Optimization of Metabolite Extraction Protocols for the Identification and Profiling of Small Molecule Metabolites from Planktonic and Biofilm Pseudomonas aeruginosa Cultures

Abstract

Background:

Metabolomics aims to characterize the metabolic phenotype and metabolic pathways utilized by microorganisms or other cellular systems. A crucial component to metabolomics research as it applies to microbial metabolism is the development of robust and reproducible methods for extraction of intracellular metabolites. The goal is to extract all metabolites in a non-biased and consistent manner; however, most methods used thus far are targeted to specific metabolite classes and use harsh conditions that may contribute to metabolite degradation. Metabolite extraction methodologies need to be optimized for each microorganism of interest due to different cellular characteristics contributing to lysis resistance.

Methods:

Three cell pellet wash solutions were compared for the potential to influence intracellular metabolite leakage of P. aeruginosa. We also compared four different extraction methods using (i) methanol:chloroform (2:1); (ii) 50% methanol; (iii) 100% methanol; or (iv) 100% water to extract intracellular metabolites from P. aeruginosa planktonic and biofilm cultures.

Results:

Intracellular metabolite extraction efficiency was found to be dependent on the extraction method and varies between microbial modes of growth. Methods using the 60% methanol wash produced the greatest amount of intracellular material leakage. Quantification of intracellular metabolites via ¹H NMR showed that extraction protocols using 100% water or 50% methanol achieved the greatest extraction efficiencies, while addition of sonication to facilitate cell lysis to the 50% methanol extraction method resulted in at least a two-fold increase in signal intensities for approximately half of the metabolites identified. Phosphate buffered saline (PBS) was determined to be the most appropriate wash solution, yielding little intracellular metabolite leakage from cells.

Conclusion:

We determined that washing in 1X PBS and extracting intracellular metabolites with 50% methanol is the most appropriate metabolite extraction protocol because (a) leakage is minimal; (b) a broad range of metabolites present at sufficiently high concentrations is detectable by NMR; and (c) this method proved suitable for metabolite extraction of both planktonic and biofilm P. aeruginosa cultures.

1. INTRODUCTION

As an analytical approach, metabolomics most directly characterizes the metabolic phenotype of a sample and reports on metabolic pathways utilized by a given microorganism when grown under certain conditions [1]. Most commonly used techniques for metabolite detection and identification include mass spectrometry (MS) and nuclear magnetic resonance (NMR). MS has the advantages of being more sensitive than NMR and can detect metabolites of wide ranging abundance [2]. However, mapping mass spectral features to specific metabolites and metabolite quantification by LC-MS is quite challenging. While NMR best detects more abundant metabolites (1 mM to 10 μM reliably) in a sample, identification of metabolites using spectral databases of small molecules such as those found in the Chenomx^TM compound libraries [3] is more straightforward. In addition, the NMR approach is highly reproducible, non-destructive, and permits accurate quantification of metabolite concentrations in a given sample [2]. The Chenomx^TM software [3] uses a sophisticated linear analysis approach that allows the specific contribution of each metabolite to be assessed in an additive manner [4] until complete or nearly complete spectral intensity contribution, splitting patterns, and chemical shift matching have been realized.

A critical process in metabolomics research is the establishment of robust and reproducible methods for metabolite extraction. A key issue is how to handle and harvest cells prior to extracting intracellular metabolites. This step involves the separation of cells from growth medium via centrifugation followed by a series of washing procedures to separate and remove residual extracellular materials from pelleted cells. This wash step is particularly important for the biofilm phenotype due to the complex extracellular matrix that often contains extracellular DNA, proteins and, in the case of P. aeruginosa, alginate among other exopolysaccharides that serve as a protective barrier against harsh environmental conditions and as a structural scaffold [5]. A common approach used to wash cell pellets is 60% methanol, which also serves as a metabolic quenching step [6]. Chen et al. [7] observed a significant leakage of intracellular metabolites when Lactobacillus bulgaricus cells were quenched with both 60% and 80% methanol. It has also been demonstrated that low ionic strength solutions, such as deionized water, cause extensive metabolite leakage from Gram-negative bacterial cells [8]. As expected, such leakages result in a vast underestimation of intracellular metabolite levels present and skewed metabolite profiles. Significant intracellular ATP loss (>10%) also takes place upon cell exposure to 60% methanol, which is particular striking because ATP, ADP, and AMP concentrations are often used to quantify the metabolic state of the cell via energy charge calculations [9].

Following cell culture harvesting and washing to remove residual extracellular materials, including extracellular small molecules, intracellular metabolites must be extracted efficiently. The ultimate goal of any extraction method is to extract all metabolites in a non-biased, reproducible, and consistent manner. However, most extraction methods used thus far are designed to target specific metabolite classes and often employ harsh conditions, such as high temperatures, that may lead to significant metabolite degradation [10].

The quintessential goal of an efficient extraction protocol is to be suitable for metabolomics studies of a variety of microbial cells; however, this is often not feasible due to different cellular characteristics that result in varying resistance to lysis across diverse microbial species. Therefore, metabolite extraction methodologies need to be optimized and evaluated for each species of interest. In the present study, several different cell wash solutions and metabolite extraction methods were tested and evaluated for their efficiency to most comprehensively profile the metabolome of P. aeruginosa, a Gram-negative bacterium that is of particular importance and concern in many areas of human health [11].

2. MATERIALS AND METHOD

2.1. Bacterial Strain, Growth Conditions, and Sampling

This study utilized the well-characterized and genome sequenced PAO1 strain of P. aeruginosa. Growth media for both planktonic and biofilm cultures consisted of 10% BHI (Brain Heart Infusion; BD Bacto^TM). Inocula for both planktonic and biofilm growth conditions consisted of closed, batch cultures grown in 10% BHI at 37 °C shaking with 220 rpm to 20 hrs post inoculation (~2.2 × 10⁹ CFUs/ml). Aliquots (10 μ1) were collected for serial dilution, drop plating, and calculation of colony forming units (CFUs).

For planktonic studies, inoculum cultures were diluted 1:1000 in fresh 10% BHI and cultured at 37 °C in 250 ml flasks shaking at 220 rpm. Planktonic cultures were grown under aerobic conditions with flask-to-medium volume ratios of 5:1. Samples were harvested at 24 hrs post inoculation. Biofilm growth was cultured as previously described [12, 13]. In brief, tissue culture inserts (Millipore Millicell, 0.4 μ1 pore size) were inoculated with five 10 μ1 droplets of overnight inoculum culture (~10⁹ CFUs/ml) and grown for 96 hrs at 37 °C. To maintain biofilm viability, the growth media was refreshed every 24 hrs. Biofilms were collected once they had reached 96 hrs post inoculation. Spent supernatant was collected from the plate wells and biofilm cells were harvested from the inserts by gently pipetting with 1 ml of sterile 1X PBS to dislodge the biofilms. Samples were immediately centrifuged at 5000 rpm for 10 min at 25 °C to pellet the cells. For both planktonic and biofilm growth conditions, samples were harvested in technical triplicates.

2.2. Extraction Procedures and NMR Sample Preparation

NMR samples were prepared in triplicate for each metabolite extraction method. For ¹H NMR analysis, dried samples were resuspended in 700 μ1 of NMR buffer [25 mM NaH₂PO₄/Na₂HPO₄ containing 0.25 mM 4,4-dimethyl-4- silapentane-1-sulfonic acid (DSS) in 100% D₂O, pH 7] and transferred to 5 mm Wilmad NMR tubes.

2.2.1. Methanol/Chloroform Extraction

Cell pellets were resuspended twice in 600 μ1 of 60% ice-cold methanol and centrifuged at 5,000 rpm for 10 min. Each of the methanol washes were dried under vacuum (SpeedVac) and stored at −20 °C prior to analysis. Washed cell pellets were resuspended in 1 ml of ice-cold 2:1 methanol/chloroform and transferred to 8 ml glass tubes. Cells were lysed by sonication (BioLogics Inc. Model 3000 Ultrasonic Homogenizer; 30% of maximum instrument power, ~50 pulse); samples were pulsed 10 times for 3 cycles. 300 μ1 of each layer of a 1:1 aqueous chloroform solution was added to the samples and mixed by inversion, followed by centrifugation at 5,000 rpm for 10 min. The top aqueous phase of each sample was transferred to a clean microcentrifuge tube and dried under vacuum (SpeedVac) at room temperature overnight and then stored at −20 °C until further use.

2.2.2. 50% Methanol Extraction

Cell pellets were resuspended and washed twice in 1 ml of 1X PBS and centrifuged at 5,000 rpm for 10 min. PBS washes were dried under vacuum (SpeedVac) and stored at −20 °C prior to analysis. Washed cell pellets were resuspended in 1.5 ml of 50% methanol and placed at −80 °C to freeze for a minimum of 2 hours. Cells were thawed in a circulating cold water bath and transferred to glass tubes. Samples were then sonicated for 30 sec. (30% power, ~50 pulse). This freeze-thaw, sonication cycle was repeated 2 additional times. Cells were centrifuged at 14,000 x g for 5 min. at 4 °C, and the supernatants were transferred to separate microcentrifuge tubes. The cell debris was resuspended in 1 ml of deionized H₂O (diH₂O) and centrifuged at 14,000 x g for 5 min. at 4 °C, and the supernatants were pooled with the initial extract. Chloroform (1 ml) was added to the pooled supernatant and mixed by vortexing briefly. Phase separation was driven by centrifugation at 5,000 rpm for 5 min. The top aqueous phase of each sample was transferred to a new microcentrifuge tube, and dried under vacuum (SpeedVac) at room temperature overnight and then stored at −20 °C until further use.

2.2.3. 100% Methanol Extraction

This method is similar to the 50% methanol extraction protocol; however, washed cell pellets were resuspended in 1.5 ml of 100% methanol instead of 50% methanol prior to beginning freeze-thaw, sonication cycles.

2.2.4. 100% Water Extraction

This method is similar to the 50% methanol extraction protocol; however, washed cell pellets were resuspended in 1.5 ml of sterile diH2O instead of 50% methanol prior to the start of the freeze-thaw, sonication cycles.

2.3. NMR Analysis

One dimensional ¹H NMR spectra were acquired as previously described [13]. Processing and analysis of spectra was performed using Chenomx^TM NMR software (version 8.0, Chenomx, CA). Each ¹H NMR spectrum was phased, baseline-corrected, and line-broadened (0.5 Hz) prior to metabolite quantification and identification. The DSS internal standard was used to quantify identified metabolites. Compound identification was carried out by fitting NMR spectral patterns for each sample manually using the Chenomx^TM spectral database of small-molecules for 600 MHz (¹H Larmor frequency) magnetic field strength NMR spectrometers. From all the ¹H 1D NMR spectra acquired on biofilm and planktonic PAO1 cell extracts, an overall number of ~83 compounds were identified.

3. RESULTS AND DISCUSSIONS

3.1. Cell Pellet Wash

In this optimization study of metabolite extraction in P. aeruginosa, three different wash solutions were evaluated for their impacts on intracellular metabolite leakage. The 60% methanol washes produced the greatest amount of leakage for nearly all of the metabolites identified whereas washes with PBS were determined to be optimal because no metabolites were found in the second wash conducted with PBS. Therefore, metabolites obtained from the first PBS wash were determined to be predominantly extracellular, and did not significantly result from leaked intracellular fractions. This finding is consistent with previous reports [8] in that the lower ionic strength of solutions such as 60% methanol and deionized water have a significant effect on intracellular metabolite leakage from cells. The ¹H NMR metabolite profile and corresponding spectral intensities resulting from the first PBS wash was used to normalize the NMR data obtained from other wash protocols in order to strictly identify those metabolites that may have been lost due to cellular leakage (Fig. 1). Metabolite concentrations significantly impacted as a result of cell leakage from the 60% methanol washes included those from glutamate, betaine, and glycine. Loss of these intracellular metabolites may confound downstream characterization of different cellular phenotypes. This metabolite leakage is also reflected in the metabolite extraction efficiencies (attomoles/CFU) identified for the method using this wash solution – 2:1 methanol/chloroform – compared to other extraction methods (Table 1).

Fig. (1).

Comparison of metabolite leakage caused by 60% methanol and water washes of P. aeruginosa cell pellets; cellular metabolite leakage is reported as metabolite abundance detected in addition to that observed in the first 1X PBS wash.

Table 1.

Sample metabolite sets (attomoles/CFU) extracted from P. aeruginosa cell pellets using different methods and detected by ¹H NMR. Metabolites were identified and quantified via Chenomx NMR Suite (v8.0); I (50% methanol); II (100% methanol); III (100% water); IV (2:1 methanol:chloroform); nd = not detected. All extraction methods were performed in triplicate. Amino acids are designated using their respective 3 letter abbreviations; GABA = 4-aminobutyrate; ADP = adenosine diphosphate; AMP = adenosine monophosphate; NAD⁺ = nicotinamide adenine dinucleotide; NADP⁺ = nicotinamide adenine dinucleotide phosphate; UMP = uridine monophosphate.

Compound	Planktonic				Biofilm
	IV	I	II	III	IV	I	II	III
GABA	nd	0.372±0.022	0.159±0.013	0.488±0.050	0.300±0.004	0.352±0.025	0.226±0.035	1.229±0.101
ADP	0.209±0.040	0.047±0.001	0.097±0.005	0.083±0.015	0.023±0.016	0.158±0.031	0.117±0.007	0.229±0.043
AMP	0.369±0.056	0.318±0.004	0.110±0.003	0.168±0.020	0.624±0.016	0.528±0.016	0.375±0.048	0.734±0.132
Acetate	0.016±0.003	0.591±0.002	0.195±0.034	1.12±0.09	0.085±0.012	5.532±0.090	6.409±2.072	31.16±0.05
Ala	0.601±0.030	1.54±0.36	0.484±0.083	3.42±0.92	0.204±0.016	2.423±0.313	0.307±0.007	3.716±0.632
Betaine	0.182±0.041	0.614±0.000	0.393±0.021	0.299±0.018	0.198±0.008	0.312±0.021	nd	nd
Choline	0.020±0.004	0.056±0.004	0.010±0.001	0.02±0.00	0.018±0.000	0.065±0.009	0.022±0.006	0.102±0.002
Formate	0.037±0.001	1.42±0.10	0.393±0.040	0.075±0.006	2.61±0.05	3.265±0.335	1.310±0.243	2.467±0.086
Glu	1.22±0.17	3.23±0.22	1.20±0.03	2.61±0.59	2.48±0.09	5.566±0.369	3.043±0.263	6.540±0.677
Gln	0.400±0.001	0.296±0.114	0.432±0.080	0.706±0.185	nd	nd	nd	nd
Gly	0.278±0.013	0.837±0.133	0.114±0.019	1.04±0.30	0.213±0.021	0.944±0.079	0.260±0.052	1.619±0.282
His	0.036±0.001	0.096±0.025	0.05±0.02	0.095±0.003	nd	0.142±0.040	nd	0.498±0.119
Ile	0.191±0.009	0.912±0.257	0.407±0.068	1.50±0.47	0.082±0.016	1.328±0.182	0.158±0.001	1.796±0.661
Leu	0.463±0.022	1.78±0.51	1.02±0.15	3.94±0.94	0.242±0.004	2.940±0.392	0.497±0.041	4.614±1.440
Mannose	0.131±0.032	0.400±0.000	0.183±0.011	0.077±0.012	0.178±0.062	0.107±0.021	nd	nd
Methanol	0.093±0.043	0.116±0.007	0.02±0.01	0.030±0.008	0.172±0.037	0.976±0.124	0.135±0.005	0.219±0.063
Met	0.065±0.009	0.309±0.082	0.189±0.030	0.808±0.206	0.108±0.004	0.529±0.074	0.085±0.009	0.842±0.184
NAD+	0.117±0.001	0.167±0.059	0.173±0.004	nd	0.554±0.016	0.598±0.053	0.584±0.027	0.449±0.072
NADP+	0.118±0.012	0.095±0.000	0.05±0.01	0.039±0.005	0.149±0.004	0.165±0.019	0.174±0.008	0.146±0.032
Trp	nd	0.133±0.036	nd	nd	nd	0.196±0.059	nd	0.425±0.202
Tyr	0.111±0.008	0.465±0.147	0.250±0.046	1.04±0.27	0.088±0.016	0.714±0.053	0.119±0.024	1.046±0.321
UMP	0.106±0.027	0.097±0.002	0.02±0.00	0.063±0.008	0.140±0.033	0.111 ±0.014	0.074±0.011	0.443±0.075
Val	0.304±0.024	1.49±0.44	0.704±0.137	3.00±0.78	0.073±0.020	1.978±0.258	0.217±0.003	2.607±0.897

3.2. Metabolite Extraction

An optimal extraction method aims to recover as many intracellular metabolites as possible at adequate concentrations for quantification, accompanied by minimal degradation and cellular leakage. However, it has been previously demonstrated that an extraction methodology optimized for a given microbe may not be suitable for the efficient extraction of metabolites originating from other microorganisms [8, 14]. Therefore, the development of an optimized extraction protocol for the microbial species of interest is crucial for an appropriate examination of the intracellular metabolome. Initially, a 2:1 methanol/chloroform protocol previously adapted for the extraction of metabolites from Staphylococcus aureus [15] and effective for both planktonic and biofilm S. aureus phenotypes [13] was used in this study. At the time, this method seemed appropriate for P. aeruginosa because it yielded a broad range of intracellular metabolites at sufficient concentrations for ¹H NMR quantification in the planktonic phenotype. Despite its efficacy with batch culture P. aeruginosa, the method produced inadequate and inconsistent results for the biofilm phenotype (Fig. 2, Table 1). As a result, three extraction methods adapted from the original protocol were assessed for their efficacy on planktonic and biofilm P. aeruginosa cultures.

Fig. (2).

Spectral overlay of intracellular metabolites extracted from planktonic (black) and biofilm P. aeruginosa (gray) using the 2:1 methanol:chloroform extraction method.

The extraction methods used here involved cell lysis with repeated freeze-thaw cycles in different solvents (100% methanol, 50% methanol, and 100% water) in combination with sonication. Efficiency of each extraction method was evaluated by recording a series of 1D ¹H NMR spectra and determining the concentrations of intracellular metabolites obtained per viable bacterial cell count (CFU). Addition of a sonication step to the freeze-thaw extraction protocol did not yield any unique metabolites when compared to samples prepared without sonication; however, adding this mechanical cell disruption step to the extraction protocol resulted in a 2-fold increase in the concentrations of ~50% of the identified metabolites in the biofilm phenotype (Fig. 3). This finding is consistent with previous studies [16, 17] indicating that sonication can be used to enhance cell lysis for the recovery of intracellular molecules.

Fig. (3).

Fold changes in metabolite yield of P. aeruginosa biofilm samples occurring when sonication is added to the 50% methanol freeze-thaw extraction method; data are reported as the mean of duplicate biological samples and standard deviation (SD) error bars.

However, sonication may lead to the degradation of some metabolites. We compared metabolite concentrations using 30 vs. 60 sonication pulses during optimization of the 50% methanol extraction method. Increasing the number of sonication pulses showed a significant decrease in certain metabolite concentrations. Amino acids, such as isoleucine, leucine, methionine, tyrosine, and valine were reduced by at least 50%, and the amino acid tryptophan was not detectable by ¹H NMR following doubling of sonication pulses (data not shown). However, the concentrations of compounds such as ADP, acetate, choline, formate, methanol, and UMP were increased at least 25% upon additional sonication (data not shown). Although metabolite stability is an issue in extraction methods utilizing sonication, the most pressing issue is to be able to maximize metabolite extraction in a reproducible fashion (Table 1, Fig. S1).

We have also found metabolite stability to be dependent upon the extraction solvent used. ¹H NMR analysis of 7-day old metabolite samples revealed that biofilm metabolites extracted with methanol maintain their abundance levels better than those extracted with water. The concentrations of metabolites such as GABA, formate, and UMP were reduced by at least 10% in water-extracted samples; furthermore, NAD⁺ and NADP⁺ were absent from these samples after 7 days of storage (data not shown). The only metabolites degraded by more than 10% in methanol-extracted biofilm samples were NAD⁺ and UMP (data not shown). One possible explanation for these results is that having methanol present during extraction may aid in the inactivation of degradative enzymes that could affect metabolite levels.

Metabolite extractions were performed in triplicate for each method and metabolite concentrations were determined and normalized to total viable cell counts (CFUs) (Table 1). The 100% water method yielded the highest concentrations for the majority of the metabolites identified; the 50% methanol method yielded slightly lower metabolite concentrations, but permitted the identification of five and eight additional metabolites that were not observed in the 100% water extraction of the planktonic and biofilm samples, respectively (Fig. 4).

Fig. (4).

Venn diagram of metabolites identified with each of the methods developed for a) planktonic and b) biofilm P. aeruginosa cell cultures.

The greater metabolite concentrations observed for the 100% water method could be due to the more ordered and compact ice crystal formation during freeze-thaw cycles. Integrity of the cellular membrane of another Gram-negative bacterium, Escherichia coli, has been shown to be compromised following freezing [18], and the majority of cells that experience intracellular ice formation at a rapid freezing rate, including P. aeruginosa, are damaged if not ruptured as a result of ice formation [19]. The method using 100% methanol yielded the lowest metabolite concentrations due, in all likelihood, to insufficient cell lysis as samples produced using the 100% methanol protocol did not freeze during the −80 °C freeze-thaw cycles. The 50% methanol extraction was determined to be the optimal method for the profiling of the intracellular metabolome of P. aeruginosa as it yielded high metabolite recovery for both planktonic and biofilm phenotypes and proved to be very reproducible (Table 1, Fig. S1). Furthermore, the reduced polarity of methanol in comparison to water may enhance the solubility and subsequent extraction of marginally soluble metabolites; this may in part allow the 50% methanol method to yield a broader range of metabolites following extraction. In addition, methanol may aid in the release of metabolites from proteins present in the crude extracts.

CONCLUSION

In this study, an effective washing and metabolite extraction method was developed for metabolite profiling of P. aeruginosa grown both in planktonic and biofilm cultures. This optimized protocol includes washing with PBS to remove extracellular materials and extracellular small molecules while causing no detectable leakage of intracellular metabolites. In contrast to the 60% methanol procedure, the washing method developed here limits the loss of intracellular metabolites that are otherwise almost completely lost when using 60% methanol. For example, we observed loss of essential amino acids such as betaine, glycine, methionine, and tyrosine. In addition, this study demonstrated that the 50% methanol extraction method yields the largest number of metabolites, as well as several additional metabolites not seen with other protocols, at adequate concentrations for identification and quantification by 1D ¹H NMR. Inclusion of sonication following each freeze-thaw cycle enhanced the metabolite extraction efficiency for nearly all metabolites of interest, including approximately half of those exhibiting at least a 2-fold concentration increase compared to metabolite profiles from samples prepared without a sonication step. The metabolite extraction method developed in this study has been optimized for planktonic and biofilm P. aeruginosa and demonstrates that it is crucial to develop and customize extraction methods for each microorganism and metabolic phenotype of interest, prior to undertaking extensive quantitative metabolite profiling studies.

Biography

Mary C.B. Ammons

Valérie Copié