3D TOCSY-HSQC NMR for Metabolic Flux Analysis Using Non

Feb 5, 2016 - Home · Browse the Journal ... Department of Environmental Sciences, Washington State ... [email protected]. ... is an open access article...
2 downloads 0 Views 593KB Size
Download PDF
Article pubs.acs.org/ac

3D TOCSY-HSQC NMR for Metabolic Flux Analysis Using NonUniform Sampling P. N. Reardon,*,† C. L. Marean-Reardon,†,‡ M. A. Bukovec,∥ B. E. Coggins,§ and N. G. Isern† †

Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, 3335 Innovation Boulevard, Richland, Washington 99352, United States ‡ Department of Environmental Sciences, Washington State University, Richland, Washington 99354, United States § Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710, United States ∥ Department of Chemical, Paper and Biomedical Engineering, Miami University, Oxford, Ohio 45056, United States S Supporting Information *

ABSTRACT: 13C-Metabolic Flux Analysis (13C-MFA) is rapidly being recognized as the authoritative method for determining fluxes through metabolic networks. Site-specific 13 C enrichment information obtained using NMR spectroscopy is a valuable input for 13C-MFA experiments. Chemical shift overlaps in the 1D or 2D NMR experiments typically used for 13 C-MFA frequently hinder assignment and quantitation of site-specific 13C enrichment. Here we propose the use of a 3D TOCSY-HSQC experiment for 13C-MFA. We employ NonUniform Sampling (NUS) to reduce the acquisition time of the experiment to a few hours, making it practical for use in 13 C-MFA experiments. Our data show that the NUS experiment is linear and quantitative. Identification of metabolites in complex mixtures, such as a biomass hydrolysate, is simplified by virtue of the 13C chemical shift obtained in the experiment. In addition, the experiment reports 13C-labeling information that reveals the position specific labeling of subsets of isotopomers. The information provided by this technique will enable more accurate estimation of metabolic fluxes in large metabolic networks.

T

overcome this limitation.6 Importantly, in these studies additional position-specific enrichment information obtained from tandem MS resulted in improved metabolic flux estimations, demonstrating the utility of this information.6 NMR spectroscopy can provide detailed isotopic labeling information for each specific carbon in a metabolite of interest. Various NMR experiments, such as 1D 13C NMR, 1D 1H NMR, 2D 1H NMR, and 2D 1H−13C HSQC NMR, have been used for 13C-MFA.7,8 One-dimensional NMR experiments have the advantage of being relatively fast compared to higher dimensional experiments; however, significant chemical shift overlap can complicate identification and quantification of isotopomers. Two-dimensional experiments can alleviate this problem by providing an additional chemical shift dimension to reduce chemical shift overlap. However, chemical shift overlap is often still a problem in 2D NMR experiments, especially in experiments where the large one-bond proton-carbon Jcouplings are used to identify and quantify isotopomer labeling. 3D NMR has been proposed as a method to further reduce chemical shift overlap and improve identification and

he powerful tool known as 13C-Metabolic Flux Analysis (MFA) enables the analysis of fluxes through a mapped metabolic reaction network by following the fates of the carbon atoms in 13C-enriched substrates through observation of the labeling patterns of resulting metabolites.1 This method is being applied in a number of fields to better understand metabolic energy transduction in living systems.2 The increased understanding gained from MFA analysis provides key insights into metabolic networks. These insights can be leveraged by synthetic biology-based engineering efforts to modify living systems for the production of metabolites or other products of interest, such as biofuels or fine chemicals.3 MFA relies on accurate identification of metabolic isotopomers and subsequent quantification of 13C labeling ratios within those isotopomers.4,5 Nuclear magnetic resonance spectroscopy (NMR) and mass spectrometry are the two major analytical techniques used to characterize isotopomers. To date, mass spectrometry (MS) has become more widely used for MFA, primarily due to its high throughput and high sensitivity. However, the typical gas chromatography mass spectrometry (GC/MS) methods employed for 13C-MFA often provide incomplete information about isotopomer distributions, with position-specific labeling being particularly difficult to obtain. In recent studies, tandem MS methods have been used to partially © XXXX American Chemical Society

Received: November 30, 2015 Accepted: February 5, 2016

A

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Whole cell hydrolysates were prepared from BW25113 Escherichia coli grown under steady-state 13 C labeling conditions.1 Cells were grown in modified M9 minimal media with 20% U-13C glucose and 80% natural abundance glucose as the carbon sources. Cells were maintained in log phase by diluting cells 100-fold into fresh media upon reaching an OD600 of ∼1.0 and repeating this process for a total of three dilutions. For the final dilution and growth, the cells were diluted 100fold into 400 mL of fresh media and grown to an OD600 of ∼1.0. Cell cultures were split into two 200 mL aliquots; after harvesting by centrifugation at 9,000×g, the resulting cell pellets were lyophilized. For acid hydrolysis, lyophilized cell pellets (∼120 mg dry weight) were suspended in 2 mL of 6 M HCl. Samples were incubated at 105 °C in a sand bath for 16 h. Following incubation, samples were allowed to cool and then were dried under a stream of dry nitrogen. Hydrolyzed samples were dissolved in 0.5 mL of 99.9% D2O and filtered with a 0.2 μm spin filter. Samples were prepared for NMR by adding d6DSS and NaN3 to a final concentration of 0.5 mM and 0.02%, respectively. NMR samples were analyzed in 3 mm NMR tubes due to high salt concentrations in the samples. NMR Data Collection and Processing. The 3D TOCSYHSQC can easily be modified for 13C MFA analysis by removing the 13C refocusing pulse that is normally used to refocus the heteronuclear J-couplings during t1. This results in the observation of 1H, 13C J-couplings in the F1 dimension and an otherwise typical HSQC in the F2, F3 dimensions. Data were collected on either an 800 MHz Agilent VNMRS spectrometer or a 600 MHz Agilent VNMRS spectrometer, both equipped with HCN triple resonance cryogenic probes. The 3D NMR data were collected using random NUS sampling of ∼20% in each dimension, with a total sampling of 3.9%. We collected the 3D TOCSY-HSQC experiments with a resolution equivalent to 256 increments in F1 and 128 increments in F2. A DIPSI-2 mixing scheme was used with a field strength of 5995 Hz (600 MHz), 7995 Hz (800 MHz) and total mixing time of 50 ms. Other NMR parameters were spectral windows of 10 ppm in F1, 100 ppm in F2 and 16.3 ppm in F3, 1024 complex points in F3, 4 scans per increment, and a recycle delay of 1s. Acquisition of the data with these parameters and a randomized NUS pattern results in the reduction of the acquisition time from ∼7 days to ∼6.5 h. All 3D NMR data were Fourier transformed, apodized, phased, and baseline corrected using nmrPipe.26 Data were apodized using a shifted sine squared function and zero filled to twice their original size. Spectra were automatically baseline corrected in nmrPipe. Spectra were phased manually in nmrDraw. Reconstruction of the NUS data was carried out using the SCRUB software package, which has been shown to quantitatively reconstruct NUS data.22 Integration of peak volumes was carried out using Sparky and Gaussian line fitting. NMR data was visualized using either Sparky or NMRviewJ.27

quantification of isotopomers. In particular, 3D J-resolved 1 H−1H experiments, such as J-COSY, have been implemented using spatio temporal encoding.9 However, experiments that rely solely on 1H chemical shifts for identification are still limited in their ability to uniquely identify metabolites. Carbon chemical shifts are well suited for metabolite identification due to their much broader chemical shift distribution. The combination of proton and carbon chemical shifts provided by 2D HSQC experiments have been shown to provide highquality resolution and identification of metabolites in complex mixtures.10 2D HSQC experiments can also be used to detect 13 C labeling of neighboring carbons, based on the multiplet structure observed in the indirect carbon dimension, if the experiment is collected at sufficiently high resolution.11 However, deconvolution of these multiplet structures can be difficult, especially at the resolutions typically used in the indirect dimension of an HSQC experiment. Typically, only the labeling of neighboring carbons is determined using HSQC experiments due to the relatively small magnitudes of the 2bond and higher C−C J-couplings. The 3D TOCSY-HSQC experiment 12 combines the enhanced identification of metabolites from the HSQC experiment with the relatively straightforward isotopomer analysis of the 2D COSY or 2D TOCSY. However, conventional NMR data collection techniques normally preclude the use of this experiment for 13C-MFA because the time required to collect the 3D spectrum is on the order of 3 to 7 days, which is often prohibitive. Recent developments in fast NMR technologies, such as non-uniform sampling (NUS), now provide methods to bypass this limitation.13−21 NUS strategies can be used to dramatically reduce the time required to collect 3D and higher-dimensional NMR spectra. This time reduction has been used to improve protein structure determination or enable new experimental applications, such as in-cell NMR spectroscopy.22−25 In the case of 13C MFA, data collection must be repeated for a suitable number of biological replicates, making the use of conventional 3D NMR impractical. Therefore, we have employed NUS technology to evaluate the feasibility of using 3D TOCSY-HSQC to obtain position specific 13C enrichment information for use in 13C MFA studies.

■

MATERIALS AND METHODS Preparation of Samples. Initial testing of the experiment was carried out on either a mixture of 1,2-13C glutamate and U-13C glutamate or a mixture of 2-13C alanine and U-13C alanine. The concentrations of 1,2-13C glutamate and U-13C glutamate were varied to yield mixtures with isotopomer ratios between 10% and 90%. The total concentration of glutamate in these mixtures was 10 mM. Solutions of alanine were prepared with a 1:1 combination of 2-13C alanine and U-13C alanine. Samples were dissolved in 1 mL of 6 M HCl and incubated for 16 h in a 105 °C sand bath. Samples were allowed to cool, dried under a stream of dry nitrogen, and dissolved in 99.9% D2O. A final concentration of 0.5 mM d6-4,4-dimethyl-4-silapentane-1sulfonic acid (d6-DSS) was included as a chemical shift reference and 0.02% NaN3 was included to inhibit microbial growth. The NaN3 and d6-DSS was purchased as a 10X stock solution (Chenomx, Alberta, CA). A final concentration of 0.2 mM gadolinum diethylene triamine pentaacetic acid (GdDTPA) was added to reduce T1 in the glutamate and alanine samples.

■

RESULTS AND DISCUSSION A typical 3D TOCSY-HSQC uses a carbon refocusing pulse during t1 to refocus the 1H,13C 1J-couplings. Removing this pulse allows evolution of the 1H, 13C 1J-couplings resulting in peaks in the F1 dimension that are split by these couplings. Analysis of the multiplet structure of the NMR peaks in the F1 dimension enables quantitation of the extent of 13C enrichment of the directly bonded carbon. The NMR peak is split into a doublet for protons attached to a 13C, while protons attached to 12 C appear as singlets, as long as those protons are located in a B

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Schematic representation of the 3D-TOCSY-HSQC spectrum of alanine. A. 2D F2,F3 projection of the 3D spectrum. This spectrum is equivalent to a 2D-HSQC spectrum. The Cα carbon is labeled 1, and the Cβ carbon is labeled 2. B. Strip plot showing the F1 dimension at the carbon chemical shift corresponding to carbon 2. Carbon 2 appears as a doublet due to the HSQC element in the pulse sequence. TOCSY cross peaks will appear as either singlets or doublets in the strip, depending on the labeling state of the directly bonded carbon. The schematic is showing a mixture of labeling at the carbon 1 position, as shown by the presence of both the singlet and the doublet in the F1 dimension. C. Strip plot showing the F1 dimension at the carbon chemical shift corresponding to carbon 1. The schematic shows a mixture of labeling at the carbon 2 position, as shown by the presence of both the singlet and doublet in the F1 dimension. D. Enrichment patterns with corresponding spectral characteristics. For each enrichment pattern, it is indicated if the pattern would be detected by the experiment and at which spectral location. The letters and numbers in the spectral location text refer to the figure panel in this figure and the corresponding number in the referred to figure panel.

spin system that contains a proton directly bonded to 13C. The relative contribution of the integrated volumes of the singlet or doublet to the total integrated volume of the multiplet corresponds to the percent labeling of the attached carbon with 12C or 13C, respectively. Peak overlap in F1 (proton) is reduced by virtue of the separation afforded by F2 (carbon) and F3 (proton) dimensions. An interesting aspect of this experiment for tracking 13C labeling is that this position-specific labeling information is only read out from carbons that are themselves 13C labeled. This is a consequence of the HSQC element in the experiment, which selects for protons directly bonded to 13C and results in only their detection in the F3 dimension. A schematic representation of the NMR spectrum and associated data interpretation is given in Figure 1. To test the accuracy and precision of this experiment for the analysis of isotopic enrichments, we used mixtures of 1,2 13C labeled glutamate and uniformly labeled glutamate, which varied from 10% uniformly 13C labeled to 90% uniformly 13C

labeled, and a total glutamate concentration of 10 mM. A representative set of subspectra from the 60% U-13C labeled and 40% 1,2-13C labeled sample is shown in Figure 2B. These data show the expected doublets and singlets for protons directly bonded to 13C and 12C, respectively. Analysis of the isotope labeling for the standard mixture and comparison to the expected isotope labeling based on the amount of each isotopomer added shows that the 3D TOCSY-HSQC experiment is accurate with good linearity. Figure 2A summarizes these observations. The slope of the linear regression near 1, yintercept of near 0, and good r2 indicate that the method is linear and quantitative. The strip plots in Figure 2B highlight the ability of this experiment to detect 13C enrichment and differentiate between sets of metabolite isotopomers. The HSQC element of the pulse sequence selects for 13C attached to the detection (F3) proton, and consequently only a subset of isotopomers is quantifiable for a given set of resonances. In the Cα strip (55.1 C

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Summary of the NUS 3D-TOCSY-HSQC data obtained on various mixtures of 13C labeled glutamate. A. Plot of expected 13C enrichment vs the 13C enrichment determined using the NUS 3D-TOCSY-HSQC experiment. The red line is the linear regression line fit to the data. B. Representative strip plots from the mixture of 60% U-13C-glutamate, 40% 1,2-13C-glutamate. The proton F1 dimension chemical shift is shown on the vertical axis, the proton F3 dimension chemical shift is shown on the horizontal axis, and the carbon F2 dimension chemical shift is noted at the top of the strip. The strip at 55.1 ppm in carbon corresponds to the resonance for the Cα of glutamate. The strip at 32.2 ppm in carbon corresponds to the resonance for the Cγ of glutamate. Error bars are ± one standard deviation.

analysis are shown in Table 1. The measured standard deviations of the position-specific 13C labeling percentages

ppm carbon) shown in Figure 2B, we are detecting enrichment of the Cγ carbon when Cα is enriched, because this set of resonances arises from detection of Hα and Cα in F3 and F2, respectively. The singlet for 12C at Cγ is only observed in this mixture when Hα is the detection proton. This is because the 1,2-13C glutamate is only 13C labeled at Cα and the carbonyl carbon, which does not have a directly bonded proton. Therefore, the resonances from the 1,2-13C glutamate spin system are only detected when Hα is detected in F3. When Hγ is detected in F3, only the uniformly labeled glutamate is 13C enriched at the Cγ position, resulting in the detection of resonances from only the uniformly labeled glutamate spin system. As expected, only doublets are observed in the Cγ strip (32.2 ppm carbon) in Figure 2B. Thus, the experiment selectively measures the 13C enrichment of a subset of isotopomers based on the presence of 13C directly bonded to the proton detected in the F3 dimension. Our NMR data show that the 3D TOCSY-HSQC experiment is capable of accurately and precisely determining patterns of isotope labeling which would be useful for metabolic flux analysis. To achieve sufficiently high resolution for this purpose we used an NUS sampling strategy. In this case, we chose to use a random NUS strategy that acquired a total of ∼3.9% of the data normally required using conventional sampling. Several reconstruction algorithms have been proposed for reconstruction of NUS data. We chose to use the SCRUB algorithm for data reconstruction in our experiments, since it has been shown to quantitatively reconstruct NOE data at low sampling percentages.22 Our above analysis shows that this strategy is well-suited for our application. Having shown that the NUS 3D TOCSY-HSQC enabled quantitative analysis of 13C labeling, we tested the experiment on E. coli hydrolysates prepared for a typical 13C MFA experiment.1 First we examined the repeatability of the NUS 3D TOCSY-HSQC by performing the experiment multiple times on the same hydrolysate sample. The results of this

Table 1. 13C Enrichment Determined from 3 3D-TOCSYHSQC Experiments Collected on an E. coli Biomass Hydrolysatea percent 13C enrichment ACβ‑Cα ACα‑Cβ DCβ‑Cα DCα‑Cβ ECγ‑Cα FCβ‑Cα ICδ‑Cβ ICγ2‑Cβ KCδ‑Cε KCβ‑Cα LCβ‑Cα LCβ‑Cδ LCα‑Cδ PCγ‑Cα RCβ‑Cα RCβ‑Cδ TCγ‑Cβ TCα‑Cβ

87.4% 89.0% 62.7% 62.5% 15.2% 93% 1.4% 90.7% 77.4% 77.7% 14.8% 51.0% 6.4% 5.1% 59% 6% 46.7% 24.0%

SD 0.3% 0.1% 0.5% 0.6% 0.9% 4% 0.5% 0.4% 0.1% 1.5% 1.5% 1% 0.7% 0.9% 2% 2% 0.1% 0.5%

SE 0.2% 0.1% 0.3% 0.4% 0.5% 2% 0.3% 0.3% 0.1% 0.9% 0.9% 0.6% 0.4% 0.5% 1.3% 1.3% 0.1% 0.3%

a The first subscript denotes the detection proton/carbon pair, and the second subscript denotes the correlated proton/carbon pair. Standard error is standard error in the mean.

ranged from a low of 0.1% to a high of 4%. The average of the standard deviations was 1%, indicating that the error associated with the NUS 3D TOCSY-HSQC is generally small. Second, we applied the experiment to 3 hydrolysates that were grown in parallel. A set of representative data is shown as a strip plot in Figure 3. The results from these experiments are summarized in D

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Multidimensional NMR techniques, especially 3D or higher, are often considered insensitive techniques. The high concentrations that can be attained in biomass hydrolysates largely overcome this issue for many 13C-MFA experiments. However, the question of sensitivity is still important, especially for studies that do not rely on biomass hydrolysates. Advances in cryogenic probe technologies have significantly improved sensitivity, suggesting that the use of 3D NMR on lower concentration samples may be feasible. Therefore, we examined the ability of the 3D TOCSY-HSQC to quantify the isotope labeling pattern of an amino acid at reduced concentrations. We examined three concentrations of alanine, ranging from 100 μM to 1 mM at 600 and 800 MHz. Strip plots of this data collected at 800 MHz are shown in Figure 4. In all cases, the measured

Figure 3. Representative NMR data from a NUS 3D-TOCSY-HSQC spectrum of an E. coli biomass hydrolysate sample. Spectra showing the crosspeaks from (A.) Hβ-Hδ-Cδ of isoleucine; (B). Hβ-Hγ2-Cγ2 of isoleucine; (C.) Hβ-Hα-Cα of aspartate; (D.) Hγ-Hβ-Cβ of threonine; (E.) Hβ-Hα-Cα of alanine; (F.) Hα-Hβ-Cβ of alanine.

Table 2. The standard deviations ranged from 0.2% to 9% with an average of 2%. There is an increase in standard deviation when using samples from multiple cultures, which is likely explained by biological variability between the cultures.

Figure 4. Qualitative analysis of sensitivity at 800 MHz. Three mixtures of 50% U-13C alanine and 50% 2-13C alanine were examined using the NUS 3D-TOCSY-HSQC experiment. Three total concentrations of alanine were analyzed, A. 100 μM, B. 500 μM, and C. 1 mM. Contours were scaled for best visualization of the resonances.

Table 2. 13C Enrichment Determined from 3 Parallel Biomass Hydrolysates Using the 3D-TOCSY-HSQCa percent 13C enrichment ACβ‑Cα ACα‑Cβ DCβ‑Cα DCα‑Cβ ECγ‑Cα FCβ‑Cα ICδ‑Cβ ICγ2‑Cβ KCδ‑Cε KCβ‑Cα LCβ‑Cα LCβ‑Cδ LCα‑Cδ PCγ‑Cα RCβ‑Cα RCβ‑Cδ TCγ‑Cβ TCα‑Cβ

88.4% 89.2% 63.4% 63.8% 16.3% 91% 1.9% 90.6% 77.5% 75% 14% 51.5% 6.3% 5.1% 58% 5% 46.9% 29%

SD

SE 0.6% 0.6% 1% 0.9% 1% 2% 1% 0.3% 0.2% 2% 4% 1.2% 0.4% 1.4% 3% 2.% 0.6% 9%

amount of 13C label agreed within ∼1% of the expected amounts of labeled amino acids, summarized in Table S1. These data show that it is fundamentally possible to quantify labeling patterns for compounds present at submillimolar concentrations using this 3D NMR experiment in a reasonable time. As with most NMR experiments, overall sensitivity can be modulated by the amount of signal averaging employed. In addition, the TOCSY mixing time can be altered to optimize the observation of different protons in the indirect dimension.28 Examination of the E. coli biomass hydrolysate data reveals that the 3D TOCSY-HSQC can provide additional labeling information that may not be easily obtained from GC/MS or 1 H NMR methods. An example of the information provided by the 3D TOCSY-HSQC is shown by the labeling observed for isoleucine. In this case, the 13C enrichment at the Cβ position is significantly different in isotopomers where Cδ is enriched when compared to isotopomers where Cγ2 is enriched. The data shows that when Cδ is labeled, Cβ is only labeled at 1.9%. In contrast, when Cγ2 is labeled, Cβ is labeled at 90.6%. In isoleucine biosynthesis, Cγ2 and Cβ originate from the same molecule of pyruvate, while the other carbons originate from a molecule of α-ketobutyrate. Thus, the experiment is reporting

0.4% 0.3% 0.6% 0.5% 0.6% 1.3% 0.6% 0.2% 0.1% 1.4% 2% 0.7% 0.2% 0.8% 1.6% 1.4% 0.3% 5%

a The first subscript denotes the detection proton/carbon pair, and the second subscript denotes the correlated proton/carbon pair. Standard error is standard error in the mean.

E

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry on the differences in 13C labeling of the two precursor molecules that contribute to isoleucine biosynthesis. The labeling of leucine provides another example of the detailed 13C isotopic labeling information that can be obtained using the 3D TOCSY-HSQC. The data show that when Cα is labeled, an average of 6.3% of the Cδ’s is labeled, and when Cβ is labeled, an average of 51.5% of the Cδ’s is labeled. These differences in 13C labeling are related to the different precursor pathways that contribute to labeling at each of these positions in leucine. The Cα carbon is derived from acetyl-CoA, while the Cδ’s and Cβ carbons are derived from two different molecules of pyruvate. The differences in the above labeling patterns provide additional information that can be incorporated into metabolic flux models. Typically, 13C MFA relies on the analysis of amino acids derived from biomass via acid hydrolysis. However, there is growing interest in incorporating intracellular metabolites and their respective labeling patterns into 13C MFA studies.29 Intracellular metabolite mixtures are often extremely complex and can exhibit significant chemical shift overlap when analyzed using 1H NMR spectroscopy, which inhibits unambiguous identification of metabolites. This chemical shift overlap is often exacerbated in 13C MFA experiments because of the large 1 H−13C J-coupling constants, which are used to quantify the extent of 13C labeling. For metabolomics data, the inclusion of both proton and carbon chemical shifts, such as in a 13C-HSQC experiment, has been shown to improve metabolite identification.10 The 3D TOCSY-HSQC experiment will also benefit from this reduction in chemical shift overlap and subsequent improvement in metabolite identification, because the 13C and proton observe dimensions produce a spectrum identical to the 2D 13C-HSQC. In the case of the biomass hydrolysates analyzed here, the availability of proton and carbon chemicals in the 3D experiment render the identification of amino acid resonances trivial. In conclusion, this work demonstrates the utility of 3D heteronuclear NMR for the analysis of site-specific 13C enrichment in biological mixtures. We show that NUS strategies are well-suited to this application and can dramatically reduce the time required to collect these data sets when compared to conventional NMR methods. Importantly, processing of the data uses existing software tools that are readily available to the NMR community, and the data are analyzed using the same methods as conventionally sampled data. Our experiments show that a 3D TOCSY-HSQC can provide valuable site-specific 13C enrichment information. The paired 13C enrichment information obtained from this experiment is complementary to that obtained with homonuclear 1H NMR methods that can detect protons without direct or nearby 13C enrichment. Furthermore, this information will provide complementary data to GC/MS and tandem MS methods and result in improvements to the certainty of flux estimations.

■

■

filtering; and Table S1, percent labeling for alanine at three total alanine concentrations (PDF)

AUTHOR INFORMATION

Corresponding Author

*Phone: 509-371-7673. Fax: 509-371-6564. E-mail: Patrick. [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research was performed using EMSL, a DOE Office of Science User Facility sponsored by the Office of Biological and Environmental Research and located at Pacific Northwest National Laboratory. We thank Dr. Hector Garcia-Martin for providing the BW25113 E. coli strain and Dr. Evgeny Tishchenko for assistance with pulse sequence programing. Funding for this work was provided in part by the William Wiley Postdoctoral Fellowship from EMSL to P.N.R. Additional funding was provided by the Development of an Integrated EMSL MS and NMR Metabolic Flux Analysis Capability In Support of Systems Biology: Test Application for Biofuels Production intramural research project from EMSL to N.G.I.

■

ABBREVIATIONS d6-DSS d6-4,4-dimethyl-4-silapentane-1-sulfonic acid Gd-DTPA gadolinum diethylene triamine pentaacetic acid HSQC heteronuclear single quantum coherence TOCSY total correlation spectroscopy NUS non-uniform sampling GC/MS gas-chromatography mass spectrometry MS mass spectrometry MFA metafolic flux analysis NMR nuclear magnetic resonance

■

REFERENCES

(1) Zamboni, N.; Fendt, S. M.; Ruhl, M.; Sauer, U. Nat. Protoc. 2009, 4, 878−892. (2) Gebreselassie, N. A.; Antoniewicz, M. R. Metab. Eng. 2015, 31, 132−139. (3) Dahl, R. H.; Zhang, F.; Alonso-Gutierrez, J.; Baidoo, E.; Batth, T. S.; Redding-Johanson, A. M.; Petzold, C. J.; Mukhopadhyay, A.; Lee, T. S.; Adams, P. D.; Keasling, J. D. Nat. Biotechnol. 2013, 31, 1039− 1046. (4) Wiechert, W. Metab. Eng. 2001, 3, 195−206. (5) Antoniewicz, M. R. J. Ind. Microbiol. Biotechnol. 2015, 42, 317− 325. (6) Choi, J.; Antoniewicz, M. R. Metab. Eng. 2011, 13, 225−233. (7) Szyperski, T. Eur. J. Biochem. 1995, 232, 433−448. (8) Szyperski, T. Q. Rev. Biophys. 1998, 31, 41−106. (9) Boisseau, R.; Charrier, B.; Massou, S.; Portais, J. C.; Akoka, S.; Giraudeau, P. Anal. Chem. 2013, 85, 9751−9757. (10) Xia, J.; Bjorndahl, T. C.; Tang, P.; Wishart, D. S. BMC Bioinf. 2008, 9, 507. (11) Choudhary, M. K.; Yoon, J. M.; Gonzalez, R.; Shanks, J. V. Biotechnol. Bioprocess Eng. 2011, 16, 419−437. (12) Cavanagh, J.; Fairbrother, W. J.; Palmer, A. G., III; Rance, M.; Skelton, N. J. In Protein NMR Spectroscopy, 2nd ed.; Cavanagh, J.,

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b04535. Figure S1, pulse sequence diagram; Figure S2, qualitative analysis of sensitivity at 600 MHz; Figure S3, comparison of 3D-TOCSY-HSQC with and without zero quantum F

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Fairbrother, W. J., Palmer, A. G., Rance, M., Skelton, N. J., Eds.; Academic Press: Burlington, 2007; pp 533−678. (13) Barna, J. C. J.; Tan, S. M.; Laue, E. D. J. Magn. Reson. 1988, 78, 327−332. (14) Coggins, B. E.; Venters, R. A.; Zhou, P. J. Am. Chem. Soc. 2004, 126, 1000−1001. (15) Coggins, B. E.; Zhou, P. J. Magn. Reson. 2006, 182, 84−95. (16) Coggins, B. E.; Zhou, P. J. Biomol. NMR 2008, 42, 225−239. (17) Hoch, J. C.; Maciejewski, M. W.; Filipovic, B. J. Magn. Reson. 2008, 193, 317−320. (18) Hyberts, S. G.; Heffron, G. J.; Tarragona, N. G.; Solanky, K.; Edmonds, K. A.; Luithardt, H.; Fejzo, J.; Chorev, M.; Aktas, H.; Colson, K.; Falchuk, K. H.; Halperin, J. A.; Wagner, G. J. Am. Chem. Soc. 2007, 129, 5108−5116. (19) Kazimierczuk, K.; Orekhov, V. Y. Angew. Chem., Int. Ed. 2011, 50, 5556−5559. (20) Kazimierczuk, K.; Stanek, J.; Zawadzka-Kazimierczuk, A.; Kozminski, W. Prog. Nucl. Magn. Reson. Spectrosc. 2010, 57, 420−434. (21) Mobli, M.; Stern, A. S.; Hoch, J. C. J. Magn. Reson. 2006, 182, 96−105. (22) Coggins, B. E.; Werner-Allen, J. W.; Yan, A.; Zhou, P. J. Am. Chem. Soc. 2012, 134, 18619−18630. (23) Reardon, P. N.; Spicer, L. D. J. Am. Chem. Soc. 2005, 127, 10848−10849. (24) Reardon, P. N.; Sage, H.; Dennison, S. M.; Martin, J. W.; Donald, B. R.; Alam, S. M.; Haynes, B. F.; Spicer, L. D. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 1391−1396. (25) Robinson, K. E.; Reardon, P. N.; Spicer, L. D. Methods Mol. Biol. 2012, 831, 261−277. (26) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A. J. Biomol. NMR 1995, 6, 277−293. (27) Johnson, B. A. Methods Mol. Biol. 2004, 278, 313−352. (28) Cavanagh, J.; Chazin, W. J.; Rance, M. J. Magn. Reson. 1990, 87, 110−131. (29) Millard, P.; Massou, S.; Wittmann, C.; Portais, J. C.; Letisse, F. Anal. Biochem. 2014, 465, 38−49.

G

DOI: 10.1021/acs.analchem.5b04535 Anal. Chem. XXXX, XXX, XXX−XXX