Abstract
Near infrared fluorescence (NIRF) optical imaging is a technique particularly powerful when studying in vivo processes at the molecular level in preclinical animal models. We recently demonstrated liver irradiation under the additional stimulus of partial hepatectomy as being an effective primer in the rat liver repopulation model based on hepatocyte transplantation. The purpose of this study was to assess optical imaging and the feasibility of donor cell expansion tracking in vivo using a fluorescent probe. Livers of dipeptidylpeptidase IV (DPPIV)-deficient rats were preconditioned with irradiation. Four days later, a partial hepatectomy was performed and wild-type (DPPIV+) hepatocytes were transplanted into recipient livers via the spleen. Repopulation by transplanted DPPIV+ hepatocytes was detected in vivo with Cy5.5-conjugated DPPIV antibody using the eXplore Optix™ System (GE HealthCare). Results were compared with nontransplanted control animals and transplanted animals receiving nonspecific antibody. Optical imaging detected Cy5.5-specific fluorescence in the liver region of the transplanted animals, increasing in intensity with time, representing extensive host liver repopulation within 16 weeks following transplantation. A general pattern of donor cell multiplication emerged, with an initially accelerating growth curve and later plateau phase. In contrast, no specific fluorescence was detected in the control groups. Comparison with ex vivo immunofluorescence staining of liver sections confirmed the optical imaging results. Optical imaging constitutes a potent method of assessing the longitudinal kinetics of liver repopulation in the rat transplantation model. Our results provide a basis for the future development of clinical protocols for suitable fluorescent dyes and imaging technologies.
Introduction
Imaging in vivo is currently a very active field of research. Repetitive monitoring of reporter gene expression in intact living animals is crucial to many applications, including cell trafficking, gene therapy studies, and transgenic or mutant animal models (25). Furthermore, imaging techniques play a prominent role in the field of oncology. The method represents a potential clinical application not only for detecting tumors, but also for subtyping them based on their distinct receptor expression (12).
Several imaging technologies are being studied for noninvasive imaging and quantization of gene expression in living subjects. Some of the imaging modalities and established reporter gene trafficking methods include single photon emission computed tomography (SPECT), positron emission tomography (PET), and magnetic resonance imaging (MRI) (24). Nevertheless, these techniques are technically complex. Of the imaging techniques currently available clinically, only PET and SPECT have the sufficient sensitivity. However, it is difficult to label and image cell surface receptors or reporter genes, and the techniques require the use of radionuclides (16). Optical imaging not only offers the necessary sensitivity but also the added advantages of increased spatial resolution and the absence of ionizing radiation (4). The basis of the technique is fluorescence or bioluminescence, and constitutes a low-cost alternative for the real-time analysis of gene expression particularly suited to small-animal models. Unfortunately, a limitation of optical imaging is the restricted tissue penetration of light or fluorescence, respectively. However, near infrared fluorescence (NIRF) dyes can propagate through two or more centimeters of tissue, if careful preparation and sensitive detection techniques are chosen (18).
Recent studies have demonstrated that NIRF optical imaging technology may be used to monitor cell surface receptors (9,12) or antigens (21). The NIRF dye Cy5.5 can be conjugated to a targeting component such as an antibody. Once applied to a living rodent (e.g., by IV injection), the labeled antibody binds to the target antigen in vivo and, on excitation, the fluorescence may then be detected. We considered this approach as potentially applicable to our recently published irradiation rat transplantation model, in which the aim was to identify and track genetically encoded hepatocytes within a negative background liver (5). Hepatocytes isolated from dipeptidylpeptidase IV (DPPIV)-positive wild-type Fisher 344 rats were transplanted into DPPIV-negative hosts following 30% partial hepatectomy (PH) and prior treatment with external beam irradiation at a dose of 25 Gy. Transplanted DPPIV+ hepatocytes integrate rapidly into the irradiated liver tissue and proliferated extensively, finally repopulating the DPPIV- host liver.
The aim of the present study was to establish a method of tracking donor hepatocytes proliferating in the host in vivo. Therefore, liver repopulation by transplanted DPPIV+ hepatocytes was detected in vivo with Cy5.5-conjugated DPPIV antibody using the eXplore Optix™ System by GE HealthCare. We presented evidence of fluorescence specific to Cy5.5 in the liver region of the transplanted animals increasing in intensity with time up to the end of the study. Additionally, monitoring of hepatic repopulation in vivo was validated ex vivo by immunofluorescence analyses.
Materials and Methods
Materials
Chemicals and reagents were supplied by Sigma-Aldrich (Munich, Germany) unless specified otherwise. Mouse monoclonal anti-DPPIV (OX-61) and IgG antibody were obtained from BD Transduction Laboratories (Heidelberg, Germany). The anti-DPPIV (labeling grade 3.5) as well as the nonspecific control IgG (labeling grade 2.0) were conjugated with Cy5.5 by Squarix Biotechnology (Marl, Germany). Rabbit polyclonal anti-connexin 32 (Cx32) was from Sigma-Aldrich (Munich, Germany). Alexa Fluor 568 goat anti-mouse IgG and Alexa Fluor 488 goat anti-rabbit IgG were purchased from Molecular Probes (Germany).
Animals
As recipients, a strain of DPPIV-deficient Fisher 344 rats was established in our animal care facilities. Syngeneic donor Fisher 344 rats were purchased from Charles River Germany. The animals were housed under 12-h light/12-h dark cycles with standard rodent feed and water available ad libitum. All animal breeding, care, and experimentation procedures were in accordance with German national and regional legislation on animal protection. Irradiation and surgical procedures were performed under sevofluran anesthesia. Female recipients with a mass of 250–300 g underwent external beam, computed tomography (CT)-based partial liver irradiation 4 days prior to PH, and hepatocyte transplantation. Control animals were subjected to the same procedure without hepatocyte transplantation. Optical imaging to detect the extent of liver repopulation at 1, 4, and 11 days, and 4, 8, 12, and 16 weeks following hepatocyte transplantation was implemented on three rats of the experimental group. Thereafter, rats were sacrificed for tissue analysis. Tissue samples from each liver lobe were excised and snap frozen in 2-methylbutane at −70°C. Cryosections of 5 μm thickness were fixed in ice cold acetone for 10 min.
Irradiation
External beam irradiation of the liver was performed as described recently (5). The target volume was irradiated with 6 MV photons (dose rate of 2.4 Gy/min) using a Varian Clinac 600 C accelerator. A total dose of 25 Gy was delivered using an anterior-posterior and posterior-anterior treatment technique.
Cell Isolation
Hepatocytes were isolated employing two-step in situ collagenase perfusion of the liver first described by Seglen (27). The obtained hepatocyte and nonparenchymal cell suspension was further purified using Percoll density gradient centrifugation (Pharmacia, Uppsala, Sweden). Freshly isolated hepatocytes (purification grade ~98%) displaying a vitality greater than 90% (tested with trypan blue exclusion) and cell attachment greater than 70% proved to be sufficient for further transplantation experiments. In preparation of cytospinning, 3 × 104 cells were centrifuged onto glass slides and fixed in ice cold acetone for 10 min. Immunohistological assessment of PECAM as an endothelial marker and CK7 as a bile duct marker ruled out any contamination with nonparenchymal cells (data not shown). However, a few blood cells (erythrocytes and leucocytes) were detected.
Partial Hepatectomy and Hepatocyte Transplantation
Four days after irradiation, a partial hepatectomy was carried out, followed by hepatocyte transplantation. The left liver lobule was removed by central ligation (1/3 hepatectomy). Thereafter, the spleen was mobilized, and 15 × 106 freshly isolated DPPIV+ hepatocytes were slowly injected. The transplanted cells subsequently effused into the portal venous system, from which they migrated into the host liver parenchyma. The laparotomy incision was closed by continuous suture.
Optical Imaging and Image Analysis
The in vivo optical imaging experiments were performed using the eXplore Optix™ system developed by GE HealthCare (London, UK). This system is a timedomain fluorescence imager. In contrast with most other systems available on the market, the eXplore Optix™ uses pulsed monochromatic laser diodes for excitation. The system scans a grid in a single pass, with a maximum resolution of 0.5 mm. A large number of excitation pulses are applied at every grid position. The fluorescence intensity is detected using a time correlated single photon counter. The detector and the pulsed excitation enable the creation of a time-of-flight histogram for the measured fluorescence photons, referred to as temporal time spread function (TPSF). The TPSF contains not only intensity information, but also the information of the mean depth and concentration of the fluorescence dye below the grid point, which cannot be determined with other systems. In addition, the lifetime of the fluorescence transition may be calculated from the TPSF decay slope. This value allows us to discriminate between the fluorescence dye and the autofluorescent background of the animal. All scans were performed with an excitation wavelength of 680 nm, and a 700-nm-long pass filter. The results are displayed as overlay to a camera image obtained directly before the measurement. All images were scaled to the maximum fluorescence intensity measured in normalized photon counts (NPC). For each rat, the total fluorescence intensity was determined by operator-defined region of interest (ROI) respecting the anatomical site of the rat liver.
Three weeks prior to the imaging experiments, animals received a special diet [GLP 3893 (Nafag 890) from Provimi Kliba, Kaiseraugst, Switzerland] simply to avoid nonspecific fluorescence from the digestive system, as no component of this particular diet fluoresces in the region of the spectrum examined. The change in diet has no observable effect otherwise. Furthermore, the rats were shaved in the ROI (ventral part of the abdomen) as the fur causes unnecessary scattering and absorption.
Cy5.5-conjugated anti-DPPIV and nonspecific control antibodies were administered intravenously via the tail vein (1.16 mg/kg). Nontransplanted animals injected with Cy5.5-conjugated specific antibody and transplanted animals injected with nonspecific Cy5.5-conjugated IgG antibody served as controls in all experiments. Measurements were performed 7 h after antibody injection (see Results section).
Immunofluorescence Colocalization Studies
The transplanted animals were sacrificed 16 weeks following transplantation, immediately after the last course of in vivo imaging. Cryostat sections were washed in Tris Buffer and incubated with a mixture of anti-Cx32 at 1:5000 and anti-DPPIV at 1:200, 4°C overnight. Immunolabeling was performed using Alexa 488-conjugated goat anti-rabbit IgG and Alexa Fluor 568 goat anti-mouse IgG (1:400, 1 h at RT). Slides were finally covered with Vectashield® mounting medium with DAPI (1 μl/ml) (Vector Laboratories, UK). Multiple immunofluorescence-conjugated specimens were evaluated in a confocal microscope (LEICA DM IRE2, Bensheim, Germany).
Results
Determination of Optimum Measurement Settings
No treatment-related morbidity or mortality was observed in the rats undergoing vivo imaging. First of all, the experimental settings had to be determined. Therefore, transplanted rats (2 months of repopulation) were injected via the tail vein with Cy5.5-conjugated specific antibody and subjected to optical imaging. The fluorescence detected proved to be highly specific. The signal to noise ratio was 14.6 (Fig. 1A), which allows a high degree of discrimination between signals in the ROI and the inherent background noise of the acquired picture (20). Additionally, when labeled tissue is irradiated with a short duration light pulse, the reporter does not emit instantaneously, but over a characteristic lifetime (1). In our experiments, the fluorescence lifetime of the Cy5.5-conjugated antibody at the liver site was mapped to 1.67 ns (Fig. 1C), demonstrating that the rat tissue in the target liver region had adequate optical properties.
Fluorescence intensity specific to Cy5.5-conjugated anti-DPPIV in the liver region of transplanted animal 7 h after antibody injection with signal (liver) to noise (background) ratio of 14.6 (A). At this point in time, only minor levels of fluorescence were detected in control animals [transplanted animals injected with conjugated unspecific IgG-antibody (B) and nontransplanted animals injected with conjugated specific antibody (D)]. (C) The life time of Cy5.5. The scale in (A), (B), and (D) represents fluorescence intensity measured as normalized photon counts (NPC). The kinetics of total liver fluorescence intensity from 1 to 10 h are summarized in (E). After elimination of untrapped conjugates, the optimal time interval between antibody application and measurements was found to be 7 h; the controls did not reveal any significant fluorescence during the course of experiment.
A time delay between injection of the fluorescent probe and imaging helps to clear the untrapped probe and to differentiate the target-specific uptake of fluorescence. Therefore, transplanted and control rats were injected with Cy5.5-conjugated specific as well as nonspecific antibody and subjected to optical imaging at various time points between 1 and 10 h following injection (Fig. 1E). We observed an initially high level of total fluorescence intensity in the region of the liver in repopulated animals administered with specific antibody, as seen 1 h after injection. As a result of redistribution in the plasma space, the emission rapidly dropped to its lowest level after 2 h. Thereafter, positive fluorescent conjugates were found to be enriched in the liver region (specific uptake) and could be clearly distinguished from the surrounding background tissue up to at least 10 h. However, the untrapped probe was significantly cleared from the liver between 5 and 7 h, with a drop in total fluorescence intensity. Thereafter, a steady state of antibody distribution and stable emission in the liver region was detectable between 7 and 10 h. Although the Cy5.5-conjugated specific antibody follows pharmacokinetic rules including excretion in the enterohepatic circulation (20), we could not detect any interference with the specific fluorescence signals in the liver up to 24 h. However, there was noticeable excretion via the urinary system with accumulation of fluophores in the urinary bladder from 5 h onwards (data not shown).
As a negative control, nontransplanted animals that had received the full protocol of pretreatment with irradiation and partial hepatectomy were also injected with the Cy5.5-conjugated specific antibody (Fig. 1D, E). The total fluorescence in the liver region was very low during the course of observation and could be clearly distinguished from the transplanted animals from 3 h onwards. As a second control experiment, transplanted animals received an injection of conjugated nonspecific control IgG and only very low intensities of specific fluorescence signals could be detected at any point in time in the liver region (Fig. 1B, E). In summary, we defined the optimal interval between antibody application and measurement of fluorescence intensity for further experiments as being 7 h.
In Vivo Imaging of Liver Repopulation
Optical imaging of recipient animals was performed by assessing a ROI measuring 54 mm2 in the area of greatest fluorescence intensity of the liver 1, 4, and 11 days, and 4, 8, 12, and 16 weeks after hepatocyte transplantation (Fig. 2A). After IV injection of Cy5.5-conjugated anti-DPPIV, optical images revealed a discernible signal in the recipient liver as early as after 1 day. Continuously increasing fluorescence intensities were observed in all three animals during the course of experiment, indicating the progressive repopulation of the livers by the DPPIV+ donor cells and their descendents. Although the protocols for the pretreatment of the host liver and cell transplantation were identical in all three animals, the data revealed differences in the total levels of fluorescence intensity with a maximum absolute variation of 191 NPC. At the endpoint of the experiment, we observed a total fluorescence intensity of 1,314 NPC in the region of the liver for animal 1, 1,386 NPC for animal 2, and 1,195 NPC for animal 3.
Representative optical imaging of transplanted animals at 1, 4, and 11 days, and 4, 8, 12, and 16 weeks; fluorescence scale as indicated; d = days, w = weeks (A). Increasing levels of fluorescence intensity were detected in all three animals during the course of host liver repopulation. The kinetics of fluorescence intensity monitoring liver repopulation are visualized in (B). The maximum values at 16 weeks were normalized to 100%, to enable a better comparison of the respective repopulation curves.
For comparison, Figure 2B summarizes the kinetics of fluorescence intensities, which were normalized to the maximum values at 16 weeks. In general, all three animals demonstrated ever increasing, although varying, fluorescence intensities, representative of the different growth curves of the donor hepatocytes. However, a common pattern could be detected. During the first 11 days, the multiplication of transplanted cells was comparatively fast. Thereafter, liver repopulation seemed to slow down, reaching a plateau phase after approximately 12 weeks.
Confirmation of Liver Repopulation by Immunofluorescence Microscopy
The isolated donor hepatocyte suspension was determined to have cell vitality of greater than 90%. Incubation with the anti-DPPIV antibody detected by Alexa Fluor 568 revealed that all vital hepatocytes displayed a strong immunoreactivity to the donor specific antigen (Fig. 3A).
Extensive repopulation of the host liver by transplanted DPPIV+ hepatocytes following preparative stimulus of external beam irradiation and partial hepatectomy. Multilayer immunofluorescence micrographs with DPPIV identified in red, gap junction protein Cx32 in green, and cell nuclei counterstaining with DAPI in blue. Expression of donor-specific and membrane-bound DPPIV in isolated donor cells (A). Massive repopulation of a major liver section as seen 16 weeks after transplantation (B). Large and in part confluent clusters of cells derived from donors are detectable; the image contains six merged adjacent microscopic fields. Representative area showing donor cell clusters emerging from the portal fields (C). Selected area at a higher magnification (D): DPPIV is expressed in the basolateral bile canalicular membrane domains of hepatocytes derived from donors revealing a pattern of star and garland-like structures. DPPIV staining is flanked by well-dotted labeling of Cx32, indicating highly differentiated hepatocytes. PV = portal vein, CV = central vein. Original magnifications as indicated by the scale bars.
Immunofluorescence colocalization studies were performed to confirm the large degree of liver repopulation detected by optical imaging. Therefore, animals were sacrificed after the last course of optical imaging 16 weeks following transplantation. Representative specimens from all liver lobules were subjected to tissue analysis and immunofluorescence microscopy. The overall liver structure appeared undamaged without signs of inflammation, fibrosis, or displacing growth. Large clusters of hepatocytes expressing the donor-specific antigen were distributed throughout the negative recipient parenchyma (Fig. 3B, C). The DPPIV+ cell clusters appeared to have emerged from the portal fields, the original location of donor cell integration. The combined stimulus of irradiation and partial hepatectomy continuously triggered liver repopulation, and donor cells progressively extended out into the parenchyma, finally forming clusters that grew to confluence. After 16 weeks, these clusters were between 100 and 500 cells in diameter. In more detail, donor cells and their descendents expressed DPPIV in a bile canalicular pattern unique to fully intact and metabolically active hepatocytes. Multiple layer immunofluorescence studies revealed that cells derived from donors coexpressed connexin 32 (Cx32), the major hepatic gap junction protein, enabling direct cell-to-cell communication, indicative of an exceptionally high degree of cell differentiation (Fig. 3D).
In summary, liver repopulation monitored in vivo was confirmed by immunofluorescence staining ex vivo. We demonstrated substantial proliferation of transplanted cells within 16 weeks. At this point in time, donor cells and their numerous descendents were fully integrated into the recipient hepatic parenchyma and displayed metabolic features of mature hepatocytes.
Discussion
To date, ex vivo techniques, such as morphometric analysis based on histochemistry or immunolabeling, or the detection of DPPIV in tissue extracts, for example using enzymatic or molecular assay and FACS analysis, represented the gold standard to determine the extent of liver repopulation in the established DPPIV+/- transplantation model (8,11,13,14,28). However, these techniques require organ harvesting before analysis, and numerous animals are necessary to perform the necessary measurements at various points in time. Furthermore, a large number of samples from one liver have to be processed and analyzed before extrapolation of results can actually represent the efficiency of liver repopulation in the whole organ. Moreover, the kinetics of donor cell multiplication are documented inadequately, as different animals are examined during the course of the study. Aiming to track organ reconstitution by transplanted cells in a single animal in particular, one may consider needle biopsies from the liver as the alternative. However, this approach is rather demanding, being technically difficult to perform on small animals. Furthermore, only small tissue samples may be retrieved with the potential risk of significant sampling errors occurring (22). Optical imaging not only provides exact data in longitudinal studies, the in vivo detection method also provides the unique opportunity to reduce the number of animals required in the experimental setting.
Considering the obstacles of established ex vivo techniques mentioned above, we developed an optical imaging technique based on the detection of a fluorescence-conjugated monoclonal antibody, enabling longitudinal studies to identify liver repopulation in the same animal. An external energy source is required for the excitation of the fluorescent dye with the associated limitation of reduced tissue depth penetration when compared with scintigraphy and MRI. However, biological tissues have minimal absorbance in the NIRF window (650–900 nm), thus allowing efficient photon penetration into and out of tissue with low intratissue scattering (3,12). In our experiments, we further reduced scattering and autofluorescence by shaving the animals prior to investigation, thus enabling deeper and more specific tissue imaging. Additionally, a special animal diet was provided to reduce the associated background noise usually prominent in the upper gastrointestinal tract.
Initial studies determined an optimum time interval for the application of the fluorescent conjugate. Seven hours after its IV injection, the antibody was well targeted with high fluorescence intensity in the liver region. However, it was important to demonstrate that the fluorescence was specific to Cy5.5 and caused by the specific binding of the antibody to its target antigen. Firstly, the signal lifetime proved to be specific for Cy5.5. Secondly, we excluded the possibility of the measured fluorescence coming from nonspecific antibody binding by performing two important control experiments: nontransplanted animals were injected with specific Cy5.5-conjugated antibody and only very small levels of fluorescence were detectable in the liver. Additionally, when transplanted animals received injections of Cy5.5-conjugated IgG, no significant fluorescence was obtained. Thirdly, we could clearly demonstrate a high uptake of the specific fluophores in all transplanted animals and more significantly in a time-dependent manner, with fluorescence increasing in parallel with liver repopulation. Finally, we were able to demonstrate a strong correlation between the Cy5.5-related fluorescence of repopulated liver and the visualization of donor cell clusters on cryo-sections utilizing the immunoreactivity of DPPIV at the end of the experiment. For the first time, optical imaging presented us with the ability to compare the growth curves of donor cells in different animals. Indeed, the kinetics of liver repopulation proved to be inhomogeneous. Nevertheless, a general pattern of donor cell multiplication emerged with an initially accelerating growth curve and late plateau phase.
The noninvasive, real-time analysis of molecular events in intact living mammals is an active area of current research (36). Optical imaging modalities are rapid, easy to use, and can be readily applied to study disease or regeneration processes in vivo. In contrast to SPECT and PET, optical imaging based on NIRF has the advantage of not requiring radioactive agents. MRI has been applied successfully to track magnetically labeled cells with high anatomical resolution (6,32,35). However, the sensitivity of MRI-based cell tracking procedures is limited. Taking into account that the procedure is time consuming, technically complicated, and expensive, the feasibility of serial follow up imaging has to be questioned (16). In contrast, optical imaging provides us with a convenient method of monitoring biological processes in real time. Furthermore, the technology is highly sensitive, operationally simple, amenable to miniaturization, and potentially mobile. Thus, optical imaging could prove to be well suited as an imaging method for patients.
The workgroup of Landis also presented with a sophisticated, noninvasive method for the evaluation of liver repopulation (15). They used a transgenic mouse transplantation model, in which donor hepatocytes were monitored via the production of phosphocreatine, permitting 31P magnetic resonance spectroscopic imaging of liver. The major advantage of this approach is that no further agent has to be applied to mediate the optical detection of hepatocyte engraftment and repopulation. However, the approach is based on expression of the transgenic protein for imaging. In our transplantation model, donor hepatocytes and descendents were visualized by conjugated antibodies detecting a reporter cell surface antigen (DPPIV). Although there is a need for the application of foreign proteins, the eXplore Optix™ imaging technique is nonetheless conceivable in a clinical setting. In oncology, the diagnostic and therapeutic use of signaling agents and tumor-targeting molecules such as antibodies, peptides, and ligands have been investigated extensively and are used routinely nowadays (23,30,31). For instance, highly specific monoclonal antibodies to receptors on the surface of cancer cells are used as contrast agents in PET imaging [e.g., octreotide in neuroendocrine tumors ((2)] or as treatment of breast and prostate cancer (radioimmunotherapy) (19) as well as advanced stages of colorectal cancer (blockage of cell signaling) (10,26).
In further experiments, we would like to correlate the relative result of total fluorescence intensities from transplanted livers with the actual number of repopulating donor cells. Therefore, the livers of transplanted animals may have to be digested enzymatically and the isolated cell suspension be quantitatively analyzed for DPPIV+ hepatocytes by FACS (11). In addition, we are aiming to combine molecular optical imaging with its relatively limited anatomical resolution with flat panel volumetric computed tomography of the liver (33). This approach may provide us with high-resolution information concerning the exact anatomical distribution of cells shortly after transplantation and reveal differences in the repopulation kinetics in the liver lobules. It could even go so far as to reveal differences in repopulation efficiencies when comparing various preparative regimens for liver repopulation (7,34).
Although there have been significant advances in hepatocyte transplantation recently, the need for a robust noninvasive method for the longitudinal and quantitative evaluation of donor cell engraftment, survival, and proliferation is of central importance to improve experimental and likewise clinical strategies. Optical imaging after injection of a fluorescent probe (Cy5.5-conjugated anti-DPPIV) certainly has the potential to monitor liver repopulation following hepatocyte transplantation in the rat transplantation model. We believe that these findings could well encourage further research into the development of clinically applicable fluorescent dyes and imaging technologies aiming to monitor transplanted patients (17,29).
Footnotes
Acknowledgments
The authors would like to thank Sabine Wolfgramm and Sabrina Goldmann for their excellent work in preparing the immunolabeling figures for this article.