Surface Characterization of Gram-Negative Bacteria and their Vesicles
Using Fourier Transform Infrared Spectroscopy and Dynamic Light Scattering
Sofia Carlsson
Degree Thesis in Chemistry 30 ECTS Master’s level
Report passed: 07 2012
Supervisors: Madeleine Ramstedt
Abstract
Gram-negative bacteria are known to produce vesicles, why and how is still relatively unknown. Gram-negative bacteria are widely studied but still there are large areas in the field covering the bacterial chemistry that are unknown.
This study is a pilot study aimed to complement the studies on surface properties for Gram-negative bacteria and their vesicles. The study is focused on the bacteria Escherichia coli (E.coli). Differences and similarities between bacteria and vesicle surface characteristics and chemical composition are presented using Attenuated Total Reflectance Fourier Transformed Infrared Spectroscopy (ATR-FTIR) and Dynamic Light Scattering (DLS).
There does not seem to be any direct correlation between bacteria and vesicles regarding the surface characteristics. However, this study provides interesting data regarding the production of vesicles and vesicle characteristics as well as information regarding the surface characteristics of bacteria.
II
III
List of abbreviations
ε the dielectric constant
f (κa) a function of the ratio between the radius of the particle and the thickness of the electrical double layer.
η The viscosity of the solution ATR Attenuated total reflectance
D Translational diffusion coefficient DLS Dynamic light scattering
DNA Deoxyribonucleic acid E.coli Escherichia coli
FTIR Fourier transform infrared
Ge Germanium
HCl Hydrochloric acid IR Infrared spectroscopy
k Boltzmann constant
LPS Lipopolysaccharides
OM Outer membrane
OMP Outer membrane protein OMV Outer membrane vesicle
R Hydrodynamic radius of a particle (or molecule)
SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis T Absolute temperature
TEM Transmission Electron Microscope Z Zeta potential
ZnSe Zinc selenide
XPS X-ray photoelectron spectroscopy
U(E) Electrophoretic mobility
IV
V
Table of contents
Abstract ... I List of abbreviations ... III Table of contents ... V
Why study vesicles from Gram-negative bacteria? ... 1
Bacteria vs. vesicle ... 1
Gram-negative bacteria ... 1
Vesicles ... 2
Techniques used for analysis ... 4
Infrared (IR) spectroscopy ... 4
Fourier Transformed Infrared (FTIR) Spectroscopy ... 5
Attenuated Total Reflectance (ATR) ... 5
Microbial cells ... 6
Dynamic light scattering, DLS ... 6
Size measurement ... 6
Zeta potential measurements ... 7
Method ... 9
Samples ... 9
Buffer preparation ... 10
Preparation of bacteria ... 10
Preparation of outer membrane vesicles ... 10
Sodium dodecyl sulfate polyacrylamideb gel electrophoresis (SDS-PAGE) ... 10
DLS analysis ... 10
ATR-FTIR spectroscopic analysis ... 11
Results and discussion ... 11
Optimization of the method ... 11
DLS analyses ... 14
ATR-FTIR spectroscopic analysis ... 17
Other data ... 23
SDS-PAGE ... 24
X-ray photoelectron spectroscopy, XPS ... 24
Conclusions ... 26
Acknowledgement ... 26
Bibliography ... 28
VI
1
Why study vesicles from Gram-negative bacteria?
So far little is known about the outer membrane vesicles (OMV) emerging from the cell surface. Why are they produced, how do they interact and what are their characteristics? There are many theories and studies but so far few answers (1), (2), (3), (4), (5). This study tries to give some answers regarding the surface of vesicles compared to the surface of bacteria. Are there some similarities? Are the characteristics of vesicle surfaces consistent or random? And if similar to bacteria, can we use them to further explore the bacterial surface and its interaction with its surroundings?
So why do we want to explore the surface of bacteria? Due to its direct contact with the external environment bacteria are in need of a cell surface that is adaptable to rapid changes in, for example, environmental components, osmotic pressure and nutrients. But it still needs to be strong and stable to survive different kinds of threats (1). This need combined with flexibility have developed unique cell surface properties where still much is unknown. A greater extent of knowledge could develop the usage of bacteria and enable the usage of bacterial vesicles, for example, in the delivery of different drugs to target cells (1). The focus of our research is to gain a greater knowledge of the characteristics of the bacterial surface in the purpose to reach a more comprehensive understanding of bacterial adhesion.
Bacteria vs. vesicle
Gram-negative bacteria
A Gram-negative bacterium differs from a Gram-positive bacterium mainly when it comes to the outer membrane (OM) and the periplasm which are both missing in the Gram-positive bacteria.
The OM consists of lipopolysaccharides (LPS), phospholipids and outer membrane proteins (OMPs). The phospholipids form a bilayer structure where the LPS are attached to the outer leaflet. This structure gives the OM a hydrophilic outer surface and a hydrophobic core. The OMPs forms hydrophilic pores that can act as a link, or channel, between the periplasm and the external environment (1).
The periplasm is a macromolecular gel consisting of peptidoglycan and includes periplasmic proteins and other macromolecules. It has a high, fixed polyelectrolyte concentration which makes it isosmotic with the cytoplasmic membrane (1).
The outline of Gram-Negative cell wall can be seen in Figure 1 (1).
Figure 1. Schematic image of Gram-negative bacteria.
2 Bacteria are very small organisms with a size range in microns; the size varies between different species. If looking at the transmission electron microscope (TEM) image, Figure 2, it can be seen that E.coli mutant ∆waaG has a size of approximately 1.0-1.5µm.
Furthermore, due to this small size, bacteria have a very high surface to volume ratio, approximately 100 000 times (6). The high surface to volume ratio indicates that the membranes of the bacteria is a large contributing factor to the dry mass of the bacteria and understanding this part of the cell should therefore contribute greatly, to the complete understanding of cellular function and -interaction.
From previous studies it seems that bacterial size and shape of the cells as well as the buoyant density differ during the growth, they all seem to increase in the exponential phase. However, the zeta potential seems to be constant (7). This indication is reinforced if looking at E.coli bacteria in a TEM. Figure 2a shows an E.coli bacterium that probably is in the beginning of the exponential phase, prior to cell division. This can be compared to Figure 2b that shows a mutant of Pseudomonas aeruginosa that is just about to divide.
a b
Figure 2.TEM image of E.coli (a) and Pseudomonas aeruginosa (b). In the red circle in image (a) there is a bacterium in the exponential phase, prior to cell division. In the red circle in image (b) there is a bacterium in the exponential phase during cell division (8).
Vesicles
Gram-negative bacteria naturally produce OMVs; small closed bubbles released from the bacteria surface. They are suggested to be formed as a bulge arising from the outer membrane as can be seen in Figure 3 (2), (3). The surface of OMVs are suggested to consist of bacterial outer membrane including LPS and OMPs and it is proposed that they are filled with periplasmic components such as proteins, enzymes and toxins (3), (9). The full mechanism of vesicle production and its components are still not fully understood. In electron microscope they appear as spherical structures with a diameter of 50-250 nm depending on strain (2), (3). It is noteworthy that vesicles are only produced by live bacteria (3).
3 Figure 3. Formation of membrane vesicles.
The purpose of vesicle production is still not fully understood although they seem to be used as transport vehicles for Gram-negative bacteria components to eukaryotic and prokaryotic cells in the surrounding environment of the bacteria (2). They are suggested to function as both a channel of waste handling as well as a delivery system. The range of components that the OMVs are suggested to transport is very broad including virulence factors, antimicrobial agents, DNA, toxins, proteins, antibiotics etc. (3). It seems the components transported can be either fully functional substances with a purpose and/or a target cell or waste material such as missfolded proteins (4). It also seems they can act as virulence factors, carrying substances that damage the target cell in different ways (3). One of the advantages suggested for vesicle production is the possibility to reach host cells located deep in tissues that cannot be reached by the bacteria themselves (2).
Vesiculation seems to occur in various environments such as solid or liquid cultures, in biofilms, planktonic cultures as well as natural environments (2), (3), (4). This also includes infected tissues (5). Approximately 1 % of the OM materials in a culture of E.coli are vesicles (5).
Interaction of Gram-negative OMVs with bacteria is suggested to occur by attachment to the membrane via adherence or fusion (5). After adherence the vesicles can be internalized into the host cell (5). For Gram-negative bacteria, this is quite a straight forward process that does not need complete reorganization of the outer membrane due to the fact that the OMVs are made of the outer membrane and consist of similar components as the bacterial cell wall (3). When fused into the membrane, the content of the vesicle can be delivered directly into the periplasm of the bacteria (3). The interaction mechanism between Gram-negative OMVs and Gram-positive bacteria is still unknown (3).
Due to the fact that the mechanism, composition and production of vesicles are so unknown, there are a number of theories regarding vesiculation although they are not fully comprehensive and restricted to a few species of bacteria.
Regarding surface characterization there has been studies on Pseudomonas aeruginosa suggesting that the growth phase of bacteria only affects the zeta
4 potential of the vesicles produced, giving a more negative zeta potential for vesicles when the bacteria are in the stationary phase. The size and the shape are suggested to remain constant (7). Studies, presented in this report, on E.coli vesicles will complement earlier findings in this field.
Other interesting studies suggest that there are also differences in the inner content of the vesicles depending on when the vesicles are produced. For example, it is suggested that vesicles produced in the exponential growth phase have a higher amount of proteins (7). It was also suggested that the vesicles produced in the exponential phase have a higher density and are more translucent (7). The amount vesicle produced is suggested to be affected by the charge of the LPS of the bacteria (3), (9) and the rate is suggested to be dependent on the bacterial growth phase with a maximum rate during the late exponential phase (2), (4). All of these things are tested on other bacterial stains than E.coli and they will not be directly studied in the scope of this project.
Moreover, early studies made the assumption that the surface charge of bacteria were homogeneously distributed over the surface although newer studies indicates that in reality there is a higher concentration of surface sites at the poles of the bacteria (10).
If so, and if vesicles are produced as suggested than there will most likely be a difference in vesicle structure due to where on the bacterial surface it is spun off from. This topic is also outside the scope of this project.
Techniques used for analysis
Infrared (IR) spectroscopy
Each atom and each molecule has a motional freedom of three i.e. in x, y and z direction. These motional freedoms can be divided into three different kinds of movements; translational, rotational and vibrational as can be seen in Table 1. The vibrational mode of motional freedom is often referred to as the normal mode of vibration (11), (12).
Table 1. Motional freedoms of molecules. N stands for numbers of atoms (11), (12).
Types of degrees of
freedom Linear molecule Non-linear molecule
Translational 3 3
Rotational 2 3
Vibrational 3N-5 3N-6
Total 3N 3N
Furthermore, the vibrational modes can then be divided into stretching (change in bond length) and bending modes (change in bond angle). The stretching can be divided into symmetric (in-phase) or asymmetric (out-of-phase) stretching.
Moreover, the bending modes can be divided according to type of bending;
scissoring, rocking, wagging and twisting. In addition, they can also be divided into in-plane (scissoring and rocking) or out-of-plane (wagging and twisting) (11), (12).
The theory behind IR spectroscopy is based on the vibrational modes of motional freedom of the atoms in a molecule and how these vibrations will affect neighboring atoms through chemical bonds (12). However, not all vibrations give rise to an IR spectrum, the vibration has to create a change in the dipole moment of the molecule.
The larger the change, the more intense will the absorption bands be (11).
A spectrum is obtained by letting infrared radiation pass through the sample and then measuring which fraction of the incident radiation is absorbed at a specific energy (11). Due to the vibrations different substances will absorb different amount of
5 radiation in different regions giving rise to individual spectra that give us information regarding the bonds within a molecule (11).
The main advantage of IR spectroscopy is that almost any sample in almost any state can be analyzed with different techniques (11). However, there is one special feature that the analyzed samples have to possess namely, the molecule has to have an electrical dipole moment that changes during vibration (11). An IR spectrum is obtained when a sample absorbs radiation in the region of infrared electromagnetic radiation (400-4000 cm-1 for mid-infrared). When the radiation is absorbed, a quantum mechanical transition occurs between two vibrational energy levels (the ground level and the first excited level for mid-infrared). The difference in energy is directly related to the frequency of the radiation (12).
Fourier Transformed Infrared (FTIR) Spectroscopy
The basic idea behind the FTIR spectroscopy is that the inference between two beams of radiation will yield an interferogram; a function of the change of path length between the two beams. This information is then transformed by a mathematical method called Fourier-transformation. The basic outline of the method can be seen in Figure 4. The main advantage of FTIR spectroscopy is that the sample is scanned with the whole spectral range at each time (12).
Figure 4. Basic outline for FTIR spectroscopy.
The crucial part in this setup is the interferometer, where the Michelson interferometer is the most common one. It consists of two mirrors, one fixed and one moveable, and a beam-splitter. The two mirrors are positioned 90 ° angle from each other. The radiation hits the beam-splitter and then 50 % of the beam goes to each mirror. The mirrors reflect the beams back to the beam splitter where they recombine and interfere; this beam is called the transmitted beam and is the one detected by the FTIR spectroscopy (12), (11).
The moving mirror will create an optical path difference between the two mirrors. If both mirrors are at the same position they will reflect both beams to the detector.
During these conditions, the beams will be in phase when they hit the detector, creating a perfect constructive interference and the radiation signal will be at maximum intensity (12). If not at the same position, there will be an interference pattern created by the difference in optical path length between the two mirrors (11).
During each scan the sample will absorb some radiation, lowering the intensity of the incident radiation (12). The second mirror is moved between each scan giving rise to a different interference pattern each time. It is these interference pattern that give rise to the raw data, also called interferogram, that are transformed by the mathematical algorithm Fourier transformation (12).
Attenuated Total Reflectance (ATR)
Different reflectance methods are used when the samples are too difficult to analyze using normal transmittance methods. For ATR the sample is placed on a crystal and then radiated with a beam of IR light. The method is based on total internal reflection i.e. an optical phenomenon where, if the incoming angle of radiation is bigger than the critical angle, all light is reflected. The critical angle is a function of the refractive indices of the crystal surfaces (11). A schematic image of the ATR setup can be seen in Figure 5.
Source Interfero-meter Sample Detector Amplifier Analog-to dogital
converter Computer
6 The crystal used is of materials that have low solubility in water and have high reflective index. This includes for example zinc selenide (ZnSe) and Germanium (Ge) (11).
Figure 5. Example of an ATR setup.
Microbial cells
When analyzing microbial cells like bacteria using IR spectroscopy, a large number of peaks can be expected. Proteins represent a major part of the bacterial mass and consequently some peaks can be overshadowed and the major peaks in the spectra will correspond to vibrations in proteins (11).
Due to the large penetration depth of the IR beam, in the range of a few µm, the whole cell will be analyzed and consequently, FTIR spectroscopy cannot distinguish between the cell wall and the inner cell (11), (10).
Dynamic light scattering, DLS Size measurement
All particles in a liquid possess something called Brownian motion; a random motion created when the surrounding medium push the particles around in the solution. The speed of motion is directly proportional the particle size, the larger the particle the slower the speed (13). The speed of motions is also proportional to the temperature of the solution, as can be seen in Equation 1 (14).
The size measurements of a particle in a solution starts with illumination of the solution with a laser and then the light will scatter due to the particles in the solution.
How they scatter is dependent on the size and this is measured by measuring how fast the intensity of the scattered light fluctuates. Small particles will cause the intensity to change more often than larger particles. The velocity of Brownian motion for a particle is defined by a property called the translational diffusion coefficient (D) and it is correlated from the intensity variations of the scattered light by a correlation program (14). The D-value is then used to recalculate the hydrodynamic radius using Stoke-Einstein equation, as can be seen in Equation 1, to give a value of the size.
Although, the size calculations are based on the particles hydrodynamic radius i.e. the size reported is for a sphere that has the same D-value as the particle.
For a non-spherical particle the size measured will be either larger or smaller than the real size depending on the particle shape (14). This also gives that small changes in shape may alter the hydrodynamic size although not all changes are detected. For
7 example if the particle is rod-shaped then small changes in length will be detected but small changes in thickness will most likely not (14).
Equation 1. The Stoke-Einstein equation; where D is the diffusion coefficient, k is the Boltzmann constant, T is the absolute temperature, η is the viscosity of the solution and R is the hydrodynamic radius of the particle (or macromolecule).
When measuring size, the temperature needs to be stable so that the motions measured are Brownian motions and not flow motions in the sample created by temperature differences, and that the correct T is used in calculations (14).
Furthermore, knowledge regarding the temperature also gives knowledge about the viscosity of the solution which is required in the hydrodynamic size measurements.
The size measurement is also affected by the ionic strength of the solution i.e. the conductivity. In a low ionic strength solution, the double electric layer will be more extended than in high ionic strength (11).
Another factor that can affect the hydrodynamic size is the structure of the surface of the particle; changes to the surface structure may alter the diffusion speed and therefore also alter the hydrodynamic size of the particle (14).
Zeta potential measurements
A charged particle in solution will have a layer of ions of opposite charge bounded to it, a so called Stern layer. Moreover, a second layer of ions are surrounding the particle, a so called diffuse layer. The zeta potential is a physical property of a particle in a suspension, and it is created in the slipping plane between the Stern layer and the diffuse layer as can be seen in Figure 6 (15).
8 Figure 6. Schematic representation of zeta potential.
The zeta potential can indicate the degree of interaction between particles in a colloid system, and consequently the stability of the same system. A near neutral zeta value (-10 to 10 mV) or a zeta value of zero (isoelectric point) is an indication of an unstable system i.e. the particles will flocculate. In contrast to a strongly negative (lower than - 30 mV) or strongly positive (higher than 30 mV) zeta potential that indicates a stable system i.e. the particles will repel each other and remain in solution (13), (15). It can also be used to investigate the surface charge of a particle since the zeta potential is dependent on the surface charge. However, the absolute zeta potential value will be lower than the surface charge, due to the ions in the double electrical layer (13), (15) (14).
The zeta potential is measured by measuring the electrophoretic mobility of the particle. As the particle move the scattered light is phase shifted and this phase shift is proportional to the zeta potential of the particle. The data is correlated to zeta potential via the Henry equation as can be seen in Equation 2 (15).
9 Equation 2. Henrys equation (15).
Several different factors, such as the strength of the electric field, the dielectric constant (ε) of the medium, the viscosity (ŋ) of the medium and the zeta potential (Z) of the particle, affect the electrophoretic mobility (U(E)) (13), (15). The Henry function f (κa) is a function of the ratio between the radius of the particle and the thickness of the electrical double layer. For aqueous media with moderate electrolyte concentration this value will be 1.5 and these conditions are referred to as the Smoluchowski approximation (15).
The zeta potential is affected by several factors such as pH and conductivity of the medium. For pH effects it can be seen that a particle in an alkaline solution generally will have a more negative zeta potential and a particle in an acid solution will have a more positive zeta potential (13). Furthermore, earlier in this report it was described that the hydrodynamic size was affected by ionic strength. The same can be said regarding the zeta potential although, in this case it is the interaction between the particle surface and surrounding ions that can change the zeta potential (15).
Method
Samples
Table 2. Bacteria strains used in this study.
E.coli strain Modified gene
BW25113 WT
RN103 ΔwaaF
RN104 ΔwaaG
RN105 ΔwaaL
RN106 ΔwaaP
RN107 ΔgalE
BW25113/pMF19 WT-o-antigen E.coli strain without flagella Modified gene
BW25113 WT/flhD
RN101 ΔwaaC
RN102 ΔwaaE
RN103 ΔwaaF/flhD
RN104 ΔwaaG/fhlD
RN105 ΔwaaL/flhD
RN106 ΔwaaP/flhD
RN107 ΔgalE/flhD
Strains for vesicle production Modified gene
BW25113 WT/flhD
RN103 ΔwaaF/flhD
RN104 ΔwaaG/fhlD
RN105 ΔwaaL/flhD
RN106 ΔwaaP/flhD
RN107 ΔgalE/flhD
10 Figure 7. LPS-differentiated E.coli mutations (8).
Buffer preparation
Acetate buffer with an ionic strength of 20 mM and a pH of 5.5 was prepared by mixing 1.6402 g of CH3COONa with 244 µl CH3COOH and then water was added to the final volume of 1000 ml. Phosphate buffer with an ionic strength of 20 mM and a pH of 7 was prepared by mixing 0.0.5531 g of NaH2PO4*2H20 with 1.8442 g of NaHPO4*12H2O and then water was added to the final volume of 1000 ml. Finally, the Tris buffer with an ionic strength of 20 mM and a pH of 8 was prepared by mixing 5.2048 g of TRISMA base with 2.1 ml of HCl and then water was added to the final volume of 1000 ml.
Preparation of bacteria
Bacteria were grown in 2 ml of ISO-SENSITEST liquid medium (Oxoid), collected by centrifugation at 10 000 rpm for 3 min and washed in 2 ml of an appropriate buffer.
Samples were stored on ice and measured within 3 hours after preparation.
Preparation of outer membrane vesicles
Outer membrane vesicles (OMV) were prepared from E. coli BW25113 and mutants of this strain deficient in different steps of LPS biosynthesis. Additionally, these strains had a transposon insertion in flhD gene rendering them free of flagella. Such strains were chosen as flagella are known to contaminate OMV preparation.
Bacteria were grown in 500 ml of ISO-SENSITEST liquid medium (Oxoid) containing kanamycin 50 µg/ml and collected by centrifugation at 5 000 rpm for 30 min at 4°C.
Supernatant containing OMVs was collected and filtered through a sterile filter with pore size of 0.2 µm (Sarstedt). The filtrate was centrifuged in ultra centrifuge at 29 000 rpm for 3 hours at 4°C. The pellet was re-suspended in 600 µl 20 mM Tris-HCL, pH 8.0.
Sodium dodecyl sulfate polyacrylamideb gel electrophoresis (SDS-PAGE) OMV samples were diluted in x2 loading buffer and separated for 1 h at 160V on 12%
ready-made SDS-PAGE (Invitrogen). 15µl of a sample was loaded to each lane of the gel. After this the gel was stained with colloidal Coomassie stain overnight and destained in water.
DLS analysis
Particle size and zeta potential were measured at 25 °C using a Malvern Nano ZS zetasizer (16). Purified bacteria were diluted in Tris-, phosphate- or acetate buffer prior to analysis so that the suspension had an optical density of 0.5 at a wavelength of 600 nm. Purified OMV’s were diluted 10 times in Tris buffer For hydrodynamic size measurement the dynamic light scattering method was used and the raw data were converted by the Strokes-Einstein relationship. For zeta potential measurements laser Doppler Micro-electrophoresis method was used and the raw
hldE waaC waaF waaG waaL waaP galE BW25113
Core OS Lipid
11 data were converted by the Smoluchowski approximation. For both size and zeta measurements a clear disposable zeta cell was used during analysis.
ATR-FTIR spectroscopic analysis
The chemical composition was analyzed using a Bruker Vertex 80v spectrometer (17) with a 3 reflection ZnSe crystal plate placed on a MIRacle base optics assembly (18).
Bacteria and OMV samples were centrifuged and the suspension removed. The pellet was then placed on the ATR crystal. For OMVs the samples had to be frozen with liquid nitrogen to enable transfer from the sample tube and placement on the crystal, this was due to the small amount of purified OMVs and also because of the slimy texture of the OMVs.
Buffer spectra were also obtained and later subtracted from the other spectra.
Results and discussion
Optimization of the method
The project started by developing an analysis method for bacteria as well as for vesicles. First of all, the lowest possible ionic strength of the buffer solution had to be established to ensure that the conductivity remained as low as possible but also to guarantee that the bacteria stay intact during the analysis. During some initial sample preparation there were indications of bacterial leakage for Pseudomonas aeruginosa at the ionic strength of 10 mM. The same leakage was not detected for E. coli indicating a more robust cell. Although to ensure that the most suitable ionic strength were used, tests with different ionic strength of phosphate buffer were performed. As can be seen in Figure 8, the zeta potential gets less negative the higher the ionic strength. This is due to that there are more ions in the solution that can screen the surface charge and therefore lowering the absolute zeta potential value. However, for ΔwaaC at ionic strength of 10 mM and 20 mM, it can be seen that the zeta potential is slightly higher at 20 mM indicating that there might be some instabilities in E.coli at 10 mM ionic strength. Analysis of viability did not show a large difference between 10 mM and 20 mM. However, for the rest of the measurements an ionic strength of 20 mM was used.
Figure 8. Variations in zeta potential, for ∆waaC and WT, due to alterations regarding the ionic strength of the phosphate buffer.
Growth conditions for the bacteria were also optimized; we compared properties of bacteria grown on blood agar plate or in liquid media. Both size and zeta potential values showed that the media did not have substantial influence on the results although standard deviation values were somewhat larger for the bacteria grown on
‐70
‐60
‐50
‐40
‐30
‐20
‐10 0
10 mM 20 mM 50 mM 100 mM
Zeta potential (mV)
Ionic strength
WaaC WT
12 blood agar plate, as can be seen in Figure 9a and b, so the rest of the measurements were done with bacteria grown in liquid ISO-SENSITEST media.
a b
Figure 9. Effects from growing sample on different culture media (blood plate or liquid media), on hydrodynamic size (a) and on zeta potential (b).
The vesicle production was a time-consuming process so to save time as well as sample the vesicles were stored in a freezer after DLS analysis and then used for IR after defrosting. To ensure that the vesicles stay stable during freezing, DLS analyses were preformed before and after freezing and the result can be seen in Figure 10a and b. There was some effect although it did not follow any trend and it was relatively small. The source of the difference could be from either the freezing process or the defrosting of the samples. The difference could also be due to uncertanties in the measuring method; giving a higher uncertanty of the absolut zeta potential value when the zeta potential was close to zero. Due to the small difference, the conclusion was made that the vesicle samples could be stored in the freezer.
a b
Figure 10. Effects on hydrodynamic size (a) and zeta potential (b) from freezing vesicle sample.
Three different buffers were used with three different pH; one acid, one alkaline and one neutral solution. These buffers were used during analyses of bacteria to establish the most suitable buffer for analyses of bacteria and vesicles. As can been seen in Figure 11, bacteria in all buffers follow the same trend. Although bacteria in phosphate buffer tended to have the most negative zeta potential and bacteria in actetate buffer tended to have the least negative zeta potential. This result does not follow the expected trend were the most alkaline solution would have the most negative zeta potential, although, these differences seem to be mainly within the standard deviation.
From these analyses only, it is difficult to make a conclusion on which buffer would be the most suitible for vesicle analysis as they only differ in zeta value within the same strain but follow the same trend between the strains. But we can make the conclusion that neither pH nor chemical composition of the buffer influence the zeta
0 1000 2000 3000
waaC‐LB waaC‐blood Hydrodynamic size (d.nm)
Bacterial strain ‐65
‐63
‐61
‐59
‐57
‐55
waaC‐LB waaC‐blood
Zeta potential (mV)
Bacteria strain
0 50 100 150 200 250 300 350
ΔwaaF WT
Hydrodynamic size (d.nm)
Bacteria strain
Fresh sample Defroste d sample
‐10
‐8
‐6
‐4
‐2 0
ΔwaaF WT
Zeta potential (mV)
Bacteria strain
Fresh sample Defroste d sample
13 potential of bacteria to any greater extend. It does not change the correlation between two strains in the same kind of buffer.
Figure 11 can also give an indication of how different alteration to the LPS chains will effect the zeta potential. As can be seen there are two groups of bacteria, those with long LPS chain (∆waaL, ∆waaP, ∆galE and WT) and those with short LPS chain (∆waaC, ∆waaE, ∆waaF and ∆waaG). From the result there can be seen that there is no indication of effects from flagella, mutants with and without flagella seem to have the same zeta potential. There can also be seen that the mutants with long LPS chain have a lower zeta potential and vice versa. And if comparing the mutants with long LPS chain with WT there is only small differences in zeta potential indicating that small changes to the LPS chain do not effect the zeta potential to any greater extend.
For the WT with additional o-antigen, the LPS chain is even longer since the o- antigen is added to the LPS chain. Consequently, it will have a less negative zeta potential than WT as can be seen in Figure 11. Exacly how the different components in the LPS chain is affecting the zeta potential is not in the scope of this project, to establish that, further reseach will be needed. But from the data it can be concluded that larger changes in length of LPS influences the zeta potential.
Figure 11. Zeta potential for different bacteria strain, in three different buffers, each with different pH.
From the OPLS-DA multivariate Scores plot of ATR-FTIR spectroscopic data of the bacterial strains (Figure 12a, 13a and 14a) it can be seen that the higher the pH the more clear is the separation of the samples, based on the type of LPS mutations.
Instead of Scores plot, the model overview an be used to gain more comprehensive information about the separation. A model overview shows if there is any significant difference between the different components and if so, how many components that exhibit a significant differences between them. The more components that can be differentiated from each other, the better is the separation of the difference substances. From this result, there can be seen that the most suitable buffer for analysing bacteria and vesicles would be Tris buffer, due to that seven different components could be identified in the multivarite analysis, as can be seen in Figure 12b, 13b and 14b. Bacteria in acetate buffer and phosphate buffer are less easily separable, most likely due to the bacterial surface is protonated or changed, in the buffers of lower pH which makes the different strains more similar to each other and/or more different within the biological replicates of a particlular strain. One reasonable explanation would be that the phosphates and amines on protein residues and the LPS chain get protonated and/or that the cells release something to the surface. Since the tris buffer seemed to best separate the different samples this buffer
‐70
‐60
‐50
‐40
‐30
‐20
‐10 0
Zeta potential (mV)
Bacteria strain
Acetate Phosphate Tris
14 was chosen for the remaining work. From now on in this report all samples shown were prepared and alaysed in Tris buffer.
a b
Figure 12. Scattered plot (a) and model overview (b) of bacteria in acetate buffer.
Numbers of components = 1+1, Number of observations = 16, Number of classes
= 6, R2X = 0,827, R2Y = 0,155, Q2 = -0,135.
c d
Figure 13.Scattered plot (a) and model overview (b) of bacteria in phosphate buffer. Numbers of components = 2+0, Number of observations = 16, Number of classes = 8, R2X = 0,759, R2Y = 0,218, Q2 = -0,0716.
e f
Figure 14. Scattered score plot (s) and model overview (b) of bacteria in Tris buffer. Numbers of components = 7 + 2, Number of observations = 16, Number of classes = 8, R2X = 0,993, R2Y = 0,899, Q2 = -0,276.
DLS analyses
One hypothesis was that those E. coli mutants and their vesicles that have the shortest LPS chain would have a smaller hydrodynamic size. All sizes measured in this report are the hydrodynamic diameter of the particle and measured in nm. If Figure 15a and b is compared to Figure 7, it can be seen that the hydrodynamic size does not follow the expected trend for either vesicles or bacteria. For bacteria the
-25 -20 -15 -10 -5 0 5 10 15 20 25
-50 -40 -30 -20 -10 0 10 20 30 40 50
to[1]
t[1]
WT____AB_0 WT____AB_1
WC____AB_0 WC____AB_1
WE____AB_0 WE____AB_1
WF____AB_0 WF____AB_1
WG____AB_0 WG____AB_1 WL____AB_0
WL____AB_1 WP____AB_0
WP____AB_1
gE____AB_0 gE____AB_1
-0,1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9
Comp[1]P
SofiaC_Bacteria_2pointBaseline_new5.M24 (OPLS/O2PLS-DA) R2Y(cum) Q2(cum)
-20 -10 0 10 20
-40 -30 -20 -10 0 10 20 30 40
t[2]
t[1]
WT____PB_0 WT____PB_1 WC____PB_0
WC____PB_1
WE____PB_0 WE____PB_1 WF____PB_0
WF____PB_1 WG____PB_0
WG____PB_1
WL____PB_0
WL____PB_1 WP____PB_0
WP____PB_1 gE____PB_0gE____PB_1
-0,1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9
Comp[1]P Comp[2]P
SofiaC_Bacteria_2pointBaseline_new5.M25 (OPLS/O2PLS-DA) R2Y(cum) Q2(cum)
-25 -20 -15 -10 -5 0 5 10 15 20 25
-40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40
t[2]
t[1]
WT____TB_0 WT____TB_1
WC____TB_0 WC____TB_1 WE____TB_0
WE____TB_1 WF____TB_0
WF____TB_1 WG____TB_0 WG____TB_1
WL____TB_0 WL____TB_1
WP____TB_0 WP____TB_1 gE____TB_0
gE____TB_1
-0,1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9
Comp[1]P Comp[2]P Comp[3]P Comp[4]P Comp[5]P Comp[6]P Comp[7]P
Comp No.
SofiaC_Bacteria_2pointBaseline_new5.M23 (OPLS/O2PLS-DA) R2Y(cum) Q2(cum)
15 expected trend is followed with just two exception; WT and ∆galE, that were expected to be more similar to ∆waaL and ∆waaP. If looking at the size measurements for bacteria there can be seen that the hydrodynamic size is relatively large in the range of 4000 - 8000 nm although if comparing that to the size observed in the TEM images in Figure 2 (1000-1500 nm) it can be seen that the hydrodynamic size measured is much larger than the size of one bacterium. Indicating that the bacteria, in the suspention used in DLS measurement, are aggregating. There were also variation between data points for the bacterial samples, an additional indication of aggregation.
For vesicles there is no similarity in trend compared to bacteria and they do not follow the expected trend at any point. For example, vesicles from ∆waaF/flhD and
∆waaG/flhD was expected to be smaller than the other mutants. These results indicate that the size of the vesicles do not correlate to the size of the bacterial aggregates Another observation regarding the size of vesicles is that the measurements show that they are larger than expected. As mentioned previously, earlier studies indicate that the vesicle size would be 50-250 nm, the result from this study indicate that vesicles from E.coli are more likely to have a hydrodynamic size in the range of 200-400 nm (unless they also aggregate as the bacterial cells do).
Due to that the bacteria seem to have a higher tendency to cluster, fewer reliable DLS data points could be collected and therefore is there a higher uncertainty, regarding the hydrodynamic size values, for bacteria than for vesicle.
a b
Figure 15. Trend in hydrodynamic size for bacteria (a) vs. vesicle (b). Samples are in Tris buffer.
Our hypothesis regarding the zeta potential of bacteria compared to vesicles was that vesicles from one specific mutant would have the same zeta potential as the bacteria and/or that the zeta potential would follow the same trend as for bacteria. If comparing the bacteria and their vesicles regarding zeta potential, it can be seen that this hypothesis were proved incorrect and that there is no similarity in values or trend, as can be seen in Figure 16. Indicating that the zeta potential of the vesicle is not correlated to the zeta potential of the bacteria, unless aggregation affects the zeta potential.
0 2000 4000 6000 8000 10000
Hydrodynamic size (d.nm)
Bacteria strain
0 100 200 300 400 500 600 700
Hydrodynamic size (d.nm)
Bacterial strain
16 Figure 16. Trend in zeta potential for bacteria vs. vesicle. Samples are in Tris buffer.
Moreover, from the Figures 15 and 16, it can be seen that vesicle hydrodynamic size and zeta potential seem to follow some kind of weak trend, namely that, a bigger vesicle will have a slightly more negative zeta potential (although it is within the standard deviation). However, this kind of trend can not be found regarding the hydrodynamic size and the zeta potential of bacterial aggregates, see Figure 15 and 16. Due to that both hydrodynamic size and zeta potential are affected by pH and ionic strength it could be expected that there would be some correlation between size and zeta potential of bacteria. For example, it could be expected that bacteria with lower zeta potential will have a higher tendency to aggregate, i.e. an opposite trend from the trend for vesicles.
Another aspect that was needed to be investigated was how the zeta potential and the hydrodynamic size were affected by vesicles ageing. The results indicates that the vesicles gets smaller with time and that the zeta potential perhaps gets slightly more negative as can be seen in Figure 17 a and b. The change in hydrodynamic size can be due to different factors such as the vesicles shrink or that there is a change in surface structure and/or chemical composition. One explanation to the alternation in size could be that the vesicles rearrange into smaller vesicle. The change in zeta potential could also be due to rearrangement of the LPS chains. It could also be a result from concentration of the surface charge as a result of the shrinking of the vesicle.
However, since the decrease in zeta potential is within the standard deviation it is very uncertain that it is decreasing, more measurements would be needed to see if this is a trend or not.
a b Figure 17. Effects on hydrodynamic size and zeta potential, of vesicle, after
letting the sample grow old. Samples are in Tris buffer.
‐70
‐60
‐50
‐40
‐30
‐20
‐10 0
Δwaaf/fhlD ΔwaaG/fhlD ΔwaaL/flhD ΔwaaP/flhD WT/flhD ΔgalE/flhD
Zeta potential (mV)
Bacteria strain
Bacteria Vesicle
0 100 200 300 400 500
ΔwaaG ΔgalE Hydrodynamic size (d.nm)
Bacteria strain
Day 1 Day 2
‐25
‐20
‐15
‐10
‐5 0
ΔwaaG ΔgalE
Zeta potential (mV)
Bacteria strain
Day 1 Day 2
17 There were also changes in zeta potential and hydrodynamic size of vesicles due to the growth state of bacteria. As can be seen in Figure 18a and b, bacteria in the exponential phase (7h) produce bigger and less negative vesicles. Compared to in the stationary phase (15h) where vesicles seem to be smaller but more negative.
As mentioned, earlier studies on vesicles from Pseudomonas aeruginosa indicate that there is no change in size but in zeta potential with growth phase (3). For E.coli, both zeta and hydrodynamic size were found to be affected by the bacterial growth phase. Interesting is also that vesicles produced after 7 hours had a size larger than the earlier documented range of size of approximatly 50-250 nm and that vesicles produced after 15 hours were in that size range. Previously there does not seem to have been any studies of the size after specific time periods in the growth curve of E.coli.
a b
Figure 18. Variations on vesicle zeta potential and hydrodynamic size due to bacterial growth phase. Samples are in Tris buffer.
As can be seen in Figure 19 a and b there is no change in zeta potential of the bacteria during the growth curve although there are changes in size. Changes in size would be expected during a living organisms growth curve. They grow bigger in the exponential phase, which is the phase where they reproduce through cell division. The observed changes in hydrodynamic size can also be due to that bacteria have different tendency to aggregate in different state of its growth curve. And the non existing change in zeta potential is probably due to that the bacteria can regulate its surface charge to be constant, maybe by vesiculation. This regulation can not be performed by vesicles due to that they are not living organisms.
a b Figure 19. Variations on bacterial zeta potential and hydrodynamic size due to
bacterial growth phase. Samples are in Tris buffer.
ATR-FTIR spectroscopic analysis
IR spectra from both vesicles and bacteria can be seen in Figure 20. For all the IR spectra most peaks are overshadowed by the peaks from water, mainly at around 3300 cm-1, 2100 cm-1 and 1635 cm-1, indicating that the buffer spectra were
0 100 200 300 400
ΔwaaF WT Hydrodynamic size (d.nm)
Bacterial strain 7 h 15 h
‐25
‐20
‐15
‐10
‐5 0
ΔwaaF WT
Zeta potential (mV)
Bacteria strain
7 h 15 h
0 2000 4000 6000 8000
WT waaC
Hydrodynamic size (n.dm)
Bacteria strain
Exp.
Late exp.
Stat.
‐70
‐60
‐50
‐40
‐30
‐20
‐10 0
WT waaC
Zeta potential (mV)
Bacterial strain
Exp.
Late exp.
Stat.
18 incompletely subtracted. Furthermore, there is also a region around 2900 cm-1 that corresponds to hydrocarbons mainly from lipids. Moreover, for E.coli bacteria, it can be seen a region between 1550 cm-1 and 1080 cm-1 that corresponds to vibrations from the proteins in the bacteria. Other components in this region will be overshadowed by the protein peaks, due to general abundance of proteins in bacteria.
When comparing with the IR spectra of vesicles, the peaks corresponding to proteins are less intense or completely absent and there is another region in the area of 1100- 800 cm-1, most likely corresponding to carbohydrates, that is more intense. This region of peaks is most likely present in bacteria spectra as well but due to the large amount of peptides the carbohydrate peaks will be relatively small. Another possible contribution to these peaks might be DNA or RNA although, due to the high amount of sugar in the vesicles, the DNA and RNA vibrations are most likely less pronounced.
If looking even closer at the region between 1100 cm-1 and 1200 cm-1, in Figure 21, it can be seen that there are four main peaks; three of them differ from the spectra of bacteria. It is the peak at 1027, 1105 cm-1 and 1152 cm-1, the peak at 1082 cm-1 is represented in both vesicle spectra and bacteria spectra and most likely correspond to phosphate and sugar vibrations.
Figure 20. ATR-FTIR spectra for bacteria (red) and vesicles (black). Numbers indicates wavenumber (cm-1) of the peaks. Samples are in Tris buffer.
1546
1082 1027 1152
1105
2111
1456 1233 3287
1634
1397
1000 1500
2000 2500
3000 3500
Wavenumber cm-1
0.00.20.40.60.8
Absorbance Units
19 Figure 21. FTIR spectra of different vesicles compared to a ∆waaL bacterium spectra (light blue). The different vesicles spectra are as followed; ∆galE (pink),
∆waaF (black), ∆waaG (red), ∆waaL (blue), ∆waaP (green) och WT (grey).
Numbers indicates wavelength (cm-1) of the peaks. Samples are in Tris buffer.
Comparisons of spectrum from IR analyses of bacteria and vesicles have been made using OPLS-DA multivariate analysis. The results are presented by Scores plots, Loadings plots and the original spectra. The three different presentations are all correlated. In the Loadings plot the areas at the positive side of the spectra will be more intensive in the samples on the positive side of the Scores plot and vice versa.
Furthermore, there can also be seen which peaks in the original spectra that the peaks in the Loadings plot originates from. The P(corr) value in the Loadings plot is a value that indicates how much the samples differs at a specific wavenumber in the IR spectra. If the P(corr) value is strongly positive or negative then there is a strong difference between the samples. Conversely, if peak values in the Loadings plot are close to zero, there is a weak difference between the samples.
Noteworthy, the data is total area normalized in OPUS; meaning that the intensity of the peaks is relative to each other and that changes in one area will alter the rest. This will have a consequence in the Loadings plot where big differences in small peaks will be overshadowed by small differences in large peak areas.
Prior to the multivariate analysis, all spectra were baseline corrected and the buffer spectrum was subtracted from the other spectra.
A comparison between bacteria and vesicle can be seen in Figures 22-25. The bacteria mutants are those without flagella due to that these are the ones that the vesicles originate from and it will also ensure that there is no influence from flagella on the results. In the Scores plot, Figure 22, a clear distinction between bacteria and vesicles can be seen.
Comparing the Scores plot in Figure 22 to the Loadings plot in Figure 21, there can be seen that the peaks in the positive area of the Loadings plot corresponds to the samples marked black in the Scores plot, i.e. to the vesicle samples. Comparing the Loadings plot to the original spectra in Figure 24 and 25, there can be seen that the negative area, in the Loadings plot, correspond to the area marked black in the IR spectra. Comparing the Scores plot to the IR spectra, there can be seen that the area marked black in the IR spectra, is the peaks that differentiate vesicles from bacteria.
The multivariate analysis result confirms the previous indications that the main component in the vesicle that differs from bacteria is the area of peaks in the region
