- Email: [email protected]
Analysis of bacterial polyhydroxybutyrate production by multimodal nanoimaging
Biotechnology Advances 31 (2013) 369–374
Contents lists available at SciVerse ScienceDirect
Biotechnology Advances journal homepage: www.elsevier.com/locate/biotechadv
Analysis of bacterial polyhydroxybutyrate production by multimodal nanoimaging Céline Mayet, Ariane Deniset-Besseau, Rui Prazeres, Jean-Michel Ortega, Alexandre Dazzi ⁎ Laboratoire de Chimie Physique, Université Paris-Sud, 91405 ORSAY, France
a r t i c l e
i n f o
Available online 22 May 2012 Keywords: Infrared nanospectromicroscopy Transmission electron microscopy Rhodobacter capsulatus Polyhydroxybutyrate
a b s t r a c t In this paper, we will employ two microscopy techniques, transmission electron microscopy and infrared nanospectromicroscopy, to study the production of polyhydroxybutyrate in Rhodobacter capsulatus and to evaluate the influence of glucose and acetone on the production yield. The results overlap which leads us to a consistent conclusion, highlighting that each technique brings specific and complementary information. By using electron microscopy and infrared nanospectromicroscopy we have proved that both glucose and acetone had a positive effect on the biopolymer production, although the first study done by Fourier transform infrared spectroscopy only identified the effect of acetone. In conclusion, we have now established a method to be able to perform fast diagnostic for PHB production. © 2012 Elsevier Inc. All rights reserved.
1. Introduction Polyhydroxyalcanoates (PHAs) are known as natural polyesters produced by bacteria with a great potential in biodegradable plastic applications (Doi, 1990). The scientific interest of the production optimization has come from the knowledge of the PHA biosynthetic pathways, the numerous promising bacterial species and finally by the genetic engineering and biochemical techniques efficiencies (Hahn et al., 1998; Kranz et al., 1997; Wong and Lee, 1998; Zhao and Cheng, 2004). Polyhydroxybutyrates (PHB) are part of this PHAs family. Produced by bacterial fermentation from carbon sources such as sugars or fatty acids, they are then degraded in soil, sludge, or seawater, under optimal conditions, at an extremely fast speed. Furthermore PHB has mechanical properties similar to those of thermoplastic synthetic polymers derived from petroleum, such as polypropylene or polyethylene. Nevertheless, the PHB bacterial production was labeled by industrials as only a prospective project because the biotechnological processes (fermentation, extraction) had too high costs compared to petroleum products. Recently, many scientists have refocused their studies on those polyesters by integrating the industrial constraints and have proposed attractive solutions to optimize PHB production yield in different strains of Alcaligenes, Azotobacter, Pseudomonas or recombinant Escherichia coli (Hahn et al., 1998; Naik et al., 2008; Wong and Lee, 1998). In this context, the study of PHB at the single cell level is crucial for the PHB biosynthesis efficiency to be evaluated. Usually, the PHB yield is first evaluated by biochemical analysis and then confirmed by transmission electron microscopy (TEM). But these techniques are time-consuming, and real-time diagnosis for PHB production is not possible.
⁎ Corresponding author. Tel.: + 33 1 69 15 32 76. E-mail address: [email protected] (A. Dazzi). 0734-9750/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.biotechadv.2012.05.003
Consequently, we propose a combination of infrared spectroscopy and microscopy, to estimate the production of bio-polymer to rapidly diagnose the efficiency of a bacterial strain as well as a culture condition. The ester carbonyl group of the PHB shows a specific mid-infrared response at 1740 cm− 1 which differs from the proteins carbonyl group (Amide I at 1650 cm− 1) (Misra et al., 2000; Naumann, 1998), making it possible to determine, by classical Fourier transformed infrared spectroscopy (FTIR), the mean production of PHB and to analyze the general behavior of a cell population. In order for this study to be performed at the sub-cellular level, a second technique is required with a nanometric resolution. We used an original set-up called AFMIR that couples an atomic force microscope (AFM) and a pulsed IR laser (Dazzi et al., 2005). This technique allows spectromicroscopy at the 10 nm scale with representative local spectra (Dazzi et al., 2008, 2010; Mayet et al., 2010). Our study was performed on Rhodobacter capsulatus whose PHB production is efficient (Madigan, 1990; Madigan et al., 2001; Pantazopoulous and Madigan, 2000). This study was done for different culture conditions that may increase the production yield. The study of PHB is difficult as PHBs are stored in the form of insoluble vesicles inside the bacteria, so it involves high-resolution microscopy such as TEM or our IR nanospectromicroscopy AFMIR. As we need to evaluate the influence of glucose and acetone on the production yield, the PHB vesicle size will be determined by TEM and AFMIR microscopy. Finally the advantages and the limitations of each technique will be discussed. 2. Materials and methods 2.1. Rhodobacter capsulatus cultures 2.1.1. Standard culture Wild type Rhodobacter capsulatus bacteria were grown in malate yeast medium supplemented with kanamycin (20 mg/ml) and tetracycline
370
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
(4 mg/ml) at 50% of the total volume. Cultures were cultivated darkness at 30 °C and kept for 48 h on a gyratory shaker (140 rpm).
2.1.2. Glucose culture The preparation of the culture was similar to the standard one, but 80% of the malate volume was substituted by glucose.
2.1.3. Acetone culture In this case all the malate was substituted by acetone. In order to be studied in infrared, the cell suspension of each culture was centrifuged at 1000 rpm for 15 min. The supernatant was removed and the cell pellet diluted in distilled water. This process was repeated three times in order to obtain suitably clean samples. Finally, a droplet of the cellular suspension was deposited on ZnSe (transparent in midinfrared) coverglass for FTIR spectroscopy or ZnSe prisms for AFMIR studies and dried at room temperature.
2.2. FTIR analysis The infrared spectra was acquired with a Bruker Vertex 70 FTIR spectrometer between 4000 and 400 cm − 1 with a spectral resolution of 4 cm − 1 using an MCT detector with a liquid nitrogen cooling system. The biomass can be directly characterized by the amplitude of the Amide I absorption band; hence we have normalized all FTIR spectra by this band so that we can compare the amount of PHB production for the different culture condition.
2.3. Principle of the nanospectromicroscopy AFMIR The AFMIR instrument combines an Atomic Force Microscope (AFM) with an infrared pulsed tunable laser (Fig. 1). A sample is placed on a ZnSe prism and then irradiated with the laser. When the laser wavelength is tuned on an absorption band of the sample, the absorbed laser light causes a photothermal effect and a temperature rise. This increase in temperature creates a local thermal expansion monitored by the tip of the AFM. The expansion of the sample generates a rapid impulse that leads to the oscillation of the cantilever on its eigenmodes. The oscillations amplitudes, on the 4-quadrant detector, are directly proportional to the energy absorbed. Hence, we can reconstruct the entire absorption spectrum of the sample by tuning the incident laser wavenumber, and comparing it with those obtained by classical IR spectroscopy techniques like FTIR (Dazzi et al., 2010). The main advantage of this technique is its resolution of less than 100 nm. It allows a chemical study of the samples at the subcellular level. This has already been demonstrated in microbiological (Dazzi et al., 2007, 2008; Mayet et al., 2008) and nanophotonic domains (Houel et al., 2007).
2.4. Sample preparation for TEM imaging For all different cultures, bacteria were fixed with glutaraldehyde 2% and paraformaldehyde 2% in a buffer of sodium cacodylate 0.1 mol/l (pH 7.4) and then post-fixed in 1% buffered osmium tetroxide. The bacteria was then completely dehydrated with ethanol at room temperature and then embedded in epoxy resin (polymerization at 60 °C for 48 h). Ultrathin sections of 50 nm thickness were cut with a diamond knife, deposited on copper grids (mesh 200) and then stained with uranyl acetate aqueous solution (20 min) and lead citrate (5 min). Samples were observed at 80 kV with Philips EM208 transmission electron microscope (80 kV, mode light background) equipped with a CCD camera. 2.5. AFMIR images analysis The AFMIR imaging analysis was performed with ImageJ software. ImageJ software is an open source software commonly used by scientist in imaging domain. This software is useful to visualize imaging data and offers the possibility of applying mathematical tools for the analysis. Each chemical mapping of PHB (already normalized by the input laser power) was renormalized by the corresponding topography (absorption signal is divided by the corresponding height), giving us the absorption density value. The signal coming from the bacteria was isolated using the threshold filter (removing the image pixels corresponding to the surface). The maximum of absorption density was then extracted for each bacterium. We assume that the maximum of these maxima corresponds to a full diameter of the vesicle also ensuring that the absorption signal is not an isolated point but is linked to a large area of absorption. As the maximum absorption density value and the volume of each bacterium are known, the theoretical maximum absorption signal if the bacterium was only full of PHB is easy to calculate. Then the real values of absorption density can be integrated (given by the chemical mapping divided by topography) on only the pixels corresponding to the surface of the bacterium to have the integral of density absorption. Finally, by dividing this last value with the theoretical maximum absorption we obtain the % of PHB absorption inside the volume of the bacterium. This process of analysis was applied for tens of bacteria for each culture condition to obtain the average PHB absorption (%). This parameter does not directly give the % of PHB volume occupation per bacterium but is a very good indicator of this value. The error comes from buried vesicles that give a smaller integrated signal than the vesicles close to the surface. So the % of PHB absorption is lower than the % of volume occupation. 3. Results 3.1. Global study by FTIR spectroscopy For each culture (standard, glucose and acetone), an FTIR spectrum has been recorded between 1850 and 1550 cm − 1 (Fig. 2) after
Fig. 1. Scheme of the AFMIR experimental setup. The sample is deposited on the upper side of the ZnSe prism and illuminated by the infrared pulsed laser. The resulting photothermal expansion generates a brief force under the AFM tip inducing the ringing of the AFM cantilever. The oscillation signal is recorded and stored in an oscilloscope.
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
371
Fig. 2. FTIR spectra of different culture types. The common culture is represented in plain line, the glucose in dotted line and the acetone in dashed line.
48 h of cultivation, which is sufficient for the culture to reach the steady state. This range allows the identification of both the ester carbonyl absorption band (C O, 1740 cm − 1) of the PHB compound and the Amide I band associated to bacteria proteins (1650 cm − 1). The results showed that the ester carbonyl band (1740 cm − 1) is prominent for the acetone culture medium whereas for the standard and glucose culture condition FTIR spectra show a poor absorption at this wavenumber. The band ratios of the ester_carbonyl/Amide_I are the following: 0.39 for acetone culture, 0.09 for glucose and 0.07 for standard. From these measurements, we can conclude that using glucose has no detectable effect on the PHB production whereas substituting malate by acetone is a relevant strategy to improve this production. However, this approach gives us only the relative increase of ester carbonyl vibrations in function of the carbonyl of the proteins. Furthermore, even if the study on the entire culture of bacteria gives us a good indication of PHB production, the effects of the culture medium substitution remain unclear at a single cell level. More precisely, the size, the shape and the repartition of the vesicles inside the bacteria as well as the increase of the bacteria volume which have an effect of the PHB production are unknown. Microscopy techniques are a good way to verify these points.
the bacteria, it is not possible to estimate correctly the % of PHB per bacteria. Therefore it is necessary for an intricate method to be developed of keeping the sections in the cutting serial order to be able, when using the TEM microscope, to reconstruct a complete bacterium by imaging the consecutive sections. We suggest the use of a second technique which is less constraining to assess, more accurately, the size of the mature vesicles.
3.2. TEM imaging
3.3. AFMIR PHB mapping
For each culture condition, over 200 bacteria (Fig. 3) were analyzed and their vesicle sizes were systematically measured. The measurements are given Fig. 4 where the size histograms for each culture have been compared. On each histogram the shape of the distribution has been added to highlight their maxima. For each case, the histogram shows a maximum size distribution centered around 60–80 nm. This corresponds to the smallest vesicles being more numerous and clustered around the periphery of the bacteria. This confirms the hypothesis of pre-maturation with two steps: first an accumulation of small vesicles in the membrane and then a migration and an accumulation of the vesicles towards the center of the bacteria (Thomson et al., 2010). The second peak corresponds to the mature vesicles size and is centered around 170 nm for the standard culture and 250–300 nm for the two other cultures. In conclusion, the presence of glucose or the acetone substitution in the culture medium has a similar effect and increase the size of mature vesicles. This local analysis is clearly efficient in obtaining an accurate estimation of small size of the vesicles and a good statistic evaluation. However, as the TEM technique requires a thin section (50–100 nm) to analyze
The relevance of our IR nanospectromicroscope AFMIR in identifying and detecting PHB vesicles with a size of 50 to 300 nm inside Rhodobacter has already been demonstrated (Mayet et al., 2010). In this paper, we have shown that taking the full width at the half maximum of the absorption signal will give a good estimation of the vesicle size. The laser is tuned at the ester carbonyl band (1740 cm− 1) wavenumber and each sample (dried bacteria) is scanned by the AFM tip to acquire simultaneously the topography and the chemical mapping of a single cell (Fig. 5). For each culture, we scanned tens of bacteria for an initial insight of the average PHB production. After the acquisition of the chemical mapping, systematic analysis of each images is done by imageJ software in order for the size of the vesicles to be determined. This analysis indicates that in a standard culture the size of PHB vesicles are around 200–300 nm (Fig. 5a, b) and are not present in all bacteria (approximately half posses a least one vesicle). For the glucose culture the size of vesicles is bigger, up to 400–500 nm (Fig. 5c, d) with at least one vesicle per bacterium. In the case of the acetone culture, it is evident that the size was around 800 nm (the larger size of bacteria) with more than one vesicle per bacterium (Fig. 5e, f). Moreover, for all
Fig. 3. Representative TEM images for the different bacteria cultures where bacteria appear in gray (contrast is proportional of the density of matter) and PHB vesicles in white. (a) Common culture. (b) Glucose culture. (c) Acetone culture. The black bar corresponds to 500 nm.
372
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
of the culture conditions the PHB vesicles seem to be localized at the apex the bacteria after maturation, which strongly agrees with the observation done by fluorescence microscopy on Beijerinckia indica bacteria (Jendrossek et al., 2007). This IR nanospectromicroscopy analysis allows us not only to easily detect the PHB, but also to estimate, by looking at the absorption distribution of the vesicles inside the bacterium, the percentage of PHB absorption inside its volume. The calculation of the average PHB absorption (on 20 bacteria per culture type) gives: Standard 11.2%, Glucose 12.5%, Acetone 22%. 4. Discussion The multimodal imaging analysis has given a good insight of the influence of glucose and acetone on the PHB production in Rhodobacter capsulatus. The FTIR analysis shows that only the addition of acetone in the culture medium allows the bacteria to produce more PHB (ester_carbonyl/Amide_I ratio of 0.39 for acetone and 0.09 for glucose). With the microscopy AFMIR or TEM, we were able to observe at the subcellular level, that both acetone and glucose influence the size of the vesicles. Indeed for these two culture conditions, bacteria produced bigger vesicles (500 nm for glucose and 800 nm for acetone) but the number of vesicles per bacterium is different. In the glucose culture, there is on average one large vesicle per bacterium while there are two for the acetone culture. Moreover, looking at the size of bacteria for acetone culture it seems that the total volume of the bacteria is slightly bigger than the other culture. That means that the amount of PHB per bacteria is larger for acetone than for glucose which agrees well with the FTIR measurements. However, the absorption mapping of PHB for the glucose culture indicates that many bacteria show a weak region of absorption or smaller vesicles (in violet or magenta on the mapping). This could indicate that the vesicles are still maturing and they have not yet reached their maximum size after this cultivation time (72 h). Perhaps the biggest vesicle sizes appear after 72 h, indicating that the assimilation of glucose is slower than the assimilation of acetone. TEM has given a good estimation of the small pre-mature vesicles but has underestimated the number of biggest size vesicles. As the method for TEM imaging is to section the sample in layer of 50–100 nm, the histogram of size is wrong for vesicles larger than this value. The likelihood of cutting a border of a vesicle is greater than cutting straight through the full diameter. The correction of the TEM histograms taking account this specific random error is too complex to establish. By evaluating the maximum size measured on TEM images (maximum size standard 400 nm, glucose 600 nm, acetone 600 nm), the results concur with those measured with the AFMIR technique. The AFMIR technique gives an indication of the absorption distribution inside the bacterium and efficiently detects the larger sized vesicles, although unfortunately, this technique is not able to detect the smaller sized vesicles (100–50 nm) buried deep under the surface of the bacterium. Consequently, with the AFMIR technique, the number of pre-mature vesicles cannot be estimated and compared to the number of bigger mature vesicles. TEM imaging behaves in the opposite way: small pre-mature vesicles can be evaluated but not the bigger ones. Finally, AFMIR mappings allow us to get the average PHB absorption (%) for a giving culture condition. This parameter is a good way to indicate the best process of production as only the largest vesicles are of interest. Furthermore, this technique does not require complex preparation as only washing of the bacteria before letting the droplet dry is sufficient for the detection of PHB. Fig. 4. Sizes histogram of PHB vesicles obtained from TEM images analysis. For each culture, the distribution can be separated in two parts, the first one corresponding to the pre-mature vesicles (small sizes) and the second one corresponding to the mature vesicles (bigger than 100 nm). (a) Distribution of sizes for the common culture. (b) Distribution of sizes for the glucose culture. (c) Distribution of sizes for the acetone culture.
5. Conclusion In this study, we have shown that the FTIR analysis was pertinent in detecting the presence of PHB in a population of cell for a given
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
373
Fig. 5. AFM topographies (yellow color corresponds to 400 nm) and corresponding AFMIR mappings of PHB (1740 cm− 1). (a–b) represent respectively the topography and the PHB mapping of 2 bacteria for a common culture. (c–d) are (topography and mapping) a representative result of 2 bacteria for the glucose culture. (e–f) represent 3 bacteria for the acetone culture. In all pictures the white bar corresponds to a length of 1 μm.
culture condition, but was not sufficient to evaluate the influence of acetone and glucose on the PHB production inside the cell. The TEM imaging is a really powerful tool for knowing the structural information on bacteria, but due to the fine cutting in the sample preparation, the size of the object detected may bare errors. Nevertheless, the visualization of vesicles is easy and gives access to the distribution of the pre-mature vesicles. This approach has confirmed the mechanism involved in the initial production of the polymer with the membrane proteins. IR nanospectromicroscopy AFMIR has demonstrated that it was easy to obtain a good estimation of the PHB production in a bacterial culture. But due to the detection principle (thermo-mechanical) the sensitivity was limited to big vesicles or small one close to the membrane. Consequently a multi-modal analysis is crucial for a correlation of all the results for a consistent conclusion. To conclude, AFMIR microscopy is a faster technique compared to the TEM. The bacteria are sampled directly from the Petri dish or the liquid medium, washed and centrifuged three times. This suspension is dried on the prism surface (few minutes), and the image scan takes 15–20 minutes. It is possible to obtain a detailed analysis of the sample after 1 day. Even if it is fast in producing the images with TEM, a few minutes after the region of interest has been localized, the preparation time for the samples into epoxy resin and the preparation of the resin slices on the copper grid is a long process (usually a total of 2 weeks). Moreover, AFMIR technique does not need to modify the sample to obtain chemical images. The other advantage of AFMIR is to directly identify the PHB by spectroscopy (absorption at 1740 cm− 1) while
the TEM images only show white regions corresponding to the location of the vesicles. A future project could combine FTIR and AFMIR to analyze the process of big PHB vesicles production by studying at different times of growing (12 h, 24 h, … , 96 h) of a specific culture to follow its size evolution (with a statistical approach) during the cell development. By knowing the oscillator force of ester carbonyl and Amide I bands, the average production of PHB (with the FTIR measurements) could be quantified by using the PHB volume occupation (estimated by at the absorption and topography images).
Acknowledgements We gratefully acknowledge Pierre Sebban and Valérie Derrien for their precious help with bacterial culture and Yolanda Ohene for her suggestions to the writing of the article.
References Dazzi A, Prazeres R, Glotin F, Ortega J-M. Local infrared microspectroscopy with subwavelength spatial resolution with an atomic force microscope tip used as a photothermal sensor. Opt Lett 2005;30(18):2388–90. Dazzi A, Prazeres R, Glotin F, Ortega J-M. Analysis of nano-chemical mapping performed by an AFM-based (AFMIR) acousto-optic technique. Ultramicroscopy 2007;107:1194–200. Dazzi A, Prazeres R, Glotin F, Ortega J-M, Al-Sawaftah M, De Frutos M. Chemical mapping of the distribution of viruses into infected bacteria with a photothermal method. Ultramicroscopy 2008;108:635–41.
374
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
Dazzi A, Glotin F, Carminati R. Theory of infrared nano-spectroscopy by PhotoThermal Induced Resonance. J Appl Phys 2010;107. Doi Y. Microbial Polyesters; 1990. Hahn SK, Ryu HW, Chang YK. Comparison and optimization of poly(3-hydroxybutyrate) recovery from Alcaligenes eutrophus and recombinant Escherichia coli. Korean J Chem Eng 1998;15(1):51–5. Houel J, Sauvage S, Boucaud P, Dazzi A, Prazeres R, Glotin F, et al. Ultraweak absorption microscopy of a single semiconductor quantum dot in the midinfrared range. Phys Rev Lett 2007;99:2174041–4. Jendrossek D, Selchow O, Hoppert M. Poly(3-hydroxybutyrate) granules at the early stages of formation are localized close to the cytoplasmic membrane in Caryophanon latum. Appl Environ Microbiol 2007;73(2):586–93. Kranz R, Gabbert K, Locke T, Madigan M. Polyhydroxyalkanoate production in Rhodobacter capsulatus: genes, mutants, expression, and physiology. Appl Environ Microbiol 1997;63(8):3003–9. Madigan M. Photocatabolism of acetone by nonsulfur purple bacteria. FEMS Microbiol Lett 1990;71:281–6. Madigan M, Jung D, Resnick S. Growth of the purple bacterium Rhodobacter capsulatus on the aromatic compound hippurate. Arch Microbiol 2001;175:462–5. Mayet C, Dazzi A, Prazeres R, Allot F, Glotin F, Ortega J-M. Sub-100 nm IR spectromicroscopy of living cells. Opt Lett 2008;33(14):1611–3.
Mayet C, Dazzi A, Prazeres R, Ortega J-M, Jaillard D. In situ identification and imaging of bacterial polymer nanogranules by infrared nanospectroscopy. Analyst 2010;135:6. Misra AK, Thakur MS, Srinivas MS, Karanth NG. Screening of polyhydroxybutyrateproducing microorganisms using Fourier transform infrared spectroscopy. Biotechnol Lett 2000;22:1217–9. Naik S, Gopal SKV, Somal P. Bioproduction of polyhydroxyalkanoates from bacteria: a metabolic approach. World J Microbiol Biotechnol 2008;24:2307–14. Naumann D. FT-IR and FT-NIR Raman spectroscopy in biomedical research. The Eleventh International Conference on Fourier Transform Spectroscopy; 1998. Pantazopoulous P, Madigan M. Primary alcohols and di-alcohols as growth substrates for the purple nonsulfur bacterium Rhodobacter capsulatus. Can J Microbiol 2000;46:1166–70. Thomson N, Summers D, Sivaniah E. Synthesis, properties ans uses of bacterial storage lipid granules as naturally occuring nanoparticules. Soft Matter 2010;6:13. Wong HH, Lee SY. Poly-(3-hydroxybutyrate) production from whey by high-density cultivation of recombinant Escherichia coli. Appl Microbiol Biotechnol 1998;50: 30–3. Zhao Q, Cheng G. Preparation of biodegradable poly(3-hydroxybutyrate) and poly (ethylene glycol) multiblock copolymers. J Mater Sci 2004;39:3829–31.
Contents lists available at SciVerse ScienceDirect
Biotechnology Advances journal homepage: www.elsevier.com/locate/biotechadv
Analysis of bacterial polyhydroxybutyrate production by multimodal nanoimaging Céline Mayet, Ariane Deniset-Besseau, Rui Prazeres, Jean-Michel Ortega, Alexandre Dazzi ⁎ Laboratoire de Chimie Physique, Université Paris-Sud, 91405 ORSAY, France
a r t i c l e
i n f o
Available online 22 May 2012 Keywords: Infrared nanospectromicroscopy Transmission electron microscopy Rhodobacter capsulatus Polyhydroxybutyrate
a b s t r a c t In this paper, we will employ two microscopy techniques, transmission electron microscopy and infrared nanospectromicroscopy, to study the production of polyhydroxybutyrate in Rhodobacter capsulatus and to evaluate the influence of glucose and acetone on the production yield. The results overlap which leads us to a consistent conclusion, highlighting that each technique brings specific and complementary information. By using electron microscopy and infrared nanospectromicroscopy we have proved that both glucose and acetone had a positive effect on the biopolymer production, although the first study done by Fourier transform infrared spectroscopy only identified the effect of acetone. In conclusion, we have now established a method to be able to perform fast diagnostic for PHB production. © 2012 Elsevier Inc. All rights reserved.
1. Introduction Polyhydroxyalcanoates (PHAs) are known as natural polyesters produced by bacteria with a great potential in biodegradable plastic applications (Doi, 1990). The scientific interest of the production optimization has come from the knowledge of the PHA biosynthetic pathways, the numerous promising bacterial species and finally by the genetic engineering and biochemical techniques efficiencies (Hahn et al., 1998; Kranz et al., 1997; Wong and Lee, 1998; Zhao and Cheng, 2004). Polyhydroxybutyrates (PHB) are part of this PHAs family. Produced by bacterial fermentation from carbon sources such as sugars or fatty acids, they are then degraded in soil, sludge, or seawater, under optimal conditions, at an extremely fast speed. Furthermore PHB has mechanical properties similar to those of thermoplastic synthetic polymers derived from petroleum, such as polypropylene or polyethylene. Nevertheless, the PHB bacterial production was labeled by industrials as only a prospective project because the biotechnological processes (fermentation, extraction) had too high costs compared to petroleum products. Recently, many scientists have refocused their studies on those polyesters by integrating the industrial constraints and have proposed attractive solutions to optimize PHB production yield in different strains of Alcaligenes, Azotobacter, Pseudomonas or recombinant Escherichia coli (Hahn et al., 1998; Naik et al., 2008; Wong and Lee, 1998). In this context, the study of PHB at the single cell level is crucial for the PHB biosynthesis efficiency to be evaluated. Usually, the PHB yield is first evaluated by biochemical analysis and then confirmed by transmission electron microscopy (TEM). But these techniques are time-consuming, and real-time diagnosis for PHB production is not possible.
⁎ Corresponding author. Tel.: + 33 1 69 15 32 76. E-mail address: [email protected] (A. Dazzi). 0734-9750/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.biotechadv.2012.05.003
Consequently, we propose a combination of infrared spectroscopy and microscopy, to estimate the production of bio-polymer to rapidly diagnose the efficiency of a bacterial strain as well as a culture condition. The ester carbonyl group of the PHB shows a specific mid-infrared response at 1740 cm− 1 which differs from the proteins carbonyl group (Amide I at 1650 cm− 1) (Misra et al., 2000; Naumann, 1998), making it possible to determine, by classical Fourier transformed infrared spectroscopy (FTIR), the mean production of PHB and to analyze the general behavior of a cell population. In order for this study to be performed at the sub-cellular level, a second technique is required with a nanometric resolution. We used an original set-up called AFMIR that couples an atomic force microscope (AFM) and a pulsed IR laser (Dazzi et al., 2005). This technique allows spectromicroscopy at the 10 nm scale with representative local spectra (Dazzi et al., 2008, 2010; Mayet et al., 2010). Our study was performed on Rhodobacter capsulatus whose PHB production is efficient (Madigan, 1990; Madigan et al., 2001; Pantazopoulous and Madigan, 2000). This study was done for different culture conditions that may increase the production yield. The study of PHB is difficult as PHBs are stored in the form of insoluble vesicles inside the bacteria, so it involves high-resolution microscopy such as TEM or our IR nanospectromicroscopy AFMIR. As we need to evaluate the influence of glucose and acetone on the production yield, the PHB vesicle size will be determined by TEM and AFMIR microscopy. Finally the advantages and the limitations of each technique will be discussed. 2. Materials and methods 2.1. Rhodobacter capsulatus cultures 2.1.1. Standard culture Wild type Rhodobacter capsulatus bacteria were grown in malate yeast medium supplemented with kanamycin (20 mg/ml) and tetracycline
370
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
(4 mg/ml) at 50% of the total volume. Cultures were cultivated darkness at 30 °C and kept for 48 h on a gyratory shaker (140 rpm).
2.1.2. Glucose culture The preparation of the culture was similar to the standard one, but 80% of the malate volume was substituted by glucose.
2.1.3. Acetone culture In this case all the malate was substituted by acetone. In order to be studied in infrared, the cell suspension of each culture was centrifuged at 1000 rpm for 15 min. The supernatant was removed and the cell pellet diluted in distilled water. This process was repeated three times in order to obtain suitably clean samples. Finally, a droplet of the cellular suspension was deposited on ZnSe (transparent in midinfrared) coverglass for FTIR spectroscopy or ZnSe prisms for AFMIR studies and dried at room temperature.
2.2. FTIR analysis The infrared spectra was acquired with a Bruker Vertex 70 FTIR spectrometer between 4000 and 400 cm − 1 with a spectral resolution of 4 cm − 1 using an MCT detector with a liquid nitrogen cooling system. The biomass can be directly characterized by the amplitude of the Amide I absorption band; hence we have normalized all FTIR spectra by this band so that we can compare the amount of PHB production for the different culture condition.
2.3. Principle of the nanospectromicroscopy AFMIR The AFMIR instrument combines an Atomic Force Microscope (AFM) with an infrared pulsed tunable laser (Fig. 1). A sample is placed on a ZnSe prism and then irradiated with the laser. When the laser wavelength is tuned on an absorption band of the sample, the absorbed laser light causes a photothermal effect and a temperature rise. This increase in temperature creates a local thermal expansion monitored by the tip of the AFM. The expansion of the sample generates a rapid impulse that leads to the oscillation of the cantilever on its eigenmodes. The oscillations amplitudes, on the 4-quadrant detector, are directly proportional to the energy absorbed. Hence, we can reconstruct the entire absorption spectrum of the sample by tuning the incident laser wavenumber, and comparing it with those obtained by classical IR spectroscopy techniques like FTIR (Dazzi et al., 2010). The main advantage of this technique is its resolution of less than 100 nm. It allows a chemical study of the samples at the subcellular level. This has already been demonstrated in microbiological (Dazzi et al., 2007, 2008; Mayet et al., 2008) and nanophotonic domains (Houel et al., 2007).
2.4. Sample preparation for TEM imaging For all different cultures, bacteria were fixed with glutaraldehyde 2% and paraformaldehyde 2% in a buffer of sodium cacodylate 0.1 mol/l (pH 7.4) and then post-fixed in 1% buffered osmium tetroxide. The bacteria was then completely dehydrated with ethanol at room temperature and then embedded in epoxy resin (polymerization at 60 °C for 48 h). Ultrathin sections of 50 nm thickness were cut with a diamond knife, deposited on copper grids (mesh 200) and then stained with uranyl acetate aqueous solution (20 min) and lead citrate (5 min). Samples were observed at 80 kV with Philips EM208 transmission electron microscope (80 kV, mode light background) equipped with a CCD camera. 2.5. AFMIR images analysis The AFMIR imaging analysis was performed with ImageJ software. ImageJ software is an open source software commonly used by scientist in imaging domain. This software is useful to visualize imaging data and offers the possibility of applying mathematical tools for the analysis. Each chemical mapping of PHB (already normalized by the input laser power) was renormalized by the corresponding topography (absorption signal is divided by the corresponding height), giving us the absorption density value. The signal coming from the bacteria was isolated using the threshold filter (removing the image pixels corresponding to the surface). The maximum of absorption density was then extracted for each bacterium. We assume that the maximum of these maxima corresponds to a full diameter of the vesicle also ensuring that the absorption signal is not an isolated point but is linked to a large area of absorption. As the maximum absorption density value and the volume of each bacterium are known, the theoretical maximum absorption signal if the bacterium was only full of PHB is easy to calculate. Then the real values of absorption density can be integrated (given by the chemical mapping divided by topography) on only the pixels corresponding to the surface of the bacterium to have the integral of density absorption. Finally, by dividing this last value with the theoretical maximum absorption we obtain the % of PHB absorption inside the volume of the bacterium. This process of analysis was applied for tens of bacteria for each culture condition to obtain the average PHB absorption (%). This parameter does not directly give the % of PHB volume occupation per bacterium but is a very good indicator of this value. The error comes from buried vesicles that give a smaller integrated signal than the vesicles close to the surface. So the % of PHB absorption is lower than the % of volume occupation. 3. Results 3.1. Global study by FTIR spectroscopy For each culture (standard, glucose and acetone), an FTIR spectrum has been recorded between 1850 and 1550 cm − 1 (Fig. 2) after
Fig. 1. Scheme of the AFMIR experimental setup. The sample is deposited on the upper side of the ZnSe prism and illuminated by the infrared pulsed laser. The resulting photothermal expansion generates a brief force under the AFM tip inducing the ringing of the AFM cantilever. The oscillation signal is recorded and stored in an oscilloscope.
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
371
Fig. 2. FTIR spectra of different culture types. The common culture is represented in plain line, the glucose in dotted line and the acetone in dashed line.
48 h of cultivation, which is sufficient for the culture to reach the steady state. This range allows the identification of both the ester carbonyl absorption band (C O, 1740 cm − 1) of the PHB compound and the Amide I band associated to bacteria proteins (1650 cm − 1). The results showed that the ester carbonyl band (1740 cm − 1) is prominent for the acetone culture medium whereas for the standard and glucose culture condition FTIR spectra show a poor absorption at this wavenumber. The band ratios of the ester_carbonyl/Amide_I are the following: 0.39 for acetone culture, 0.09 for glucose and 0.07 for standard. From these measurements, we can conclude that using glucose has no detectable effect on the PHB production whereas substituting malate by acetone is a relevant strategy to improve this production. However, this approach gives us only the relative increase of ester carbonyl vibrations in function of the carbonyl of the proteins. Furthermore, even if the study on the entire culture of bacteria gives us a good indication of PHB production, the effects of the culture medium substitution remain unclear at a single cell level. More precisely, the size, the shape and the repartition of the vesicles inside the bacteria as well as the increase of the bacteria volume which have an effect of the PHB production are unknown. Microscopy techniques are a good way to verify these points.
the bacteria, it is not possible to estimate correctly the % of PHB per bacteria. Therefore it is necessary for an intricate method to be developed of keeping the sections in the cutting serial order to be able, when using the TEM microscope, to reconstruct a complete bacterium by imaging the consecutive sections. We suggest the use of a second technique which is less constraining to assess, more accurately, the size of the mature vesicles.
3.2. TEM imaging
3.3. AFMIR PHB mapping
For each culture condition, over 200 bacteria (Fig. 3) were analyzed and their vesicle sizes were systematically measured. The measurements are given Fig. 4 where the size histograms for each culture have been compared. On each histogram the shape of the distribution has been added to highlight their maxima. For each case, the histogram shows a maximum size distribution centered around 60–80 nm. This corresponds to the smallest vesicles being more numerous and clustered around the periphery of the bacteria. This confirms the hypothesis of pre-maturation with two steps: first an accumulation of small vesicles in the membrane and then a migration and an accumulation of the vesicles towards the center of the bacteria (Thomson et al., 2010). The second peak corresponds to the mature vesicles size and is centered around 170 nm for the standard culture and 250–300 nm for the two other cultures. In conclusion, the presence of glucose or the acetone substitution in the culture medium has a similar effect and increase the size of mature vesicles. This local analysis is clearly efficient in obtaining an accurate estimation of small size of the vesicles and a good statistic evaluation. However, as the TEM technique requires a thin section (50–100 nm) to analyze
The relevance of our IR nanospectromicroscope AFMIR in identifying and detecting PHB vesicles with a size of 50 to 300 nm inside Rhodobacter has already been demonstrated (Mayet et al., 2010). In this paper, we have shown that taking the full width at the half maximum of the absorption signal will give a good estimation of the vesicle size. The laser is tuned at the ester carbonyl band (1740 cm− 1) wavenumber and each sample (dried bacteria) is scanned by the AFM tip to acquire simultaneously the topography and the chemical mapping of a single cell (Fig. 5). For each culture, we scanned tens of bacteria for an initial insight of the average PHB production. After the acquisition of the chemical mapping, systematic analysis of each images is done by imageJ software in order for the size of the vesicles to be determined. This analysis indicates that in a standard culture the size of PHB vesicles are around 200–300 nm (Fig. 5a, b) and are not present in all bacteria (approximately half posses a least one vesicle). For the glucose culture the size of vesicles is bigger, up to 400–500 nm (Fig. 5c, d) with at least one vesicle per bacterium. In the case of the acetone culture, it is evident that the size was around 800 nm (the larger size of bacteria) with more than one vesicle per bacterium (Fig. 5e, f). Moreover, for all
Fig. 3. Representative TEM images for the different bacteria cultures where bacteria appear in gray (contrast is proportional of the density of matter) and PHB vesicles in white. (a) Common culture. (b) Glucose culture. (c) Acetone culture. The black bar corresponds to 500 nm.
372
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
of the culture conditions the PHB vesicles seem to be localized at the apex the bacteria after maturation, which strongly agrees with the observation done by fluorescence microscopy on Beijerinckia indica bacteria (Jendrossek et al., 2007). This IR nanospectromicroscopy analysis allows us not only to easily detect the PHB, but also to estimate, by looking at the absorption distribution of the vesicles inside the bacterium, the percentage of PHB absorption inside its volume. The calculation of the average PHB absorption (on 20 bacteria per culture type) gives: Standard 11.2%, Glucose 12.5%, Acetone 22%. 4. Discussion The multimodal imaging analysis has given a good insight of the influence of glucose and acetone on the PHB production in Rhodobacter capsulatus. The FTIR analysis shows that only the addition of acetone in the culture medium allows the bacteria to produce more PHB (ester_carbonyl/Amide_I ratio of 0.39 for acetone and 0.09 for glucose). With the microscopy AFMIR or TEM, we were able to observe at the subcellular level, that both acetone and glucose influence the size of the vesicles. Indeed for these two culture conditions, bacteria produced bigger vesicles (500 nm for glucose and 800 nm for acetone) but the number of vesicles per bacterium is different. In the glucose culture, there is on average one large vesicle per bacterium while there are two for the acetone culture. Moreover, looking at the size of bacteria for acetone culture it seems that the total volume of the bacteria is slightly bigger than the other culture. That means that the amount of PHB per bacteria is larger for acetone than for glucose which agrees well with the FTIR measurements. However, the absorption mapping of PHB for the glucose culture indicates that many bacteria show a weak region of absorption or smaller vesicles (in violet or magenta on the mapping). This could indicate that the vesicles are still maturing and they have not yet reached their maximum size after this cultivation time (72 h). Perhaps the biggest vesicle sizes appear after 72 h, indicating that the assimilation of glucose is slower than the assimilation of acetone. TEM has given a good estimation of the small pre-mature vesicles but has underestimated the number of biggest size vesicles. As the method for TEM imaging is to section the sample in layer of 50–100 nm, the histogram of size is wrong for vesicles larger than this value. The likelihood of cutting a border of a vesicle is greater than cutting straight through the full diameter. The correction of the TEM histograms taking account this specific random error is too complex to establish. By evaluating the maximum size measured on TEM images (maximum size standard 400 nm, glucose 600 nm, acetone 600 nm), the results concur with those measured with the AFMIR technique. The AFMIR technique gives an indication of the absorption distribution inside the bacterium and efficiently detects the larger sized vesicles, although unfortunately, this technique is not able to detect the smaller sized vesicles (100–50 nm) buried deep under the surface of the bacterium. Consequently, with the AFMIR technique, the number of pre-mature vesicles cannot be estimated and compared to the number of bigger mature vesicles. TEM imaging behaves in the opposite way: small pre-mature vesicles can be evaluated but not the bigger ones. Finally, AFMIR mappings allow us to get the average PHB absorption (%) for a giving culture condition. This parameter is a good way to indicate the best process of production as only the largest vesicles are of interest. Furthermore, this technique does not require complex preparation as only washing of the bacteria before letting the droplet dry is sufficient for the detection of PHB. Fig. 4. Sizes histogram of PHB vesicles obtained from TEM images analysis. For each culture, the distribution can be separated in two parts, the first one corresponding to the pre-mature vesicles (small sizes) and the second one corresponding to the mature vesicles (bigger than 100 nm). (a) Distribution of sizes for the common culture. (b) Distribution of sizes for the glucose culture. (c) Distribution of sizes for the acetone culture.
5. Conclusion In this study, we have shown that the FTIR analysis was pertinent in detecting the presence of PHB in a population of cell for a given
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
373
Fig. 5. AFM topographies (yellow color corresponds to 400 nm) and corresponding AFMIR mappings of PHB (1740 cm− 1). (a–b) represent respectively the topography and the PHB mapping of 2 bacteria for a common culture. (c–d) are (topography and mapping) a representative result of 2 bacteria for the glucose culture. (e–f) represent 3 bacteria for the acetone culture. In all pictures the white bar corresponds to a length of 1 μm.
culture condition, but was not sufficient to evaluate the influence of acetone and glucose on the PHB production inside the cell. The TEM imaging is a really powerful tool for knowing the structural information on bacteria, but due to the fine cutting in the sample preparation, the size of the object detected may bare errors. Nevertheless, the visualization of vesicles is easy and gives access to the distribution of the pre-mature vesicles. This approach has confirmed the mechanism involved in the initial production of the polymer with the membrane proteins. IR nanospectromicroscopy AFMIR has demonstrated that it was easy to obtain a good estimation of the PHB production in a bacterial culture. But due to the detection principle (thermo-mechanical) the sensitivity was limited to big vesicles or small one close to the membrane. Consequently a multi-modal analysis is crucial for a correlation of all the results for a consistent conclusion. To conclude, AFMIR microscopy is a faster technique compared to the TEM. The bacteria are sampled directly from the Petri dish or the liquid medium, washed and centrifuged three times. This suspension is dried on the prism surface (few minutes), and the image scan takes 15–20 minutes. It is possible to obtain a detailed analysis of the sample after 1 day. Even if it is fast in producing the images with TEM, a few minutes after the region of interest has been localized, the preparation time for the samples into epoxy resin and the preparation of the resin slices on the copper grid is a long process (usually a total of 2 weeks). Moreover, AFMIR technique does not need to modify the sample to obtain chemical images. The other advantage of AFMIR is to directly identify the PHB by spectroscopy (absorption at 1740 cm− 1) while
the TEM images only show white regions corresponding to the location of the vesicles. A future project could combine FTIR and AFMIR to analyze the process of big PHB vesicles production by studying at different times of growing (12 h, 24 h, … , 96 h) of a specific culture to follow its size evolution (with a statistical approach) during the cell development. By knowing the oscillator force of ester carbonyl and Amide I bands, the average production of PHB (with the FTIR measurements) could be quantified by using the PHB volume occupation (estimated by at the absorption and topography images).
Acknowledgements We gratefully acknowledge Pierre Sebban and Valérie Derrien for their precious help with bacterial culture and Yolanda Ohene for her suggestions to the writing of the article.
References Dazzi A, Prazeres R, Glotin F, Ortega J-M. Local infrared microspectroscopy with subwavelength spatial resolution with an atomic force microscope tip used as a photothermal sensor. Opt Lett 2005;30(18):2388–90. Dazzi A, Prazeres R, Glotin F, Ortega J-M. Analysis of nano-chemical mapping performed by an AFM-based (AFMIR) acousto-optic technique. Ultramicroscopy 2007;107:1194–200. Dazzi A, Prazeres R, Glotin F, Ortega J-M, Al-Sawaftah M, De Frutos M. Chemical mapping of the distribution of viruses into infected bacteria with a photothermal method. Ultramicroscopy 2008;108:635–41.
374
C. Mayet et al. / Biotechnology Advances 31 (2013) 369–374
Dazzi A, Glotin F, Carminati R. Theory of infrared nano-spectroscopy by PhotoThermal Induced Resonance. J Appl Phys 2010;107. Doi Y. Microbial Polyesters; 1990. Hahn SK, Ryu HW, Chang YK. Comparison and optimization of poly(3-hydroxybutyrate) recovery from Alcaligenes eutrophus and recombinant Escherichia coli. Korean J Chem Eng 1998;15(1):51–5. Houel J, Sauvage S, Boucaud P, Dazzi A, Prazeres R, Glotin F, et al. Ultraweak absorption microscopy of a single semiconductor quantum dot in the midinfrared range. Phys Rev Lett 2007;99:2174041–4. Jendrossek D, Selchow O, Hoppert M. Poly(3-hydroxybutyrate) granules at the early stages of formation are localized close to the cytoplasmic membrane in Caryophanon latum. Appl Environ Microbiol 2007;73(2):586–93. Kranz R, Gabbert K, Locke T, Madigan M. Polyhydroxyalkanoate production in Rhodobacter capsulatus: genes, mutants, expression, and physiology. Appl Environ Microbiol 1997;63(8):3003–9. Madigan M. Photocatabolism of acetone by nonsulfur purple bacteria. FEMS Microbiol Lett 1990;71:281–6. Madigan M, Jung D, Resnick S. Growth of the purple bacterium Rhodobacter capsulatus on the aromatic compound hippurate. Arch Microbiol 2001;175:462–5. Mayet C, Dazzi A, Prazeres R, Allot F, Glotin F, Ortega J-M. Sub-100 nm IR spectromicroscopy of living cells. Opt Lett 2008;33(14):1611–3.
Mayet C, Dazzi A, Prazeres R, Ortega J-M, Jaillard D. In situ identification and imaging of bacterial polymer nanogranules by infrared nanospectroscopy. Analyst 2010;135:6. Misra AK, Thakur MS, Srinivas MS, Karanth NG. Screening of polyhydroxybutyrateproducing microorganisms using Fourier transform infrared spectroscopy. Biotechnol Lett 2000;22:1217–9. Naik S, Gopal SKV, Somal P. Bioproduction of polyhydroxyalkanoates from bacteria: a metabolic approach. World J Microbiol Biotechnol 2008;24:2307–14. Naumann D. FT-IR and FT-NIR Raman spectroscopy in biomedical research. The Eleventh International Conference on Fourier Transform Spectroscopy; 1998. Pantazopoulous P, Madigan M. Primary alcohols and di-alcohols as growth substrates for the purple nonsulfur bacterium Rhodobacter capsulatus. Can J Microbiol 2000;46:1166–70. Thomson N, Summers D, Sivaniah E. Synthesis, properties ans uses of bacterial storage lipid granules as naturally occuring nanoparticules. Soft Matter 2010;6:13. Wong HH, Lee SY. Poly-(3-hydroxybutyrate) production from whey by high-density cultivation of recombinant Escherichia coli. Appl Microbiol Biotechnol 1998;50: 30–3. Zhao Q, Cheng G. Preparation of biodegradable poly(3-hydroxybutyrate) and poly (ethylene glycol) multiblock copolymers. J Mater Sci 2004;39:3829–31.