ScienceSoft, LLC

CHAPTER 6: Tutorial IV: Advanced Structure Elucidation

The previous tutorials: CHAPTER 3: "Tutorial I: Using NMRanalyst", CHAPTER 4: "Tutorial II: Setting Analysis Parameters", and CHAPTER 5: "Tutorial III: Combining Analysis Results" cover the basic NMRanalyst spectral analysis. This tutorial covers generating a generic line list for nitrogen resonances (SECTION 6.2: "Generation of a Generic 15N Resonance List"), positioning nitrogen atoms in determined molecular carbon skeletons (SECTION 6.3: "15N HMBC Spectrum Analysis"), using additional spectra to reduce the number of possible molecular skeletons (SECTION 6.5: "Using DQF-COSY" and SECTION 6.6: "Using DQ 1,1-ADEQUATE"), and placing unobserved heteroatoms in determined structures (SECTION 6.7: "Adding Heteroatoms").

Dr. Péter Sándor (Varian Deutschland GmbH) graciously provided 500 MHz datasets of 20 mg strychnine1 (MF: C21H22N2O2) acquired in CDCl3 using a 3 mm tube. The provided 1D proton (acquisition time: 1.1 min), 1D carbon (39.4 min), multiplicity edited 13C gHSQC (39 min), 13C gHMBC (1 h 17 min), 15N gHMBC (3 h 12 min), gDQF-COSY (1 h 36 min), and DQ 1,1-ADEQUATE (19 h 21 min) spectra are used in this tutorial. Copy the supplied strychnine dataset containing the experimental data. Delete its ADEQUATE, DQF-COSY, HMBC, HSQC, and N15_HMBC subdirectories. This removes the provided NMRanalyst parameter settings for this compound. In NMRanalyst, start [Edit] [Preferences...], set Mode: to [Full NMRanalyst] and deselect the [Show All Input Fields] switch.

6.1 1D Carbon, 1D Proton, Edited HSQC, and HMBC Analyses

The analysis of these spectrum types is covered by previous tutorials. Switch NMRanalyst to the copied strychnine directory. Select the HMBC spectrum type. Change to the 1D Analysis workwindow. Load from directory carbon.fid the file procpar and run the 1D Analysis workwindow. Load HMBC/1d_analysis.log to add the table of obtained reference corrected numerical signal descriptions with CDCl3 solvent resonances marked for removal. Rerun the workwindow. Display the shown carbon spectrum (file carbon.plot).

Switch to the HSQC spectrum type and select the 1D Analysis workwindow for the 1D proton spectrum analysis. Load file proton.fid/procpar. Run the analysis. Display the created and shown proton.plot spectrum.

Switch to the FFT workwindow. Load C13_ghsqc.fid/procpar. Set F1: Start of Spectrum to 2533.119 (taking into account the 6.052 Hz correction from the 1D carbon spectral analysis), and run the workwindow. Switch to the 2D Analysis workwindow. After optimization, as described in the previous tutorial, the final 2D Analysis parameters are: Thresholds: Integral of 66.917, [F1 Phase] and [F2 Phase] switches in Mapping and Detection deselected, F1 Phase = -1.32221 + -0.0020997 * ppm, and F2 Phase = 0.38455 + 0.0548430 * ppm. Run the 2D Analysis and it runs the Report workwindow. In the Graphic workwindow, specify the same phase functions as above and display the shown hsqc.spec file.

The resulting shown structure.plot file illustrates the detected correlations in graphical form.

Select spectrum type HMBC. Switch to the FFT workwindow, and load file C13_ghmbc.fid/procpar. Increase F1: Start of Spectrum to 2313.039 and run the FFT. Switch to the 2D Analysis workwindow. The strongest HMBC correlations result from 2-bond and 3-bond couplings. Often the 2D Analysis workwindow Integral threshold can be set so that a sufficient number of 2-bond and 3-bond correlations are detected without detection of longer-range correlations. But longer-range HMBC correlations are indistinguishable from 2-bond and 3-bond correlations. Set the Integral threshold high enough to exclude noise and spectral distortions from detection. Set this field now to 1.5, run the 2D Analysis workwindow, and it auto-runs the Report workwindow. From the Graphic workwindow, display the shown hmbc.spec.

6.2 Generation of a Generic 15N Resonance List

A quality 1D proton spectrum can be acquired in seconds. But for nuclei such as carbon or nitrogen, acquiring the 1D spectrum tends to be slower than acquiring HSQC and HMBC spectra. For strychnine, the 15N HMBC spectrum was acquired without a corresponding 1D nitrogen spectrum. Similar to SECTION 5.5: "Structure Identification Without 1D Carbon Spectrum", a generic line list needs to be generated by NMRanalyst to describe the nitrogen frequencies. Select spectrum type N15_HMBC and switch to its FFT workwindow. Load file N15_ghmbc.fid/procpar. Run the 15N HMBC spectrum transform.

To generate the generic resonance list for this nitrogen dimension, switch to the 1D Analysis workwindow. Set the Input File Format menu to [Generate Generic List] which hides input sections that do not apply. Set Use Relaxation Time [s] to 0.015. Copy the four values from the FFT workwindow F1 dimension to the 1D Analysis workwindow fields of the same name: Observe Frequency is 50.655 MHz, Spectral Width is 9117.154 Hz, Start of Spectrum is -261.032 Hz, and Number of Points is 320. Run the 1D Analysis workwindow and the generic 15N line list consisting of 78 resonances is generated.

6.3 15N HMBC Spectrum Analysis

With selected N15_HMBC spectrum type, switch to the 2D Analysis workwindow. Set F1 1D Analysis Output File to nitrogen.out. Due to the use of the generic 15N line list, set Map F1 Frequencies ± to 58 Hz to cover the 1D Analysis 115 Hz line spacing. Set Thresholds: Integral to 1.5. In the Report workwindow, set Redetermined F1 Resonance List to nnitrogen.out. Run the 2D Analysis and it runs the Report workwindow. From the Graphic workwindow, display the n15_hmbc.spec file with analyzed correlation locations (n15_hmbc_areas.plot) as shown. Determined nitrogen frequencies are saved in the specified nnitrogen.out, so an exhaustive search of the nitrogen axis is no longer needed.

In the 2D Analysis workwindow, set the F1 1D Analysis Output File to nnitrogen.out and delete the Map F1 Frequencies ± entry. In the Report workwindow delete the Redetermined F1 Resonance List entry to not overwrite the newly created line list. Now run the 2D Analysis again and it auto-runs the Report workwindow. Switch to the AssembleIt workwindow. Change the 1D: Nitrogen field to nnitrogen.out and start the workwindow. Display the created NC_corrs.plot file. It is a circle diagram. As only a few correlations were derived, click the NMRgraph [View] [Redo Layout] menu item to obtain the shown layout.

6.4 Strychnine Structures Derived From Previous Analysis Results

From the previous analysis results, the possible strychnine structures can be derived. Return to spectrum type HMBC (not N15_HMBC) and select the AssembleIt workwindow. Select the COMBINE NMR ANALYSIS RESULTS section. Set 1D: Nitrogen to nnitrogen.out. Delete the 1,1-ADEQUATE, DQF-COSY, and INADEQUATE field entries. For faster structure generation, the HMBC Weak field should be specified. To see how to choose this value, deselect the AssembleIt: ELUCIDATE MOLECULAR STRUCTURE FROM NMR DATA section in this input screen and start the workwindow. The middle of the output screen shows the detected 127 HMBC correlations:

 HMBC 'HMBC/report.log': 499.87 MHz  F1Diff=0.155026  F2Diff=0.016914
                         F1-Range=+/-9.74  F2-Range=+/-6.50
    1:  #   225   C1=  3  H2= 35  F1= 140.032  F2=  4.092  I= 20.120
    2:  #   889   C1= 10  H2= 34  F1=  77.391  F2=  4.090  I= 16.070
  127:  #   770   C1=  9  H2= 10  F1= 116.005  F2=  7.198  I=  1.507

The strongest HMBC correlation has an integral of 20.120 and the weakest one an integral of 1.507. A consistent way to acquire HMBC spectra and experience in analyzing such spectra will yield insights on how to set this Weak threshold. Set the AssembleIt workwindow HMBC Weak field to 2.2, declaring the bottom one-sixth of these correlations to potentially contain longer-range correlations. A similar Weak threshold could be set for N15_HMBC. As only two nitrogens were detected, the obtainable speed improvement is limited and considering all of these correlations to be potential longer-range correlations is fine.

Now select the AssembleIt: ELUCIDATE MOLECULAR STRUCTURE FROM NMR DATA section. Typically about 20% of expected 2-bond HMBC correlations are not observed due to unfavorable 2JC,H coupling constants. Set Evaluate: Unobserved Bonds to 10 for AssembleIt to try to derive up to this number of unobserved bonds from longer-range correlations. For aromatic systems, 3-bond correlations are typically stronger than 2-bond ones. Set the Evaluate: option menu to [Bonds Over sp2-?-?-sp2] to more likely recover such bonds. Finally, set 4-Bond HMBC Correlations to 4 and Long Range HMBC Correlations to 1 to have AssembleIt consider up to this number of longer-range correlations below the Weak threshold. Start the workwindow and in about 8 minutes (on a 3.06 GHz PC) AssembleIt reports 6 structures consistent with the specified strychnine data. They are saved in the file CC_corrs.plot.1.

Without the AssembleIt exhaustive listing of all possible structures, the first found structure might be regarded as the correct one. But display the CC_corrs.plot.1 file of possible strychnine structures. All structures have the correct number of carbons and nitrogens, correct atom valences, and explain the observed HSQC and HMBC correlations within the specified limits. One of these structures is likely the correct carbon and nitrogen skeleton of strychnine. But further information is helpful to identify which one is the correct one.

6.5 Using DQF-COSY

The HMBC 2-bond (2JC,H), 3-bond (3JC,H), and longer-range (nJC,H) correlation ambiguity cannot be resolved experimentally. A valuable source of unambiguous correlation information is an additional DQF-COSY spectrum. HMBC can detect a bond if one of its bonded carbons is protonated. For DQF-COSY bond detection, both carbons need to be protonated. But then their vicinal coupling identifies their carbon-carbon bond.

Switch to spectrum type DQF-COSY and its FFT workwindow. Load gdqfcosy.fid/procpar and run its transform. In the 2D Analysis workwindow, set Thresholds: Integral to 1, deselect the [F1 Phase] switch for Mapping and Detection, set F1 Phase = 1.56517 + -0.0017787 * ppm, and F2 Phase = 0.06599 + 0.0905206 * ppm. Run the 2D Analysis and it auto-runs the Report workwindow. In the Graphic workwindow, set the same phase functions as above, and display the dqf-cosy.spec file and the correlation locations are also displayed as shown.

Run the AssembleIt workwindow. From the Graphic workwindow display the created CC_corrs.plot. Reformat the circle diagram with the NMRgraph [View] [Redo Layout] and switch off the correlation labels through [View] [Hide Correlation Labels] to obtain the shown correlations display.

Return to spectrum type HMBC and the AssembleIt workwindow to add this DQF-COSY information. Set 2D: DQF-COSY to DQF-COSY/report.log. Run this workwindow. Two possible strychnine structures result in a few seconds.

6.6 Using DQ 1,1-ADEQUATE

The first two tutorials cover the carbon-carbon bond detection from a 2D INADEQUATE spectrum. This would resolve remaining ambiguities. But it is over an order of magnitude less sensitive than HSQC, HMBC, and DQF-COSY spectra. The ADEQUATE acronym stands for Adequate sensitivity DoublE-QUAnTum spEctroscopy. The ADEQUATE experiment detects H-C-C moieties. It is known as 1,1-ADEQUATE due to the use of 1JH,C and 1JC,C couplings. Like 2D INADEQUATE, 1,1-ADEQUATE depends on both carbons involved being 13C. Using proton excitation and detection, ADEQUATE can be several times more sensitive than 2D INADEQUATE and this spectrum type can be acquired on an indirect detection probe.

Switch to the 1,1-ADEQUATE spectrum type. In the FFT workwindow load adequate_dq.fid/procpar and adjust F1: Start of Spectrum to 14597.577 (considering the +6.052 Hz referencing correction from the 1D carbon spectrum). Run this workwindow. In the 2D Analysis workwindow, set Thresholds: Integral to 1200, to only consider stronger resonances. Run the workwindow and it auto-runs the Report workwindow. Load file ADEQUATE/report.log in the Graphic and 2D Analysis workwindows to set the phase functions. Rerun the 2D Analysis workwindow and it auto-runs the Report workwindow.

Display the adequate.spec file with correlation locations using the Graphic workwindow, as shown. Unaliased correlation areas are shown as black and F1 aliased correlations as green bounding boxes. Like 2D INADEQUATE, the 1,1-ADEQUATE signal-to-noise ratio is too limited to make spectral 1D projections useful. So the 1D proton spectrum is shown along the F2 axis. The 1D carbon spectrum does not have the double-quantum format of the F1 axis and is not shown.

Run the AssembleIt workwindow and display the created CC_corrs.plot. Reformat the circle diagram with the NMRgraph [View] [Redo Layout] and switch off the correlation labels with [View] [Hide Correlation Labels] for the shown results.

Return to spectrum type HMBC and its AssembleIt workwindow. Set 2D: 1,1-ADEQUATE to ADEQUATE/report.log and run this workwindow. The structure determination takes under one second and two structures remain. Display the shown CC_corrs.plot.1. Both structures are reported with four 4-Bond, one Long-Range, and seven Unobserved-Bond correlations.

6.7 Adding Heteroatoms

The AssembleIt derived strychnine structure represents the possible carbon and nitrogen skeletons. From the edited HSQC, the resonance frequency and number of bonded hydrogen for each carbon are known, though they are not displayed. But it is known which carbons have remaining unexplained free valences. Free valences result from bonds to heteroatoms and higher bond multiplicities, which are not detected by the used NMR spectra. Based on the obtained molecular carbon chemical shift assignments, further predictions about bonded heteroatoms and higher bond orders can be derived from their expected effect on observed carbon chemical shifts.

To add heteroatoms to the previously displayed structures, select in NMRgraph [Prediction] [Place Heteroatoms...]. In the started popup, confirm O (oxygen) is selected as heteroatom to consider and click [OK]. Oxygen atoms are placed in the structure to minimize the difference between observed and predicted carbon chemical shifts. The resulting structures are sorted by shift agreements.

Select [Prediction][13C Shift Agreements...]. The average differences between the predicted and observed carbon shifts are 0.160 and 7.992 ppm for the two structures. So the shown top structure is the most likely one and it agrees with the complete strychnine structure shown at the beginning of this tutorial.

6.8 Identifying the Molecular Structure Through Shift Prediction

Using only HMBC and HSQC derived ambiguous carbon-carbon correlations tends to become compute intensive for larger molecules where numerous candidate structures result. The chemical shift represents the chemical environment of each atom. So placing heteroatoms and sorting structures according to agreement of the carbon shifts with prediction can identify the most likely structure. In SECTION 6.4: "Strychnine Structures Derived From Previous Analysis Results", six possible structures result from HMBC and HSQC correlations. Placing heteroatoms and sorting these structures in carbon shift agreement order identifies the correct strychnine structure. In SECTION 6.5: "Using DQF-COSY", adding DQF-COSY information decreases the possible strychnine skeletons to two. Adding oxygen atoms and sorting the structures in carbon shift agreement order also identifies the correct strychnine skeleton. So 1,1-ADEQUATE information often is not needed for the structure elucidation of smaller molecules.

1Strychnine is an alkaloid extract obtained from the dried ripe seeds of Strychnos nux vomica, a small tree of the East Indies. In the past, strychnine has been used as an antiseptic, stomach tonic, circulatory stimulant, central nervous system stimulant, and medication for the relief of constipation. Strychnine is still in limited use today as a bird, mammal, and insect control agent.

ScienceSoft, LLC