Identification of L-Tryptophan by down-field ¹H MRS: A precursor for brain NAD⁺ and Serotonin syntheses

Abstract

Purpose:

To explore the presence of new resonances beyond 9.4 ppm from the human brain, down-field proton magnetic resonance spectroscopy (DF ¹H MRS) was performed in vivo in the human brain on six healthy volunteers at 7.0T.

Methods:

To maximize the SNR, a large voxel was placed within the brain to cover the maximal area in such a way that sinus cavities were avoided. A spectrally selective 90° E-BURP pulse with an excitation bandwidth (BW) of 2 ppm was used to probe the spectral chemical shift range between 9.1 ppm and 10.5 ppm. The E-BURP pulse was integrated with PRESS spatial localization to obtain non-water suppressed ¹H MR spectra from the desired spectral region.

Results:

In the DF ¹H MRS obtained from all the volunteers scanned, we have identified a new peak consistently resonating at 10.1 ppm. Proton associated with this resonance is in cross-relaxation with the bulk water as demonstrated by the water saturation and deuterium exchange experiments.

Conclusion:

Based on the chemical shift, this new peak was identified as indole (-NH) proton of L-Tryptophan (L-TRP) and was further confirmed from phantom experiments on L-TRP. These promising preliminary results potentially pave the path for investigating the role of cerebral metabolism of L-TRP in healthy and disease conditions.

Introduction

Recently published studies on the in vivo detection of nicotinamide adenosine dinucleotide (NAD⁺) in the ¹H NMR spectrum by de Graff, et al ^1,2 has re-ignited a lot of interest in determining the characteristics of the DF ¹H MRS found at chemical shifts >4.7 ppm ^3–5. A challenge of DF ¹H MRS is that many of the metabolite peaks that resonate in this region are either in direct chemical exchange or have significant cross-relaxation with water ^6–8. Consequently, water suppression may inadvertently reduce the signal from these metabolites so that they are difficult if not impossible to detect and quantify. However, without water suppression, the water signal itself may also complicate the detection and quantification of these down-field (DF) metabolites ^1,2,6. These challenges in the DF ¹H MRS have prompted the design of a new scheme of pulses for selective excitation in the DF region of interest by neither suppressing nor exciting the water signal. To address these challenges, de Graff, et al., ² deployed spectrally selective 1D LASER excitation to acquire the frequency selective excitation DF unsuppressed water spectrum using a surface coil. They selected a 20 mm thick slice covering the occipital cortex on healthy volunteers for detection of the NAD⁺ metabolites which are in the order of a few hundred micromolar concentration ². Further progress has been made to obtain a localized, water unsuppressed spectra for detection of NAD^{+ 9,10} by employing a spatially selective excitation using the E-BURP pulse sequences ^11,12 as published recently by Bagga, et al ³.

Thus far, DF ¹H-MRS studies have employed spectroscopy techniques and have investigated the DF metabolites <10.0 ppm, yet no other metabolite peaks have been observed beyond 9.4 ppm in the human brain. In this study, we sought to determine if there were additional metabolite peaks that exists in the DF beyond the 9.4 ppm. To maximize SNR of potentially low signal metabolites, we employed a large voxel size and varied the center frequency of the spectrally selective excitation with an aim to detect other metabolites especially with lower concentrations than NAD⁺.

Methods

All the magnetic resonance imaging (MRI) and DF ¹H MRS acquisitions were performed on healthy volunteers (n=6; 4 males and 2 females). The study protocol was explained to all the subjects, and their consent was obtained under an approved Institutional Review Board (IRB) study. The mean age of the healthy volunteers was 32.67 ± 6.86 Y (range 26–43 Y).

MRI acquisition protocol:

All the acquisitions were done on the 7.0 T MAGNETOM Terra scanner (Siemens Healthcare, Erlangen, Germany) using a single channel transmit and 32-channel receive phased array head radio frequency coil (Nova Medical, Wilmington, MA, USA). The study protocol consisted of a localizer, MPRAGE ¹³, single voxel stimulated echo sequence for reference voltage calibration followed by DF ¹H MRS acquisitions. For reference voltage calibration, a non-water suppressed STEAM sequence was used, with a range of transmit B₁ voltages. The water signal amplitudes were then fit to a sin^3(theta) curve to determine the optimal transmit reference voltage. For DF ¹H MRS acquisitions, a large voxel was placed with in the brain to cover a maximal area avoiding the sinus cavities as shown in the Fig. 1, and this volume varied from 200–350 cm³ across the individuals. For all the volunteers, a spectrally selective 90° E-BURP pulse with excitation center placed at 10.0 ppm with an excitation bandwidth (BW) of 600Hz (1ppm on either side of the excitation center) was applied for DF ¹H MRS acquisition along with a three narrow spatially selective refocusing 180° Shinnar-Le Roux (SLR) pulses for localization with a BW of 800Hz as mentioned in our previous study ³. Repetition time (TR) and time to echo (TE) were 1000 ms and 18 ms, respectively. Number of averages were 256. Total time for each down-field spectroscopy acquisition including the water reference was 7 min 5 s. Additionally, on one of the volunteers, DF ¹H MRS acquisitions were acquired with and without water saturation at a TR of 1340 ms using the same sequence with excitation center set at 10.0 ppm.

Figure 1.

Representative anatomical images from one of the healthy volunteers, showing the positioning of the voxel colored in red within the brain for the in vivo down-field proton MRS (DF ¹H MRS).

A phantom consisting of 10 mM L-TRP was prepared in phosphate buffer saline and was used for acquisition of spectroscopy experiments with the same sequence and excitation parameters as above at 7.0 T.

NMR sample preparation:

A solution of L-TRP at 50 mM concentration was prepared in a solution of phosphate buffer saline (PBS) and the pH was adjusted to 7.0. A capillary tube containing deuterium oxide (D₂O: for the lock signal) with 0.1% trimethylsilylpropanoic acid (TSP: for chemical shift referencing) was placed in a 5 mm NMR tube containing the sample. This was done to prevent any chemical exchange between the D₂O and the labile protons on tryptophan. Additionally, another phantom containing 50 mM of L-TRP was prepared in a mixture of 50%. PBS and 50% D₂O at pH 7.0. The ¹H NMR spectra were collected at 37°C on a 9.4 T vertical bore high resolution spectrometer (Varian).

The water unsuppressed ¹H NMR spectrum was acquired with 16K data points, a spectral width of 6410 Hz, 1024 averages, and a recycle delay of 1.0 s (For L-TRP phantom in 50% D₂O, the averages were 128 and recycle delay was 2.0s). The ¹H water-suppressed NMR spectrum was acquired with same parameters as above except that the averages were 512, recycle delay was 2.0 s, and saturation delay was 6.0 s. For spectral processing of in vitro data acquired at 9.4 T, automatic phase correction along with an automatic Bernstein polynomial baseline correction ¹⁴ was applied after a line-broadening of 2.0 Hz, and zero-filling to 64 K data points to all spectra using MestreNova (Mestrelabs, Spain). The water suppressed ¹H NMR spectrum was overlaid on the non-water suppressed spectrum and the signal intensity was normalized to the TSP peak and the peak of interest is labeled.

Results

A representative voxel from the brain of one of the volunteers from which the DF ¹H MRS was acquired in vivo is shown in the Fig. 1. Scan time was ~7 min including the water reference scans. Representative spectra from one of the volunteers that were scanned, when the excitation center was placed at 10.0 ppm, we detected a new resonance occurring at around 10.1 ppm (Fig. 2). This was consistently observed from the spectra of all the volunteers and their DF ¹H MRS when the excitation is centered at 10.0 ppm is shown in Supporting Information Figure S1. Another observation is the well resolved NAD⁺ peaks when the excitation frequency was centered at 10.0 ppm as shown in Fig. 2. To determine whether this new resonance at 10.1 ppm has any chemical exchange or cross-relaxation effect with water we have acquired one dataset without and with water suppression on a sixth volunteer with excitation center placed at 10.0 ppm. In this spectrum with water suppression, we found that the peak could not be detected at 10.1 ppm, implying that it has significant cross-relaxation or chemical exchange with water (Fig. 3B).

Figure 2.

Representative brain DF ¹H-MR spectra from one of the healthy volunteers in vivo with the RF excitation center placed at 10.0 ppm. Notice that the tryptophan indole (-NH) resonance is clearly seen at 10.1 ppm. Abbreviation: L-TRP, L-tryptophan; NAD⁺, nicotinamide adenosine dinucleotide.

Figure 3.

The DF ¹H-MR spectra with RF excitation center at 10.0 ppm without (A) and with (B) water suppression from the brain of one of the healthy volunteer in vivo. Notice that the tryptophan indole (–NH) resonance disappears when water suppression is applied, indicating that it is in chemical exchange or in cross-relaxation with water.

Based on an extensive search of the literature for in vivo brain metabolites, we hypothesized that the resonance observed at 10.1 ppm may be attributed to L-TRP ¹⁵. To confirm this, we performed additional experiments to determine the in vitro ¹H NMR spectra of L-TRP without and with water suppression (scaled with respect to TSP) at 37°C as shown in Fig. 4. The sidechain indole (-NH) proton peak is clearly shown resonating at 10.12 ppm and this peak was not detected when a water suppression scheme was applied, which was in close agreement with characteristics of this indole (-NH) proton of L-TRP under in vivo conditions as shown in Fig. 3.

Figure 4.

The ¹H-NMR spectra of the 50-mM L-TRP phantom, without (brown) and with (blue) water suppression. Note that the L-TRP indole (-NH) proton, as indicated with a red box, has a resonance peak at 10.1 ppm. The peak vanishes with water suppression, indicating that it is in direct exchange or in cross-relaxation with water.

To further validate the characteristics of this indole (-NH) proton we performed an additional in vitro water unsuppressed ¹H NMR spectra on phantom containing L-TRP in a mixture of 50% D₂O and 50% PBS. This peak at 10.1 ppm was intact even in 50% D₂O as shown in Supporting Information Figure S2, indicating that the disappearance of this peak when a water suppression is applied is mainly due to cross-relaxation rather than the chemical exchange with water.

In order to make sure that the observed signal is not an artifact we have performed additional in vitro experiments on a 10 mM L-TRP phantom, where water side bands are observed at 9.4 ppm and 0 ppm when frequency excitation was centered at 4.7 ppm as shown in Fig. 5A. This was not the case when frequency excitation was centered at 10.0 ppm, where the observed resonance was from indole -NH proton of L-TRP at 10.1 ppm and no signs of any water sidebands neither at 9.4 ppm nor at 0 ppm as shown in Fig. 5B. The results from the in vivo data were also identical as shown in Fig. 5C, with water side bands occurring at same chemical shifts as observed from L-TRP phantom when excitation frequency was centered at 4.7 ppm, whereas when excitation frequency was centered at 10.0 ppm, we clearly see the L-TRP resonance at 10.1 ppm and on the mirror side of water resonance at -0.7ppm we have not observed any resonance as shown in Fig. 5D indicating that the resonance we have observed at 10.1 ppm is not an artifact.

Figure 5.

The DF ¹H-MR spectra when excitation frequency was centered at 4.7 ppm (A,C) and 10.0 ppm (B,D) from in vitro (10mM L-TRP phantom) and in vivo (from one of the healthy volunteers) experiments at 7 T, respectively. Notice that the water side bands are indicated in red arrows for both the data when excitation frequency was centered at 4.7 ppm, but when centered at 10.0 ppm, only the metabolites resonance was seen. On the mirror side of water (blue arrows), no resonance was seen at −0.7 ppm, indicating that the 10.1-ppm peak we observed was not an artifact.

Discussion

This is for the first time in the down-field ¹H MRS that a new resonance at 10.1 ppm has been identified in vivo in human brain. Based on a literature review of in vivo metabolites, we suggest that the new resonance at 10.1 ppm may be attributed to L-TRP. L-TRP is an essential amino acid in the human brain with reported concentrations of 0.02 – 0.06 μmol/g wet weight of brain ¹⁶ and is also a major precursor for production of NAD⁺ through de novo synthesis pathway ^17–20. It also serves as a precursor for the production of the important brain neurotransmitter serotonin and a hormone melatonin, the latter of which is involved in regulation of circadian rhythm ^21,22. Our in vitro phantom data also supports that this new resonance could be from side chain indole (-NH) proton of L-TRP. Based on both phantom and in vivo data, we demonstrated that this indole (-NH) proton of L-TRP has significant cross-relaxation with water rather than chemical exchange.

A recent chemical exchange saturation transfer (CEST) study of in vitro phantoms of L-TRP and the downstream intermediate metabolites in serotonin synthesis showed a z-spectrum with a small reduction in the water magnetization varying from 5.11 to 5.47 ppm for these metabolites with respect to water ¹⁵. The DF absolute chemical shift resonances for the indole (-NH) proton of these metabolites reported in that study was 10.17, 10.02, 9.99 and 9.81 ppm for L-Tryptophan, 5-Hydroxytryptophan, Serotonin and 5-Hydroxyindoleaceticacid, respectively ¹⁵. Given the similarities of chemical shifts for indole (-NH) resonance from all the molecules mentioned above, it is evident that all contribute to the 10.1 ppm resonance except 5-Hydroxyindoleacetic acid. Based on this study and our in vitro findings and the in vivo concentrations of all the metabolites mentioned above, we believe that the major potential contributor to the 10.1 ppm resonance is from L-TRP. Additionally, we suggest that there may be minor contributions from the other metabolites as mentioned above involved in the serotonin biosynthetic pathway although to a lesser extent. The fact that the serotonin and its downstream metabolites contribute to less extent to 10.1 ppm resonance might also be supported by the published studies that ~90–95% of L-TRP is catabolized by de novo synthesis pathway to NAD⁺ while less than 3% of L-TRP is utilized for serotonin synthesis throughout the body with only 1% of L-TRP utilized for serotonin synthesis in brain ^23–25. Although the literature data and our phantom results suggests that L-TRP may be the major contributor to the observed peak, more work is needed to elicit other possible resonances at this chemical shift.

Further optimization of our pulse sequences and post-processing methods for fitting of this new resonance at 10.1 ppm along with the repeatability studies are needed for robust quantification. This new resonance identified as the indole (-NH) proton of essential amino acid L-Tryptophan, also the lowest concentration amino acid present in the brain could encourage to explore it as new biomarker in the study of many neuropsychiatric disorders such as anxiety ^26–28, depression ^29–31, autism ^32–34, etc., to name a few associated with serotonin receptors in the human brain.

Conclusions

To conclude, this is the first study to report the new resonance in vivo at 10.1 ppm in human brain by employing a spectral selective excitation and spatial selective localization pulse sequence. Based on our in vivo human brain and in vitro phantom data, we suggest that this new resonance at 10.1 ppm is most likely attributed to proton from indole moiety of L-Tryptophan, although the possibility of a minor contribution from downstream metabolites during serotonin synthesis cannot be ruled out.

Supplementary Material

Supporting Document

Supporting Figure S1. Representative DF ¹H-MR spectra from the rest of the four healthy volunteers brain in vivo when the RF excitation center was placed at 10.0 ppm. Notice that the tryptophan indole (-NH) resonance is clearly seen at 10.1 ppm in all the volunteers.

Supporting Figure S2. Water unsuppressed ¹H NMR spectra of the 50mM L-Tryptophan phantom without (brown) and with (blue) 50% D₂O. Note that the L-TRP indole proton as pointed with red box has resonance peak at 10.1 ppm and does not change even with the addition of 50% D₂O, indicating that it is in cross-relaxation with water rather than the chemical exchange.