Purpose

Hyperpolarization has enabled imaging of real-time metabolic activity, yet is limited by the rapid decay of magnetization via T₁ relaxation. While ¹³C nuclei are generally limited to T₁’s of about one minute or less, protected ¹⁵N positions are known to exhibit T₁’s as long as several minutes¹. However, ¹⁵N detection is intrinsically insensitive and longer-lived sites typically exhibit only tiny chemical-shift changes upon metabolic conversion, limiting the applicability of ¹⁵N for in vivo metabolic investigations. Thus, a potentially ideal approach is storage of hyperpolarization at ¹⁵N followed by polarization transfer to J-coupled protons for detection with enhanced sensitivity and chemical-shift separation among metabolites². Carnitine is a quaternary ammonium compound with key biochemical roles as a buffer for excess acetyl units as well as transport of long-chain fatty acids into mitochondria via CPT1. The purpose of this study was to investigate hyperpolarization and optimized ¹H-based detection of HP L-[¹⁵N]carnitine-d₉ (Fig. 1), a new metabolic imaging probe whose ultra-long T₁ (210s in H₂O / 160s in vivo) and excellent safety profile are ideally suited for in vivo studies and eventual clinical translation.

Methods

DNP: Selective trimethyl deuteration provided added protection from T₁ relaxation by ¹H-¹⁵N dipolar coupling, while preserving transfer targets in the rest of the molecule, in particular the methine resonance with ³J_NH~3.6Hz. L-[¹⁵N]carnitine-d₉ (Cambridge Isotopes) was dissolved in water/DMSO to 4M, with trityl radical OX063. Each sample was loaded into the GE SPINLab hyperpolarizer (5T/0.8K), irradiated @ 140GHz and rapidly dissolved in superheated H₂O, giving a 16mM aqueous solution of HP [15N]carnitine. HP carnitine solution was transferred to a Varian/Agilent 4.7T small animal scanner located one floor below the polarizer. MR Hardware: During sample transfer (~30s), the ambient magnetic field around the sample was controlled at ~50 Gauss using a home-built, hand-held solenoidal electromagnet (Fig. 2A). A home-built 5cm dual-tuned ¹⁵N/¹H (20.2MHz/199.3MHz) square transceiver RF surface coil was also constructed for this study (Fig. 2B). MR experiments: ¹⁵N pulses were calibrated using a 5mL 10M [¹⁵N₂]urea vial phantom. Aqueous and in vivo ¹⁵N T₁’s were estimated by fitting dynamic HP ¹⁵N data obtained by repetitive low flip angle, hard-pulse excitation (5 every 30s). For the in vivo study, a normal rat was anesthetized and injected with 2.5mL of the 16mM HP carnitine solution via tail vein, with RF coil placed anteriorly over the rat liver. For developing optimized ¹H-based detection, a vendor-supplied refocused INEPT sequence was modified with added 10:1 coherence selection gradients for specifically detecting polarization transferred from ¹⁵N to ¹H (Fig. 3), and tested on a syringe containing 3mL of 16mM HP carnitine. In anticipation of ¹H-detected in vivo metabolic studies, the ¹H chemical shift between carnitine and acetylcarnitine at the methine position was measured by ¹H MR of a reference solution and compared against spectral databases.

Results and Discussion

Estimated T₁’s determined from fitting dynamic low flip angle HP [¹⁵N]carnitine data (Fig. 4A) acquired at 4.7T were 210s (in H₂O) and 160s (in vivo). The DNP enhancement factor was estimated to be ~57,000 at 4.7T, based on comparison with [¹⁵N₂]urea phantom data, corresponding to a polarization level of 11%. Remarkably, the [¹⁵N]carnitine coherence was directly detectable in rat liver for over 5 minutes (Fig. 4B), despite the relatively low dosage (~7mg carnitine injected) and the intrinsically low sensitivity of ¹⁵N. Injection was tolerated well, consistent with the excellent safety profile of L-carnitine. Applying the new INEPT sequence with gradient-based coherence selection eliminating water and any other ¹H resonances, ¹⁵N hyperpolarization was successfully transferred to protons in carnitine (Fig. 5), including some inadvertent transfer to both sets of methylene protons. The ¹H chemical-shift difference between carnitine and acetylcarnitine in an aqueous reference solution was measured to be 1.0ppm (which can readily be resolved in proton spectroscopic imaging), in good agreement with spectral databases. Large quantities of acylcarnitines, especially acetylcarnitine, are known to be formed within minutes after intravenous injection of carnitine at similar dose levels used for hyperpolarized MRI³, indicating strong potential for metabolic studies using this new probe. In comparison with the gold standard MR metabolic imaging probe [1-¹³C]pyruvate, the T₁ of [¹⁵N]carnitine is 3x longer in H₂O and ~4-5x longer in vivo. A limitation of this study that will be addressed soon was the use of surface coils, as uniform B₁ transmission is highly desirable for maximally efficient polarization transfer experiments.

Conclusion

We have achieved hyperpolarization and ¹⁵N->¹H hyperpolarization transfer for improved detection of HP L-[¹⁵N]carnitine-d₉, a new MR metabolic imaging probe with ultra-long T₁.

Acknowledgements

The authors are grateful to JJH Ackerman for his valued consultation and enthusiastic encouragement, as well as the support of the Mallinckrodt Institute of Radiology. Grants: NIH S10OD023580, K01DK099451, R01DK115987.