This paper details methodology for radiolabeling a human-specific anti-CD19 monoclonal antibody and how to use it to quantify B cells in the central nervous system and peripheral tissues of a mouse model of multiple sclerosis using in vivo PET imaging, ex vivo gamma counting, and autoradiography approaches.
Multiple sclerosis (MS) is the most common demyelinating central nervous system (CNS) disease affecting young adults, often resulting in neurological deficits and disability as the disease progresses. B lymphocytes play a complex and critical role in MS pathology and are the target of several therapeutics in clinical trials. Currently, there is no way to accurately select patients for specific anti-B cell therapies or to non-invasively quantify the effects of these treatments on B cell load in the CNS and peripheral organs. Positron emission tomography (PET) imaging has enormous potential to provide highly specific, quantitative information regarding the in vivo spatiotemporal distribution and burden of B cells in living subjects.
This paper reports methods to synthesize and employ a PET tracer specific for human CD19+ B cells in a well-established B cell-driven mouse model of MS, experimental autoimmune encephalomyelitis (EAE), which is induced with human recombinant myelin oligodendrocyte glycoprotein 1-125. Described here are optimized techniques to detect and quantify CD19+ B cells in the brain and spinal cord using in vivo PET imaging. Additionally, this paper reports streamlined methods for ex vivo gamma counting of disease-relevant organs, including bone marrow, spinal cord, and spleen, together with high-resolution autoradiography of CD19 tracer binding in CNS tissues.
MS is an immune-mediated neurological disorder; the unique presentation in each patient can make management challenging for both patients and clinicians1. The disease itself is characterized by the presence of demyelinating lesions and immune cell infiltration in the brain and spinal cord, resulting in physical and cognitive impairment2. The traditional paradigm that MS is a T cell-mediated disease was first challenged in a landmark phase II clinical trial of rituximab3, a therapy targeting the CD20+ subset of B cells. Additional B cell therapies have since been developed that target CD194, a pan B cell biomarker that is expressed on a broader range of B cells, which can be both diagnostically and therapeutically advantageous. Moreover, existing methods to assess treatment efficacy (i.e., monitoring the number of relapses and magnetic resonance imaging [MRI] activity) do not afford early measures of response-thus placing patients at significant risk of CNS damage due to suboptimal therapy selection and optimization. Hence, there is a critical need for strategies to monitor specific immune cells, such as CD19+ B cells, in real time in the CNS and periphery of MS patients.
PET imaging is a robust imaging technique that allows in vivo, whole-body visualization of a given target of interest, such as CD19. While blood draws, records of relapse rates, and lesion monitoring via MRI provide snapshots into treatment efficacy, PET imaging can allow researchers and clinicians to monitor effectiveness of a therapy across the entire body. This proactive approach to therapeutic monitoring allows clinicians to assess medication effectiveness in real time, enabling rapid adjustments as needed. Monitoring the location and density of cell populations associated with disease also permits longitudinal assessment of severity using patient-specific anatomic information. It is thus essential to establish reproducible analytical methods to reliably utilize the full potential of PET imaging in clinical and preclinical settings.
This paper describes methods (Figure 1) to perform PET imaging, ex vivo gamma counting, and autoradiography (ARG) of CD19+ B cells with a 64Cu-labeled anti-human CD19 monoclonal antibody (mAb), known as 16C4-TM (64Cu-hCD19-mAb), in the experimental autoimmune encephalomyelitis (EAE) mouse model of MS induced in transgenic mice expressing human CD19 (hCD19) using human recombinant myelin oligodendrocyte glycoprotein 1-125 (MOG1-125). We also provide methods to assess radiotracer binding accurately and reproducibly in the brain and spinal cord, both critical sites of pathogenesis often severely affected in this and other neurodegenerative models. These techniques allow for non-invasive investigation of the role of B cells in disease pathology and have the potential to be clinically translated to assess the efficacy of anti-B cell therapies in MS.
Figure 1: Study design. An overview of key techniques in this article. (A) Placing the mice in the scanning bed on their backs reduces motion in the spine. (B) PET/CT imaging of the mice. (C) Make an incision down the dorsal side of the animal to expose the spinal column. (D) Bisect the spinal column into cervical/thoracic and lumbar portions and remove the sections following the five indicated cuts. (E) Use a syringe to remove the spinal cord from the spinal column by making a seal with the syringe and spinal column and flushing from the cranial and caudal ends of the spinal column as shown. (F) Isolated cervical/thoracic and lumbar spinal cord segments. Abbreviation: PET/CT = Positron emission tomography/computed tomography. Please click here to view a larger version of this figure.
All animal studies were carried out in accordance with the Administrative Panel on Laboratory Animal Care (APLAC) at Stanford University, a program accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC International). Mice were acclimated to the vivarium for at least 7 days prior to start of study to minimize stress on the mice, as stress can affect EAE induction.
1. EAE induction in female humanized CD19 mice
2. Animal care and scoring in EAE mouse model
3. mAb conjugation, radiolabeling, and characterization
4. Dose preparation
NOTE: Before handling the dose, wear proper PPE, including lab coat, body and finger dosimeters, and gloves.
5. Cannulation and Injection
NOTE: See previously described methods6 for intravenous cannulation of mice for injection of the radiotracer6.
6. PET/CT imaging
7. Dissection for ex vivo gamma counting and autoradiography
8. Ex vivo gamma counting
9. Ex vivo autoradiography (ARG) of CNS tissue
10. Analysis of biodistribution data
11. PET image analysis
12. Ex vivo autoradiography analysis
The hCD19-mAb was DOTA-conjugated and radiolabeled with 64Cu as shown in Figure 2. EAE and naïve mice underwent PET/CT scanning (Figure 3) 18-24 h after injection with 64Cu-DOTA-hCD19-mAb. PET/CT images were co-registered using the PET analysis software, and the CNS tissues were analyzed using manual ROIs or a semi-automated 3D brain atlas. Radiotracer binding in ROIs (Figure 4) was higher in EAE mice than in naïve mice. Ex vivo gamma counting and ARG showed increased binding in the spinal cord (both lumbar and cervical thoracic segments) and brain (ARG only) of EAE mice compared to naïve (Figure 5 and Figure 6). Ex vivo gamma counting of perfused mice also showed decreased radiotracer binding in peripheral organs, including spleen, femur, and bone marrow (Figure 5), consistent with B cells leaving the periphery and infiltrating the CNS in this EAE model.
Figure 2: Conjugation and radiolabeling scheme for generating 64Cu-labeled human-specific CD19 monoclonal antibody, 16C4-TM mAb (64Cu-DOTA-hCD19-mAb), in addition to quality control data. (A) Reaction of DOTA-NHS-ester with hCD19 monoclonal antibody to produce hCD19-DOTA conjugate (not to scale) and radiolabeling with 64Cu-CuCl3 to produce 64Cu-DOTA-hCD19-mAb. (B) Representative ITLC chromatograph. The peak at 40-60 cm is the radiolabeled antibody; unbound 64Cu-CuCl3 travels with the mobile phase and would be present from 200 to 240 cm. There is no detectable free 64Cu-CuCl3 in this chromatograph. (C) The quality control specifications of the radiolabeled antibody. Abbreviations: DOTA-NHS ester = 1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetraacetic acid mono-N-hydroxysuccinimide ester; ITLC/HPLC = instant thin layer chromatography/high-performance liquid chromatography; MALDI/LC-MS = matrix-assisted laser desorption/ionization/liquid chromatography-mass spectrometry; CPM = counts per minute. Please click here to view a larger version of this figure.
Figure 3: Photographs demonstrating how to secure mice in a 3D-printed bed within the PET scanner to enable high quality imaging of the spinal cord and brain while minimizing motion. (A) 3D printed four-mouse scanner bed (also known as "mouse hotel") equipped with heating elements and anesthesia tubing. (B) Anesthetized mice in a supine position to maximize straightness of the spine; the bed position of each mouse is recorded. (C) Mice taped securely across their head to minimize motion in the brain and across the belly to minimize motion from breathing, without affecting breathing. (D) Mouse bed positioned within the scanner and taped to the scanning bed. Anesthesia tubing was connected from scanner to bed and isoflurane set to 2%. Mouse breathing was monitored to ensure appropriate isoflurane level before closing the scanner door. Abbreviation: PET = Positron emission tomography. Please click here to view a larger version of this figure.
Figure 4: Spinal cord image and brain analysis and results using PET analysis software. (A) i) ROIs (pink and tan) drawn on spine to separate lumbar from the thoracic and cervical vertebrae and prepare image for Otsu thresholding. ii) Spinal vertebrae (turquoise and red) were segmented out using Otsu Thresholding. iii) Vertebrae were then made immutable in the 3D ROI menu, and the spinal cord divided into cervical/thoracic (purple) and lumbar (navy) ROIs. iv) The vertebral ROI was removed, leaving spinal cord ROIs and representative brain atlas applied. (B) Representative analysis of PET results from various CNS regions represented as %ID/g normalized to the ROI of the heart within each animal. PET acquisition was a 10 min static scan via PET/CT imaging. Brain regions quantified using a semi-automated brain atlas approach, shown in panel A. iv) Representative results show either significance or trending toward significant increase in tracer binding in the brain and thoracic spinal cord. Statistics performed using Student's t-test (*: p < 0.0332). Abbreviations: PET = Positron emission tomography; ROIs = regions of interest; CNS = central nervous system; CT = computed tomography; %ID/g = percent injected dose per gram of tissue; EAE = experimental autoimmune encephalomyelitis. Please click here to view a larger version of this figure.
Figure 5: Representative quantification of ex vivo gamma counting in various organs in EAE and naïve mice expressed as %ID/g. Post-PET scan, mice were perfused with PBS to remove the radiotracer present in the blood, either free or bound to blood-resident CD19+ B cells, and organs quickly dissected and weighed to have an accurate weight of each organ. Tracer binding is significantly decreased in the spleen and bone marrow in EAE mice compared to naïve ones. Increased radiotracer binding is observed in both the lumbar and cervical/thoracic spinal cord segments of EAE mice. The brain does not show significant increase in radiotracer signal, though it is trending toward significant increase. Statistics performed using Student's t-test (*: p < 0.0332; ****: p < 0.0001). Abbreviations: PET = Positron emission tomography; %ID/g = percent injected dose per gram of tissue; EAE = experimental autoimmune encephalomyelitis; PBS = phosphate-buffered saline. Please click here to view a larger version of this figure.
Figure 6: Ex vivo ARG images depict 64Cu-DOTA-hCD19-mAb binding in sagittal brain sections and whole spinal cords from EAE compared with naive mice. Digital phosphor storage films were scanned using a phosphor imager after being exposed to radioactive tissue samples for approximately 10 half-lives (127 h or 5 days). The resulting images reveal visually higher signal in the brain of EAE mice compared to brain sections from naïve mice, which is expected due to the regions known to contain B cells in this model5. Specifically, there is increased tracer signal in the brain stem, cerebellum, and ventricles of EAE mouse brain sections. This increase in signal for EAE mouse brain sections mirrors what was found for in the whole-brain PET quantification detailed above. Similarly, there is an increase in radiotracer binding in both the cervical/thoracic and lumbar spinal cord segments compared to naïve spinal cords, reflecting what was found using ex vivo gamma counting. Abbreviations: PET = Positron emission tomography; EAE = experimental autoimmune encephalomyelitis; Vent = ventricles; Cb = cerebellum; BS = brain stem; TSc = thoracic and cervical spinal cords combined; LSc = lumbar spinal cord. Please click here to view a larger version of this figure.
Supplemental Figure S1: Staining of CNS tissues of naïve and EAE mouse brain tissue with CD45R/B220. B cells are observed in brainstem, meninges, and white matter of EAE mice (n = 7 EAE, n = 5 naïve mice, average four slices per animal). This figure is from 5. Scale bars = 5 mm (low magnification [1x]) in sagittal brain images, 100 µm (high magnification [20x]) in brainstem, meninges, and cerebellar white matter. Please click here to download this File.
This article describes a streamlined method for imaging human-CD19+ B cells in a mouse model of MS using CD19-PET. Due to the heterogeneous presentation of MS and varying responses to treatments, its management in the clinic can be challenging and new approaches for therapy selection and monitoring are greatly needed. PET imaging could serve as a powerful tool for monitoring disease progression and individual response to B cell-depleting therapy. In addition to MS, CD19-PET imaging could be used to monitor B cell depletion after treatment in subtypes of lymphomas and leukemia or other B cell mediated diseases. This protocol and representative data show the utility of imaging B cells in neurological diseases.
To study human CD19+ B cells in the context of MS, we chose the B cell-dependent MOG1-125 EAE model7. Similar to other EAE models, this model presents with symptoms of progressive paralysis and infiltration of immune cells into the CNS. However, the MOG1-125 model is unique in that it is a B cell-driven model: mice contain varying numbers of B cells in the subarachnoid space in the meninges, brain stem, parenchyma, and ventricles. These lymphocytes can be sparsely scattered throughout these regions and/or form follicle-like structures, which are also observed in humans with MS8,9. In addition to using naïve mice as controls, a complete Freund's adjuvant (CFA)-only induction kit may be used (i.e., an identical induction emulsion to what is given to EAE mice sans the MOG protein). In the EAE mouse model, the blood brain barrier (BBB) is dysfunctional and allows larger entities, such as antibodies, to cross. The CD19-mAb radiotracer will only bind and remain in the CNS if B cells are present; the tracer will circulate back into the blood pool if B cells are not present. We have demonstrated this using gamma counting and ex vivo autoradiography of CNS tissues by perfusing before measuring the radioactivity levels in the tissues. We have also demonstrated this in earlier publications reporting the use of mAb-based PET radiotracers (i.e., immunoPET imaging approaches) for detecting B cells in the CNS1,2.
The DOTA chelator was used since it has been used in clinical PET imaging with copper-64 labeled peptides and antibodies, and we aim to translate the hCD19-mAb for clinical imaging of MS patients. DOTA has adequate binding affinity to copper-64 in vivo. The in vivo stability is very important because free 64Cu goes to the liver and can obscure the signal of bound radiotracer; thus, it is important to measure the signal in the liver to calculate the relative signal compared to other organs. Muscle is typically taken as a control tissue, but in the case of EAE, there can be inflammation present in the muscles. The half-life of 64Cu is 12.7 h, which affords ample time for the DOTA-hCD19-mAb to bind to its target while ensuring signal can be measured by PET. When preparing the conjugate, small-scale (75-125 µg) test reactions should be performed to determine the amount of DOTA to add to mAb to produce the desired DOTA/mAb ratio (e.g., a reaction of 6-10-fold excess DOTA-NHS-ester per mol mAb may afford a conjugate of 1-2 DOTA/mAb). The reaction time and temperature (e.g., 2-4 h or overnight at 4 °C or room temperature) also influences the DOTA/mAb ratio and should be optimized. A titration with nonradioactive copper can be performed to calculate the number of DOTAs per mAb; however, we recommend performing MALDI-MS and/or LC-MS for more reliable and accurate results.
The calculated DOTA/mAb ratio is an average value for a particular sample and some variation is expected. For MALDI, several shots are taken per sample for the conjugated and unconjugated mAbs. We then calculate the ratio of conjugated to unconjugated to determine the average number of DOTA/mAb. The DOTA/mAb ratio is important because too many chelators will disrupt antibody binding and too few will lead to inconsistent radiolabeling and low signal. The ratio should be very close between batches of conjugate to maintain consistent signal intensity and binding kinetics; ideally, the same batch of conjugate should be used for all experiments within a particular study. A promising technique to reduce potential effects on immunoreactivity due to possible overconjugation is to use site-specific conjugation10 whereby the chelator conjugation is site-selective on the heavy chain glycans of the antibody, thus guaranteeing the addition of 1 chelator per mAb.
The radiolabeling reaction conditions should be optimized to ensure the highest labeling efficiency and yield since differences in antibody, DOTA/mAb ratio, and 64Cu molar activity, among other conditions, will impact radiolabeling. Using the optimal 64Cu to mAb conjugate ratio may allow the radiotracer to be used without purification, reducing the time required for radiolabeling and loss due to the gravity flow column and radioactive decay. A consistent and reliable molar activity may also be achieved when the same 64Cu to mAb conjugate ratio is used, which is especially important when comparing results across multiple cohorts of mice or imaging studies. The ITLC conditions may also be modified to suit each user. If purification is necessary, an aliquot should be saved for either HPLC and/or UV/Vis spectrophotometry so that the molar activity can be calculated.
It is important to note that using radiolabeled antibodies for imaging can be challenging. It is essential that the antibody used for the radiotracer be biologically inert so as not to have a physiological effect. Moreover, since antibodies have a long blood residence, one must wait long enough for circulation, binding, and clearance of a given mAb to ensure a suitable signal-to-background without compromising image quality. Typically waiting for 20-48 h for a 64Cu-labeled mAb is sufficient but one should image at 2, 4, 6, 12, 24, 48 h post injection when assessing a new mAb PET tracer to determine the best time point for imaging in a given rodent model. The same is true for acquiring ARG images with the highest signal-to-background ratio. The representative images in this protocol were taken at 18-20 h post injection, though other time points may be used depending on the radioisotope used. Different antibodies binding to different epitopes of CD19 will produce varying results and should be rigorously characterized.
When analyzing the spinal cord signal, it is important to position mice on their backs in the scanning bed to reduce movement caused by breathing. Additionally, supine placement can help straighten the spine in mice that have increased spinal curvature due to the progression of EAE disease. Another important aspect to consider when aiming to detect signal in the spine and spinal cord is to avoid injecting MOG1-125 on the flank as the injection sites can bind the tracer due to the associated immune response in those areas. The close proximity of the injection site can interfere with spinal cord analysis; thus, injections in the chest are preferable for the application described herein.
The image analysis techniques used are specific for CNS imaging. A brain atlas tool within the image analysis software affords reproducible and reliable results as long as the registration of PET and CT is accurate. Using the semi-automated 3D brain atlas and adjusting it to fit the skull of each mouse allows for consistent ROIs between animals. Since there is currently no automated or semi-automated approach for analyzing the signal in spinal cords, manual ROIs must be drawn. Notably, when quantifying CD19+ B cells (or any cell type present in both the bone marrow and spinal cord), it is critical to eliminate the signal arising from the spinal column and bone marrow as much as possible. The reason for this is that naïve mice are known to contain more CD19+ B cells in their bone marrow than EAE mice, in which B cells leave the periphery to infiltrate the CNS5,11. This bone marrow signal can obscure the true signal in the spinal cord.
To delineate true spinal cord signal while minimizing the contribution of signal from the spinal column and bone marrow, Otsu thresholding of the CT image can be used to make an immutable ROI for the spinal column. A separate spinal cord ROI can then be easily drawn within the spinal column. The same technique can also be applied to measure bone marrow in the femur. This is a very useful method to gain insights into tracer binding in the spinal cord. However, due to the relatively low spatial resolution of PET and issues pertaining to the partial volume effect when scanning small anatomical regions of mice, use of additional ex vivo confirmatory techniques (e.g., gamma counting, ARG) enables the validation of radiotracer binding in the spinal cord without the presence of blood, cerebrospinal fluid, or spillover signal from the spinal column.
Signal in the cervical/thoracic spinal cord tends to vary in the EAE mice depending on the severity of disease and how many B cells infiltrate during the adaptive immune response. This variation in the number of B cells that infiltrate, as well as the small amount of B cells in the CNS compared to those in the pelvic/spinal bone marrow of naïve mice, can make in vivo quantitation of spinal cord tissue challenging in mice. Given the spatial resolution of PET in small animal imaging, signal from the bone marrow can spill over onto the spinal cord signal. Ex vivo biodistribution and autoradiography completed here aid in validating the PET signal of the vertebrae versus spinal cord tissue. Mice are perfused prior to dissection to remove any unbound tracer in the blood pool so that the gamma counting and autoradiography results reflect the tracer that is actually bound in each organ rather than the tracer that is in the blood pool in that organ.
Radiotracers circulate through the blood, and with antibody tracers, specifically, there is often unbound radiotracer present in the blood for weeks after the initial injection. Since we are imaging the brain and spinal cord, which have many blood vessels, it is important to understand what portion of the signal is truly due to tracer binding in brain/tissue of interest versus that present in the blood pool. It is thus necessary to divide the brain signal by signal in the heart/blood pool. In the clinical setting, the same image analysis techniques of Otsu thresholding of vertebrae and ROIs of spinal cord tissues can be used for quantification. Given the larger tissue volumes in humans compared to mice, there should be significantly less impact from partial volume effects, leading to improved accuracy and negating the need for ex vivo techniques to confirm in vivo findings. The use of PET in the clinic will allow clinicians to personalize therapy for each patient depending on their individual B cell burden.
ARG is particularly useful for acquiring high-resolution images to enable more accurate delineation of the spatial location of tracer binding in small regions such as brain stem and cerebellum. The same sections and/or adjacent sections can be saved for immunohistochemical stains to confirm the presence of B cells. We have previously stained CNS tissues with CD45R/B220 (Supplemental Figure S1) to correlate the number of B cells with PET and ARG signal5,9. The staining can then be compared spatially to the ARG results to verify that the radiotracer signal matches the staining pattern. B cells can be present in clusters or diffusely throughout the brain stem; PET sensitivity is sufficiently high to measure the signal, which is encouraging for clinical translation. For spinal cord ARG, removing the spinal cord from the vertebrae ensures that the signal measured is due to tracer binding in the spinal cord tissue rather than the bone marrow and/or blood, which can obscure PET images due to partial volume effects.
Similar to ARG, ex vivo gamma counting enables the quantification of radioactive signal in individual organs. For this particular technique, it is important to measure the wet weight of tissues and ensure they are at the bottom of their respective tubes before placing the tubes in the gamma counter. The tubes must be labeled with the mouse number and tissue, so that the correct tube is used; the tube is then weighed on a calibrated balance and organs are inserted to the nearest tenth of a microgram (0.0001 mg). Some tissues are extremely small and the difference in the tube mass before and after will be in the order of 0.0001 mg. The tissues should be weighed immediately following dissection to prevent loss of moisture, which results in a lower mass. After weighing, the brain and spinal cord tubes should be filled with PBS to prevent from drying before freezing these tissues for ARG.
The authors have nothing to disclose.
We are grateful for the support from the SCi3 small-animal imaging facility at Stanford and Dr. Frezghi Habte for his technical assistance with the PET/CT. LC-MS is performed by the core staff at the Stanford University Mass Spectrometry (SUMS) core facility and we appreciate the staff for providing this service. We thank Horizon Therapeutics for very kindly providing the hCD19-mAb and Jodi Karnell in particular for her technical guidance and support. This work was funded by the NIH NINDS (1 R01 NS114220-01A1).
0.5 mL 50 kDa MWCO Centrifugal filter | MiliporeSigma | UFC505008 | centrifugal filter |
64Cu-CuCl3 | Washington University in St. Louis; University of Wisonsin, Madison; or another vendor | ||
AR-2000 Radio-TLC Imaging Scanner | Eckert & Ziegler | AR-2000 | |
Autoradiography cassette | Cole Palmer | EW-21700-34 | Aluminum, 8" x 10" |
Autoradiography film | GE Life Sciences | 28-9564-78 | Storage Phosphor Screen BAS-IP SR 2025 E Super Resolution, 20 x 25 cm, screen only |
Butterfly Needle Catheter | SAI Infusion Technologies | BLF-24 | |
DOTA-NHS-ester | Macrocyclics | B-280 | |
EAE Induction Kit | Hooke Laboratories | EK-2160 | |
Geiger Counter | Ludlum | 14C | |
GNEXT PET/CT Scanner | Sofie | GNEXT | |
Hidex Automatic Gamma Counter | Hidex | AMG | |
HPLC Column | Phenomenex | 00H-2146-K0 | 5 μm SEC-s3000 400 Å, 300 x 7.8 mm |
Illustra NAP-5 column | Cytiva | 17085301 | DNA gravity column |
Image J | NIH | ARG analysis software | |
Low Protein Binding Collection Tubes (1.5 mL) | Thermo Scientific | PI90410 | |
NanoDrop Lite Spectrophotometer | Thermo Scientific | 840281400 | UV-Vis micro/nano-spectrophotometer |
PCR tubes 0.2 mL, for DNA grade | Eppendorf | 30124707 | |
Typhoon phosphor imager 9410 | GE Healthcare | 8149-30-9410 | |
VivoQuant | Invicro | Version 4 Patch 3 | PET Analysis Software; must purchase brain atlas add-on |
Zeba Spin Desalting Columns, 7K MWCO, 0.5 mL | Thermo Scientific | PI89882 | Desalting column |