Background
Immunoreactive microglia have been reported in both human brains and animal models of Parkinson’s disease [
1‐
7]. Multiple studies of autopsy brains have shown microgliosis in not just the substantia nigra but also the striatum, hippocampus and other cortical areas [
5‐
13]. Reports using rodent models of disease also demonstrate increased microgliosis correlating with dopaminergic neuron loss using the rat 6-hydroxydopamine (6-OHDA) toxin injection model [
3]. The LPS-injected rodent model of Parkinson’s disease (PD) also produces microgliosis [
2]. Similarly, MPTP injection into rodents also results in striato-nigral microgliosis correlating with loss of dopaminergic cells [
4]. Nevertheless, it is not yet clear whether the reactive microglia are a cause or effect of neuron loss during disease progression. Collectively, histologic findings from human brains and temporal analysis from rodent studies continue to support the idea that microglia activation and the associated inflammatory changes are part of the disease process.
In addition to the brain pathology, numerous studies report altered peripheral immune cell responses during disease suggesting a possible communication between the peripheral immune system and the central nervous system [
14,
15]. A number of clinical studies have suggested that select inflammatory markers in the blood of PD patients may correlate with the degree of severity of the disease or disease-associated symptoms such as depression, anxiety, fatigue and sleep patterns changes [
16‐
23].
In spite of the studies of human autopsy brains, clinical reports involving PD patients, and rodent models of the disease all having correlative findings, a number of conflicting reports exist. For example, animal models for PD have not been able to fully define the temporal changes of inflammatory response with respect to disease progression and neuron loss. Ideally, the rodent models of disease should accurately reflect the human condition to allow for mechanistic dissection. This is demonstrated concretely by the fact that only after characterization of human TNF-α overexpressing mice was TNF-α fully realized as a valid arthritis target [
24].
In order to determine the contribution of neuroinflammatory changes to the neuron loss and motor dysfunction associated with PD, we employed a mouse humanized CD34+ line designed to contain engrafted multi-lineage human immune cell populations following injection of CD34+ hematopoietic stem cells [
25]. The humanized CD34+ mice, along with age-matched C57BL/6 mice, were injected with MPTP to produce PD-like symptoms with or without intravenous injection of the immunosuppressive calcineurin/NFAT inhibitor drug, FK506, as a neuroprotectant. As expected, MPTP injections resulted in loss of dopaminergic neuron tyrosine hydroxylase (TH) immunoreactivity, decreased motor strength as seen from pole tests and grip strength tests, and increased microgliosis/astrogliosis. Interestingly, FK506 was able to attenuate the behavioral problems, cytokine levels, TH loss, and microgliosis/astrogliosis only in the humanized CD34+ mice. The mouse compared to human serum and brain cytokine profiles were also different across the two strains of mice. This study demonstrates for the first time that MPTP induces PD-like symptoms in female humanized CD34+ compared to female C57BL/6 mice. We also report a valid therapeutic target, calcineurin/NFAT activity, for treating motor impairment and inflammation in the humanized immune system model by using a clinically available drug, FK506. Finally, we validated that several of the peripheral cytokine changes observed in the MPTP-injected humanized CD34+ mice but not the C57BL/6 mice paralleled changes observed in human PD plasma suggesting that these mice may be a more accurate model for understanding human disease.
Methods
Animals
The humanized NOD scid gamma (NSG) mice and age-matched wild type mouse line, C57BL/6, were purchased from the Jackson Laboratory (Bar Harbor, Maine). For generation of humanized NSG mice, female NSG mice are injected with human hematopoietic stem cells (hCD34+) [
25,
26]. Engraftment of mature human white blood cells (hCD45+) is confirmed 12 weeks after the injection [
27]. Mice with more than 25% hCD45+ cells were considered successfully humanized and housed in a BSL2 negative pressure facility before shipping to UND. The mice obtained for this study were engrafted with hematopoietic stems cells from one individual with total hCD45+ cells engraftment ranging from 56 to 71%. The mice were 16 weeks of age at the time of arrival and allowed to acclimate for 7 days before experimentation. The mice were randomly divided into 4 groups: saline injection, MPTP injection, FK506 injection or MPTP and FK506 injection.
Animal use
All animal use was approved by the University of North Dakota Institutional Animal Care and Use Committee (UND IACUC). Mice were provided food and water ad libitum and housed in a 12 h light/dark cycle. The investigation conforms to the National Research Council of the National Academies Guide for the Care and Use of Laboratory Animals (8th edition).
Antibodies and reagents
Anti-TH antibody was purchased from EMD Millipore (Billerica, MA). Anti-Iba-1 antibody and anti-GFAP antibody were from Wako Chemicals (Richmond, VA) and Cell Signaling Technology, Inc. (Danvers, MA), respectively. The horseradish peroxidase conjugated secondary antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse TNF-α ELISA kit was obtained from R&D Systems (Minneapolis, MN). Elite Vectastain ABC avidin and biotin kits, biotinylated anti-rabbit, anti-mouse, and anti-rat antibodies and the Vector VIP kits were from Vector Laboratories Inc. (Burlingame, CA). The anti-CD68 antibody was obtained from AbD Serotec (Raleigh, NC). Human specific anti-CD68 and anti-HLA-DR (LN3) antibodies were from Bio-Rad (Hercules, CA). The human specific anti-CD45 antibody was purchased from Dako (Carpinteria, CA).
MPTP and FK506 treatments
The 16-week old female CD34+ mice and the age-matched C57BL/6 mice were given 3 intraperitoneal (i.p.) injections of saline vehicle or MPTP-HCl (18 mg/kg of free base) at 2 h intervals for a total of 3 injections. For FK506 treatments, mice were given saline vehicle or 10.0 mg/kg/day starting 30 min after the first MPTP injection and continuing through 4 additional days after the last MPTP injection, totaling 5 days of FK506 injections.
Pole test
Following the MPTP and FK506 injections, mice were housed for an additional 3 days and behaviorally tested on day 8. Each animal was administered the pole test to assess locomotor activity as a measure of dopaminergic neuron function following the MPTP injections [
28]. Briefly, mice were placed head-upward on the top of a vertical rough-surfaced pole (diameter 8 mm, height 55 cm) with a base that was positioned on a flat surface. The time until the mouse descended to the bottom of the pole/cage floor (locomotor activity time, TLA) was recorded with a maximum of 120 s. Mice were returned to their home cages after testing and the pole was wiped clean with 50% ethanol in between mice and allowed to dry before the next trial.
Grip strength test via Kondziela’s inverted screen test
To test gross strength of the four limb muscles in mice, mice were challenged with the inverted screen test as previously described by Deacon
et al [
29]. A 43 x 43 cm
2 square wire mesh frame was made for this purpose. The mesh was12 x 12 mm
2 square formed by 1-mm diameter wires. The frame was 4 cm deep wooden beading to prevent mice from climbing to the other side of the mesh. After a 15 min rest following the pole test, each mouse was placed onto the center of the mesh square frame and the screen was rotated over the course of 2 s to an inverted position with the mouse head declining first. The screen was held at 40-50 cm over a clean surface, and the time it took each mouse to let go of the screen was measured until the maximum of 120 s was reached. After testing, mice were returned to their home cages.
Tissue and serum collection
Following the behavioral tests, the mice were euthanized by i.p. injection of a mixture of 100 mg/kg ketamine/16 mg/kg xylazine, followed by cardiac exsanguination and PBS perfusion. Blood was collected and serum separated by centrifugation at 2000 x g for 10 min at 4 °C for ELISA arrays. The brain, spleen and intestines were collected from each mouse. The striatum was dissected out of the right cerebral hemisphere and flash frozen in liquid nitrogen. A portion of the ileum and spleen were also flash frozen. The left cerebral hemisphere and remaining spleen and ileum samples were immersion-fixed using 4% paraformaldehyde in PBS. The fixed tissue was cryoprotected through 2 successive 30% sucrose changes before sectioning.
Human plasma and tissue
Age matched human plasma (Parkinson’s disease and healthy controls) were obtained from the Harvard Biomarkers Study (
http://neurodiscovery.harvard.edu/biomarkers-core). Human normal adult spleen frozen sections were purchased from BioChain Institute, Inc. (Newark, CA). Slides were first processed for antigen retrieval by boiling in Tris-EDTA, pH 9.0, for 20 min, then immunostained along with mouse spleen and intestine sections as described below.
Immunostaining mouse brains
The left cerebral hemispheres for humanized CD34+ and age-matched C57BL/6 mice (saline, FK506, MPTP and MPTP + FK506) were sectioned using a freezing microtome. Briefly, multiple paraformaldehyde-fixed and sucrose-equilibrated tissues were embedded in a 15% gelatin (in 0.1 M phosphate buffer, pH 7.4) matrix to form a sample block for simultaneous processing. The block was immersed in a 4% paraformaldehyde solution for 3-4 days to harden the gelatin matrix, followed by replacing the solution with 30% sucrose every 3-4 days each for 2 weeks. The blocks were then flash frozen using dry-ice/isomethylpentane, and 40 μm serial sections were cut using a freezing microtome. Serial sections (960 μm apart) were immunostained using anti-TH antibody (1:1000 dilution), anti-Iba-1 antibody (1:500 dilution) or anti-GFAP antibody (1:1000 dilution). The antigens were visualized using Vector ABC kit and Vector VIP as the chromogen (Vector Laboratories, Inc., Burlingame, CA) according to the manufacturer protocols. The slides were coverslipped using VectaMount (Vector Laboratories) following a standard dehydrating procedure through a series of ethanol concentrations and Histo-Clear (National Diagnostics, Atlanta, GA). Photomicrographs were taken using an upright Leica DM1000 microscope and Leica DF320 digital camera system. Quantitation of immunostaining was performed as previously described [
30]. Briefly, optical densities from the striatum, substantia nigra, and hippocampus regions from the same serial sections were measured using Adobe Photoshop software (Adobe Systems, San Jose, CA). The values for each section were averaged (4 sections/brain, 4-5 brains per condition) and compared.
Immunostaining peripheral tissue
Spleen and ileum samples were serially cut (10 μm) onto gelatin subbed slides using a cryostat. These sections, along with human spleen slides (positive controls), were immunostained using anti-CD68 (1:500), anti-hCD68 (1:500), anti-HLA-DR (LN3) (1:500) and anti-hCD45 (1:250) antibodies. The immunostaining was visualized using Vector VIP as the chromogen.
ELISA arrays
Striatal samples and serum from all groups of humanized CD34+ and C57BL/6 mice were flash frozen following collection. Pre-weighed samples from mice and human plasma samples were used for multi-analyte ELISA arrays (Qiagen, Valencia, CA) and the levels of various cytokines were determined as per the manufacturer protocol. Human and mouse Inflammatory Cytokines Multi-Analyte ELISArray kits from Qiagen were chosen in order to obtain expression of a wide array of inflammatory and anti-inflammatory cytokines. Cytokine concentrations were normalized by tissue weight.
Design-based stereological quantification of Iba-1 positive cells
Paraformaldehyde-imbedded sections from humanized CD34+ and age-matched C57BL/6 mice (saline, FK506, MPTP and MPTP + FK506) were cut at 40 μm intervals and immunolabled for Tyrosine Hydroxylase (TH), Iba-1 and GFAP. Sections from three different mice for each condition of mice were assessed for TH, GFAP or Iba-1-positive nuclei with the section interval set at 960 μm. Non-biased quantification of immuno-positive nuclei was conducted using the optical fractionator method originally developed by West and colleagues [
31] and adapted from previous studies with other neural populations [
32,
33]. TH, GFAP or Iba-1-positive cells in the temporal cortex layer were counted in every 24th section using StereoInvestigator 11.06.2 (Microbrightfield Inc., Williston, VT) on an Olympus BX51WI equipped with a motorized x, y, and z stage. In each animal, immuno-positive cells were counted from 21 to 30 randomly and systematically selected frames in every serial section using the 40x objective. For cell counting, the contour of the Substantia Nigra was delineated with 2× magnification based on standard anatomical markers and the grid size set following sub-sampling optimization. In brief, the number and location of counting frames and the counting depth for that section were determined by entering parameters for the grid size (300 X 300 μm), the thickness of top guard zone (2 μm) and the optical dissector height (26 μm). The guard zones were set at 2 μm above and below the counting depth for each section with regional thickness and variation in section integrity taken into consideration and the max and min z-depth determined at each sampling location prior to counting. The optimal counting frame width (100 μm) and height (100 μm) were determined with an initial oversampling of a representative set of sections from control and treated samples. The cell bodies of the TH, GFAP or Iba-1-positive cells were counted if they were entirely within the 26 μm depth of the counting frame and only at the z-depth at which the nucleus was in focus. The StereoInvestigator software calculated the total number of Iba-1, TH, and GFAP-positive cells in the temporal cortex using the 3D optical fractionator formula designated:
N = t/h X 1/ssf X 1/asf X 1/hsf X ΣQ-. For the calculations, “t/h” is the section mounted thickness over the counting frame height; “ssf” is the section sampling fraction; “asf” is the area sampling fraction, which was calculated by dividing the area sampled with total area of the temporal cortex (in our tracings, the sum of temporal cortex areas sampled in every 24th section); “hsf” is the height sampling fraction, which was determined by dividing the height of the counting frame (26 μm) by the section thickness determined for each sampling site (37-39 μm on average); and “ΣQ”- denotes the total count of particles sampled for the entire temporal cortex. The sampling was optimized for maximal efficiency, with a final mean coefficient of error (CE, Gundersen) of less than 10% for all sections sampled. One-way analysis of variance (ANOVA; GraphPad Prism, 6.0c; LaJolla, CA) was used to analyze positive nuclei and all data are presented as mean ± SD for
n = 3 samples.
Statistical analysis
Data are presented as mean ± standard deviation. Values statistically different from controls were determined using one-way ANOVA (or two-way ANOVA where required). The Turkey-Kramer multiple comparisons post-test was used to determine p values.
Statistical analysis for human plasma cytokine ELISA arrays
A total of 140 plasma samples were analyzed to determine cytokine level differences in 4 groups: female Parkinson’s disease donors, female healthy controls, male Parkinson’s disease donors and male healthy controls. Initial sample size for all 4 conditions was n = 35. Following ELISA experiments for 12 cytokines, the samples that were below the detection limit were excluded from the analyses. The samples that did not represent the population for a 95% confidence interval i.e. the values that were 2 times standard deviations above or below the mean were also excluded from the analyses. This resulted in a sample size ranging from 14 to 34 for all the cytokines but IL-4 (n = 5-7). Values statistically different between the groups were determined using two-way ANOVA with Holm-Sidak post-hoc test.
Discussion
Our findings suggest that female C57BL/6 and female humanized CD34+ mice can both be used with the MPTP model of PD. Moreover, we observed profound strain as well as species differences with respect to MPTP-induced changes and their response to anti-inflammatory FK506 treatment. MPTP injection produced the expected impairment of motor performance, increased microgliosis, increased astrogliosis, elevated brain cytokine levels, and reduced TH immunoreactivity in the substantia nigra and striatum of C57BL/6 mice. However, there were no changes in serum cytokine levels in the female C57BL/6 mice following MPTP treatment and FK506 provided no significant protective effects as assessed by behavior, TH immunoreactivity, Iba-1 microglial immunoreactivity, or brain and serum cytokine levels. This suggests, for reasons unclear, that the drug did not have a robust anti-inflammatory effect in the C57BL/6 mice. A similar decrease in TH immunoreactivity, increased microgliosis, astrogliosis and impaired motor performance was observed in the MPTP injected humanized CD34+ mice demonstrating conservation of toxin-mediated neuron death and disease modeling in this line. In addition, these mice had an increase in select mouse cytokines in the both blood and brains following MPTP injection quite different from changes observed in the C57BL/6 mice. More importantly, both blood and brain levels of specific human cytokines were elevated in the MPTP injected mice demonstrating a possibility that the blood cytokine level profile may be used as a biomarker of brain cytokine levels and neuronal loss. This idea is further supported by cytokine analyses of human PD plasma compared to healthy controls. Plasma from female PD patients and the MPTP-injected humanized CD34+ mice showed an increase in IL-1α, IL-2, and IL-6 compared to controls. This suggests that these cytokines could be further validated as potential biomarkers for the disease, possibly in a sex selective fashion. In addition, it would be very interesting to further study the direct effects of these cytokines on brain changes in the humanized CD34+ mice. Finally, unlike the C57BL/6 findings, FK506 treatment had a dramatic effect on improving behavioral performance, increasing TH immunoreactivity, decreasing microglial Iba-1 immunoreactivity, decreasing GFAP astroglial immunoreactivity, and lowering serum cytokine levels in the CD34+ humanized mice suggesting potential for human translation. These data suggest that the humanized immune system mice provide a unique resource for characterizing the contribution of inflammatory changes to disease.
Human CD34+ hematopoietic stem-cell engrafted NSG mice develop multi-lineage human immune cells and have become a popular, valid platform for infectious disease research and cancer biology, especially with respect to immune-oncology studies [
36]. Engraftment of human hematopoietic stem cells is much more efficient in female scid mice as compared to males [
37]. Therefore humanized CD34+ mice from Jackson Laboratories can only be obtained as females. In this study, age-matched female C57BL/6 mice were used for comparing differences between the human and mouse systems. Many murine MPTP-related studies are typically performed on male mice to avoid the increased risk of fatalities observed with females [
38]. However, even more surprising to us was the fact that the combined MPTP + FK506 treatment group experienced higher levels of toxicity in the C57BL/6 strain totaling near 50% of animal loss in this group and not the MPTP or FK506 alone mice. In addition, with the limited numbers surviving in this group there were no drug-dependent apparent anti-inflammatory effects associated with reduction of cytokines in the serum or brains. The dose of FK506 was chosen based on previously reported work that demonstrated protection against MPTP-induced dopamine depletion in the striatum of MPTP-injected C57BL/6 mice using 10 mg/kg/day FK506 [
39]. The lack of protective effects of FK506 in MPTP-injected C57BL/6 mice was surprising. One possible explanation for this is that the MPTP injection paradigm is typically performed in male mice and our study was necessarily performed in female mice in order to be able to compare to the female humanized CD34+ mice. The fact that FK506 did not provide the benefits predicted may be due to sex differences. However, we cannot rule out the possibility that a different concentration of FK506 without overt toxicity may have also provided benefits in the female C57BL/6 mice. Due to limitations in the number of available humanized CD34+ mice for the study design, alternative FK506 dosage interventions were not possible. This increased toxicity susceptibility of C57BL/6 female mice to the dosage of FK506 used is, in itself, interesting and worth pursuing in future work. Understanding this vulnerability may give some insight into understanding the reason for drug-induced nephrotoxicity in some patients. Although we can only hypothesize, we expect that the loss of TH immunoreactivity in the C57BL/6 mice is primarily due to the more classic model of direct neurotoxicity by MPP
+. Conversely, a significant portion of the loss of TH immunoreactivity in the CD34+ humanized mice involves an inflammatory component that is preventable by FK506 treatment.
Even though the specific changes in temporal phenotype of immune cells in both the brain and periphery during PD remain unclear, it is well-known that either of these cell types rely upon secretion of a number of cytokines, such as IL-1β, IL-6 and TNF-α to act via both paracrine and autocrine cellular mechanisms [
15,
40]. In an effort to identify particular brain cytokines that were exclusive to the humanized immune system, we quantified an array of cytokines expression profiles in C57BL/6 mice vs. humanized CD34+ mice and serum cytokines vs. brain cytokines. We also compared the plasma cytokine profiles from human PD patients and healthy controls to identify a few cytokines such as IL-1α, IL-2, IL-4, and IL-6 as unique and robust serum biomarkers of human disease that were replicated in the brains and serum of the humanized CD34+ mice. This suggests opportunity for further studies involving immunosuppressive therapeutic targets and even enticing possibilities of small scale clinical trials using FK506 in order to quantify changes in serum levels of these cytokines to reflect an anti-inflammatory effect in the brain in correlation with behavioral improvement in PD patients.
Since the humanized CD34+ mice have not been very well characterized with regards to brain biology, it was difficult to know beforehand whether or not there would be any presence of human cells in the brains of 12 weeks post-engrafted mice. However, immunostaining against human specific macrophage antigens demonstrated the presence of sparse meningeal-like human CD45 and human HLA-DR specific immunoreactivity. The immunostaining was clearly not as robust in the brain as compared to the spleen and intestine, there is a possible role of parenchymal brain microgliosis and cytokine secretion in these mice. Nevertheless, we also cannot rule out the possibility that increases in human specific cytokines in both serum and brains of MPTP injected humanized CD34+ mice and detection of human macrophage markers such as CD68 and CD45 in peripheral organs such as intestine and spleen suggests that the peripheral immune system in these mice is effectively populated with human immune cells and that the cytokines produced by these cells may be infiltrating the brain from the periphery to perhaps drive the gliosis, behavioral deficits, and loss of TH staining. Indeed, recent work demonstrates the presence of lymphatic vessels in the brain that might allow for the cytokine or lymphocyte influx [
41]. It would be very exciting to repeat this work with the humanized CD34+ mice and determine whether iv delivery of select human cytokines that were elevated by the MPTP treatment and in human PD plasma, IL-1α, IL-2, IL-4, or IL-6, are sufficient to produce the brain changes observed (i.e. gliosis, motor deficit, loss of TH staining, and elevated brain cytokines). Conversely, selective cytokine receptor antagonists or neutralizing antibodies may be able to attenuate the adverse effects of MPTP injection in these mice. This may provide significant insight into understanding whether a peripheral immune change can drive the brain disease in humans. Overall, this study demonstrates for the first time, the MPTP induced PD model in female humanized CD34+ mice and a possible therapeutic effect of FK506 in treating the MPTP-associated behavior deficits as well as inflammatory responses observed during the disease.
Conclusions
These data demonstrate for the first time, induction of Parkinson’s disease-like symptoms in female humanized CD34+ mice using MPTP. There are distinct species differences between the cytokine profiles of humanized CD34+ mice and C57BL/6 mice. MPTP was able to produce loss of tyrosine hydroxylase immunoreactivity, loss of motor strength, and increase in proinflammatory cytokines in both strains of mice. However, these effects were attenuated using an immunosuppressant drug, FK506, only in the humanized CD34+ mice. Our data suggest that these mice may represent a more accurate model for assessing inflammatory changes in PD and developing anti-inflammatory therapeutics.
Acknowledgments
We are grateful for Dr. Clint Hosford for advice regarding statistical analysis. We thank the Drs. Adrian Ivinson, Dr. Bradley T. Hyman, the HBS Investigators, as well as all patients participating in the Harvard Biomarkers Study for their invaluable contributions.