Validating a minipig model of reversible cerebral demyelination using human diagnostic modalities and electron microscopy

Abstract

Background: Inflammatory demyelinating diseases of the central nervous system, such as multiple sclerosis, are significant sources of morbidity in young adults despite therapeutic advances. Current murine models of remyelination have limited applicability due to the low white matter content of their brains, which restricts the spatial resolution of diagnostic imaging. Large animal models might be more suitable but pose significant technological, ethical and logistical challenges.

Methods: We induced targeted cerebral demyelinating lesions by serially repeated injections of lysophosphatidylcholine in the minipig brain. Lesions were amenable to follow-up using the same clinical imaging modalities (3T magnetic resonance imaging, ¹¹C-PIB positron emission tomography) and standard histopathology protocols as for human diagnostics (myelin, glia and neuronal cell markers), as well as electron microscopy (EM), to compare against biopsy data from two patients.

Findings: We demonstrate controlled, clinically unapparent, reversible and multimodally trackable brain white matter demyelination in a large animal model. De-/remyelination dynamics were slower than reported for rodent models and paralleled by a degree of secondary axonal pathology. Regression modelling of ultrastructural parameters (g-ratio, axon thickness) predicted EM features of cerebral de- and remyelination in human data.

Interpretation: We validated our minipig model of demyelinating brain diseases by employing human diagnostic tools and comparing it with biopsy data from patients with cerebral demyelination.

Funding: This work was supported by the DFG under Germany's Excellence Strategy within the framework of the Munich Cluster for Systems Neurology (EXC 2145 SyNergy, ID 390857198) and TRR 274/1 2020, 408885537 (projects B03 and Z01).

Keywords: Electromagnetic-guided navigation system; Inflammatory-demyelinating brain disease; In vivo minipig model; Lysophosphatidylcholine; PET/MRI; Scanning electron microscopy.

Fig. 1

MiniSWINE methodology overview. a) Demyelination induction and remyelination follow-up timeline (d = days). Timepoint "0": baseline imaging using CT and Gd-MRI. Timepoints "10-20-30 days post-injection": every ten days induction of a new demyelinating lesion at a distinct localisation in the centrum semiovale and follow-up of the previous lesions. Gradient colour schematic reflects successive de- and remyelination stages. SEM = scanning electron microscopy. b) LPC chemical formula; R¹, R² = variable fatty acid chains. c) Overview photographs and photomicrographs of successive stages of MiniSWINE. White blurring obscures content with potentially strong emotional impact. The scale bar in insets "MRI imaging", "¹¹C-PIB-PET imaging", and "Histology pipeline" is 25 mm, and the scale bar in inset "SEM pipeline" is 2.5 μm. EMTS = electromagnetic tracking system.

Fig. 2

MiniSWINE: electromagnetic-tracking system (EMTS) and surgical procedure. a) MiniSWINE setup for EMTS. b) The reference tracker was rigidly attached to the minipig forehead. The tracked guide sleeve was essential for the precise alignment of the drill. c) Cannula assembly. The CED cannula was trackable as long as it contained the sensor wire. Once on target, the wire was replaced with the single-use 20G epidural catheter. d) Sequential cannula development, cannula tip profiles, and empirical spread functions of CED in 0.2%–0.6% agarose gel. Evolution from a modified human brain biopsy cannula with a side-opening near the tip (far left) up to our definitive "telescopic" design (far right) with a round bevelled edge. Notice the near-ideal spherical spread function of the latter. e) Injection "hands-on" robotic system assembly consisting of a control unit and a small robot for the controlled and tracked insertion of the cannula. f) MiniSWINE software in operator view, 1st row and corresponding surgical overview, 2nd row. White dotted circles in 1st row mark system feedback to the operator. White blurring obscures content with potential emotional impact. g) Mean in vitro deviations from the target in a phantom test of 0.23 ± 0.03 mm (n = 25 trials). h) Scatterplot of 3D deviations from target in vivo in all 3D planes; black "target" set in the common origin of the coordinate system. i) Mean in vivo deviations from the target by study group: no significant difference between acute LPC lesions (1.24 ± 0.5 mm, n = 8) and acute CTRL lesions (1.83 ± 0.5 mm, n = 8), 1-way ANOVA with df = 1, F = 0.71, p = 0.41.

Fig. 3

MiniSWINE MRI. a) Example axial tomographic planes, each row corresponding to one acute lesion from a separate minipig, scale bar = 2 cm. b) MRI signals in each stage (aLPC n = 18, aCTRL = 13; iLPC n = 11, iCTRL n = 9, sLPC n = 4, sCTRL n = 5). Statistical significance and p values were determined by the Kruskal–Wallis test followed by Dunn's multiple comparisons test, or, given data were normal and homoscedastic, by one-way ANOVA followed by Tukey's multiple comparisons test; ∗0.01 ≤ p ≤ 0.05,∗∗0.001 ≤ p ≤ 0.01, ∗∗∗p ≤ 0.001. c) Spearman's correlograms between MRI signals: dotted lines for linear regressions, coloured areas for 95%-confidence intervals, R = Spearman correlation coefficient, p values from the Spearman's rank correlation test. d) 3D reconstructions of exemplary aLPC, and e) aCTRL lesions. Note that T2FLAIR hyperintense signals take up the largest volume, followed by T1Gd and, depending on the lesion localisation, DIR signals. In contrast, the SWI signals are mostly confined to the needle trajectory through the tissue.

Fig. 4

Multimodal validation of de- and remyelination. a) Example 3D lesion reconstruction. Dotted planes correspond to localisations in the following: 1—aLPC, 2—aCTRL, 3—iLPC, 4—iCTRL. b) Correlative imaging and histopathology. Dotted white circles on PET/MRI correspond to black dotted rectangles in LFB photomicrographs. c) Higher magnification photomicrographs (note different scales for presentation purposes because LPC lesions are more extensive than CTRL) corresponding to the areas above. d) High magnification micrographs of exemplary aspects (from left to right): LFB-scarce WM in aLPC; partially, diffusely remyelinated WM in iLPC, corresponding to a region from c), rows 3–4; normal-appearing WM in aCTRL; well delineated (black line) border of demyelination in aLPC; additional indicators of fresh demyelination in the form of myelin phagocytosing macrophages (foam cells) in aLPC. e) Equivolumetric sphere size distribution for LPC (V = volume). f) Spearman's correlogram between the T2FLAIR volume and ¹¹C-PIB-uptake in aLPC/aCTRL. g) Spearman's correlogram between SWI susceptibility volumes and ¹¹C-PIB-uptake in aLPC/aCTRL. h) Spearman's correlogram between T2FLAIR volume and the histopathologically detectable mean demyelination areas in aLPC/aCTRL. i) Slide-by-slide sum (Σ) of demyelinated areas across groups. Statistical significance and p values determined by Kruskal–Wallis test followed by Dunn's multiple comparisons test, ∗0.01 ≤ p ≤ 0.05. j) Spearman's correlogram between time post-induction and mean remyelination area. R = Spearman correlation coefficient, p values are calculated from Spearman's rank correlation test.

Fig. 5

Immunohistochemical characterisation of LPC lesions across stages. a) Acute stage (aLPC), b) Intermediate (iLPC), c) Subacute (sLPC). Common denominators of a–c): From left to right: Schematic corresponding to Fig. 1 of the lesion stage; LFB (Luxol fast blue) overview of the entire coronal slice (scale bar = 5 mm, black square delineates ROI magnified on the right and in each low magnification inset of the following stainings). Cell-specific marker stainings containing paired low-magnification (15×, left, scale bar = 1 mm, black square delineates ROI magnified on the right) and high magnification insets (right, 60×, scale bar = 100 μm) for following cellular markers: Olig2 (oligodendrocyte lineage cells), Iba1 (microglia), NeuN (neuronal somata), CD3 (lymphocytes), CD68 (monocytes/macrophages). d) From left to right: areas of LFB signal loss (i.e., demyelination), Olig2+ cell densities per 300 × 300 μm² ROI, proportion of area containing activated GFAP+ astrocytes in 1 × 1 mm² ROI, proportion of area containing reactive Iba1+ microglia in 1 × 1 mm² ROI, NeuN+ cell densities per 300 × 300 μm² ROI, CD3+ cell densities per 300 × 300 μm² ROI. We refrained from CD68 quantification because of a meagre signal-to-noise ratio, presumably due to low cross-species primary antibody affinity. Statistical tests were performed using the Kruskal–Wallis-Test, followed by Dunn's multiple comparisons test. n = 4, ∗0.01 ≤ p ≤ 0.05, ∗∗0.01 ≤ p ≤ 0.001, ∗∗∗p ≤ 0.001.

Fig. 6

Comparison of MiniSWINE and human demyelinating CNS disease biopsy ultrastructure. a) 1st row: Pairwise T2FLAIR and T1Gd axial MRI slices illustrating, in the case of the minipigs (from the left, columns 1 and 3), the areas where the autopsy was performed and, in the case of the humans (from the left, columns 2 and 4), the areas where biopsies were performed. White/black bounding boxes indicate approximate areas from which a small sample was analysed with SEM (see also Methods). 2nd row: SEM-micrographs from aLPC, aHum, sLPC, sCTRL and, respectively, sHum groups, scale bar = 10 μm. 3rd row, 4th row: High magnification SEM-micrographs, scale bar = 2 μm. White arrows = thin myelin; hollow white arrows = lipid droplets in foam cells; black arrows with white outline: astrocytic lipid droplets and lysosomal inclusions; black arrows: axonal pathology. b) Quantification of axonal and myelin pathology of axonal diameter (above), as well as g-ratio (below) in all study groups, including aHum and sHum. c) Quantification of myelin pathology. Numbers of, from the left, normally myelinated, thinly myelinated and demyelinated axons per 10 × 10 μm ROI. b and c) Kruskal–Wallis test followed by Dunn's multiple comparisons test, ∗0.01 ≤ p ≤ 0.05, ∗∗0.01 ≤ p ≤ 0.001, ∗∗∗p ≤ 0.001. d) Comparisons of the predictive power of linear regression models of axonal thickness and g-ratio from MiniSWINE (the linear model itself as a straight line) compared to the human linearised data, represented by the dotted straight line. The nearer the trajectory of the coloured line was to one of the dotted lines, the better the prediction of MiniSWINE relative to human data was. Statistical significance and p values were determined by likelihood ratio tests, ∗∗∗p ≤ 0.001, n.s. = not significant.