In the central nervous system, myelin is a component of oligodendrocytes defined by its ultrastructure of multiple lamellae of protein-rich lipid bilayers; myelin insulates the axons of certain nerves to facilitate saltatory conduction and provides trophic support (Morell and Quarles, 1999a). The extent of myelination modulates the function and health of axons and is dynamically regulated throughout the lifespan and in response to a variety of neurologic conditions (Young et al., 2013; Duncan and Radcliff, 2016).

The development of robust and specific biomarkers of myelination would facilitate further advancements in the diagnosis, monitoring, and treatment of demyelinating diseases. For instance, there has been recent interest in the development of promoters of remyelination as complementary to the current immunosuppressive treatment of multiple sclerosis (Magalon et al., 2012; Deshmukh et al., 2013; Plemel et al., 2017; Lubetzki et al., 2020), an autoimmune demyelinating disease that most commonly affects young adults and ultimately results in progressive functional impairment, cognitive deficits, and early mortality (Lucchinetti et al., 2005; Popescu et al., 2013). An imaging-based biomarker of myelin would be crucial for the development of these potential remyelinating agents at both the pre-clinical and clinical stages.

Myelin imaging is challenging because the T₂ of myelin protons (T₂ from several μs to a few 100's of μs) is much shorter than the minimal echo times (TEs; no less than several ms) of conventional sequences (Horch et al., 2011; Wilhelm et al., 2012; Du et al., 2014a; Sheth et al., 2016). In addition, the longer T₂ water protons comprise over 90% of the MRI signal in the brain (Fan et al., 2018). We have previously shown that adiabatic inversion recovery prepared ultrashort echo time (IR-UTE) sequences with nominal TEs as short as 8 μs are able to robustly suppress long-T₂ water signals and generate a high contrast myelin image on a clinical 3T scanner (Du et al., 2014a,b). This short T₂ signal persists after removing most of the myelin-associated water fraction by D₂O exchange (Fan et al., 2017, 2018; Seifert et al., 2017), suggesting that the IR-UTE sequence is detecting signal from semisolid myelin protons. Myelin is comprised of ~40–45% lipids/cholesterol, 10–15% protein, and 40% water (Morell and Quarles, 1999b). The majority of the myelin protons detected by UTE sequences are thought to originate from the long-chain methylenes of the bilayers, with additional contributions from cholesterol, choline, and proteins (Horch et al., 2011; Wilhelm et al., 2012). The IR-UTE sequence has also been shown to detect demyelinated MS lesions as confirmed by autopsy (Sheth et al., 2016, 2017).

This study was designed to test the capability of the IR-UTE sequence in detection of demyelination using an animal model. In adult C57BL/6 mice, the copper chelator cuprizone induces demyelination with well-characterized temporal and regional dynamics, notably resulting in pronounced demyelination of the caudal corpus callosum (CC) after 5–6 weeks of exposure (Hiremath et al., 1998; Mason et al., 2001; Matsushima and Morell, 2001; Taylor et al., 2010). Additional acute changes are likely reactive to demyelination including mild edema, microgliosis, astrocytosis, and axonal injury; however there is minimal inflammation compared to models such as experimental autoimmune encephalitis (Matsushima and Morell, 2001; Gudi et al., 2009). Thus, the C57BL/6 cuprizone model is well-suited for determining the sensitivity of IR-UTE for acute demyelination.

In this study, eight 8-week-old C57BL/6 mice were fed cuprizone (n = 4) or control chow (n = 4) for 5 weeks and then imaged by 3D IR-UTE MRI and compared with conventional diffusion tensor imaging (DTI). The differences in IR-UTE signal were compared in the major white matter tracts in the brain and correlated with the Luxol Fast Blue (LFB) histochemical marker of myelin.

Representative images of the optimized IR-UTE sequence of a control mouse brain are shown in Figure 1. The IR-UTE image in Figures 1B, C demonstrates high signal intensity in the major white matter tracts, including the corpus callosum, internal capsule, dorsal and ventral hippocampal commissures, and deep cerebellar white matter; intermediate signal in mixed white and gray matter structures such as the basal ganglia and both superior and inferior colliculi; and minimal signal from cortical gray matter and cerebral spinal fluid (CSF). The high signal in the calvarium seen only on IR-UTE and not FSE images reflects ultrashort T₂ signal from cortical bone, and there is additional signal from retrobulbar fat. The later echo time of TE = 2 ms in Figure 1D has dramatically reduced signal intensities, notably with the majority of the white matter tracts now demonstrating less signal than residual CSF, indicating that the majority of the white matter IR-UTE signal has ultrashort T₂. The retrobulbar fat, bone marrow fat, and minimal CSF signal are more apparent in the TE = 2 ms image due to longer T₂s.

The IR-UTE signal visually decreased in the cuprizone-treated mice in several white matter tracts that are known to be sensitive to cuprizone-induced demyelination (Figure 2). Axial images through the ventral hippocampal commissure and splenium of the corpus callosum demonstrate the greatest decrease in signal intensity in the corpus callosum and to a lesser extent in the ventral hippocampal commissure and white matter of the basal ganglia. In contrast, there was no significant signal change in the internal capsule, which is known to be resistant to cuprizone (Yang et al., 2009). These findings are confirmed to be statistically significant with quantitative analysis using ROIs (Figure 3).

DTI imaging also was able to detect a difference between the control and cuprizone-treated mice. Demyelination is thought to be associated with a decrease in total fractional anisotropy, specifically with an increase in radial diffusivity as the loss of myelin bilayers allows water molecules to diffuse in a radial direction from the axon tracts (Xie et al., 2010). In contrast, axial diffusivity is more sensitive to axonal damage, which transiently changes during early treatment with cuprizone (Sun et al., 2006). As expected and in line with prior studies of DTI imaging of cuprizone-treated mice, a decrease in fractional anisotropy and an increase in radial diffusivity were seen in the splenium of the corpus callosum, whereas there was no statistically significant change in axial diffusivity. The IR-UTE signal correlated with both fractional anisotropy and radial diffusivity, but only had a weak correlation with axial diffusivity that was not statistically significant (Figure 4).

Both the IR-UTE and DTI imaging results are corroborated by semi-quantitative histochemical analysis of myelin using LFB staining (Figure 5). All of the cuprizone-treated mice had decreased myelin staining in the genu of the corpus callosum compared to the control mice; however complete demyelination was not achieved in this study. Additionally, T₁ mapping demonstrates that cuprizone treatment did not significantly change the T₁ of the ventral hippocampal commissure (Figure 6). Therefore, the TI_null of the long T2* components does not significantly change in this model. Otherwise, it could have resulted in artifactual signal intensity changes.

This study demonstrated that the 3D IR-UTE sequence is able to detect demyelination with decreased myelin signal in an animal model of acute demyelination. The ability of the IR-UTE sequence to detect the ultrashort T₂ signal and suppress the long T₂ signal was demonstrated in mouse brains, similar to prior work using in vitro phantoms and human volunteers (Du et al., 2014a; Sheth et al., 2017; Fan et al., 2018). The mice treated with cuprizone demonstrated loss of IR-UTE signal in white matter tracts known to be sensitive to cuprizone but no changes in white matter tracts that are known to be resistant, whereas the IR-UTE signal in white matter tracts of control mice was unchanged. The IR-UTE signal was also correlated with findings using DTI MRI. The 3D IR-UTE measurement of demyelination was further confirmed by LFB staining. These findings demonstrate that the IR-UTE signal was sensitive to myelin loss in the cuprizone mouse model.

Many of the existing methods for myelin imaging and quantification detect myelin indirectly through its interactions with myelin-associated water, and have been shown to detect demyelination in the cuprizone model and other animal models. These include magnetic transference (MT)-based imaging methods (Zaaraoui et al., 2008; Varma et al., 2015; Khodanovich et al., 2017), T₂ relaxometry/myelin water fraction imaging (Thiessen et al., 2013; Wood et al., 2016), diffusion based imaging methods (Song et al., 2005; Sun et al., 2006; Wang et al., 2011), quantitative susceptibility mapping (Wang et al., 2019), and others. However, these methods may also be sensitive to other pathologic changes including microgliosis, edema, and mild axonal injury that accompany demyelination in this model (Wood et al., 2016). Directly imaging myelin protons using IR-UTE would be expected to improve specificity for myelin loss in the setting of heterogeneous pathological changes, or may complement these other methods. The indirect measures of myelin are sensitive to B₁ and B₀ inhomogeneities and may be complicated by edema and iron deposition. Unlike IR-UTE, conventional MRI techniques also cannot measure myelin relaxation times (e.g., T1 and T2^* relaxation times), which may allow for a direct assessment of myelin quality.

Recently, another UTE-based method was found to correlate with histological markers of myelin in the cuprizone model better than the myelin water fraction and RD and similar to the MT-based macromolecular fraction (Soustelle et al., 2019). This Diff-UTE sequence uses diffusion gradients for suppression of long T₂ water signals, which allows relative preservation of the ultrashort T₂ signal at a short TR. In contrast, the IR-UTE sequence uses IR preparation for long T₂ signal suppression, allowing for recovery of the ultrashort T₂ signal using a longer TR and multispoke acquisition per IR preparation (Carl et al., 2016; Ma et al., 2020b). The IR-UTE sequence is more robust to B₁ inhomogeneity and does not require assumptions about the T₁ and T₂ of myelin for gradient tuning, which is required in Diff-UTE imaging of myelin for appropriate signal nulling of diffusive long-T2 components of the white matter (Soustelle et al., 2019); however, inversion time needs to be carefully determined for IR-UTE imaging. We have recently designed a Double-Echo Sliding Inversion Recovery Ultrashort Echo Time (DESIRE-UTE) method which allows image reconstruction at a wide range of inversion times so that the optimal inversion time does not need to be chosen prospectively (Ma et al., 2020c). Another approach is a short-TR adiabatic inversion-recovery UTE (STAIR-UTE) method which allows robust long T₂ signal suppression with optimized short TR/TI pairs (Ma et al., 2020a). In addition, the Diff-UTE study was performed on mouse brains that had been fixed, which may alter the MR properties of myelin. The specificity of the IR-UTE and Diff-UTE sequences have not yet been tested during remyelination or in other models of demyelination.

There are several limitations of this study. First, the sample size of both the treated and control mice was small, owing to the fact that this study was designed as a proof-of-concept for future studies. Additionally, complete demyelination was not achieved, which limited the ability to examine other contributors to the ultrashort T₂ IR-UTE signal such as inflammation and gliosis. A possible reason for myelin not being as low as expected could be animals not eating enough, or variable amounts, of the cuprizone-containing diet. Future studies will be needed to test the specificity of the IR-UTE signal for myelin in the setting of both demyelination and remyelination using other models of demyelination, and to investigate the advantages and disadvantages over conventional MRI techniques for myelin quantification.

In conclusion, the 3D IR-UTE method was able to robustly detect the ultrashort T₂ components in major white matter tracts of the mouse brain with decreased IR-UTE signal of myelin during cuprizone-induced demyelination as confirmed by LFB staining. Therefore, 3D IR-UTE is a promising non-invasive method for measuring brain myelin content in mouse models of demyelination.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was approved by IACUC at UCSD and at Novartis Institutes for BioMedical Research. The study was conducted in accordance with the local legislation and institutional requirements.

ACS: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing—original draft, Writing—review & editing. YM: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing—original draft, Writing—review & editing. SrihS: Resources, Supervision, Writing—review & editing. SrinS: Resources, Supervision, Writing—review & editing. RB: Data curation, Methodology, Resources, Software, Writing—review & editing. EC: Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing—review & editing. LD: Investigation, Methodology, Resources, Writing—review & editing. AMS: Investigation, Methodology, Resources, Writing—review & editing. JD: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing—original draft, Writing—review & editing.