Department of Physics & Astronomy
Detroit, MI USA
BIOMAS: Bacteria Identification by Optical Molecular and Atomic Spectroscopy
THEORY
Laser-induced breakdown
spectroscopy. (LIBS) is an all-optical analytical technique that
utilizes a pulsed laser and optical spectroscopy to detect and identify
the elements present in a target material, including pathogenic
bacteria. During LIBS, a short pulse of laser light is focused to
a small spot on a bacteria-containing target (blood, sputum, etc.)
which creates a high-temperature (10,000–20,000 K) micro-plasma within
the focal region of the laser. During this process, the sample
illuminated by the laser is completely vaporized (“ablated”). The
sample is reduced to its constituent atomic components, which are
entrained in the micro-plasma plume. A careful spectroscopic
analysis of the light emitted from this plasma plume yields
identifiable emission lines only from those elements that were present
in the target (Cremers and Radziemski, 2006; Miziolek et al.
2006). The positive identification of many elemental lines within
the emission spectrum then provides an immediate and unique spectral
“fingerprint” which positively identifies the bacteria in the
sample. This sample could be a bacterial colony, a solid surface,
a vial of bacteria-containing water, or any clinical sample (e.g.
sputum, blood, etc.). Any target material which can be
illuminated by the laser can theoretically be tested to reveal the
presence and identity of a pathogen. Beginning in 2003, studies
began to show that the ability of LIBS to rapidly (within seconds or
minutes) detect harmful pathogens including those which cannot be
cultured within a reasonable amount of time or cannot be cultured at
all could offer a radically new paradigm to the health sciences
for the detection, identification, and control of infectious diseases.
LIBS has already been demonstrated to be useful when used in idealized settings to identify and discriminate a wide variety of microorganisms such as bacteria (Morel et al., 2003; Samuels et al., 2003; Dixon and Hahn, 2005; Kim et al., 2004, Baudelet et al., 2006a), fungi/mold spores, pollens, and proteins (Hybl et al., 2003; Samuels et al., 2003) based on the emission intensity of trace inorganic elements found in the microorganism (DeLucia, 2005). Much of the work to date has predominantly been focused on bacterial surrogates (which can be safely and easily handled) as a model for NIAID Category A and B priority pathogens, specifically Bacillus globigii and Bacillus subtilis var. niger (Kim et al., 2004) as a surrogate for Bacillus anthracis (anthrax), Escherichia coli as a surrogate for Yersinia pestis, Staphylococcus aureus as a surrogate for Staphilococcus epidermis, and Proteus mirabilis (Leone et al., 2004). Our group of Rehse and Palchaudhuri, the group of Baudelet et al. (Baudelet et al., 2006a; Baudelet et al., 2006b; Baudelet et al., 2006c), and others (Assion et al., 2003) have focused on Escherichia coli. To date our work remains the only cited reference on the analysis of pathogenic E. coli strains. In addition, we have demonstrated the ability to discriminate Pseudomonas aeruginosa from E. coli (Rehse et al., 2007) and have proven that P. aeruginosa samples cultured in nutrient media containing blood are identically identified to samples cultured in non-blood containing media. Prior to our work in this field, only one published report (Kim et al., 2004,) was conducted on live bacteria in culture, the rest using complicated and unrealistic drying and concentration of the bacteria into pellets for ease of detection.
The majority of the work described here has focused primarily on the relative concentrations of Na, Mg, Ca, P, K, Fe, and C identified in the LIBS spectra to form the basis of this discrimination. This approach has been demonstrated to be particularly useful in identifying Gram-negative bacteria. The reason that this is true is dependent on the biochemistry of the bacteria outer membrane.
We have chosen specifically to investigate Gram-negative bacteria because they have been classified traditionally on a serological basis, which is even now very much used in clinical labs and hospitals, rather than molecular biology techniques (PCR, rRNA, etc.). Because of antigenic variation, such serological classification needs further confirmation by the development of a new technology (at the atomic and molecular level). We strongly believe that the combination of optical modalities is ideally suited to this purpose.
Two types of bacterial surface structures form the basis for the serological classification system. There are the O-antigen and H-antigen of which the O-antigen of LPS (O) is more important than the H-antigen of flagella. The O-antigen identifies the serogroup and the H-antigen identifies its serotype. For example, more than 160 different serogroups of E. coli are already known, but most of them rarely cause disease, except O55.
For the bacterium, the outer membrane is first and foremost a permeability barrier, but primarily due to its polysaccharide content, it possesses many of the interesting and important characteristics of Gram-negative bacteria (Nikaido, 1973; Nikaido and Takae, 1979). The inner face of the outer membrane is composed of phospholipids similar to the phosphoglycerides that compose the plasma membrane. The outer face of the outer membrane may contain some phospholipids, but mainly it is formed by a different type of amphilic molecule which is composed of lipopolysaccharide (LPS) (Kamio and Nikaido, 1976; Raetz, 1990). Outer membrane proteins usually traverse the membrane and anchor the outer membrane to the underlying petidoglycan sheet. The LPS molecule that constitutes the outer face of the outer membrane is composed of a hydrophobic region, called lipid A, which is attached to a hydrophilic linear polysaccharide region, consisting of the core polysaccharide and the O-specific polysaccharide (Nikaido and Vaara, 1985). This outer leaflet of LPS molecules is composed of three distinct components. The innermost layer of LPS is the lipid A tail, which is anchored into the hydrophobic region of the outer membrane. The endotoxic property of LPS resides largely in the lipid A component (Stone, 1994). The core component contains an eight-carbon sugar called KDO as well as phosphates and possesses an overall electronegative charge (Al-Tahhan et al., 2000). The O-specific polysaccharide component extends tens of nanometers above the outer bilayer, which brings it into contact with the environment (Poxton, 1993; Pink et al., 2003).
a
lipid bi-layer
In such a complicated molecular system, how can atomic spectroscopy possibly give any information about the function of the organism? The key is that two specific divalent cations, Ca2+ and Mg2+, play a crucial role in stabilizing the membrane by binding adjacent LPS molecules (Leive, 1974). The exact mechanism of the stabilization of the cations is not completely clear, but we already know that the treatment of Gram-negative bacteria with a powerful chelating agent such as EDTA cause the dispersion of LPS molecules (Ibrahim et al., 1997). It is believed that Ca2+ and Mg2+ act to stabilize the entire membrane structure by forming metal ion bridges between phosphate groups of phospholipids or LPS and the membrane proteins (Asbell and Eagon, 1996;). Alternatively, it has been suggested that rather than acting as a series of intermolecular cross-bridges, the ions primary role is to perform charge neutralization of the electronegative KDO inner core (Snyder et al., 1999). Molecular dynamics simulations of the LPS/Ca2+ binding suggest that ions uniformly distributed across a small region of this inner core can accurately model measured transmembrane potentials (Lins and Straatsma, 2001). Moreover, the effect of displacement of these cation binding sites has been measured by wide angle x-ray diffraction to significantly alter the LPS packing structure from a hexagonal to a nonhexagonal lattice (Snyder et al., 1999). This could explain the changes in membrane permeability, perhaps the antigenic variation for serogrouping, as a function of cation concentration. This permeability changes as a function of cation concentration (Ca, Mg, Na, and Ba) is directly related to antibiotic efficacy against the bacteria (Pink et al., 2003; Ibrahim et al., 1997). It is these specific cations to which LIBS is particularly sensitive.
All of these works clearly demonstrate the potential usefulness of LIBS as an optical modality for the identification and discrimination of bacterial strains, as well as possibly providing the atomic basis of serological classification. The usefulness of LIBS will be greatly enhanced when combined with other optical modalities, specifically Raman spectroscopy (Burgio et al., 2000; Giakoumaki et al., 2006; Wiens et al., 2005; Noll and Fricke-Begemann, 2006).
Raman spectroscopy is a versatile molecular vibrational technique that has only recently been widely applied for identification purposes in microbiology (Maquelin et al., 2002; Navratil et al., 2006). In this application, a laser beam is non-destructively incident upon a bacterial target and the inelastically scattered light is carefully dispersed. Shifts in the scattered photon energy corresponding to vibrational modes in the molecules of the target are then measured to determine molecular composition. This technique has been effectively used on bacterial slurries (Goodacre, 1998a), to probe bacterial colonies (Choo-Smith, 2001), to identify and discriminate medically important microbes including E. coli, Klebsiella spp., Enterococcus spp. (such as E. hirae, E. durans, E. casseliflavus and E. gallinarum), Proteus mirabilis, Staphylococcus aureus, Candida strains, Pseudomonas spp, (Maquelin et al., 2002; Jarvis and Goodacre, 2004a), and even to probe individual microbial cells (Huang et al., 2004; Harz et al., 2005). Of particular interest is the ability to discriminate spectra from vancomycin-resistant strains of Enterococcus faecium from non-resistant strains (Maquelin et al., 1998) and methicillin-sensitive and –resistant strains of S. aureus (Goodacre et al., 1998b).
A very recent advance called surface-enhanced Raman scattering (SERS) has made Raman spectroscopy more widely applicable in this biophotonic application by enhancing the spectroscopic signal by factors of 103 up to 106, and by quenching the contaminating background fluorescence (Kneipp et al., 2002). This advance has made the real-time identification and discrimination of a wide variety of bacteria possible (Grow et al., 2003; Jarvis and Goodacre. 2004b; Jarvis et al. 2004). In addition, a limited number of groups worldwide have begun utilizing “hybrid” LIBS/Raman systems for a variety of applications (Boyain-Goitia et al., 2003). The reason for using “hyphenated” techniques (such as LIBS-Raman) is that the quantity of information added by using orthogonal modalities increases very rapidly. In particular, the ability to obtain both atomic and molecular information by using the LIBS-Raman technique is impossible using either technique alone.
Chemometrics. Many opponents of LIBS-Raman optical modalities continue to insist that common contaminants or “dirty” samples will forever preclude the ability of these techniques to provide useful clinical diagnosis of bacterial infection. However, previous work has already demonstrated that with appropriate statistical strategies, the optical technologies can be successfully used to discriminate the target bacteria from common contaminants and biological “interferents” (Beddows and Telle, 2005; Munson et al., 2005; Hybl et al., 2006). In particular, the use of sophisticated software to perform a chemometric analysis has made this discrimination from background signatures possible. Such techniques, such as Principal Component Analysis (PCA) and Discriminant Function Analysis (DFA) which is described in detail in the Preliminary Studies section have allowed the use of the entire optical spectrum from the bacteria to discriminate it from closely related yet different biological sample contaminants.
When using these chemometric techniques, usually as much unique data as possible from each sample is collected. The mathematic algorithms then are able to identify subtle, yet reproducible differences in spectral information that uniquely identify the organism. The combination of the two optical modalities of LIBS and Raman spectroscopy would provide complementary spectral information which, when taken as a complete spectral fingerprint, offers a much greater opportunity for identification than either modality alone.
Sensitivity. Although prior LIBS experiments have been performed on colonies grown in culture (approximately 1,000-10,0000 bacteria ablated per test) for strongest signal, experiments conducted by the PI have also been performed on dilute liquid cultures with positive results and Raman analysis has been performed on single microbial cells, indicating the possibility of applying both of these techniques to real clinical samples.
Ablated E. coli on an agar substrate
The atomic composition of E. coli. (and lines used in discrimination)
LIBS has already been demonstrated to be useful when used in idealized settings to identify and discriminate a wide variety of microorganisms such as bacteria (Morel et al., 2003; Samuels et al., 2003; Dixon and Hahn, 2005; Kim et al., 2004, Baudelet et al., 2006a), fungi/mold spores, pollens, and proteins (Hybl et al., 2003; Samuels et al., 2003) based on the emission intensity of trace inorganic elements found in the microorganism (DeLucia, 2005). Much of the work to date has predominantly been focused on bacterial surrogates (which can be safely and easily handled) as a model for NIAID Category A and B priority pathogens, specifically Bacillus globigii and Bacillus subtilis var. niger (Kim et al., 2004) as a surrogate for Bacillus anthracis (anthrax), Escherichia coli as a surrogate for Yersinia pestis, Staphylococcus aureus as a surrogate for Staphilococcus epidermis, and Proteus mirabilis (Leone et al., 2004). Our group of Rehse and Palchaudhuri, the group of Baudelet et al. (Baudelet et al., 2006a; Baudelet et al., 2006b; Baudelet et al., 2006c), and others (Assion et al., 2003) have focused on Escherichia coli. To date our work remains the only cited reference on the analysis of pathogenic E. coli strains. In addition, we have demonstrated the ability to discriminate Pseudomonas aeruginosa from E. coli (Rehse et al., 2007) and have proven that P. aeruginosa samples cultured in nutrient media containing blood are identically identified to samples cultured in non-blood containing media. Prior to our work in this field, only one published report (Kim et al., 2004,) was conducted on live bacteria in culture, the rest using complicated and unrealistic drying and concentration of the bacteria into pellets for ease of detection.
The majority of the work described here has focused primarily on the relative concentrations of Na, Mg, Ca, P, K, Fe, and C identified in the LIBS spectra to form the basis of this discrimination. This approach has been demonstrated to be particularly useful in identifying Gram-negative bacteria. The reason that this is true is dependent on the biochemistry of the bacteria outer membrane.
We have chosen specifically to investigate Gram-negative bacteria because they have been classified traditionally on a serological basis, which is even now very much used in clinical labs and hospitals, rather than molecular biology techniques (PCR, rRNA, etc.). Because of antigenic variation, such serological classification needs further confirmation by the development of a new technology (at the atomic and molecular level). We strongly believe that the combination of optical modalities is ideally suited to this purpose.
Two types of bacterial surface structures form the basis for the serological classification system. There are the O-antigen and H-antigen of which the O-antigen of LPS (O) is more important than the H-antigen of flagella. The O-antigen identifies the serogroup and the H-antigen identifies its serotype. For example, more than 160 different serogroups of E. coli are already known, but most of them rarely cause disease, except O55.
For the bacterium, the outer membrane is first and foremost a permeability barrier, but primarily due to its polysaccharide content, it possesses many of the interesting and important characteristics of Gram-negative bacteria (Nikaido, 1973; Nikaido and Takae, 1979). The inner face of the outer membrane is composed of phospholipids similar to the phosphoglycerides that compose the plasma membrane. The outer face of the outer membrane may contain some phospholipids, but mainly it is formed by a different type of amphilic molecule which is composed of lipopolysaccharide (LPS) (Kamio and Nikaido, 1976; Raetz, 1990). Outer membrane proteins usually traverse the membrane and anchor the outer membrane to the underlying petidoglycan sheet. The LPS molecule that constitutes the outer face of the outer membrane is composed of a hydrophobic region, called lipid A, which is attached to a hydrophilic linear polysaccharide region, consisting of the core polysaccharide and the O-specific polysaccharide (Nikaido and Vaara, 1985). This outer leaflet of LPS molecules is composed of three distinct components. The innermost layer of LPS is the lipid A tail, which is anchored into the hydrophobic region of the outer membrane. The endotoxic property of LPS resides largely in the lipid A component (Stone, 1994). The core component contains an eight-carbon sugar called KDO as well as phosphates and possesses an overall electronegative charge (Al-Tahhan et al., 2000). The O-specific polysaccharide component extends tens of nanometers above the outer bilayer, which brings it into contact with the environment (Poxton, 1993; Pink et al., 2003).
a
lipid bi-layerIn such a complicated molecular system, how can atomic spectroscopy possibly give any information about the function of the organism? The key is that two specific divalent cations, Ca2+ and Mg2+, play a crucial role in stabilizing the membrane by binding adjacent LPS molecules (Leive, 1974). The exact mechanism of the stabilization of the cations is not completely clear, but we already know that the treatment of Gram-negative bacteria with a powerful chelating agent such as EDTA cause the dispersion of LPS molecules (Ibrahim et al., 1997). It is believed that Ca2+ and Mg2+ act to stabilize the entire membrane structure by forming metal ion bridges between phosphate groups of phospholipids or LPS and the membrane proteins (Asbell and Eagon, 1996;). Alternatively, it has been suggested that rather than acting as a series of intermolecular cross-bridges, the ions primary role is to perform charge neutralization of the electronegative KDO inner core (Snyder et al., 1999). Molecular dynamics simulations of the LPS/Ca2+ binding suggest that ions uniformly distributed across a small region of this inner core can accurately model measured transmembrane potentials (Lins and Straatsma, 2001). Moreover, the effect of displacement of these cation binding sites has been measured by wide angle x-ray diffraction to significantly alter the LPS packing structure from a hexagonal to a nonhexagonal lattice (Snyder et al., 1999). This could explain the changes in membrane permeability, perhaps the antigenic variation for serogrouping, as a function of cation concentration. This permeability changes as a function of cation concentration (Ca, Mg, Na, and Ba) is directly related to antibiotic efficacy against the bacteria (Pink et al., 2003; Ibrahim et al., 1997). It is these specific cations to which LIBS is particularly sensitive.
All of these works clearly demonstrate the potential usefulness of LIBS as an optical modality for the identification and discrimination of bacterial strains, as well as possibly providing the atomic basis of serological classification. The usefulness of LIBS will be greatly enhanced when combined with other optical modalities, specifically Raman spectroscopy (Burgio et al., 2000; Giakoumaki et al., 2006; Wiens et al., 2005; Noll and Fricke-Begemann, 2006).
Raman spectroscopy is a versatile molecular vibrational technique that has only recently been widely applied for identification purposes in microbiology (Maquelin et al., 2002; Navratil et al., 2006). In this application, a laser beam is non-destructively incident upon a bacterial target and the inelastically scattered light is carefully dispersed. Shifts in the scattered photon energy corresponding to vibrational modes in the molecules of the target are then measured to determine molecular composition. This technique has been effectively used on bacterial slurries (Goodacre, 1998a), to probe bacterial colonies (Choo-Smith, 2001), to identify and discriminate medically important microbes including E. coli, Klebsiella spp., Enterococcus spp. (such as E. hirae, E. durans, E. casseliflavus and E. gallinarum), Proteus mirabilis, Staphylococcus aureus, Candida strains, Pseudomonas spp, (Maquelin et al., 2002; Jarvis and Goodacre, 2004a), and even to probe individual microbial cells (Huang et al., 2004; Harz et al., 2005). Of particular interest is the ability to discriminate spectra from vancomycin-resistant strains of Enterococcus faecium from non-resistant strains (Maquelin et al., 1998) and methicillin-sensitive and –resistant strains of S. aureus (Goodacre et al., 1998b).
A very recent advance called surface-enhanced Raman scattering (SERS) has made Raman spectroscopy more widely applicable in this biophotonic application by enhancing the spectroscopic signal by factors of 103 up to 106, and by quenching the contaminating background fluorescence (Kneipp et al., 2002). This advance has made the real-time identification and discrimination of a wide variety of bacteria possible (Grow et al., 2003; Jarvis and Goodacre. 2004b; Jarvis et al. 2004). In addition, a limited number of groups worldwide have begun utilizing “hybrid” LIBS/Raman systems for a variety of applications (Boyain-Goitia et al., 2003). The reason for using “hyphenated” techniques (such as LIBS-Raman) is that the quantity of information added by using orthogonal modalities increases very rapidly. In particular, the ability to obtain both atomic and molecular information by using the LIBS-Raman technique is impossible using either technique alone.
Chemometrics. Many opponents of LIBS-Raman optical modalities continue to insist that common contaminants or “dirty” samples will forever preclude the ability of these techniques to provide useful clinical diagnosis of bacterial infection. However, previous work has already demonstrated that with appropriate statistical strategies, the optical technologies can be successfully used to discriminate the target bacteria from common contaminants and biological “interferents” (Beddows and Telle, 2005; Munson et al., 2005; Hybl et al., 2006). In particular, the use of sophisticated software to perform a chemometric analysis has made this discrimination from background signatures possible. Such techniques, such as Principal Component Analysis (PCA) and Discriminant Function Analysis (DFA) which is described in detail in the Preliminary Studies section have allowed the use of the entire optical spectrum from the bacteria to discriminate it from closely related yet different biological sample contaminants.
When using these chemometric techniques, usually as much unique data as possible from each sample is collected. The mathematic algorithms then are able to identify subtle, yet reproducible differences in spectral information that uniquely identify the organism. The combination of the two optical modalities of LIBS and Raman spectroscopy would provide complementary spectral information which, when taken as a complete spectral fingerprint, offers a much greater opportunity for identification than either modality alone.
Sensitivity. Although prior LIBS experiments have been performed on colonies grown in culture (approximately 1,000-10,0000 bacteria ablated per test) for strongest signal, experiments conducted by the PI have also been performed on dilute liquid cultures with positive results and Raman analysis has been performed on single microbial cells, indicating the possibility of applying both of these techniques to real clinical samples.
Ablated E. coli on an agar substrate
The atomic composition of E. coli. (and lines used in discrimination)
|
wavelength
(nm)
|
|
|
213.618
|
P I
|
|
214.914
|
P I
|
|
247.856
|
C I
|
|
253.56
|
P I
|
|
279.553
|
Mg II
|
|
280.271
|
Mg II
|
|
285.213
|
Mg I
|
|
373.69
|
Ca II
|
|
383.231
|
Mg I
|
|
383.829
|
Mg I
|
|
393.366
|
Ca II
|
|
396.847
|
Ca II
|
|
422.673
|
Ca II
|
|
430.253
|
Ca I
|
|
518.361
|
Mg I
|
|
585.745
|
Ca I
|
|
588.995
|
Na I
|
|
589.593
|
Na I
|
|
769.896
|
K I
|
DFA Discriminates Bacteria from Growth Medium and Ablation Substrate
E. coli Strain Discrimination
E. coli vs. P. aeruginosa Discrimination
Posters
LIBS for Counter-Bioterrorism (presented at ASM2007)
PAPERS
S.J. Rehse, J. Diedrich, and S. Palchaudhuri, “Identification and Discrimination of Pseudomonas Aeruginosa Bacteria Grown in Blood andBile by Laser-induced Breakdown Spectroscopy,” Spectrochimica Acta Part B 62, 1169-1176 (2007).
J. Diedrich, S.J. Rehse, and S. Palchaudhuri, “Pathogenic Escherichia coli Strain Discrimination Using Laser-Induced Breakdown Spectroscopy,” Journal of Applied Physics 102, 014702 (2007).
J. Diedrich, S.J. Rehse, and S. Palchaudhuri, “Escherichia coli identification and strain discrimination using nanosecond laser-induced breakdown spectroscopy,” Applied Physics Letters 90, 163901 (2007).
This site updated, Jan. 03, 2008.
Back to the Rehse Group main page