2.1 Sample preparation

The experiment was conducted in the Experimental Zoology Department of Central South University, Changsha, China, and was in line with Central South University experimental welfare ethics review standards. Male Specific Pathogen Free (SPF) Sprague Dawley rats weighing 250-280 g were used for establishing models and provided by Laboratory Animal Center of Central South University. The intraluminal suture method [2] was used to establish stroke rat models with the right middle cerebral artery (MCA) occlusion, that is, the median anterior cervical incision first, and then exposing the right common carotid artery bifurcation, inserting a standard nylon thread from the common carotid artery near the bifurcation into the internal carotid artery with a depth of approximately 18±0.5 mm. According to the description of Longa et al., neurobehavioral experiments were performed immediately after surgery in rats to evaluate the effect of occlusion [21]. Following the principle of random allocation, 48 rats were randomly assigned to 12 groups, including the sham operation group and groups with occluded right middle artery for 3 h, 4.5 h, 6 h, 12 h, 1 d, 3 d, 5 d, 7 d, 10 d, 14 d, and 28 d, respectively, based on duration of occlusion. During the feeding period, 1 rat in the 10 d group, 2 rats in the 14 d group, and 2 rats in the 28 d group died of a stroke. To ensure that at least 3 rats were involved in each group, a total of 2 rats were randomly added into the 14 d and 28 d groups. In total, 50 rats were used and 45 were included in our study. After removing the intraluminal suture from the 45 successful models, the models were anesthetized and subsequently sacrificed by perfusing with 0.9% saline and ice-cold paraformaldehyde successively. The separated intact brain was fixed with paraformaldehyde in 0.1 M phosphate-buffered saline (pH 7.4) overnight and subsequently dehydrated in 30% sucrose for 24 h. The most successful brain samples were collected in each group, with the principle of complete perfusion, no visible blood remaining, and a rather stiff surface. The collected sample was immediately frozen and was subsequently sectioned in the coronal plane using a cryostat to obtain ten 60-μm-thick slices (containing the hippocampus). For analysis by SR-μXRF, the samples were finally dried in air to a fully dehydrated state and placed on 3-μm-thick Mylar membrane. To reduce errors, the entire sample preparation process was strictly regulated to avoid contamination with Fe, Zn, calcium, magnesium, and Cu.

2.2 Synchrotron radiation-based micro-X-ray fluorescence analysis (SR-μXRF)

SR-μXRF combines fluorescence X-ray analysis and X-ray microscopy. Fluorescence analysis is based on the X-ray fluorescence generated by the photoelectric effect of the interaction between the X-ray and the substance, and the information of the sample is analyzed. The photoelectric effect is observed when the photon energy of X-rays is greater than the electron binding energy of the object atoms. After absorbing the photon energy, the inner electron of the atom is finally transferred to the high energy level orbit, and the electron of the outer layer transfers to the electron hole of the inner layer to generate a fluorescent signal [2]. Therefore, within a certain range of X-ray photon energies, the fluorescence signals emitted by different elements are different according to the different electron binding energy and electron arrangement of each elemental atom [1, 22]. In this experiment, the scanning analysis of the distribution of Fe, Zn, calcium, magnesium, and Cu in the brain tissue by SR-μXRF was performed at beamline 15U station in Shanghai Synchrotron Radiation Facility (SSRF, Shanghai, China). The semiquantitative experiment was performed in air at room temperature with a photon energy of 10 Kev and a monochromator angle of 45 degrees. The spot size of the beam on the sample stage was 50 × 50 μm², with a focusing time of 1 second per point, a step size of 50 μm for raster scanning, and a silicon detector for collecting fluorescence counting of 90 degrees. A light microscope was coupled to a computer for sample viewing. Each pixel was extracted from the energy dispersive spectrum using the GeoPIXE software (CSIRO, Australia), and the intensity map of the metal was imaged using Igor Pro software (WaveMetrics, USA) to quickly and accurately obtain two-dimensional, in situ micro-amounts of brain tissue element content and distribution. In experiments, Compton Scattering in X-ray spectra was used as an internal standard to compensate for the differences in thickness and density between the thin sections of tissue.

2.3 Data Analysis

Due to the high heterogeneity of elemental distribution in the brain-wide scale, the relative signal value of all the scanning spots (containing all the duplicate samples) was collected from the control side and the experimental side of the brain, thus dividing the data into two parts. Statistical analysis was performed using the Statistical Package for the Social Sciences version 20.0 (International Business Machines Corp., Armonk, NY). A nested t-test was conducted to compare the difference between the sham group and different post-MCAO time point groups. P<0.05 was considered to be significantly different. Data are presented as the mean±standard deviation (SD).

3.1 Iron fluorescence mapping

According to the image (Fig. 1), the distribution of Fe element in the sham group was more even and the average value was higher in the vicinity of the cortex. However, in the cerebral ischemia group, the average value of Fe was higher in the cortex and lesion area. The content of Fe in the lesion area evidently increased early and subsequently decreased over time. In the short-duration groups, the mean value at 3 h increased and slightly decreased at 4.5 h and 6 h, and distribution means at 12 h and 1 d increased in the local region, specifically in the lesion area. In the long-duration groups, the content of Fe element in 1 d to 5 d showed a recovering and decreasing trend and in 7 d–14 d continued to decrease to a smaller volume of lesion hemisphere. In the long-duration experimental groups, the fluctuation of Fe in the continuously perfused left brain tissue region was not evident when compared to the lesion side, i.e., side with interrupted blood supply.

Fig. 1Full-screen View

(Color online) Image of iron (Fe) distribution in the brain tissues at different post-middle cerebral artery occlusion (MCAO) time points by synchrotron radiation-based micro-X-ray fluorescence analysis. (a) Fe distribution in brain tissue of sham-operated rat. (b-l) Fe distribution in rat brain tissue at post-MCAO 3 h, 4.5 h, 6 h, 12 h, 1 d, 3 d, 5 d, 7 d, 10 d, 14 d, and 28 d, respectively. The sagittal plane of a rat brain demonstrated the position from which the slice was obtained. The ischemic lesion was on the right hemisphere marked with "R."

3.2 Zinc fluorescence mapping

According to the image (Fig. 2), Zn distribution in the sham group was average compared to the other groups, and the mean was lower, while Zn in the cerebral ischemia group was not evenly distributed. In the short-duration groups, the distribution of Zn was more balanced, and the mean elemental content was higher than that in the sham group. In the long-duration groups, the means of Zn in the lesion area fluctuated over time. The results showed that the Zn content in the 1d group increased, after which the Zn content in the 5 d group increased and exceeded that in the 3 d group. Subsequently, in the 7 d group, the Zn content decreased, and the Zn content in the 14 d and 28 d groups decreased successively. The Zn content in 28 d group was significantly low, specifically in the center of the lesion. In contrast, the fluctuation of the Zn content on the normal side was small.

Fig. 2Full-screen View

(Color online) Image of zinc (Zn) distribution in the brain tissues at different post-middle cerebral artery occlusion (MCAO) time points by synchrotron radiation-based micro-X-ray fluorescence analysis. (a) Zn distribution in brain tissue of sham-operated rat. (b-l) Zn distribution in rat brain tissue at post-MCAO 3 h, 4.5 h, 6 h, 12 h, 1 d, 3 d, 5 d, 7 d, 10 d, 14 d, and 28 d, respectively. The sagittal plane of a rat brain demonstrated the position from which the slice was obtained. The ischemic lesion was on the right hemisphere marked with "R."

3.3 Calcium fluorescence mapping

According to the image (Fig. 3), the distribution of Ca in the sham group was more balanced, its distribution in the cerebral ischemia group was unbalanced. The distribution of Ca was mainly on the lesion side. In the long-duration groups, the content continued to increase, and the highest content was observed in the 28 d samples. In the short-duration groups, the means and distribution of Ca did not have significant change compared with the sham group. However, the means and distribution of Ca seemed to increase slightly in the lesion area, which was close to the 1d group. The means of Ca content in the normal side were all lower than that in the lesion side in the same sample.

Fig. 3Full-screen View

(Color online) Image of calcium (Ca) distribution in the brain tissues at different post-middle cerebral artery occlusion (MCAO) time points by synchrotron radiation-based micro-X-ray fluorescence analysis. (a) Ca distribution in brain tissue of sham-operated rat. (b-l) Ca distribution in rat brain tissue at post-MCAO 3 h, 4.5 h, 6 h, 12 h, 1d, 3 d, 5 d, 7 d, 10 d, 14 d, and 28 d, respectively. The sagittal plane of a rat brain demonstrated the position from which the slice was obtained. The ischemic lesion was on the right hemisphere marked with "R."

3.4 Potassium fluorescence mapping

According to the image (Fig. 4), the distribution of K in the sham group was more even. In the experimental group (3 h, 4.5 h, 6 h, 12 h), the average values of K decreased slightly, and the distribution did not have significant changes compared with the sham group. Over time, the means of K increased significantly on the lesion side in the 1d group and decreased again in the 3d group. The content of cortex K in the 5 d, 7 d, 10 d, 14 d, and 28 d groups showed a slight increase when compared with the ahead groups, with distribution seen on the lesion areas, while the content in the lesion core area (striatum) was still relatively low. However, compared with the other elements, the increasing degree of means in every group was not significantly high.

Fig. 4Full-screen View

(Color online) Image of potassium (K) distribution in the brain tissues at different post-middle cerebral artery occlusion (MCAO) time points by synchrotron radiation-based micro-X-ray fluorescence analysis. (a) K distribution in brain tissue of sham-operated rat. (b-l) K distribution in rat brain tissue at post-MCAO 3 h, 4.5 h, 6 h, 12 h, 1d, 3 d, 5 d, 7 d, 10 d, 14 d, and 28 d, respectively. The sagittal plane of a rat brain demonstrated the position from which the slice was obtained. The ischemic lesion was on the right hemisphere marked with "R."

3.5 Copper fluorescence mapping

According to the image (Fig. 5), in the sham group, the mean of Cu was higher around the median sagittal line. In the short-duration group (3 h, 4.5 h, 6 h, 12 h, and 1 d), the expression of Cu increased extensively in the brain tissue of the experimental groups. From the 3^rd day, the overall Cu content decreased, and the lesion side was lower than the control side.

Fig. 5Full-screen View

(Color online) Image of cupper (Cu) distribution in the brain tissues at different post-middle cerebral artery occlusion (MCAO) time points by synchrotron radiation-based micro-X-ray fluorescence analysis. (a) Cu distribution in brain tissue of sham-operated rat. (b-l) Cu distribution in rat brain tissue at post-MCAO 3 h, 4.5 h, 6 h, 12 h, 1 d, 3 d, 5 d, 7 d, 10 d, 14 d, and 28 d, respectively. The sagittal plane of a rat brain demonstrated the position from which the slice was obtained. The ischemic lesion was on the right hemisphere marked with "R."

3.6 Quantitative analysis

To quantitatively evaluate the difference between the sham group and the several post-MCAO time point groups, we collected the relative signal value of the elements from all the scanning spots due to the high heterogeneity of the data in the brain-wide scale. The collected data were subsequently divided into two parts, the control side and the experimental side, to take into account the different conditions endured by the two brain hemispheres. The data were presented using a histogram (Fig. 6). The statistical evaluation showed a high SD value in almost all the groups and confirmed the heterogeneity of element distribution in the brain. In the whole-brain scale, the content of Fe, Ca, and Cu was relatively stable after MCAO operation, and even the local change was significant (Figs. 1, 3, and 5). It also indicated that K and Zn changed more significantly than other elements during the progression of MCAO. The content of K changed significantly in almost all the time points after MCAO, and Zn mainly changed after 3 d.

Fig. 6Full-screen View

(Color online) Quantification of the relative signal value of various elements at different post-middle cerebral artery occlusion (MCAO) time points. (a–e) The quantification of iron, zinc, calcium, potassium, and copper on the control side and experimental side of the brain at different time points after MCAO. "C" and "E" represent the control side and experimental side, respectively. Nested t-test was employed to test the difference between the post-MCAO groups and the sham group on the whole-brain scale. ^*P<0.05, ^**P<0.01, ^***P<0.001 vs. the sham group at each time point.

4.1 Iron

Fe, one of the essential elements in the body, is a prosthetic group of neurotransmitters and important myelin protein synthase. Fe ion is also a catalyst that significantly increases the concentration of reactive oxygen species (ROS). Studies have shown that after Fe aggregation, a large number of hydroxyl radical generated by Fenton reaction can promote the peroxidation of membrane lipid and the production of ROS, leading to cytotoxic damage. Studies have found that neurons have Fe ion metabolism disorders in Alzheimer’s disease (AD) [24]. The activation of nuclear factor kappa-light-chain-enhancer of activated B cells leads to increasing divalent metal transporter 1 expression, increasing Fe influx and aggravating neuronal damage after cerebral ischemia [25]. Moreover, with cerebral hemorrhage, hemoglobin degrades to heme Fe, thereby promoting the production of ROS [26]. Furthermore, intravenous injection of Zn (2+) into Wistar rats can reduce the Fe accumulation after cerebral ischemia, which can cause neuronal damage, and the Fe chelator deferoxamine can reduce the severity of myocardial infarction induced by Fe aggregation after cardiac ischemia-reperfusion [27]. Studies have shown that there are two main sources of Fe in the lesion area: (1) hypoxia-ischemia that damages the blood-brain barrier and the free Fe ions that directly leak into the brain instead of through the involvement of transferrin, and (2) ischemia leads brain cells to reduce the regulation of Fe storage protein, resulting in increased concentration of free Fe ions in the brain tissue. In hypoxic conditions, the free ferric ions are reduced to ferrous ions, and the ferrous ions undergo Haber–Weiss reaction to generate hydroxyl radicals, resulting in oxidative stress and subsequent cell apoptosis and tissue infarction [23, 28, 29]. Moreover, some studies have shown that the protection of the blood-brain barrier after ischemic stroke and the use of some lipid-soluble Fe chelators can reduce the infarct range and ROS production [30].

Fe content of lesions in elemental distribution map increased first and subsequently decreased, which suggests that ischemic brain tissue oxidative stress response increased first, and compensatory repair was subsequently observed. Recent studies have found that prophylactic knocking out of the Tau gene that mediates Fe export has protective benefits on brain tissues after MCAO [31]. Combined with our results, they jointly imply that early clinical intervention to increased local Fe content is of great importance.

4.2 Zinc

Various studies have shown that during cerebral ischemia, Zn has the dual role of neurotoxicity and neuroprotection. Selective death of hippocampi 1 neurons during cerebral ischemia is caused by the release of Zn-containing vesicles at the terminal axon of excitatory neurons to fragile postsynaptic neurons [32]. Zn can affect adenosine triphosphate consumption and the oxidized form of coenzyme NAD⁺ to disrupt the different phases of cellular respiration. Moreover, the use of Zn chelators reduces neuronal death [33, 34]. Successive studies have shown that in the early stages of ischemia, vulnerable neurons upregulate the expression of Zn transporter protein-1 (a local plasma membrane transport protein that promotes Zn efflux). However, contrary to chelation-based therapeutic interventions, various studies have demonstrated the neuroprotective benefits following various Zn compounds [35]; hence, Zn should have neurotoxic and neuroprotective dynamic balance in a postischemic setting. It is hypothesized that this may be related to the fluctuation of Zn content distribution in postischemic lesions (the specific changes are shown in the experimental results).

4.3 Calcium

The mechanism of Ca ion cytotoxicity is still controversial. There are mainly intracellular "calcium overload" hypothesis and "source-specific" hypothesis, both of which are related to the entry of calcium ions from the extracellular and endoplasmic reticulum, mitochondria, into the neurons’ cytoplasm after cerebral ischemia. As a neurotransmitter, glutamate develops excitotoxicity in ischemic environment. It can overstimulate receptors such as N-methyl-D-aspartic acid receptor (NMDA) and allow a large influx of Ca ions [36, 37]. Studies have shown that the use of NMDA receptor blocker BQ-869 can reduce intracellular Ca concentration and reduce the infarct range of the tissue [38]. Membrane disintegration of the endoplasmic reticulum and mitochondria or dysfunction of membrane ion channels releases Ca into the cytoplasm [36]. Increased intracellular free Ca has become an important cause of cell death. With the extension of time, in the elemental distribution map, the lesion areas have high Ca content, specifically in the 28 d group. It has been hypothesized that this distribution may be related to the formation of calcification, which may be related to ischemic necrosis and increased alkaline phosphatase of the local tissue.

4.4 Potassium

Some studies have shown that during the edema phase of the ischemia, the blood-brain barrier is destroyed, Na ions flow in, and K ions flow out in large quantities [39]. The change of Na and K ions inside and outside the neurons contributed to pure cytotoxic edema in the ischemic hemisphere [40, 41]. This may be related to the early reduction of the K content in the lesion areas of the element distribution map. This is consistent with Sentaro et al.’s findings. The blockade of K-related ion channels plays an important role in controlling neuronal hyperactivity after ischemia [42]. Ischemic stroke accounts for a higher risk of epileptogenesis [43], which may also be highly connected with K and cortical spreading depolarization. Compared to the control groups without post-stroke epilepsy, the post-stroke epilepsy group showed a significant reduction in Kv4.2, a type A K channel subunit [44]. Additionally, cortical spreading depolarization and epilepsy often occur simultaneously after ischemic stroke. The interplay between cortical spreading depolarization and epilepsy is also complicated. The contradiction of which one initiates the other electrophysiological phenomena still existed [45]. It has the potential to alter the threshold of epileptogenesis by manipulating the occurrence of cortical spreading depolarization, which may reduce the epilepsy-associated sequelae [46]. Furthermore, serum potassium levels significantly affect the prognosis of ischemic stroke patients [47]. In the elemental distribution map, the content of the lesion core is lower, while the cortex and around the lesion core shows an increasing trend, which could be, respectively, explained by the flow out theory and compensatory theory.

4.5 Copper

Cu is a strong oxidant, and its effects include phospholipid synthesis, hemoglobin formation, and electron transport. Cu is a trace element necessary for several key processes in the brain [48]. Although studies of chronic neurodegenerative diseases such as AD have shown that Cu homeostasis of the brain has been disturbed, little is known about Cu homeostasis of brain tissue in acute ischemic stroke. Recent studies indicate that Cu transmission plays an important role in the regulation of immune response in the brain. CuII (atsm), a type of Cu bis (thiosemicarbazone) complex, reduces the damage of the postischemic immune response to the brain tissue by influencing the immunogenic effects of several cells and activity of microglia [49, 50], which has a great potential for the treatment of ischemic stroke. In the elemental distribution map, the changing trend of Cu in lesions after cerebral ischemia is similar to that of Fe; thus, the oxidative stress reaction caused by Fe element may be correlated with the immune response of Cu, increasing and decreasing synchronously.

By SR-μXRF, this experimental result showed that after an ischemic stroke, the in situ steady state of Fe, Zn, Ca, K, and Cu in brain tissue is imbalanced. Combined with previous research results, the changes in these elements aim to provide a new idea and target for the study of ischemic stroke intervention. For the early increase of Fe content, clinical intervention should be performed as soon as possible, or intervention of patients with stroke prognosis should be performed in advance, to avoid a series of oxidative damages of Fe after stroke. The content of Cu relatively presented a similar trend of fluctuation compared to Fe, that is, the initial increase and the following decrease of the content of the two elements. The specific association between Cu and Fe homeostasis in the human body is complicated; hence, dyshomeostasis of the two elements is observed. It has been reported that Fe concentration decreases after feeding a Cu-deficient diet in a rat model [51]. Interestingly, a similar changing trend of Cu and Fe has been observed in our results. However, it has also been reported that the lack of ceruloplasmin, a type of Cu-binding protein, causes Fe to accumulate in the pancreas, retina, and brain [52]. This indicated that in different pathological processes, the metabolism of Cu and Fe can vary greatly, even in the opposite manner. The exact association between the two elements in the brain remains elucidated, specifically after ischemic stroke. Thus, the research about which role Cu plays to Fe during ischemic stroke is significantly valuable. If there is certainly an antagonistic effect of Cu, this can provide a new target for the clinical treatment of stroke. The experiment also found that the content of potassium, consistent with the results of other studies, decreased early, after which it has slightly increased. Potassium is an important ion involved in the resting potential and action potential of the neuron. The specific mechanism of later increased change needs to be further explored.

The statistical analysis also provided interesting information about the progression of MCAO. It confirmed that the distribution of different elements was imbalanced because of a rather high SD value in almost every group. Additionally, the nested t-test was conducted to compare the difference of element content between the sham group and the different post-MCAO time point groups. The results showed that the content of potassium significantly changed in almost all the time point groups, specifically in the early phase of MCAO (3 h, 4.5 h, 12 h), indicating its essential role in initiating the pathology of ischemic stroke. Moreover, the content of Zn also changed significantly at the time points of 3 d, 7 d, 10 d, and 28 d, suggesting its important role in the later period of ischemic stroke. Regarding the shortage of our statistical analysis, the nested t-test can only test the difference between the sham group and the experimental groups on a brain-wide scale and caused the neglect of the local information. For example, there are several "heat zones" and "cold zones" presented in Figs. 1–5, indicating the local enrichment or depletion of these elements. Our results showed that the content of Fe and Ca did not change significantly in the whole-brain scale after the MCAO. Thus, the local enrichment of Fe and Ca may be related to the functional or pathological change of the corresponding encephalic region. Additionally, the hot spots in K and Zn images may be caused by both the functional change and the change of inflow or outflow of the elements in the brain. Due to the limited sample size in our present study, further study that determines the exact association between these heat and cold zones and the functional change of specific encephalic regions should be conducted. We believed that this local information was also important to decipher the progression of ischemic stroke.

4.7 Future perspective of SR-μXRF

SR produces rather strong X-rays, which are several orders of magnitude higher in intensity than laboratory X-ray sources. It can focus on the micron or even nanometer scales and can be used to observe cellular or intracellular micro-element distribution [53, 54]. Considering its nondestructive feature to the biological tissue, the SR-μXRF method used in the analysis of metal elements in biological tissues is considered as one of the most reliable methods for quantitative and in situ analysis of trace elements in biological systems, providing strong technical support for some diseases, specifically of the nervous system [55]. However, several problems are still to be addressed to optimize the efficiency of the scanning using SR-μXRF, for instance, accelerating the acquisition process to increase data acquisition efficiency and developing smaller beam sizes to enhance resolution without reducing the visual field [56]. Additionally, the radiation damage also existed during the scanning, which should be avoided by choosing proper scanning parameters. Previous studies have mentioned significantly possible utilization that may accelerate the finding of targets. For example, Epaule et al. used SR-μXRF technology to detect and map the distribution of a nanoparticle-packaged drug and confirmed that SR-μXRF is sufficiently sensitive to achieve the results [57], which may also be used to detect the distribution of brain-targeted medicine. Another example is that Wang et al. combined the immunogold labeling and SR-μXRF to detect the distribution of certain proteins including APP and Aβ42 in the brain [58], which indicates that SR-μXRF has the potential to combine with other technologies to increase the types of detection targets. These results suggest that SR-μXRF has the potential to assist in cerebrovascular disease, including ischemic stroke, directly or indirectly. Similar to ischemic stroke in this study, the disruption of homeostasis after the onset of the disease is closely related to the successive pathophysiological processes in brain tissue, such as oxidative stress. When conducting in-depth studies in the future, we will use SR-based three-dimensional (3D) micro-fluorescence imaging combined with 3D micro-tomography of angioarchitecture methods to further explore the mechanism and correlation of metal element changes with the plasticity of microvascular networks following cerebral ischemia, thus providing new targets for effective intervention of stroke [59-61].