Xu H, Modi S J, Lacourse W R and Marten M R 2005 Biotechnol. Progr. 21 292–9 [60] Alsteens D, Dupres V, Mc Evoy K, Wildling L, Gruber H J and Dufrêne Y F 2008 Nanotechnology 19 384005–13 [61] Hoh J H and Schoenenberger C A 1994 J. Cell Sci. 107 1105–14 www.ncbi.nlm.nih.gov/pubmed/7929621 [62] Shroff S G, Saner D R and Lal R 1995 Am. J. Physiol. 269 C286–92 www.ncbi.nlm.nih.gov/pubmed/7631757 [63] Ebenstein D M and Pruitt L A 2006 Nano Today 1 26–33 [64] Fujihira M, Tani Y, Furugori M, Akiba U and Okabe Y 2001 Ultramicroscopy 86 63–73 [65] Carmona A, Cloetens P, Devès G, Bohic S and Ortega R 2008 J. Anal. At. Spectrom. 23 1083–8 [66] Guerquin-Kern J-L, Wu T-D, Quintana C and Croisy A 2005 Biochim. Biophys. Acta 1724 228–38
13