Increased Retention of Gadolinium in the Inflamed Brain after repeated administration of gadopentetate dimeglumine: A Proof-of-Concept Study in Mice Combining ICP-MS and Micro- and Nano-SR-XRF

OBJECTIVES
The aim of this study was to determine in vivo if brain inflammation leads to increased gadolinium (Gd) retention in brain tissue after repeated applications of Gd-based contrast agents (GBCAs).


MATERIALS AND METHODS
Experimental autoimmune encephalomyelitis (EAE) was induced in female SJL/J mice (n = 6). Experimental autoimmune encephalomyelitis and healthy control mice (n = 4) received 2.5 mmol/kg Gd-DTPA over 10 days (8 injections, cumulated dose of 20 mmol/kg), starting at day 14 post immunization when EAE mice reached the maximal clinical disability. In a group of mice, T1-weighted 2-dimensional RARE images were acquired before the first GBCA injection and 1 day after the last injection. Mice were killed either 1 day or 10 days after the last Gd application. From each single animal, a brain hemisphere was used for Gd detection using inductively coupled plasma mass spectrometry, whereas the other hemisphere was processed for histology and synchrotron x-ray fluorescence spectroscopy (SR-XRF) analysis.


RESULTS
Gadolinium deposition in inflamed brains was mapped by SR-XRF 1 day after the last Gd-DTPA injections, although only mild signal hyperintensity was found on unenhanced T1-weighted images. In addition, using inductively coupled plasma mass spectrometry, we detected and quantified Gd in both healthy and EAE brains up to 10 days after the last injections. However, EAE mouse brains showed higher levels of Gd (mean ± SD, 5.3 ± 1.8 μg/g; range, 4.45-8.03 μg/g) with respect to healthy controls (mean ± SD, 2.4 ± 0.6 μg/g; range, 1.8-3.2 μg/g). By means of micro-SR-XRF, we identified submicrometric Gd hotspots in all investigated samples containing up to 5893 μg Gd/g tissue. Nano-SR-XRF further indicated that Gd small hotspots had an average size of ~160 nm diameter and were located in areas of high inflammatory activity.


CONCLUSIONS
After repeated administrations of Gd-DTPA, ongoing inflammation may facilitate the retention of Gd in the brain tissue. Thus, neuroinflammation should be considered as a risk factor in the recommendation on use of linear GBCA-enhanced MRI.


1) The state of the research
Multiple sclerosis (MS) is a chronic inflammatory disease of the brain and spinal cord. It is one of the most common cause of non-traumatic disability in young adults, affecting more than two million people worldwide (Thompson, Banwell et al. 2018). MS is considered to be an autoimmune disease in which self-reactive immune cells get access to the CNS leading to myelin destruction, neuroaxonal degeneration and the subsequent formation of multifocal lesions (Weissert 2013). Consequently, the hallmarks of MS pathology are inflammation, demyelination and neurodegeneration (Lublin, Reingold et al. 2014).
Although the exact etiology of the disease is yet unclear, some risk factors that may be involved in MS pathogenesis, includes environmental exposure, genetic susceptibility (Kaminska, Koper et al. 2017), as well as age, gender, family history, race, climate, smoking, vitamin D deficiency, certain infections and autoimmune disease, such as type I diabetes or inflammatory bowel disease (Loken-Amsrud, Lossius et al. 2015). Most patients are often diagnosed between the ages of 20 and 50, with two to three times higher incidence in female than in male (Kurtzke 2005). People living in northern countries, particularly those from Northern Europe, have highest risk of developing MS, while Asian, African and native American people have lowest risk (Mechelli, Annibali et al. 2010). Low level of serum vitamin D increases the risk of developing MS and also affects disease activity in MS patients (Sintzel, Rametta et al. 2018). Besides, Epstein-Barr virus has been linked to MS, as the risk of developing MS following Epstein-Barr virus infections is four times higher after ten years of infection (Steiner, Nisipianu et al. 2001).
There are four main clinical subtypes of MS, clinically isolated syndrome (CIS), relapsingremitting MS (RRMS), secondary progressive MS (SPMS), and primary progressive MS (PPMS). The most frequent forms are RRMS, in which acute attacks are followed by complete or partial recovery, and PPMS, characterized by disease progression from onset. More than 80% of patients show a relapsing-remitting-course at the beginning of the disease, which in the majority of cases converts to a progressive disease course -SPMS -after 10-25 years (Filippi, Preziosa et al. 2016).
Clinical signs and symptoms of MS are varied and depend on the amount and the location of nerve damage. MS severity is widely measured with the extended disability status scale (EDSS), a score based on symptoms in eight functional systems, walking ability and activities of daily life, including fatigue, walking difficulties, numbness or tingling, spasticity, weakness, vision dysfunctions, dizziness and vertigo, bladder difficulties, sexual troubles, bowel problems, pain, cognitive changes, emotional changes and depression (Green, Cutter et al. 2017, Izquierdo 2017. However, diagnostic methods as the EDSS have limitations, including the small sensitivity in relatively low scores and the lack of the MS lesions activity information. Therefore, in the last two decades magnetic resonance imaging (MRI) has already become one of the most important and most frequently used imaging tools in the diagnosis and monitoring of MS. In general, T2-weighted MRI is used to determine the MS lesion burden, while active MS lesions can be visualized in vivo by enhanced T1weighted MRI using gadolinium-based contrast agents (GBCA). GBCA-based MRI is used to detect leaky blood-brain barrier (BBB) and white matter lesions, and is commonly used for diagnostic purposes as well as to monitor clinical disease progression and evaluate the patients' response to a given treatment (Geraldes, Ciccarelli et al. 2018).
In general, GBCAs are commonly and widely used to enhance MRI signals to diagnose and monitor different diseases, such as tumor, infection, and bleeding (Goulle, Cattaneo et al. 2009). They have been used for almost three decades and showed a faultless safety record, which profited from the stability depending upon the physicochemical properties.
Structurally, GBCAs are categorized into linear agents and macrocyclic agents. Linear GBCAs, such as Magnevist (gadopentetate dimeglumine) and MultiHance, are openchain and less stable, while macrocyclic GBCAs, including Dotarem, Gadavist et al, fully enclose the gadolinium iron. Linear GBCAs were found to lead to enhanced deposits of gadolinium in the brain compared to macrocyclic GBCAs (Murata, Gonzalez-Cuyar et al. 2016, Ichikawa, Motosugi et al. 2017. These facts led to suspensions or restrictions on linear agents by the European Medicament Agency (EMA) in November 2017 (EMA/625317/2017) and to the publication of new class warnings and recommendations by the FDA. Particular caution is recommended in case of pregnancy or kidney problems and if patients did already receive the MRIs with GBCAs.
As mentioned above, in central nervous system (CNS), GBCAs are extensively used to diagnose and monitor diseases such as multiple sclerosis that are characterized by alterations in the vascular density and/or permeability. However, Kanda and his colleague first reported high signal intensity appearing in the globus pallidus and dentate nucleus of brain tumors patients who had received multiple MRIs with GBCAs (Kanda, Ishii et al. 2014), and it aroused public concern. It indicated Gd deposition in brain tissue rather than washout after multiple GBCAs applications. Additionally, histological evidences in patients with multiple sclerosis, encephalitis, transient ischemic attack, intracranial hemorrhage and traumatic brain injury (McDonald, McDonald et al. 2015, Stojanov, Aracki-Trenkic et al. 2016 further confirmed this observation. However, it remains unclear, whether neuroinflammation and BBB disruption may affect the risk of gadolinium retention inside the CNS. In order to further investigate gadolinium deposition in inflamed brain tissue, in our study, we performed multiple Magnevist (linear GBCA) administrations in experimental autoimmune encephalitis (EAE) mice, the prototypical animal model for multiple sclerosis. In addition, we used T1-weighted MRI to image the brain after repeated applications of Magnevist, histology to show the inflammatory areas, inductively coupled plasma mass spectrometry (ICP-MS) to quantify the Gd retention in the brain and synchrotron-based X-ray fluorescence microscopy (SR-XRF) to map the Gd distribution.

2.1) Experimental autoimmune encephalomyelitis (EAE)
All interventions related to the in vivo EAE model (induction, scoring, weighting, sacrificing) were conducted by S. Wang.
The opportunity to obtain CNS tissue from patients is rare, therefore, animal models have been applied to investigate the pathogenesis of the disease. EAE is the prototypical animal model for MS, which is commonly used (Robinson, Harp et al. 2014). Although MS is a uniquely human disease, many histopathological features can be induced in EAE models following induction of CNS-directed autoimmunity, including demyelination, axonal and neuronal damage as well as multiple lesions distributed in time and space. As in MS, lesions are generally perivascular and more prominent in the brain stem and spinal cord (Rangachari and Kuchroo 2013, Baker and Amor 2014). Thus, EAE is an autoimmune, CD4+ T-cell-mediated disease characterized by the breakdown of BBB, mononuclear cell infiltration, demyelination and neurodegeneration (Robinson, Harp et al. 2014).
EAE can be induced in various species, including rodents and primates, by introducing specific antigens, which ultimately result in motor dysfunction. In mice, active EAE can be induced by immunization with specific myelin peptide epitopes (Stromnes and Goverman 2006). Consequently, myelin-specific T lymphocytes are activated in the periphery and migrate across the BBB into the CNS. After entering the CNS, T cells are activated by local and infiltrating antigen-presenting cells, leading to subsequent inflammatory cascades, the involvement of other cells (such as monocytes or macrophages), and ultimately demyelination and axonal damage (Glatigny and Bettelli 2018). On the other hand, passive EAE is induced by the adoptive transfer of encephalitogenic T cells, which were previously obtained from an immunized mouse and activated in vitro.
As in the MS, the disease course can follow a relapsing-remitting form and a chronic form  In this study, to better clarify the relationship between inflammation and Gd deposition, After immunization, mice were monitored daily for clinical signs and recorded in detail as described in Table 1. Clinical signs in EAE are commonly assessed on a six-stage scale of 0 to 5, with 0 being "clinically normal" and 5 "being died of EAE". To facilitate scoring, we also give mice "in-between" scores when the clinical signs lie between two defined score. The" in-between" scores are well established, and help standardized and establishing the final score, principally when signs of disease are mild (between 0 and 2), and thus, difficult to determine. These in-between grades are, however, not included in the publications to facilitate comparison with other labs around the world Furthermore, clinical signs were scored as follows (s. Table 1

Applications of GBCA were conducted by S. Wang.
To investigate whether Gd deposition is influenced by brain inflammation, healthy control (HC) and EAE mice were exposed for a total of eight applications of a linear GBCA (Gd-DTPA, Magnevist, Bayer, Germany). GBCA was applied into the tail vein at 2.5 mmol/kg during four consecutive days, followed by a two-days pause and four additional consecutive daily injections. In total each mouse was injected with a cumulated dose of 20 mmol/kg Magnevist.
Vasodilation of the tail vein was induced by submerging the tail in warm water for 5 to 10 minutes. Afterward, the intravenous injection of Magnevist was performed with the mouse placed in a sternal position on an appropriate mouse restraining device. The tail was firstly wiped with gauze dampened with 75% ethanol solution to increase the visibility of the vein and to locate one of the two lateral veins in the middle of the tail. A 27-gauge needle with the bevel facing upward was slid into the tail vein at an angle of approximately 30°.
Proper placement of the needle was confirmed by observing a flashback of the blood when applying negative pressure to the plunger. After slowly injecting Magnevist solution through the tail vein, the needle was removed, and the puncture site was gently pressed with a dry piece of gauze until the bleeding stopped. The tail was finally cleaned with disinfectant. When consecutive multiple intravenous injections were needed, strict disinfection procedure and anti-inflammatory therapy with infrared lamp illumination were performed after injection to avoid infection and inflammation of the tail.
In EAE mice, Magnevist application started on day 14 post immunization (p.i.), when EAE mice reached the maximal clinical disability. As depicted in Figure 2, the degree of Gd retention was investigated in vivo using MRI (s. paragraph 2.3). In addition, Gd quantification and location within the CNS was assessed by ICP-MS (s. paragraph 2.6) and SR-XRF (s. paragraph 2.7), respectively.

2.3) Magnetic resonance imaging (MRI)
MRI scans and analysis were conducted by S. Wang.
MRI is a non-invasive and painless technique that utilizes a powerful magnet and computer-generated radio waves to produce detailed and cross-sectional images of tissues, organs and bones of the body. Unlike X-rays and computed tomography (CT), MRI scans have no damaging ionizing radiation of X-rays. It is a common medical technology to improve the diagnosis and monitoring of diseases including blood vessel damage, brain injury, heart disease, bone infection, damage to joints and tumor in various organs (Ibrahim and Dublin 2018).
MRI uses the principle of nuclear magnetic resonance (NMR) phenomenon to detect the emitted electromagnetic waves through a gradient magnetic field based on the different attenuation of the released energy in different structural environments inside the material.
And it provides information about the position and type of nucleus that constitutes this object, which can be used to draw a structural image of the interior of the object (Stikova 2012). Furthermore, the foundation of the NMR is the interaction of a nuclear spin with an external magnetic field, B0. The dominant nucleus in MRI is the proton in hydrogen, which interacts with the external field resulting in the precession of the proton spin about the field direction. In addition, the energy of interaction with B0 depends on the direction of nuclear moments, so that the minimum energy corresponds to the state in which the moments are parallel to B0. As a result, in thermal equilibrium, most of nuclear moments are aligned along the external field. The alignment of magnetic moments gives rise to nonzero magnetization in macroscopic samples of solids, liquids or gases containing a large number of nuclei. The NMR phenomenon is observed when a macroscopic sample in a static magnetic field is irradiated by an oscillating magnetic field with a frequency that equals the frequency of precession (McGowan 2008). Additionally, the MR technique using a combination of 90° and 180° radiofrequency (RF) pulses to generate an echo signal is called the spin-echo sequence, which is one of the basic sequences performed in MRI. Spatially encoded signals in MRI are obtained by the repetitive and selective excitations of the nuclear magnetization in targeted area of the object. After each excitation, in the presence of external magnetic field gradients, the NMR signal is sampled multiple times within a short acquisition interval (Gray, Burstein et al. 2008). Figure 3 illustrates the steps of the MRI. The MRI parameters include T1 relaxation and T2 relaxation times, which are intrinsic properties of different tissues (Pooley 2005  For small animal MRI, magnets commonly operate at a high field strength, in the range of 4.7 -11.7 Tesla, as opposed to the standard clinical range of 1.5 -3 Tesla. The main advantage of the higher field strength is the increased signal-to-noise ratio (SNR), which enables higher resolution acquisition; at the same time, increased image artifacts and limitations related to the specific absorption rate make higher field strength scanning technically more challenging (Denic, Macura et al. 2011). Considering the advantages and limitations, 7-Tesla small-animal scanner is a proper instrument for this study.
Moreover, the MRI sequence should be optimized to obtain images with good resolution MRI requires that the study subjects to be completely immobilized. In order to achieve this, appropriate anesthesia is required during the scan. Thus, mice were anesthetized via face mask with 1.5-2.0% isoflurane (Forene, Abbot, Wiesbaden, Germany) in 30% O2 and 70% N2O, under continuous respiration monitoring using a press-sensitive pad placed in the thorax (Small animal instruction Inc, Stony Brook, NY). Isoflurane concentration was adjusted in order to maintain the respiration rate at about 60 breaths per minute (Ewald, Werb et al. 2011). In addition, the core temperature inside narrowbore scanners is 15-20°C, and a heating system is required to maintain the normal body temperature of mice while in the scanner (Denic, Macura et al. 2011). With the purpose of providing a suitable core temperature, the mice were placed on a bed with circulating heated water to maintain a constant body temperature at 37°C.
To determine if repetitive application of GBCA leads to retention in the brain, in a group of mice, coronal T1-weighted 2D RARE images were acquired before first Magnevist injection and one day after the last injection. Data acquisition was done with ParaVision 5.1 software (Bruker Biospin, Germany).

Mouse perfusion, organ extraction and tissue preparation were performed by S. Wang.
Mice were sacrificed either one day or ten days after the last Gd application. Mice were lethally anaesthetized with 1 ml of a mixture of ketamine (415 mg/ kg) and xylazine (9.7 mg/ kg). As soon as reflexes between the toes were no longer present, the peritoneum was opened, and the aorta was cut above the diaphragm. To proceed to perfusion, a 25gauge needle attached to a 20-cc syringe that has been filled with PBS was carefully inserted into the left ventricle of the heart. Correct placement of the needle was confirmed by minimal resistance and dark red blood beginning to flow from the right atrium. The animal was then perfused with the entire PBS (20-30 ml).
Thereafter, the brains were carefully taken out and were cut sagittally into two symmetrical halves. Half of the brain was postfixed in 4% paraformaldehyde overnight at 4°C and then PBS washed, following 30% sucrose in PBS soaking until the tissue sunk to the bottom. Meanwhile, the other half of the brain was stored in liquid nitrogen for elements trace determination (ICP-MS). Hemisphere tissues for histology and SR-XRF were embedded in Tissue Tek Optimal cutting temperature compound and frozen very fast in 2-methylbutane cooled with dry ice, then stored in -80°C. Afterwards, those tissue samples were cut horizontally in a cryostat, into 10 µm thick consecutive cryosections and stored at 4°C. In addition, three of the consecutive cryosections from each mouse were directly mounted between two ultralene foils (SPEX sample prep) for SR-XRF mapping, descripted below.

H&E staining was performed by S. Wang.
H&E staining is the most common staining technique in histopathology, with a combination of two dyes, the eosin dye staining cytoplasm and extracellular matrix red or pink and the hematoxylin dye staining nucleus a purplish blue. It is widely used to determine the infiltrating inflammatory cells. In this study, to detect brain regions showing inflammation, frozen hemisphere sections were stained with hematoxylin and eosin.
Frozen hemisphere sections were thawed at room temperature for 30 minutes, then Eventually, images were acquired by a Zeiss Axio Observer microscope.

2.6) Inductively coupled plasma mass spectrometry (ICP-MS)
The ICP-MS was performed by M. Roman using the brain hemispheres provided by S.

Wang.
ICP-MS is an analytical technique intended for elemental determinations, which is able to detect elements at very low concentrations with high accuracy. Over the past four decades, it has been established as the most reliable technique for quantifying elements in diversified samples with a wide working range of concentrations and low interferences.
ICP-MS enables very sensitive, accurate and precise multi-elemental analysis of trace and ultra-trace elements and of isotope ratios. Therefore, it has developed into one of the most important mass spectrometric techniques for studying biological components and environmental materials. (Mittal, Kumar et al. 2017) ICP-MS consists of an ion source, a sampling interface, an ion lens, a mass spectrophotometer (Balcaen, Bolea-Fernandez et al. 2015). The ion source is an ideal ionization source for mass spectrometry analysis, and it is capable of ionizing over 90% of many elements. The ions generated in the ion source are introduced into the mass analysis unit through a sampling interface, which is composed of a sampling cone and a

2013)
Furthermore, ICP-MS has evolved to become a sensitive tool for Gd-specific detection and quantification in various tissues (Frame and Uzgiris 1998, Telgmann, Faber et al. 2012, Sato, Ito et al. 2013, Sato, Tamada et al. 2015. Thus, we chose ICP-MS to identify and quantify the elements, especially Gd, in brain tissues. In this study, we used ICP-MS to quantify K, P, Na, Ca, Fe and Gd in the hemisphere tissue of four healthy control (HC) and four EAE mice, which were sacrificed either one day (n=2) or ten days (n=2) after the last injection. One hemisphere was selected for each sample, placed into a 2 mL Eppendorf tube, immediately frozen and shipped to ECSIN Lab. Hemispheres were firstly coded and dried to constant weight in vacuum at 20℃ overnight, using a Concentrator Plus (Eppendorf) in V-AQ mode. Ten volumes of ultrapure grade concentrated Nitric Acid (HNO3) were directly poured into each tube, then samples were placed into a thermostatted block (Falc) at 70℃ for 8h to achieve complete mineralization of the tissue. After cooling, the digests were transferred into 15 ml tubes, pooled with triplicate rinse of the original tubes using 1 ml of Milli-Q water each, spiked with the internal standards Sc and Rh, and brought to 10 ml with Milli-Q water (final concentration of Sc and Rh 10 ng/g). The solutions were directly analyzed for determination of K, P, Na, Ca, Fe and Gd, and further diluted with ultrapure HNO3 2% (v/v, plus 10 ng/g of the internal standards) 1:50 for the determination of Ca and Fe; or 1:1000 for the determination of K, P and Na.
Multi-elemental analyses were conducted by ICP-MS using an instrument Agilent 7900 equipped with ASX-500 autosampler ( were performed for all elements within each analysis. The wet-to-dry mass ratio of brain samples ranges between 4.1 and 4.5, consistently with the average value of human brain regions (Krebs, Langkammer et al. 2014). Total mass concentration of Gd and other elements in the hemisphere sample was reported as a ratio of the dry weight of the hemisphere sample, and the results were reported in the published paper.

S. Wang went to ESRF to image the retention of gadolinium using SR-XRF with the help of Dr. Bernhard Hesse. Analysis of SR-XRF data was mainly performed by B. Hesse.
Therefore, the details of analysis were not be include in this manteltext.
X-ray fluorescence spectroscopy is a non-destructive instrumental method of qualitative and quantitative analysis for chemical elements, particularly used in the investigation of metals, glass, ceramics and building materials, as well as in biological tissues, based on the intensities and energies of the emitted X-rays. A synchrotron machine aims to accelerate electrons to extremely high energy and then make them change direction periodically. The resulting X-rays are emitted as dozens of thin beams, each directed toward a beamline next to the accelerator. The essential workflow is displayed in Figure   4. SR-XRF is a label free and multi-elemental analyzed method with highly intensive Xrays generated by a synchrotron radiation (SR) excitation source, which can produce Xrays 100 billion times brighter than the X-rays used in hospitals. Due to its small source size, low divergence, high photo flux and linear polarization, SR is an ideal source for high-precision XRF analysis of heterogeneous and complex materials. Thus, SR-XRF is capable of focusing on a very small region of interest, and it can be used for mapping with high spatial resolution without destroy the samples, from millimeters to nanometers (Bohic, Cotte et al. 2012). In this study, we used SR-XRF to map the distribution pattern of Gd depositions and to reveal potential co-localizations with other elements in the mouse brain. We cooperated with and obtained a grant from the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. ESRF hosts one of the most intense and brilliant hard X-rays sources worldwide, and it is a high energy electron accelerator with an electron energy of 6.02 GeV, giving information about the position and motion of atoms and thus exposing the structure of the tissue. Besides, it is an extremely time-consuming method, and it took around 75 hours in ESRF to obtain the mapping images of eight mouse brains. However, each map is composed of many hundred thousand of individual spectra, each spectrum corresponds to one position of the sample.
We performed SR μXRF at the ID21 beamline, ESRF (Cotte, Pouyet et al. 2017), which e application aims to identify and localize unknown components in complex mixtures with sub-micrometric resolution and millimetric field of view. The components of ID21 beamline are presented in Figure 5 and includes four complementary end-stations and two separate branch-lines: the direct-branch-line hosts (i) the scanning X-ray microscope optimized for two-dimensional (2D) micro-spectroscopy end-station and (ii) the full-field X-ray absorption end-station. The side-branch-line hosts (iii) the scanning X-ray microscope optimized for 2D micro-X-ray diffraction (μXRD) and μXRF end-station; and (iv) the scanning micro-Fourier-transform infrared end-station is independent from the Xray source. In this study we only used the direct-branch-line hosting the scanning X-ray micro-spectroscopy.
The scanning X-ray micro-spectroscopy end-station, exploiting micro fluorescence, operates in the 2.0-9.2 KeV energy range for the acquisition of 2D μXRF mapping. This energy range gives access not only to low-Z elements (chemical elements with a low atomic number of protons in the nucleus) such as sulfur (S), but also to most of the 3d transition metals such as manganese (Mn), iron (Fe), cobalt (Co), and copper (Cu).
Heavier metals such as tin (Sn), antimony (Sb), lead (Pb) or gadolinium (Gd) can also be analyzed, through their L-or M-edges. It works under vacuum condition; samples are mounted into the scanning X-ray microscope using a transfer chamber and a transfer cane, which allows the main microscope chamber to be kept under vacuum during sample exchange. Besides, a vibration-free cryogenic stage is available for performing cryomicroscopy. Samples can be studied under cryogenic conditions, allowing the study of frozen hydrated biological specimens (cryo-sectioned tissues). It offers the 2D μXRF for elemental mapping, and single point micro-X-ray near-edge structure (μXANES) spectra for punctual chemical speciation mapping with sub-micrometric resolution. The beam can be focused down to ~0.3 μm (vertical) × 0.7 μm (horizontal), with a flux of 10 9 to 10 11 ph/s by means of Kirkpatrick-Baez mirror system. To start the experiment, the regions of interest (ROIs) were determined according to the results of MRI scans and H&E staining performed by S. Wang. Slightly enhanced signal in T1-weight images was found in the cerebellar region of the EAE after multiple GBCA applications, as well as inflammatory lesions notably located in cerebellum. As a result, we choose cerebellar area as the interest region. Then SR μXRF maps were performed using the in-vacuum scanning X-ray spectroscopy setup at ID21 beamline, ESRF.
In the beginning of the mapping prior definition of the ROI, Gd L-edge XANES from Gdreference materials was collected. The spectra collected of the Gd from the brain tissue was then related to this reference values. As reference, we also measured the original contrast agents and the contrast agent in different buffer solutions. To start the SR-XRF, the beamline was optimized for highest possible flux at an energy of 7.3 keV. This energy excited the Gd L lines well, while visualizing the relevant elements K, P, S, Ca and Fe in tissue, and its co-localization with Gd was essential for understanding the distribution of Gd and the origin of Gd cluster formation. The beam was focused down to ~0.6×0.8 µm² (vertical × horizontal). The flux was ~5×10 10 photons/s (~180 mA SR current in multibunch mode). Acquisition time per pixel was 100 milliseconds. Afterwards, eight tissue sections from the mouse brain were mounted and investigated, four from healthy mice and four from EAE mice, in two different time points after GBCA applications (1 day or 10 days). For each section, a region of about 4000×4000 µm² was scanned at 35 µm spatial resolution to allocate the ROI, where subregions were mapped at 10 µm, 3 µm and 0.5 µm resolution. In total, each section required around 8-10 hours for the beamline setup, the ROI selection, the sample mounting, and data collection. Following data were mainly analyzed by B. Hesse using PyMCA.
In addition, to better map the Gd distribution in a high resolution, mouse #3 (M3, EAE mouse sacrificed 10 days after last GBCA application) was analyzed by means of nano-XRF at the ID16B, ESRF (Martinez-Criado, Villanova et al. 2016). This part of study was performed by B. Hesse. The method and analysis details were descripted in the published paper.

3) The significant new results
We found that after repetitive application of linear GBCA into healthy and EAE mice, Gd deposited more strongly in the inflamed brains.
In addition, using micro-and nano-SR-XRF, we demonstrated that although Gd hotspots were observed in both, healthy and inflamed brains, hotspots were more abundant in the inflamed tissue, particularly inside the cerebellum. Quantification of the total amount of Gd in brains from EAE and healthy mice by ICP-MS confirmed an increased Gd retention in the inflamed CNS.
Thus, inflammation and disrupted BBB may favor Gd retention.

4) The resulting clinical applications and future prospects
Our results confirmed that after repetitive applications of linear GBCA, Gd is retained in the CNS tissue. Importantly, we show that ongoing inflammation favors the deposition. It is known that Gd could bind to other macromolecules (Robert, Fingerhut et al. 2018) via the so-called transmetallation reaction, where Gd+3 is bumped out by another endogenous cation (such as Zn+2 or Ca+2). Dissociated free Gd+3 are less stable than chelated Gd in vivo, and they are prone to combine with other endogenous anions (such as phosphate, hydroxide, carbonate or citrate) and deposit in the tissue (Morcos 2008).
Those new synthetic and unknown structured chelated Gd molecules would be continuously accumulating in tissue and highly probably be toxic. In addition, GBCA are extracellular molecules under normal conditions, but the dissociated free Gd+3 which is kicked out by other cations could bind the phosphate acid from the cell wall and get internalized over time (Hao, Ai et al. 2012). Apparently, free gadolinium (Gd3+) is toxic to mammals, but the mechanism of this toxicity is unclear.
Furthermore, a various of short-term symptoms were described related to Gd deposition, called "Gadolinium Deposition Disease", which were observed in both linear and macrocyclic GBCAs (Semelka, Ramalho et al. 2016). These symptoms include tight or sharp pain in the extremities, as well as bone pain, persistent clouded mentation, headaches, and arthralgias (McNamara and Rahmani 2018, Semelka, Ramalho et al. 2018). Those might be the toxic effect of Gd, but not identical, common or severe. Thus, the cause relationship between Gd deposition and chronic effects is not fully recognized, and the side effects of the Gd deposition are still questionable (Tedeschi, Caranci et al. 2017). The clinical and biological significance of deposited gadolinium in the brain remains unknown. There are no signs of harm from gadolinium exposure in animal models, and no behavioral changes have been reported in animals undergoing repeated gadolinium administrations (Robert, Lehericy et al. 2015, Gulani, Calamante et al. 2017. As motioned, linear GBCAs are less stable than macrocyclic GBCAs and the use of linear GBCAs are now regulated or warned in Europe (EMA) and USA (FDA), but this is not the case for many other countries, e.g. China. Therefore, the use of linear GBCA remains an alarm for the world health. This study should contribute to the worldwide prohibition of linear GBCAs and to encourage the research on potential retention and long-term side effects of the macrocyclic forms of GBCA.
Additionally, in MS, clinical relapses correlate with the development of perivascular inflammatory lesions inside the CNS (Barkhof and Scheltens 2002). Lesion monitoring is commonly required to make an accurate diagnosis and to monitor disease progression and response to the treatment. Typically, new lesions are visualized using GBCAs on MRI (Tourbah and Berry 2000). After GBCAs application, gadolinium goes inside the brain through the disrupted blood-brain barrier (BBB) caused by neuroinflammation and the MS lesions are enhanced in the MRI images (Tommasin, Gianni et al. 2017).
Generally, the more severe the inflammatory activity, the greater the burden of GBCAenhancement on post-contrast T1-weighted scans (Kaunzner and Gauthier 2017).
Therefore, GBCAs enhanced MRI is commonly used for diagnostic purposes and to evaluate the patients' response to a given treatment. Consequently, MS patients are commonly exposed to multiple lifetime doses of contrast agents.
Therefore, a comprehensive risk assessment on the use of linear GBCAs should take into consideration that patients suffering from chronic neuroinflammatory disorders may have an increased risk of gadolinium retention after being exposed to repetitive GBCA applications. Further research is needed to confirm if macrocyclic GBCAs are also increasedly retained in the inflamed brain. More to the point, for future clinic and research, in patients that need GBCAs-MRI, we should obtain the history on possible previous GBCAs administrations and evaluate the need of linear GBCAs administration even more critically than ever. Besides, there is a need to deeply investigate the underlying mechanism of the gadolinium deposition in tissue. To clarify how linear GBCAs deposited in various tissue, and to illuminate the toxicity or side effects of the deposition.