In vivo MRI is sensitive to remyelination in a nonhuman primate model of multiple sclerosis

  1. Maxime Donadieu
  2. Nathanael J Lee
  3. María I Gaitán
  4. Seung-Kwon Ha
  5. Nicholas J Luciano
  6. Snehashis Roy
  7. Benjamin Ineichen
  8. Emily C Leibovitch
  9. Cecil C Yen
  10. Dzung L Pham
  11. Afonso C Silva
  12. Mac Johnson
  13. Steve Jacobson
  14. Pascal Sati  Is a corresponding author
  15. Daniel S Reich  Is a corresponding author
  1. Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
  2. Department of Neurology and Neurological Sciences, Stanford University, United States
  3. Department of Neurobiology, University of Pittsburgh, United States
  4. Section on Neural Function, National Institute of Mental Health, National Institutes of Health, United States
  5. Department of Neuroradiology, Clinical Neuroscience Center, University Hospital Zurich, University of Zurich, Switzerland
  6. Viral Immunology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
  7. Cerebral Microcirculation Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
  8. Department of Radiology and Radiological Sciences, Uniformed Services University of the Health Sciences, United States
  9. Vertex Pharmaceuticals Incorporated, United States
  10. Neuroimaging Program, Department of Neurology, Cedars Sinai, United States

Abstract

Remyelination is crucial to recover from inflammatory demyelination in multiple sclerosis (MS). Investigating remyelination in vivo using magnetic resonance imaging (MRI) is difficult in MS, where collecting serial short-interval scans is challenging. Using experimental autoimmune encephalomyelitis (EAE) in common marmosets, a model of MS that recapitulates focal cerebral inflammatory demyelinating lesions, we investigated whether MRI is sensitive to, and can characterize, remyelination. In six animals followed with multisequence 7 T MRI, 31 focal lesions, predicted to be demyelinated or remyelinated based on signal intensity on proton density-weighted images, were subsequently assessed with histopathology. Remyelination occurred in four of six marmosets and 45% of lesions. Radiological-pathological comparison showed that MRI had high statistical sensitivity (100%) and specificity (90%) for detecting remyelination. This study demonstrates the prevalence of spontaneous remyelination in marmoset EAE and the ability of in vivo MRI to detect it, with implications for preclinical testing of pro-remyelinating agents.

Editor's evaluation

This important study combined MRI(magnetic resonance imaging) imaging and histopathology to examine the remyelination of brain lesions in an EAE marmoset model of multiple sclerosis. This work addresses in a non-human primate a missing link in the neuropathology of myelin repair, because in human MS it is virtually impossible to study the lesion dynamics by MRI (in live patients) and demyelination by histology (upon brain autopsy). The data presented are solid although the conclusions would have been stronger with further histological evidence of remyelination.

https://doi.org/10.7554/eLife.73786.sa0

Introduction

Multiple sclerosis (MS) is a debilitating inflammatory demyelinating disorder affecting millions worldwide (Reich et al., 2018). MS causes dynamic changes to myelin in the central nervous system, including the quintessential focal inflammatory destruction of myelin, as well as the phenomenon of remyelination that can follow the demyelination (Lubetzki et al., 2020; Brown et al., 2014; Chari, 2007; Lassmann et al., 1997). Demyelinated axons are susceptible to persistent damage and neurological dysfunction in MS; therefore, remyelination is a crucial aspect of tissue repair, and as such represents an important therapeutic target (Kremer et al., 2019). Most knowledge about remyelination in MS derives from postmortem studies using histochemistry and electron microscopy. This is because investigating remyelination in vivo in real time is limited by imperfect discrimination on neuroimaging modalities such as magnetic resonance imaging (MRI). Furthermore, in human beings, where collecting serial short-interval scans is challenging, it is difficult to detect and track the dynamic occurrence of remyelination. Therefore, to investigate the pathobiology of remyelination in the context of focal inflammatory demyelination, a reliable preclinical model is needed to develop techniques that can then be applied clinically.

Rodent models have been widely used to investigate various aspects of the pathobiology of demyelination. However, while toxin models in mice, including the lysolecithin and cuprizone models, display demyelination and even remyelination, they do not recapitulate the recruitment of peripherally derived adaptive immune cells that occurs in MS, which can potentially confound the MRI signal (Kipp et al., 2012; Kipp et al., 2017). Conversely, rodent experimental autoimmune encephalomyelitis (EAE) models, though driven by an immune response, are often neither focally nor profoundly demyelinating and mostly impair the spinal cord. There is no known rodent model that is characterized by multifocal inflammatory demyelination in the brain that resembles MS and is disseminated in both space and time.

However, EAE in the common marmoset (Callithrix jacchus) is a well-recognized translational model that serves as a bridge between the rodent EAE and human MS (Jagessar et al., 2016; Kap et al., 2010; Kap et al., 2016). Not only does EAE resemble MS white matter lesions (WML) both radiologically and pathologically (Sati et al., 2012; Maggi et al., 2017), but marmoset WML spontaneously remyelinate (Lee et al., 2019), as occurs in MS.

Prior studies demonstrated that certain signal changes in MRI, such as magnetization transfer ratio (MTR), correlate with demyelination and remyelination in MS lesions (Chen et al., 2008; Chen et al., 2013; Filippi et al., 2012; Absinta et al., 2016; Laule and Moore, 2018). This has also been investigated in animal models, albeit mainly in rodent cortex (Yano et al., 2018; Schmierer et al., 2010) rather than white matter. It has also been demonstrated that partial remyelination can occur and can be localized either to specific parts of the lesion (most commonly the lesion edge) or to the whole lesion (Filippi et al., 2019; Patrikios et al., 2006).

Here, we studied focal WML in marmoset EAE. We utilized serial in vivo MRI, mainly involving proton density-weighted (PDw), T1-weighted (T1w) before and after intravenous injection of a gadolinium chelate, and MTR sequences, to age and characterize lesions and predict remyelination. We further analyzed WML histopathology, focusing on myelin lipids and proteins as well as mature oligodendrocytes and their precursors, to compare and study the reliability of using various in vivo MRI sequences to predict remyelination in the context of recovery from inflammatory demyelination.

Results

Lesion characterization and categorization on MRI

EAE was induced in six marmosets as described, and the animals followed by MRI at regular intervals until sacrifice due to clinical progression. Lesions were grouped into three categories. Those categorized as ‘early active’ at histopathology typically remained hyperintense on PDw images until the terminal scan and grew rapidly to several cubic millimeters, subsequently showing minimal lesion volume change over time (example in Figure 1A–B) and were enhancing on terminal T1w MRI with gadolinium contrast. Lesions classified as ‘chronic, at least partially demyelinated,’ demonstrated areas of PDw hyperintensity that persisted until the terminal scan despite resolution of early gadolinium enhancement on T1w images (Figure 2A–B). Lesions classified as ‘remyelinated’ were initially hyperintense on PDw images, with subsequent return toward isointensity over time (Figure 3A–B). None of the lesions returning to isointensity on PDw images demonstrated gadolinium enhancement on T1w images at terminal MRI.

Example of an early active demyelinating EAE lesion with persistent hyperintensity on PDw MRI and active inflammatory demyelination on histopathology.

(A) In vivo PDw MRI acquired before EAE induction (baseline) and at various timepoints leading up to the terminal scan. Images were processed as described in Methods. Red arrows: focal white matter lesion first detected 16 weeks after immunization, which persisted through the terminal MRI 3 weeks later. Red boxes: location of magnified insets. (B) Temporal evolution of volume (blue line) and normalized PDw signal intensity (orange line) of the segmented lesion. (C) Histochemical panel magnification of the same lesion, demonstrating inflammation (Iba1+microglia/macrophage infiltration) and demyelination (loss of normal PLP staining). (D) Higher magnification images from the center of the lesion (red boxes in C) showing increased cellularity, loss of myelin lipid (LFB), partial loss of oligodendrocytes and their precursors (ASPA/Olig2), partial loss of axons (Biel), and edema (increased intercellular spaces). Scale bars = 200 µm. Hematoxylin counterstaining used for PLP and Iba1. Lesion selected from M#6. Abbreviations: EAE, experimental autoimmune encephalomyelitis; PDw, proton density-weighted; MRI, magnetic resonance imaging; PLP, proteolipid protein; LFB-PAS, Luxol fast blue–periodic acid Schiff; Biel, Bielschowsky; ASPA, aspartoacylase.

Example of a chronic, at least partially demyelinated EAE lesion core with long-lasting hyperintensity on serial PDw MRI and complete loss of myelin on histopathology.

(A) In vivo PDw MRI acquired before EAE induction (baseline) and at various timepoints leading up to the terminal scan. Images were processed as described in Methods. Red arrows: focal white matter lesion first detected 23 weeks after immunization, which persisted through the terminal MRI 32 weeks later. Red boxes: location of magnified insets. (B) Temporal evolution of volume (blue line) and normalized PDw signal intensity (orange line) of the segmented lesion. (C) Histochemical panel magnification of the same lesion, demonstrating mild inflammation (Iba1+ microglia/macrophage infiltration) as well as demyelination (loss of normal PLP staining). (D) Higher magnification images from the center of the lesion (red boxes in C) showing minimally increased cellularity, partial loss of myelin lipid (LFB), loss of oligodendrocytes and their precursors (ASPA/Olig2), and loss of axons (Biel). Scale bars = 200 µm. Hematoxylin counterstaining used for PLP and Iba1. Lesion selected from M#3. Abbreviations: EAE, experimental autoimmune encephalomyelitis; PDw, proton density-weighted; MRI, magnetic resonance imaging; PLP, proteolipid protein; LFB-PAS, Luxol fast blue–periodic acid Schiff; Biel, Bielschowsky; ASPA, aspartoacylase.

Example of a nearly complete remyelinated EAE lesion with initial hyperintensity that returned to isointensity on serial PDw MRI.

(A) In vivo PDw MRI acquired before EAE induction (baseline) and at various timepoints leading up to the terminal scan. Images were processed as described in Methods. Red arrows: focal white matter lesion first detected 27 weeks after immunization, which largely resolved on MRI and could not be reliably segmented on the terminal MRI 32 weeks later. Red boxes: location of magnified insets. (B) Temporal evolution of volume (blue line) and normalized PDw signal intensity (orange line) of the segmented lesion. (C) Histochemical panel magnification of the same lesion, demonstrating pale myelin lipid staining (LFB) and near-normal levels of myelin protein (PLP). (D) Higher magnification images from the center of the lesion (red boxes in C) showing minimal inflammation (Iba1), presence of oligodendrocytes and their precursors (ASPA/Olig2), and partial preservation of axons (Biel). Scale bars = 200 µm. Hematoxylin counterstaining used for PLP and Iba1. Lesion selected from M#3. Abbreviations: EAE, experimental autoimmune encephalomyelitis; PDw, proton density-weighted; MRI, magnetic resonance imaging; PLP, proteolipid protein; LFB-PAS, Luxol fast blue–periodic acid Schiff; Biel, Bielschowsky; ASPA, aspartoacylase.

In vivo PDw MRI is sensitive to lesion remyelination

Using in vivo MRI only, 40 focal WML were detected in the 6 EAE marmosets (Table 1). Interrater reliability for PDw MRI classification was 94%, with Cohen’s kappa of 0.89, and consensus was achieved across the raters for all lesions. Of the 40 lesions, 12 (30%) were classified as predicted early active, 10 (25%) as predicted chronic at least partially demyelinated, and 18 (45%) as predicted remyelinated. Four of the six animals demonstrated predicted remyelinated lesions (M#1–4), whereas the remaining two only had predicted acute demyelinating and chronic, at least partially demyelinated lesions (M#5–6).

Table 1
Classification of focal lesions by proton density-weighted magnetic resonance imaging.
AnimalPredicted early activePredicted chronic, at least partially demyelinatedPredicted remyelinatedTotal
M#11269
M#21168
M#31146
M#40123
M#53306
M#66208
Total12101840

Based on the histological classification criteria, 31 lesions were identified in the 5 animals with postmortem tissue (M#1 was not included in the histological analysis). Twelve of the 31 lesions (39%) were classified as early active, 10 (32%) as chronic at least partially demyelinated, and 9 (29%) as remyelinated (Table 2). All three animals with predicted remyelinated lesions on MRI (M#2–4) had remyelinated lesions on histology. M#5 and M#6 only harbored early active and chronic, at least partially demyelinated lesions, consistent with MRI findings. Two lesions in M#2 and one in M#4 were identified as predicted remyelinated on MRI but chronic, at least partially demyelinated on histology.

Table 2
Classification of focal lesions by histopathology.
AnimalEarly activeChronic, at least partially demyelinatedRemyelinatedTotal
M#21348
M#31146
M#40213
M#53306
M#67108
Total1210931

Of the 31 focal WML identified on both in vivo PDw MRI and histology, classification was concordant for 27 WML (87%). When lesions were grouped by myelination status only (i.e. early active or chronic, at least partially demyelinated vs. remyelinated), in vivo PDw MRI predicted 19 demyelinated lesions and 12 remyelinated lesions, whereas histology showed 22 demyelinated and 9 remyelinated lesions (Table 3). Relative to histology, PDw MRI prediction was therefore 100% sensitive and 90% specific for remyelination.

Table 3
Confusion matrix for PDw MRI prediction of demyelinated vs. remyelinated lesions compared to histology.
n=31Predicted demyelination by MRIPredicted remyelination by MRI
Actual demyelination by histology19322
Actual remyelination by histology099
1912

MTR is less sensitive to lesion remyelination than PDw

MTR was also used to classify lesions using a similar rater-based analysis to that applied to the PDw images. The sensitivity and specificity for predicting remyelination, relative to histology, were 82% and 79%, respectively.

T1w gadolinium enhancement as a marker of acute inflammation

Across the six animals scanned longitudinally, 82% of the lesions newly detected on PDw MRI presented T1w gadolinium enhancement. Enhancement was occasionally seen at the following timepoint (10–15 days after first detection). No chronic, at least partially demyelinated or remyelinated WML presented gadolinium enhancement on the terminal scan. M#5–6 presented at least one early active WML enhancing lesion at their terminal scan.

Lesion remyelination occurs over a 4- to 9-week period and is sensitive to lesion size

Based on longitudinal MRI imaging of EAE lesions and analysis of normalized PDw signal intensity, with comparison to histology, we found that inflammation and demyelination were dominant in lesions younger than 10 weeks of age, corroborating previous work (Lee et al., 2019; Lee et al., 2018). In four representative lesions that remyelinated, PDw signal intensity stabilized near baseline between 4 and 9 weeks after initial lesion detection on MRI (Figure 4). Lesions larger than 0.5 µL at peak, as measured on PDw MRI, did not return to isointensity and correspondingly remained at least partially demyelinated on histology. On the other hand, most lesions smaller than 0.5 µL returned to isointensity and appeared remyelinated on histology.

Evolution of in vivo PDw MRI signal intensity of remyelinating EAE lesions shows a typical 4- to 9-week time course of remyelination.

Vertical axis: mean PDw signal intensity relative to gray matter. Horizontal axis: weeks post-immunization. Blue line corresponds to mean normalized PDw signal intensity (with standard deviation) of the segmented lesion, relative to gray matter signal intensity, quantified in a region of interest drawn manually and located in the normal appearing white matter area before the lesion appeared and kept constant over time. Normal white matter displays an average normalized signal intensity of 0.65–0.75. Vertical red arrows indicate EAE immunization. Vertical red bars indicate days when corticosteroid treatments were administered (M#1 was treated twice). Horizontal red double arrows indicate the estimated period of demyelination. Horizontal green double arrows indicate the estimated period of remyelination based on the downward slope of intensity measurement, followed by plateauing of signal intensity drop. Green titles indicate that the lesion subtype was confirmed with histopathology. M# corresponds to animal number in Table 4. Abbreviations: PDw, proton density-weighted; EAE, experimental autoimmune encephalomyelitis; MRI, magnetic resonance imaging; M, marmoset; wks, weeks.

Table 4
Demographics and experimental information for the six marmosets included in this study.

Immunizations used human white matter homogenate. Experiment duration corresponds to the time between immunization and terminal MRI.

AnimalSexAge (years)First immunizationSecond immunizationCorticosteroid treatmentExperiment duration (weeks)
M#1*Male3.7100 mg200 mgYes40
M#2*Male3.7100 mg200 mgNo54
M#3Male6.2200 mgYes61
M#4Male6.2200 mgNo60
M#5Female2.8200 mgYes18
M#6Female2.8200 mgNo16
  1. Denote the three different pairs of twin animals.

Remyelination is independent of corticosteroid administration

Per protocol, three of the six marmosets were given corticosteroids for 5 consecutive days to determine whether this treatment might alter lesion fate (principally remyelination). However, we found no differences in the prevalence of predicted remyelinated lesions based on corticosteroid treatment status. On serial PDw MRI, 10 of 21 lesions (48%) were predicted to be remyelinated in corticosteroid-treated marmosets, compared to 8 of 19 (42%) in untreated animals. Based on histological analysis in animals with available tissue, 4 of 12 lesions (33%) in corticosteroid-treated marmosets were remyelinated, compared to 5 of 19 (27%) lesions in untreated marmosets. The point biserial correlation model analysis showed that steroid administration had no significant correlation with remyelination detected on MRI (p=0.8). The average experimental duration also did not differ (40 weeks in treated and 43 weeks in untreated marmosets; p=0.85). Histological analyses also did not reveal differences between treated and untreated lesions.

Histological quantification recapitulates MRI rater analysis of lesion myelin status

Assessment of proteolipid protein (PLP) staining in 31 lesions and 10 normal appearing white matter (NAWM) areas (1500 µm2 centered over the lesion core) demonstrated larger unstained areas in early active (58 ± 25%) and chronic, at least partially demyelinated (38 ± 25%) lesions compared to remyelinated lesions (4.5 ± 1.1%) or NAWM (2.6 ± 0.3%) (Figure 5A). LFB assessment showed similar results: 63 ± 26% unstained area in early active lesions, 43 ± 26% in chronic, at least partially demyelinated lesions, 15 ± 25% in remyelinated lesions, and 3.0 ± 0.4% in NAWM (Figure 5B). We observed a significantly smaller unstained PLP area in NAWM compared to remyelinated lesions (two-sample t-test, p<0.001). There were no apparent differences between PLP and LFB staining in the different lesion categories and in NAWM (two-sample t-test). Interestingly, one lesion appeared remyelinated on PLP, with less than 6% of unstained area, but demyelinated on LFB (82% of unstained area).

Histological quantification of myelin highlights different patterns across lesion categories and NAWM.

Cells were quantified in a region of interest of 1500 µm2 centered on each lesion core. Percentage of demyelination quantified for each lesion category and NAWM on PLP (A) and LFB (B). Number of ASPA and Olig2 double-positive oligodendrocytes (C) and ASPA negative, Olig2-positive OPC (D) in different lesion categories and NAWM. Data in A and B highlight more demyelination in acute and chronic lesions compared to remyelinated lesions or NAWM. Data in C and D highlight more mature oligodendrocytes in remyelinated lesions and NAWM compared to acute and chronic lesions, as well as more OPC in acute lesions compared to chronic lesions, remyelinated lesions, or NAWM. (E) Bubble plot of the 31 lesions and NAWM investigated with histopathology. Vertical axis: counts of ASPA- Olig2+ OPC. Horizontal axis: counts of ASPA+ Olig2+ oligodendrocytes. Bubble color: red; acute lesions, yellow; chronic lesions, green; remyelinated lesions, purple; NAWM. Bubble size: volume on PDw at the terminal scan in mm3. ANOVA: *p<0.05, **p<0.005. Abbreviations: NAWM, normal appearing white matter; PLP, proteolipid protein; LFB, Luxol fast blue; ASPA; aspartoacylase; ANOVA, analysis of variance; OPC, oligodendrocyte progenitor cells; PDw, proton density-weighted.

Oligodendrocyte and OPC counts are consistent with degree of demyelination in lesions

Quantitative assessment of ASPA+/Olig2+ (mature oligodendrocytes) and ASPA-/Olig2+ (oligodendrocyte precursor cell [OPC]) across the 31 lesions and 10 NAWM areas showed, as expected, more mature oligodendrocytes in remyelinated lesions and NAWM than early active or chronic, at least partially demyelinated lesions (Figure 5C). Consistent with PLP observations, we observed significantly more oligodendrocytes in NAWM compared to remyelinated lesions (two-sample t-test, p<0.001). Interestingly, more OPC were found in early active demyelinating lesions than in remyelinated lesions or NAWM (Figure 5D). Younger lesions (<10 weeks of age by MRI) had more OPC than older lesions, highlighted by a negative correlation between lesion age and OPC count (r=–0.46; p=0.009).

Discussion

In this study, we determined whether high-resolution, serial, in vivo conventional PDw MRI can effectively predict remyelination status in focal WML of marmoset EAE, a relatively faithful MS model. We found that remyelination is relatively common in this model: four of the six marmosets studied had remyelinated lesions on either MRI or histopathology, and nearly half (18) of the 40 lesions identified on MRI were predicted to be remyelinated. In people, WML repair is an age-dependent process that is variable across patients and clinical disease classification (Chari, 2007; Lassmann et al., 1997; Patrikios et al., 2006). Interestingly, the two animals that did not show remyelination by in vivo MRI, M#5 and M#6, had the shortest experimental duration (16–18 weeks, compared to 40–61 weeks for the other four animals), suggesting more aggressive lesions and/or insufficient time for remyelination.

Our characterization of remyelinated lesions showed low residual levels of inflammation (microglia/macrophages) and an extensive but incomplete (using NAWM as reference) repopulation of mature oligodendrocytes, specifically a higher number of ASPA+/Olig2+ cells in the lesion core compared to chronic, at least partially demyelinated lesions. Lower percentage of unstained area on LFB and PLP slides (<6%), consistent with data from MS (Chari, 2007; Lucchinetti et al., 2000; Lucchinetti et al., 1999; Patani et al., 2007), was also observed in remyelinated lesions. Remyelinated lesions were nearly isointense to healthy white matter at the terminal in vivo PDw MRI timepoint and showed no enhancement on post-gadolinium T1w images.

Interestingly, one lesion appeared remyelinated on PLP with less than 6% of unstained area but demyelinated on LFB (82% of unstained area) and hyperintense on PDw MRI. This result suggests potentially distinct timelines for reconstitution of myelin lipids vs. protein in the context of remyelination (Chari, 2007). As previously described, MRI seems to be more sensitive to content of myelin lipid than of myelin protein (Leuze et al., 2017).

As expected, assessment of early active and chronic, at least partially demyelinated lesions showed fewer mature oligodendrocytes in the lesion core and edge as well as a percentage of demyelinated area >20% on LFB and PLP staining. Both early active and chronic, at least partially demyelinated were hyperintense on terminal in vivo PDw MRI. Taken together, these data suggest that in marmoset EAE, lesions that initially present as hyperintense foci on PDw MRI and subsequently return toward isointensity over time are undergoing remyelination. Our data further suggest that signal intensity time courses, as well as lesion size, can be used to infer lesion status in the setting of early lesion development and recovery, despite the fact that PDw MRI is also known to be sensitive to tissue-level factors other than myelin.

The time course of marmoset EAE lesion development and repair is relatively stereotyped, as confirmed here. Previous work from our group has shown that once lesions form, the period of demyelination lasts around 6 weeks (Lee et al., 2019), consistent with data from this study (4–7 weeks; see Figure 4). The time course of marmoset remyelination is somewhat slower than toxin-induced demyelinating models in rodents, further raising potential relevance of the marmoset EAE model for preclinical testing of putative remyelination therapies in MS (McMurran et al., 2019). Our data also suggest that, during the lifespan of a lesion, if its size exceeds ~0.5 µL, the lesion will most likely not start the repair process and remain chronically demyelinated.

Of the 12 early active lesions, 7 showed an increased number of ASPA-/Olig2+ cells in the lesion vicinity (Figure 5). This pattern, never observed in remyelinated lesions or NAWM, suggests a possible proliferative OPC response, which may indicate attempted repair or potentially an inflammatory role of OPC in this setting (Harrington et al., 2020).

In this study, we compared PDw and MTR MRI, since MTR is widely used to assay remyelination in vivo in MS (Chen et al., 2008; Chen et al., 2013; Mallik et al., 2014), and found that in the context of marmoset EAE, PDw, in close connection with histopathology, appears to have higher sensitivity and specificity for remyelination status. MTR may be low in remyelinated lesions compared to healthy white matter because of the presence of incomplete or morphologically altered myelin sheaths (Brown et al., 2014; Mallik et al., 2014). Unlike MTR, which suffers from signal-to-noise reduction due to its calculation as a voxel-wise division of signal intensity measures, PDw signal intensity is directly measured by the MRI system, and thus PDw may prove more reliable for simply discriminating the presence or absence of remyelination and for characterizing its time course.

To determine whether corticosteroid treatment alters the course of lesion repair, half of the marmosets (for each twin pair, the one that showed lesions first) received 5-day treatment courses. We found no differences by corticosteroid treatment status in prevalence or time course of remyelination on either MRI or histological quantification. However, it remains possible that optimal timing of corticosteroid treatment might influence remyelination. For example, it is possible that early initiation of corticosteroid treatment (as soon as the first lesion was detected) in this study, resulting in treatment completion before lesions were even 1 week old. This might have been too early to influence lesion outcome, as the initial demyelination period typically lasts 4–7 weeks before remyelination ensues. It is also possible that reduction of inflammation via corticosteroid treatment could have hampered more rapid remyelination by slowing clearance of myelin debris, which is a prerequisite for OPC recruitment (Cunha et al., 2020) and differentiation (Gruchot et al., 2019; Rawji et al., 2016).

The main limitation of the study is its small sample size. This limitation is partially compensated by our focus on individual lesions, rather than the number of animals. Another limitation is that different EAE immunization protocols were applied across marmosets, though we did not observe any difference in lesion evolution or outcome either radiologically or histologically. Notably, previous studies from our group with a similar variety of EAE induction methods have also not shown consistent differences in disease course or lesion pathobiology (Lee et al., 2019; Lee et al., 2018). Finally, we did not obtain ultrathin sections to quantify myelin thickness, as we were interested in performing a battery of stains on our tissue, and as lesions were found at a variety of orientations relative to axons, which would complicate such quantification.

In conclusion, in vivo longitudinal PDw MRI can effectively predict remyelination in the context of marmoset EAE, with high sensitivity and specificity relative to histology. Given strong similarities between marmoset EAE and MS with respect to lesion development and repair, these results suggest a paradigm for preclinical – and possibly early-phase clinical – studies to investigate putative remyelinating therapies.

Methods

Marmoset EAE induction

Six marmosets (three pairs of twins; four males and two females, ages 2–6 at baseline; same group used for two different studies) were included in the study (Table 4). As a pilot study, two marmosets (M#1–2) first received 100 mg of human white matter homogenate followed by an additional 200 mg after no lesions were detected by in vivo MRI 2 months after the initial injection (Lee et al., 2019; Lee et al., 2018). Following protocol revision to boost the chance of early disease induction, four additional marmosets (M#3–6) received 200 mg of temporal-lobe white matter homogenate collected at autopsy. All white matter homogenates were mixed with complete Freund’s adjuvant (Difco Laboratories). Data from the two induction protocols were combined for this study, as prior marmoset EAE studies have not revealed lesion-level pathology differences (’t Hart and Massacesi, 2009), and marmosets are a scarce nonhuman primate animal resource for which data use should be maximized.

In each twin pair, per protocol, the first animal to develop a lesion, as detected by in vivo MRI, received intravenous methylprednisolone (18 mg/kg/day for 5 consecutive days), the goal of which was to reduce severity of inflammation, potentially allowing longer-term evaluation of the lesions, and to detect a potential effect of corticosteroids on lesion repair. The specific regimen was based on treatment of acute MS relapses (Goodin, 2014; Sellebjerg et al., 2005), scaled to the marmoset setting. Per protocol, experiments were terminated when animals either became paraplegic or lost more than 20% of their baseline body weight.

Marmosets were usually housed with their twin counterpart, to maximize social interactions and enhance psychosocial wellbeing. Animals were weighed and monitored daily to ensure adequate nutritional intake and physical wellbeing. All protocols were approved by the National Institutes of Neurological Disorders and Stroke (NINDS) Institutional Animal Care and Use Committee (IACUC). Specifically, the neuroethics committee of the NINDS was consulted, and formally went through the protocol prior to submission on crucial topics including minimization of pain, justification of number of animals and the sex ratio, and dosing of methylprednisolone based on human data (IACUC protocol number #1308).

Marmoset in vivo MRI

All six marmosets were scanned weekly under anesthesia, as previously described (Sati et al., 2012; Lee et al., 2019; Lee et al., 2018). We used a PDw sequence, which is sensitive to demyelination (Reich, 2017), to visualize lesions in vivo. T1w and MTR images were also obtained (Figure 6). T1w scans were repeated after injection of gadolinium-based contrast material (gadobutrol, 0.3 mmol/kg; triple the dose typically used in human clinical practice) to visualize enhancing lesions. Specific parameters for the different MRI sequences are listed in Table 5.

Figure 6 with 2 supplements see all
In vivo (baseline) multicontrast marmoset brain magnetic resonance imaging (MRI).

The T1 subtraction image was obtained by voxelwise subtraction of pre-gadolinium from post-gadolinium T1-weighted (T1w) images. The magnetization transfer ratio (MTR) image was derived voxelwise as (M0 – MSAT)/M0. Animal M#5.

Table 5
Main parameters used for the different MRI contrasts acquired in vivo.
MRI contrastPDwT1wT2wT2*wMTR*
Sequence2D RARE2D MDEFT2D RARE2D MGE3D MGE
FOV (mm)32×2432×2432×2432×2438.4×38.4
Matrix214×160214×160214×160214×160256×256
Number of slices3636363636
Slice thickness (mm)11111
TR (ms)230012.58000215020
TE (ms)164.272185
TI (ms)N/A1200N/AN/AN/A
ETL1N/A4N/AN/A
Excitation pulse (shape, FA)Sinc3, 90°Sinc3, 12°Sinc3, 90°Sinc3, 70°Sinc3, 10°
Refocusing pulse (shape, FA)Sinc3, 180°N/ASinc3, 180°N/AN/A
Preparation pulse (type, shape, FA, offset)N/AExcitation/inversion, Sech, 90°/180°N/AN/AMT, Gauss, 1500 Hz
NEX11212
AT7 min 40 s6 min 56 s13 min 20 s7 min 10 s7 min 58 s
  1. FOV = field of view; TR = repetition tine; TE = echo time; TI = inversion time; ETL = echo train length (or RARE factor); FA = flip angle; NEX = number of repetitions; AT = acquisition time; RARE = rapid acquisition with relaxation and enhancement; MDEFT = modified drive equilibrium Fourier transform; MGE = multi-gradient echo.

  2. *

    Sequence was performed twice: with (MSAT) and without (M0) the MT pre-pulse.

To minimize harm or pain in animals, all procedures, including intravenous access for gadolinium contrast material injection, were done under anesthesia. For post-anesthesia recovery, animals were gently woken with warm blankets and returned to their housing only after the animals were back to pre-anesthesia baseline, including spontaneous breathing, physical activity, and interactivity.

In vivo MRI analysis of EAE lesions

Images were postprocessed using an in-house pipeline, which included an N3 intensity correction, image cropping, a multicontrast registration aiming to align every set of images on PDw sequence at the resolution of 150 µm in-plane (1 mm thickness), skull-stripping, and intensity normalization to gray matter signal intensity, which was set to 1, over the whole set of images (Figure 6—figure supplement 1).

Focal marmoset EAE WML were detected on PDw images using an automated convolutional neural network-based segmentation algorithm (Roy et al., 2018b; Roy et al., 2018a), and segmentations were subsequently verified and edited as needed. The network was trained using PDw MRI images from three EAE marmosets scanned in a previous study along with binary manual segmentation of lesions. An atlas consists of a pair of postprocessed PDw images (baseline and a timepoint) and the binary lesion segmentation of that timepoint. The baseline is assumed and was verified to be lesion-free. Once the network was trained, it was applied to the serial PDw images collected on all six animals (Figure 6—figure supplement 2), and lesion masks were automatically generated. Any lesion smaller than 4 voxels (0.0225 μL) was considered below the threshold and was removed from analysis.

For temporal progression computation, a lesion at timepoint t=ti was identified as the same lesion at timepoint t=ti+1 if they overlapped by at least 4 voxels in 3D. All automated lesion segmentations were reviewed by an experienced rater. For subsequent MRI analysis of lesion trajectories, including intensity changes over time, all lesions were identified on every scan of each marmoset, and the average intensities were calculated using the automatically segmented lesions. For prelesion timepoints, a region of interest (ROI) of the size of the maximum lesion volume was drawn manually and centered over the white matter area where the lesion later appeared. When a lesion disappeared during the time course of the disease, an ROI of the maximum lesion size was propagated until the terminal scan.

MRI WML categorization

MRI WML categorization and prediction were set based on our prior experience and performed independently by two experienced raters, blinded to histology. WML were classified according to the terminal MRI as follows: 1 – predicted early active, described as hyperintense on PDw and enhancing on T1w scans after injection of gadolinium; 2 – predicted chronic, at least partially demyelinated, described as hyperintense on PDw and not enhancing on T1w scans after injection of gadolinium; and 3 – predicted remyelinated, described as initially hyperintense on PDw, isointense on PDw at the terminal scan, and not enhancing on T1w scans after injection of gadolinium.

Brain tissue preparation

Marmoset brains were collected immediately after death once the animals met the study endpoint. Brain tissue was processed using formalin fixation, paraffin embedding, and subsequent histopathological staining, as described previously (Lee et al., 2019; Lee et al., 2018; Luciano et al., 2016; Absinta et al., 2014). Briefly, an ultrahigh-resolution, ex vivo, 3D MRI of extracted brains was used to create individualized brain cradles with a 3D printer, which was in turn used to guide cutting of the brains into 2–4 mm slabs in an extremely close plane and axis to that of the in vivo MRI. Postmortem histological processing failed for animal M#1’s brain.

Histopathology of WML and NAWM

For visualizing myelin, Luxol fast blue (LFB) staining with periodic acid Schiff (PAS) counterstain and immunohistochemistry for myelin PLP were used. For characterizing inflammation and edema, hematoxylin and eosin (HE) and immunohistochemistry for ionized calcium-binding adaptor molecule (Iba1), CD3, and CD20, were used. Oligodendrocytes and OPC were assessed with aspartoacylase (ASPA) and oligodendrocyte transcription factor 2 (Olig2) double staining (mature oligodendrocytes are considered ASPA and Olig2 positive; OPC are considered Olig2 positive but ASPA negative). For axon staining, Bielschowsky’s silver method was used. Briefly, deparaffinized slides were covered with 20% AgNO3 and incubated at 40°C inside a dark chamber for 30 min. Slides were washed and placed in ammonia silver solution, prepared by adding concentrated ammonium hydroxide drop-by-drop into AgNO3 until brown precipitate disappeared, at 40°C for 30 min. Developer working solution was added to the slides, made with developer stock solution (37–40% formaldehyde, citric acid, and nitric acid), ammonium hydroxide, and distilled water. After all incubations, slides were washed with 1% ammonium hydroxide, washed in distilled water, and treated with 5% sodium thiosulfate solution. Detailed immunohistochemical methods are provided in Supplementary file 1.

Histological WML categorization

Histological analysis and characterization were performed by one experienced rater blinded to the MRI. The categorization was performed according to our experience with marmoset EAE lesions as detailed in previous publications (Lee et al., 2019; Lee et al., 2018; Maggi et al., 2014). The lesions were categorized as follows: 1 – early active: LFB-PAS shows prominent demyelination with LFB+, PAS+, and/or PLP+ phagocytes, indicating ingestion of myelin breakdown products (Figure 1C). PLP immunohistochemistry also demonstrates demyelination with myelin debris. ASPA/Olig2 double immunohistochemistry demonstrates loss of oligodendrocytes. Qualitative assessment highlights prominent Iba1+ cell infiltration, CD3+ and CD20+ cells in the perivascular cuff and lesion core (not shown), and loss of axons on Bielschowsky silver staining. HE staining shows edema marked by irregular clear spaces around cells. 2 – Chronic, likely at least partially demyelinated: LFB-PAS staining and PLP immunohistochemistry show areas of complete demyelination (Figure 2C). Lesions contain few Iba1+ cells, and CD3+ and CD20+ cells (not shown) are scarce and only found around vessels. There is loss of both oligodendrocytes and OPC on ASPA/Olig2 staining. There is less edema compared to early active lesions (HE staining), and there is substantial loss of axons (Bielschowsky silver stain). 3 – Remyelinated; LFB-PAS staining and PLP immunohistochemistry show nearly normal myelin structure (Figure 3C). Both oligodendrocytes and OPC are present, as demonstrated by staining with ASPA/Olig2. Inflammatory cells are less prominent, with few infiltrations of Iba1+ microglia/macrophages or CD3+ and CD20+ lymphocytes (not shown). Bielschowsky staining shows some preservation of normal axon structures.

Quantitative measurement of demyelination and remyelination was performed by a single experienced rater (MD). To obtain a quantitative measurement of demyelination and remyelination, the percentage of demyelinated area for each lesion was extracted on LFB and PLP staining. For consistency, an ROI of 1500 µm2 centered on each lesion core was placed; this ROI was large enough to include even the biggest lesion in our sample. The demyelinated area for each stain was calculated using the thresholding tool on FIJI (Schindelin et al., 2012) as follows: number of pixels with a null value*100/total number of pixels. ASPA+/Olig2+ cell count (oligodendrocytes) and ASPA-/Olig2+ cell count (OPC) were assessed using thresholding tools in FIJI45 within the same ROI for each lesion.

Statistical analysis

To evaluate the statistical sensitivity and specificity of in vivo MRI detection of chronic demyelination or remyelination, relative to histopathology, we created confusion matrices and calculated true or false positive and negative rates, as well as sensitivity and specificity. For interrater reliability of MRI-predicted remyelination, we calculated Cohen’s kappa. To test the effects of corticosteroid treatment and sex on remyelination, we used the point biserial correlation model.

Data availability

All of the 6 marmosets' serial in vivo MRI images, including all the sequences used for analysis and figure generation, were uploaded in an easily accessible format (NIFTI). The file names are titled with the corresponding animal # used in the manuscript, as well as the date of MRI acquisition. All the Iba1 and PLP immunohistochemistry stains have been uploaded as well.

References

Decision letter

  1. Jeannie Chin
    Reviewing Editor; Baylor College of Medicine, United States
  2. Timothy E Behrens
    Senior Editor; University of Oxford, United Kingdom

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "in vivo MRI of Endogenous Remyelination in a Nonhuman Primate Model of Multiple Sclerosis" for consideration by eLife. We appreciate your patience during the handling of this submission. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jeannie Chin as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this response to help you prepare a revised submission. As you will see below, there was an agreement that your work is an important step towards the MRI-based diagnosis of CNS remyelination in multiple sclerosis. However, it was also felt that the present manuscript has shortcomings in the presentation of histological data and some missing information about data acquisition. Please look for details below how to improve a revised version of your manuscript that better supports the claims made.

Reviewer #1 (Recommendations for the authors):

Similar to the serial MRI, the manuscript would be enhanced by supplementing Iba1- and PLP-stained sections of each lesion.

Reviewer #2 (Recommendations for the authors):

Before publication I think that this would be improved by consideration of the following

How do the authors know that PDw is imaging new myelin formation rather than loss of inflammation? The signal is maintained in acute and chronic inflammation, and lost in remyelination -which is also associated with reduced inflammation. It's not clear to me how the two events can be distinguished by this imaging protocol. Is the time course of inflammatory loss different to that for every myelination? The gadolinium experiments mentioned in the methods but not included in the results might address this?

How is the distinction between acute and chronic inflammatory lesions made. Is it just the time of onset after immunisation?

In figure 5 it is not clear how the transition from demyelination to remyelination was established, and so the evaluation of the time course for remyelination is not clear

Reviewer #3 (Recommendations for the authors):

If possible, it would be useful to have images of lesions (histology) at somewhat lower magnification as well, to appreciate differences in density etc. This would be very helpful particularly in the case of Luxol Fast Blue staining of remyelinated lesions so that pale staining compared to non-affected tissue can be appreciated, and for ASPA/Olig2 and Bielschowsky staining, to illustrate differences in density.

Regarding Iba1 staining, clear cellular labeling pattern is observed only in Figure 4, while Figure 2 shows high background and very few if any clearly labeled cells, which is a bit better in Figure 3.

Points to clarify:

1. Figure 5 shows the estimated timing of demyelination/remyelination. Three out of four graphs are from methylprednisolone (MP)-treated animals. It was indicated in Methods that MP administration was performed after demyelination was first detected, but the first graph shows 1 double bar, at 12 weeks (prior to the first peak), and a triple bar at 18 weeks approx. Was this animal treated twice?

2. In the same figure, 2 lesions from M#5 are shown for which, as the legend indicates, remyelination was confirmed histologically. Yet, on page 16, it is indicated that remyelination was detected in animals M#2-4 by MRI and confirmed by histology, and that M#5 had a single remyelinated lesion not seen on MRI, which seems to contradict Figure 5. Could this be clarified?

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "in vivo MRI of Endogenous Remyelination in a Nonhuman Primate Model of Multiple Sclerosis" for further consideration by eLife. Your revised article has been evaluated by Jeannie Chin (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are remaining issues that need to be addressed, as described in the individual reviewer comments at the end of this correspondence. A brief summary of those essential issues follows:

Essential revisions:

1) Related to the finding that none of the lesions returning to isointensity on PDw MRI demonstrated gadolinium enhancement on T1w scans, while the gadolinium data is very helpful, no quantification is provided. This should be added. Ideally, additional histological evidence of remyelination should be provided to strengthen the power of PDw to classify the lesions. Also, please clarify whether gadolinium enhancement is also considered as a marker of acute versus chronic lesions in this paper.

2) The authors assert that prior data shows that the changes seen occur at the same time as remyelination and don't correlate with any changes in inflammation, but this was not sufficiently addressed. Please include a more detailed analysis of remyelination and inflammation in the lesions on which the MRI was performed, rather than a comparison with older data. Of particular interest would be histological evidence of remyelination such as reduced myelin density, shorter and thinner myelin internodes and myelin protein staining of remyelinating oligodendrocyte cytoplasm.

3) A concern raised in the previous review was that it was not possible to see loss of oligodendroglia based on the image provided to illustrate acute lesions (now in Figure 1), in which there is a small oligo-free area surrounding what may be blood vessels, but the remaining PLP-free area contains many oligodendroglial cells. The question of oligodendroglial loss could be addressed by quantifying the density in perilesional area and comparing it to the intralesional density. However, the authors simply refer to a previous manuscript (JCI, 2019). Were the same animals/lesions analyzed in this study? This point needs to be clearly addressed throughout the manuscript wherever references to previous data are made. Please add quantitative data to demonstrate oligo depletion followed by repopulation in remyelinated lesion. If the quantification does not support the loss of oligodendroglia, then these claims should be removed.

4) The histological criteria for lesion classification is slightly confusing, which was brought up in the original review as well. The authors assert that experienced histopathologists blindly classified these lesions in Kuhlmann 2017, but it is not really clear that exactly the same immunolabelings were performed. Please provide a clear statement on which histological parameters (immunolabelings) were considered by the histopathologists to classify the lesions in this study.

Note: Because the results of this manuscript have important implications for pre-clinical testing of remyelinating drugs, solid data on the lesions analyzed in this manuscript should be provided to reinforce the conclusions.

Reviewer #1 (Recommendations for the authors):

There is no histological evidence of remyelination such as reduced myelin density, shorter and thinner myelin internodes and myelin protein staining of remyelinating oligodendrocyte cytoplasm. Most lesional areas are small. It is most likely that many of the MRI ROIs are due to alterations in myelinated white matter. It may be that all lesion are remyelinated to a point that they cannot be distinguished from normal myelinated white matter. It would seem unlikely that a lesion in the process of remyelination is not present in all the lesions investigated. This reduces the utility of the model to investigate remyelination. What can be measured? Unlikely that remyelination can be enhanced. I do not consider this as a reliable model for remyelination. This a relevant question based upon the species utilized in this study.

Reviewer #2 (Recommendations for the authors):

My key point of concern was

"How do the authors know that PDw is imaging new myelin formation rather than loss of inflammation? The signal is maintained in acute and chronic inflammation, and lost in remyelination -which is also associated with reduced inflammation. It's not clear to me how the two events can be distinguished by this imaging protocol. Is the time course of inflammatory loss different to that for every myelination? The gadolinium experiments mentioned in the methods but not included in the results might address this?"

This is rebutted as follows

"- This is an essential point, and we thank the reviewer for the inquiry. We agree (and have now clarified in the paper) that PDw imaging is sensitive to inflammation and other changes; the lack of specificity to myelin is indeed well known in the field. What is critical for our purposes, however, is the ability to infer changes in myelin from the combination of in-vivo MRI data and extensive knowledge of the time course of lesion development and repair, which in our hands is derived from the analysis of ~100 animals over more than a decade, as reflected in our prior publications and the current manuscript. Hence, our paper is essentially about sensitivity of PDw MRI to myelin changes in this context, not about specificity of the technique. It is worth noting, in this context, that chronic EAE lesions are typically either gliotic (chronic inactive) or remyelinated, with few inflammatory cells; thus, the difference in MRI signal between remyelinated and chronically demyelinated lesions cannot be explained purely by inflammation. Furthermore, none of the lesions returning to isointensity on PDw MRI demonstrated gadolinium enhancement on T1w scans. In the revision, we have added material under the description on remyelinated lesions in the Results section."

The gadolinium data is very helpful, although no quantification is provided. This should be added.

I take the point that the prior data argues that the changes seen occur at the same time as remyelination and don't correlate with any changes in inflammation, but feel my concern is not really addressed in the discussion despite it being recognised as "essential". At the very least I'd like to see an addition to the discussion here, although Id much prefer a more detailed analysis of remyelination and inflammation in the lesions on which the MRI was performed rather than a comparison with older data

Reviewer #3 (Recommendations for the authors):

My general feeling is that, while several issues were successfully addressed by this revision, referring to previous manuscripts has not provided answers to some fundamental questions raised in the previous review.

One of my points was that I could not see loss of oligodendroglia based on the image provided to illustrate acute lesions (now in Figure 1), in which there is a small oligo-free area surrounding what may be blood vessels, but the remaining PLP-free area contains many oligodendroglial cells. The question of oligodendroglial loss could be addressed by quantifying the density in perilesional area and comparing it to the intralesional density. However, the authors simply refer to a previous manuscript (JCI, 2019). Were the same animals/lesions analyzed in this study? This is really not clear from the manuscript. One sentence in the Discussion refers to previous analyses of oligo repopulation and remyelination, but is it the same lesions/animals? This should be clarified.

Because some acutely demyelinating lesions show oligo preservation, I do not think that showing a decrease in oligos is necessary to claim that the lesion is demyelinated, but I do think that if this claim is made, it should be supported by data, especially when the image provided does not illustrate the claim. Otherwise, the claim should be omitted. Thus, I recommend adding some quantitative data to demonstrate oligo depletion followed by repopulation in remyelinated lesions. If no depletion is evidenced in acute lesions, then it can be discussed that demyelination with oligodendroglial preservation has also been observed in early MS lesions. If these data on the same lesions are provided in previous publications, it should be clearly indicated that these are the same lesions.

A crucial issue raised in the previous round was that of PDw sequences detecting changes other than demyelination/remyelination. The authors did mention that gadolinium enhancement was present in acute but not chronic lesions, both of which were demyelinated and show similar Pdw, which, in my opinion, does suggest that changes in PDw signal are not due to inflammatory infiltrates, even though does not exclude that other events (axonal changes, astrocytes, diffuse microgliosis) might alter the signal. Including gadolinium data and potentially, quantification, would be a good argument to exclude inflammatory infiltrates as the cause of changes in PDw. This point should be reinforced in the Discussion. Ideally, additional histological evidence of remyelination should be provided to strengthen the power of PDw to classify the lesions.

I still find the histological criteria for lesion classification slightly confusing. The answer to my suggestion to clearly define lesion classification criteria applied in this paper was that experienced histopathologists blindly classified these lesions based on Tanja Kuhlmann paper from 2017, but from the data presented, it is not really clear that exactly the same immunolabelings were performed (as compared to Kuhlmann paper). From the authors' response, it can be concluded that all these analyses were performed in their previous work on marmoset lesions (again, were these the same lesions?), and thus not repeated in this study. I think that the authors need to provide a clear statement on which histological parameters (immunolabellings) were considered by the histopathologists to classify the lesions in this study.

The authors state that lesions were classified as acute if younger than 5 weeks and remyelinated or chronic if older than 5 weeks. I suppose this refers to MRI classification of the lesions, and not the histological one. As the authors also describe acute lesions as gadolinium enhancing, and chronic lesions (and remyelinated ones) gadolinium enhancement-free, was gadolinium enhancement also considered as a marker of acute versus chronic lesions in this paper?

Because the results of this manuscript have important implications for pre-clinical testing of remyelinating drugs, solid data on the lesions analyzed in this manuscript should be provided to reinforce the conclusions.

https://doi.org/10.7554/eLife.73786.sa1

Author response

Reviewer #1 (Recommendations for the authors):

Similar to the serial MRI, the manuscript would be enhanced by supplementing Iba1- and PLP-stained sections of each lesion.

We agree that supplementing all the lesions with Iba1 and PLP stains would be a useful addition and have now added them to the supplementary data section of the manuscript, under Source data 9.

Reviewer #2 (Recommendations for the authors):

Before publication I think that this would be improved by consideration of the following

How do the authors know that PDw is imaging new myelin formation rather than loss of inflammation? The signal is maintained in acute and chronic inflammation, and lost in remyelination -which is also associated with reduced inflammation. It's not clear to me how the two events can be distinguished by this imaging protocol. Is the time course of inflammatory loss different to that for every myelination? The gadolinium experiments mentioned in the methods but not included in the results might address this?

This is an essential point, and we thank the reviewer for the inquiry. We agree (and have now clarified in the paper) that PDw imaging is sensitive to inflammation and other changes; the lack of specificity to myelin is indeed well known in the field. What is critical for our purposes, however, is the ability to infer changes in myelin from the combination of in-vivo MRI data and extensive knowledge of the time course of lesion development and repair, which in our hands is derived from the analysis of ~100 animals over more than a decade, as reflected in our prior publications and the current manuscript. Hence, our paper is essentially about sensitivity of PDw MRI to myelin changes in this context, not about specificity of the technique.

It is worth noting, in this context, that chronic EAE lesions are typically either gliotic (chronic inactive) or remyelinated, with few inflammatory cells; thus, the difference in MRI signal between remyelinated and chronically demyelinated lesions cannot be explained purely by inflammation. Furthermore, none of the lesions returning to isointensity on PDw MRI demonstrated gadolinium enhancement on T1w scans. In the revision, we have added material under the description on remyelinated lesions in the Results section.

How is the distinction between acute and chronic inflammatory lesions made. Is it just the time of onset after immunisation?

We thank the reviewer for the opportunity to clarify this. The differentiation between acute and chronic is solely based on the timeline of the lesion progression, detected by the serial in vivo MRI, and not on the time after immunization.

In figure 5 it is not clear how the transition from demyelination to remyelination was established, and so the evaluation of the time course for remyelination is not clear

We thank the reviewer for pointing out the unclear time course as highlighted on Figure 5 (now Figure 4). Our evaluation was based on the MRI signal intensity change. We have clarified this point in our manuscript in the figure legend.

Reviewer #3 (Recommendations for the authors):

If possible, it would be useful to have images of lesions (histology) at somewhat lower magnification as well, to appreciate differences in density etc. This would be very helpful particularly in the case of Luxol Fast Blue staining of remyelinated lesions so that pale staining compared to non-affected tissue can be appreciated, and for ASPA/Olig2 and Bielschowsky staining, to illustrate differences in density.

As mentioned above, we agree that lower magnification images will be helpful to identify areas of thinly remyelinated regions and have included these.

Regarding Iba1 staining, clear cellular labeling pattern is observed only in Figure 4, while Figure 2 shows high background and very few if any clearly labeled cells, which is a bit better in Figure 3.

We included both a higher and lower magnification of the Iba1 staining on Figure 2 (now Figure 1) for better visualization. We hope this clarifies the issue.

Points to clarify:

1. Figure 5 shows the estimated timing of demyelination/remyelination. Three out of four graphs are from methylprednisolone (MP)-treated animals. It was indicated in Methods that MP administration was performed after demyelination was first detected, but the first graph shows 1 double bar, at 12 weeks (prior to the first peak), and a triple bar at 18 weeks approx. Was this animal treated twice?

We apologize for the lack of clarity in our description. As you mentioned, the animal was treated twice after the first steroid treatment did not result in any clinical or radiological improvement. We have clarified this in our figure (now Figure 4) legend.

2. In the same figure, 2 lesions from M#5 are shown for which, as the legend indicates, remyelination was confirmed histologically. Yet, on page 16, it is indicated that remyelination was detected in animals M#2-4 by MRI and confirmed by histology, and that M#5 had a single remyelinated lesion not seen on MRI, which seems to contradict Figure 5. Could this be clarified?

We thank the reviewer for catching this error; we have now clarified and corrected the error in our revised manuscript.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Essential revisions:

1) Related to the finding that none of the lesions returning to isointensity on PDw MRI demonstrated gadolinium enhancement on T1w scans, while the gadolinium data is very helpful, no quantification is provided. This should be added. Ideally, additional histological evidence of remyelination should be provided to strengthen the power of PDw to classify the lesions. Also, please clarify whether gadolinium enhancement is also considered as a marker of acute versus chronic lesions in this paper.

We appreciate the comment and have made substantial changes to the paper in response.

First, we would like to clarify that most of the lesions newly detected on PDw MRI showed gadolinium enhancement on T1w images. Only lesions identified as “early active” on pathology were found to be enhancing on the terminal MRI. We added the following paragraph in the Results section:

“T1w gadolinium enhancement as a marker of acute inflammation

Across the 6 animals scanned longitudinally, 82% of the lesions newly detected on PDw MRI presented T1w gadolinium enhancement. Enhancement was occasionally seen at the following timepoint (10–15 days after first detection). No chronic, at least partially demyelinated or remyelinated WML presented gadolinium enhancement on the terminal scan. M#5–6 presented at least one early active WML enhancing lesion at their terminal scan.”

The use of gadolinium enhancement to categorize lesions as acute or chronic has now been clarified in the Results section, in the section entitled “MRI WML categorization.”

Second, as suggested, we have added more histological evidence of lesion remyelination in the revision. To do so, we assessed, over 31 lesions, the percentage of unstained area on PLP and LFB stains for myelin protein and lipid, respectively. We also added 10 NAWM areas from the same animals as control. As expected, we observed a larger unstained area for demyelinated lesions vs remyelinated lesions on both PLP and LFB. We have added details to the method section as well as a detailed paragraph in results:

“Histological quantification recapitulates MRI rater analysis of lesion myelin status

Assessment of PLP staining in 31 lesions and 10 normal appearing white matter (NAWM) areas (1500 µm2 centered over the lesion core) demonstrated larger unstained areas in early active (58±25%) and chronic, at least partially demyelinated (38±25%) lesions compared to remyelinated lesions (4.5±1.1%) or NAWM (2.6±0.3%) (Figure 5A). LFB assessment showed similar results: 63±26% unstained area in early active lesions, 43±26% in chronic, at least partially demyelinated lesions, 15±25% in remyelinated lesions, and 3.0±0.4% in NAWM (Figure 5B). We observed a significantly smaller unstained PLP area in NAWM compared to remyelinated lesions (2-sample t-test, p<0.001). There were no apparent differences between PLP and LFB staining in the different lesion categories and in NAWM (2-sample t-test). Interestingly, 1 lesion appeared remyelinated on PLP, with less than 6% of unstained, area but demyelinated on LFB (82% of unstained area).”

Oligodendrocyte density was addressed with ASPA and Olig2 staining. ASPA+/Olig2+ cells, reflecting mature oligodendrocytes, and ASPA-/Olig2+ cells, potentially reflecting OPC, were counted for every lesion in comparison to 10 NAWM areas. Our analysis showed a higher ASPA+/Olig2+ cell count in remyelinated lesions compared to acute or chronic lesions as well as negative correlation between lesion volume at terminal PDw MRI and ASPA+/Olig2+ cell number. A detailed paragraph was added in the Results section:

“Oligodendrocyte and OPC counts are consistent with degree of demyelination in lesions Quantitative assessment of ASPA+/Olig2+ (mature oligodendrocytes) and ASPA-/Olig2+ (OPC) across the 31 lesions and 10 NAWM areas showed, as expected, more mature oligodendrocytes in remyelinated lesions and NAWM than early active or chronic, at least partially demyelinated lesions (Figure 5C). Consistent with PLP observations, we observed significantly more oligodendrocytes in NAWM compared to remyelinated lesions (2-sample t-test, p<0.001). Interestingly, more OPC were found in early active demyelinating lesions than in remyelinated lesions or NAWM (Figure 5D). Younger lesions (<10 weeks old by MRI) had more OPC than older lesions, highlighted by a negative correlation between lesion age and OPC count (r = -0.46; p = 0.009).”

The revision includes a newly added Figure 5, which summarizes the quantitative histological analysis. Adequate description in the method section was also added (“Histological WML categorization”). We have also added additional comments in the Discussion section.

2) The authors assert that prior data shows that the changes seen occur at the same time as remyelination and don't correlate with any changes in inflammation, but this was not sufficiently addressed. Please include a more detailed analysis of remyelination and inflammation in the lesions on which the MRI was performed, rather than a comparison with older data. Of particular interest would be histological evidence of remyelination such as reduced myelin density, shorter and thinner myelin internodes and myelin protein staining of remyelinating oligodendrocyte cytoplasm.

As requested and described in the response to essential revision #1 (above), we have added additional histological analysis to back up and quantify our findings regarding remyelination.

In addition to myelin staining, we did additional characterization of CD3+ T cells and CD20+ B cells to assess inflammation in each type of lesion. More details were added in the paragraph “Histological WML categorization” in the method section for each lesion category.

We also included enhanced discussion on those points in the appropriate section of the paper.

We added acknowledgement in the discussion that additional immunohistochemistry would be useful to further characterize myelination status, but unfortunately we were not able to achieve this in our tissue:

“Finally, we did not obtain ultrathin sections to quantify myelin thickness, as we were interested in performing a battery of stains on our tissue, and as lesions were found at a variety of orientations relative to axons, which would complicate such quantification.”

3) A concern raised in the previous review was that it was not possible to see loss of oligodendroglia based on the image provided to illustrate acute lesions (now in Figure 1), in which there is a small oligo-free area surrounding what may be blood vessels, but the remaining PLP-free area contains many oligodendroglial cells. The question of oligodendroglial loss could be addressed by quantifying the density in perilesional area and comparing it to the intralesional density. However, the authors simply refer to a previous manuscript (JCI, 2019). Were the same animals/lesions analyzed in this study? This point needs to be clearly addressed throughout the manuscript wherever references to previous data are made. Please add quantitative data to demonstrate oligo depletion followed by repopulation in remyelinated lesion. If the quantification does not support the loss of oligodendroglia, then these claims should be removed.

We appreciate the request for clarification. With respect to the second point, the lesions and animals studied in this project were different those reported in our previous publication (JCI, 2019).

Please see the response to essential revision #1 (above) for the revisions we made to quantify OPC and oligodendrocytes. The new analysis is summarized in the newly added Figure 5, and the methods are described under “Histological WML categorization.”

4) The histological criteria for lesion classification is slightly confusing, which was brought up in the original review as well. The authors assert that experienced histopathologists blindly classified these lesions in Kuhlmann 2017, but it is not really clear that exactly the same immunolabelings were performed. Please provide a clear statement on which histological parameters (immunolabelings) were considered by the histopathologists to classify the lesions in this study.

In response to this critique, we reshaped the method section to give more detail on our lesion categorization for both MRI and histopathology. The following paragraphs were added or changed:

“MRI WML categorization

MRI WML categorization and prediction were set based on our prior experience and performed independently by 2 experienced raters, blinded to histology. WML were classified according to the terminal MRI as follows: 1 — predicted early active, described as hyperintense on PDw and enhancing on T1w scans after injection of gadolinium; 2 — predicted chronic, at least partially demyelinated, described as hyperintense on PDw and not enhancing on T1w scans after injection of gadolinium; and 3 — predicted remyelinated, described as initially hyperintense on PDw, iso-intense on PDw at the terminal scan, and not enhancing on T1w scans after injection of gadolinium.”

“Histological WML categorization

Histological analysis and characterization were performed by 1 experienced rater blinded to the MRI. The categorization was performed according to our experience with marmoset EAE lesions as detailed in previous publications14,24,44. The lesions were categorized as follows: 1 — early active: LFB-PAS shows prominent demyelination with LFB+, PAS+, and/or PLP+ phagocytes, indicating ingestion of myelin breakdown products (Figure 1C). PLP immunohistochemistry also demonstrates demyelination with myelin debris. ASPA/Olig2 double immunohistochemistry demonstrates loss of oligodendrocytes. Qualitative assessment highlights prominent Iba1+ cell infiltration, CD3+, and CD20+ cells in the perivascular cuff and lesion core (not shown), and loss of axons on Bielschowsky silver staining. HE staining shows edema marked by irregular clear spaces around cells. 2 — Chronic, likely at least partially demyelinated: LFB-PAS staining and PLP immunohistochemistry show areas of complete demyelination (Figure 2C). Lesions contain few Iba1+ cells, and CD3+ and CD20+ cells (not shown) are scarce and only found around vessels. There is loss of both oligodendrocytes and OPC on ASPA/Olig2 staining. There is less edema compared to early active lesions (HE staining), and there is substantial loss of axons (Bielschowsky silver stain). 3 — Remyelinated; LFB-PAS staining and PLP immunohistochemistry show nearly normal myelin structure (Figure 3C). Both oligodendrocytes and OPC are present, as demonstrated by staining with ASPA/Olig2. Inflammatory cells are less prominent, with few Iba1+ microglia/macrophages or CD3+ and CD20+ lymphocytes (not shown). Bielschowsky staining shows some preservation of normal axon structures.

Quantitative measurement of demyelination and remyelination was performed by a single experienced rater (MD). To obtain a quantitative measurement of demyelination and remyelination, the percentage of demyelinated area for each lesion was extracted on LFB and PLP staining. For consistency, a region of interest (ROI) of 1500 µm2 centered on each lesion core was placed; this ROI was large enough to include even the biggest lesion in our sample. The demyelinated area for each stain was calculated using the thresholding tool on FIJI45 as follows: number of pixels with a null value*100/total number of pixels. ASPA+/Olig2+ cell count (oligodendrocytes) and ASPA-/Olig2+ cell count (OPC) were assessed using thresholding tools in FIJI45 within the same ROI for each lesion.”

https://doi.org/10.7554/eLife.73786.sa2

Article and author information

Author details

  1. Maxime Donadieu

    Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Methodology, Writing – original draft, Writing – review and editing
    Contributed equally with
    Nathanael J Lee
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0009-0003-7349-1648
  2. Nathanael J Lee

    1. Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    2. Department of Neurology and Neurological Sciences, Stanford University, Palo Alto, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing
    Contributed equally with
    Maxime Donadieu
    Competing interests
    No competing interests declared
  3. María I Gaitán

    Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Writing – review and editing
    Competing interests
    No competing interests declared
  4. Seung-Kwon Ha

    1. Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    2. Department of Neurobiology, University of Pittsburgh, Pittsburgh, United States
    Contribution
    Data curation, Formal analysis, Writing – original draft, Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Nicholas J Luciano

    Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Snehashis Roy

    Section on Neural Function, National Institute of Mental Health, National Institutes of Health, Bethesda, United States
    Contribution
    Data curation, Formal analysis, Validation, Methodology
    Competing interests
    No competing interests declared
  7. Benjamin Ineichen

    1. Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    2. Department of Neuroradiology, Clinical Neuroscience Center, University Hospital Zurich, University of Zurich, Switzerland, Switzerland
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  8. Emily C Leibovitch

    Viral Immunology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  9. Cecil C Yen

    Cerebral Microcirculation Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Data curation, Formal analysis
    Competing interests
    No competing interests declared
  10. Dzung L Pham

    Department of Radiology and Radiological Sciences, Uniformed Services University of the Health Sciences, Bethesda, United States
    Contribution
    Data curation, Formal analysis
    Competing interests
    No competing interests declared
  11. Afonso C Silva

    1. Department of Neurobiology, University of Pittsburgh, Pittsburgh, United States
    2. Cerebral Microcirculation Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Supervision, Validation, Methodology
    Competing interests
    No competing interests declared
  12. Mac Johnson

    Vertex Pharmaceuticals Incorporated, Boston, United States
    Contribution
    Supervision, Funding acquisition, Writing – review and editing
    Competing interests
    is a shareholder and employee of Vertex Pharmaceuticals, Inc
  13. Steve Jacobson

    Viral Immunology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Supervision, Project administration
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3127-1287
  14. Pascal Sati

    1. Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    2. Neuroimaging Program, Department of Neurology, Cedars Sinai, Los Angeles, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Validation, Investigation, Methodology, Writing – original draft, Writing – review and editing
    For correspondence
    pascal.Sati@cshs.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6763-0125
  15. Daniel S Reich

    Translational Neuroradiology Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, United States
    Contribution
    Conceptualization, Resources, Funding acquisition, Validation, Investigation, Methodology, Writing – review and editing, Supervision
    For correspondence
    daniel.reich@nih.gov
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2628-4334

Funding

National Institute of Neurological Disorders and Stroke (Intramural Research Program)

  • Maxime Donadieu
  • Nathanael J Lee
  • Seung-Kwon Ha
  • Nicholas J Luciano
  • Benjamin Ineichen
  • Emily C Leibovitch
  • Cecil C Yen
  • Afonso C Silva
  • Steve Jacobson
  • Pascal Sati
  • Daniel S Reich

Adelson Family Foundation

  • Maxime Donadieu
  • Seung-Kwon Ha
  • Daniel S Reich

Vertex Pharmaceuticals

  • Mac Johnson
  • Daniel S Reich

Swiss National Science Foundation

  • Benjamin Ineichen

National Multiple Sclerosis Society (RG-1907-34570)

  • Dzung L Pham
  • Snehashis Roy

National Institutes of Health (R21OD030163)

  • Dzung L Pham
  • Snehashis Roy

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This study was funded by the National Institute of Neurological Disorders and Stroke Intramural Research Program and through a Cooperative Research and Development Agreement with Vertex Pharmaceuticals. We thank Dr. Tianxia Wu for help with statistical analyses. We thank everyone who took care of the animals and assisted with acquisition of MRI scans.

Ethics

The study was performed under the guideline and in accordance with the National Institutes of Health IACUC. Specifically, the neuroethics committee of the National Institutes of Neurological Diseases and Stroke formally went through our manuscript prior to submission on salient topics including minimization of pain, justification of number of animals and the sex ratio, dosing of methylprednisone based on available human data. All procedures were performed under anesthesia to minimize discomfort and pain. Animals were housed in pairs or triplets to maximize social interactions and well-being. The institutional IACUC protocol number is #1308.

Senior Editor

  1. Timothy E Behrens, University of Oxford, United Kingdom

Reviewing Editor

  1. Jeannie Chin, Baylor College of Medicine, United States

Version history

  1. Received: September 10, 2021
  2. Preprint posted: October 28, 2021 (view preprint)
  3. Accepted: April 12, 2023
  4. Accepted Manuscript published: April 21, 2023 (version 1)
  5. Version of Record published: May 10, 2023 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 890
    Page views
  • 126
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Maxime Donadieu
  2. Nathanael J Lee
  3. María I Gaitán
  4. Seung-Kwon Ha
  5. Nicholas J Luciano
  6. Snehashis Roy
  7. Benjamin Ineichen
  8. Emily C Leibovitch
  9. Cecil C Yen
  10. Dzung L Pham
  11. Afonso C Silva
  12. Mac Johnson
  13. Steve Jacobson
  14. Pascal Sati
  15. Daniel S Reich
(2023)
In vivo MRI is sensitive to remyelination in a nonhuman primate model of multiple sclerosis
eLife 12:e73786.
https://doi.org/10.7554/eLife.73786

Further reading

    1. Neuroscience
    Mika Rubinov
    Review Article

    Genuinely new discovery transcends existing knowledge. Despite this, many analyses in systems neuroscience neglect to test new speculative hypotheses against benchmark empirical facts. Some of these analyses inadvertently use circular reasoning to present existing knowledge as new discovery. Here, I discuss that this problem can confound key results and estimate that it has affected more than three thousand studies in network neuroscience over the last decade. I suggest that future studies can reduce this problem by limiting the use of speculative evidence, integrating existing knowledge into benchmark models, and rigorously testing proposed discoveries against these models. I conclude with a summary of practical challenges and recommendations.

    1. Neuroscience
    Ann-Kathrin Joechner, Michael A Hahn ... Markus Werkle-Bergner
    Research Article

    The synchronization of canonical fast sleep spindle activity (12.5–16 Hz, adult-like) precisely during the slow oscillation (0.5–1 Hz) up peak is considered an essential feature of adult non-rapid eye movement sleep. However, there is little knowledge on how this well-known coalescence between slow oscillations and sleep spindles develops. Leveraging individualized detection of single events, we first provide a detailed cross-sectional characterization of age-specific patterns of slow and fast sleep spindles, slow oscillations, and their coupling in children and adolescents aged 5–6, 8–11, and 14–18 years, and an adult sample of 20- to 26-year-olds. Critically, based on this, we then investigated how spindle and slow oscillation maturity substantiate age-related differences in their precise orchestration. While the predominant type of fast spindles was development-specific in that it was still nested in a frequency range below the canonical fast spindle range for the majority of children, the well-known slow oscillation-spindle coupling pattern was evident for sleep spindles in the adult-like canonical fast spindle range in all four age groups—but notably less precise in children. To corroborate these findings, we linked personalized measures of fast spindle maturity, which indicate the similarity between the prevailing development-specific and adult-like canonical fast spindles, and slow oscillation maturity, which reflects the extent to which slow oscillations show frontal dominance, with individual slow oscillation-spindle coupling patterns. Importantly, we found that fast spindle maturity was uniquely associated with enhanced slow oscillation-spindle coupling strength and temporal precision across the four age groups. Taken together, our results suggest that the increasing ability to generate adult-like canonical fast sleep spindles actuates precise slow oscillation-spindle coupling patterns from childhood through adolescence and into young adulthood.