|Home | About | Journals | Submit | Contact Us | Français|
Since Amyloid precursor protein (APP) fragments exist in many tissues throughout the body, including the fluid compartments of blood, they have been the focus of numerous investigations into their potential as a biomarker of Alzheimer’s disease (AD). Using immunohistochemistry, immuno-electron microscopy, Western blotting and quantitative RT-PCR we examined whether APP processing in leukocytes is analogous to APP processing in the brain. We show APP immunoreactivity at light and electron microscopic levels in the cytoplasm and nucleus of peripheral blood leukocytes (PBL) yet our Western blotting data demonstrated that brain and PBL contain different APP fragments and differentially expressed APP processing enzymes. ADAM10, nicastrin and BACE2 were present in brain but were undetected in PBL. PS1 and BACE1 were detected in both tissues but showed different patterns in Western blots. Quantitative PCR results identified Neprilysin as the only processing enzyme we interrogated where Western and qPCR data coincided. Although our data on differential processing of APP in brain and PBL point to exercising caution when generalizing between blood and brain with regard to mechanisms, they have no implications regarding utility as biomarkers.
Molecules resulting from the posttranslational processing of amyloid precursor protein (APP) play a central role in the pathophysiology of Alzheimer’s disease (Hardy and Selkoe, 2002; Walsh et al., 2005; Haass and Selkoe, 2007; Shankar et al., 2007; Palop and Mucke 2010; Lublin and Gandy, 2010; Shimada et al., 2011; Jonsson et al., 2012), and are also ubiquitously present in many tissues throughout the body (e.g. Roher et al., 2009). In the search for biomarkers of AD, these two facts have provided a rationale for examining easily obtained peripheral tissues for biomarkers of AD, and APP fragments in blood have received special attention in these studies (Borroni et al., 2010; Takeda et al., 2010; Hampel et al., 2010; Mukaetova-Ladinska, et al., 2012). The APP products circulating in the fluid compartment(s) of blood have a number of potential sources, including from brain, endothelial cells, platelets and leukocytes (e.g. Konig et al., 1992; Sandbrink et al., 1994). To minimize such multiplicity of sources, we restricted our investigation to the leukocyte compartment of blood. Since it has been shown that leukocytes express APP (e.g. Konig et al., 1992; Li et al., 1999) and that peptides produced from APP may serve as AD biomarkers (e.g. Holtzman, 2011; Mayeux and Schupf, 2011), we hypothesized that APP processing in leukocytes may be similar to APP processing in the brain. In view of data showing transport of a C-terminal fragment of APP to translocate to neuronal nuclei (Kimberly et al., 2001) and to modulate transcription (Cao and Sudhof, 2001, 2004; Muller et al., 2008; Buoso et al., 2010), we used immunohistochemistry, immuno-electron microscopy, Western blotting and quantitative RT- PCR to examine APP in cytoplasm and nucleus of peripheral blood leukocytes (PBL) as well as the expression of selected enzymes involved in the processing of APP. Selected leukocyte data were compared to brain. We conclude that although APP fragments appear to distribute similarly in both neurons and PBL, the APP fragment species in these two tissues are not completely comparable. Our data suggest this lack of comparability may be related to differential expression of APP processing enzymes in brain and in leukocytes. Previously demonstrated differential splicing of APP message in these two tissues (Konig et al., 1992) may also contribute to the observed lack of comparability. This lack of comparability does not signify leukocyte gene products cannot provide useful biomarkers. It does, however, suggest a need for caution in generalizing between blood and brain with regard to mechanisms.
Samples were obtained from a total of twenty-six individuals. All specimens were collected with informed consent and under IRB-approved protocols.
For Western blots, immunohistochemistry (IHC) and immunoelectron microscopy, blood samples were obtained from four Alzheimer’s disease, three age matched non-demented controls, three young controls (in their 20’s), and one Parkinson’s disease. The AD, PD and matched controls came from the Alzheimer’s Disease Center at the University of Rochester Medical Center and the Cleo Roberts Clinic at Banner Sun Health Research Institute. Young controls were laboratory personnel. Blood samples were drawn by an experienced phlebotomist and processed for IHC, immuno-electron microscopy and Western blotting. Postmortem human superior frontal gyrus was obtained from two Alzheimer’s and two age matched controls from the brain bank at the University of Rochester Alzheimer’s Disease Center.
For quantitative RT-PCR, blood samples were obtained from two Alzheimer’s, two aged matched non-demented controls, two young controls (in their 20’s) and one Parkinson’s disease case. The AD, PD and matched controls for qPCR were from the Cleo Roberts Clinic at Banner Sun Health Research Institute. Young controls were laboratory personnel. Postmortem human superior frontal gyrus (SFG) for qPCR was obtained from two Alzheimer’s and two aged matched controls from the brain bank at Banner Sun Health Research Institute.
For IHC, blood was drawn into EDTA tubes. Leukocytes were then isolated by lysing RBC with 1.5M ammonium chloride, 0.1M sodium bicarbonate and 10mM EDTA, centrifuged at 2000 rpm for 5 minutes and after a series of washes in 1X PBS, resuspended in 1X PBS and counted on a hemocytometer. 1.5–2.0 × 105 cells were applied onto poly-L-lysine coated glass slides and allowed to air dry and stored at RT until use. Slides were washed with PBS then fixed for 15 minutes with 2% paraformaldehyde in PBS. After rinsing with PBS and PBS + 0.1% Triton X-100 (Sigma) and blocking with normal goat serum or normal horse serum, the slides were incubated overnight in a moist chamber at 4°C in primary antibodies targeted at the C-terminus of APP (rabbit polyclonal- Calbiochem) 1:500. After washing, bound primary antibodies were detected by Alexa 488-conjugated secondary antibodies (Molecular Probes). Slides were also treated with Hoechst Nuclear stain (1:10,000) and Sudan Black (0.3 %) to suppress autofluorescence.
Leukocytes were isolated as stated above and then applied onto Superfrost Plus glass slides (VWR) and allowed to air dry and stored at RT until use. Slides were washed with PBS then fixed for 10 minutes with 2% paraformaldehyde in PBS. After rinsing with PBS and PBS + 0.1% Triton X-100 (Sigma) and blocking with 1% BSA (Sigma), the slides were incubated overnight in a moist chamber at 4°C in primary antibodies targeted at the C-terminus of APP (rabbit polyclonal- Calbiochem) 1:500. After washing, bound primary antibodies were detected by Alexa 488-conjugated secondary antibodies (Molecular Probes). Slides were also treated with red-fluorescent 7-aminoactinomycin D (7-AAD) nuclear stain (Molcular Probes) at 1:50 and Sudan Black (0.3 %) to suppress autofluorescence. The microscope used was an Olympus IX70 with argon and krypton lasers. Imaging done with Fluoview (Olympus) software.
Buffy coats were fixed in 0.1M sodium phosphate buffered 4.0% paraformaldehyde for 3 hours, rinsed in phosphate buffer, pelleted into 2.0% agarose (low temperature melting), dehydrated in a graded series of methanol to 100% at −20°C, transitioned into Lowicryl K4M resin at −20°C and polymerized in capsules for 5 days under ultraviolet light at −20°C. Thin sections (70–80nm) were placed onto formvar coated nickel grids, blocked with a 1/10 dilution of PowerBlock (Biogenex, San Ramon, CA), rinsed (5 minutes × 4) in distilled H2O and incubated in a 1:100 dilution of rabbit polyclonal anti-Amyloid Precursor Protein, C-terminal (Calbiochem) overnight at 4°C. The grids were rinsed in phosphate buffered saline (PBS) (5 minutes × 5) and incubated 30 minutes in a 1/20 dilution of Protein A conjugated to 15nm gold. Negative controls substituted the primary with PBS only and were incubated in a 1/20 dilution of Protein A 15nm gold. The grids were rinsed in PBS (5 minutes × 4 times) and incubated 10 minutes in 2.5% glutaraldehyde to stabilize the gold labeled antigen. The grids were rinsed in distilled H2O (5 minutes × 4), stained for 8 minutes in aqueous uranyl acetate and 10 minutes in lead citrate. The labeled sections on the grids were photographed using a Hitachi 7100 transmission electron microscope with an attached MegaView III(Soft-Imaging System, Lakewood, CO) digital camera.
For Western blotting, blood was drawn into EDTA tubes. Leukocytes were then isolated by lysing RBC with 1.5M ammonium chloride, 0.1M sodium bicarbonate and 10mM EDTA and a series of washes in 1X PBS and spun at 1600 rpm for 10 minutes. Leukocyte pellet was resuspended in lysis buffer containing 20mM Tris, pH 7.5, 0.5% Nonidet P40, 1mM EDTA, 0.1M NaCl, 1 mM PMSF and protease inhibitor cocktail (Sigma P8340) and phosphatase inhibitor cocktails (Sigma P2850, P5726). After resuspension, the tubes were placed on ice for 20 minutes and vortexed intermittently during the incubation then centrifuged at 15,000 × g at 4°C for 5 minutes. (Jenkins et al., 2001). Total protein concentration of supernatant was determined by DC Protein Assay (Bio-Rad 500-0112) based on the Lowry method. Brain tissue was homogenized in the same lysis buffer (as above) and total protein concentration determined as above.
Proteins were separated by SDS-PAGE on Tris-HCl, 10–20% gradient gels (Bio Rad) and transferred to PVDF membranes (Immuno-Blot; Bio Rad), incubated in primary antibodies’ as detailed in Table 1, then in HRP-goat anti rabbit secondary (Jackson ImmunoResearch) and detected by enhanced chemiluminescence technique (GE Healthcare).
Each blot contained the same samples in the same order at the total protein concentrations listed (Table 1). Equal loading of gels (10 ug) was confirmed by Coomassie stain (data not shown). Table 1 also lists the antibodies and concentrations used. All Western protocols were tested and optimized for primary and secondary antibody concentrations and protein loading amounts.
2.5 mL of fresh blood was drawn into PAXgene ™ Blood RNA Tubes (Qiagen, Valencia, CA, USA) and total RNA extracted using PAXgene™ Blood miRNA Kit (Qiagen). Total RNA was extracted from brain tissue using the RNeasy Mini Kit (Qiagen) according to manufacturers instructions. RNA integrity and concentration was determined by Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). The average RIN for all samples used was 8.8. The quantitative RT-PCR procedure was as follows: 500 ng of total RNA was reverse transcribed with the High Capacity RNA-to-cDNA Kit (Applied Biosystems, Foster City, CA, USA). 2ul of a 1:5 dilution of cDNA was combined with TaqMan® Gene Expression Master Mix (Applied Biosystems) and the TaqMan® Gene Expression Assay (Applied Biosystems) in a 30 ul reaction. The quantitative RT-PCR amplifications were run in triplicate on an iCycler iQ™ (Bio-Rad, Hercules, CA, USA). Thermal cycling conditions were: 2 minutes at 60 degrees C, 10 minutes at 95 degrees C and 40 cycles of denaturation at 95°C for 15 seconds and annealing and extension at 60 degrees C for 1 minute.
TaqMan® Assays were for APP (amyloid beta precursor protein), ADAM10, BACE1, BACE2, NCSTN (nicastrin), PSEN1 (presenilin 1), and NEP (neprilysin). GUSB (beta glucuronidase) was used as an endogenous control since its expression was found to be uniform across all samples. Normalized data were then determined as a ratio using the (2−delta delta CT) method (Livak and Schmittgen, 2001). Leukocyte gene expression fold change was calculated relative to gene expression in the SFG. Primers for the PCR reaction were selected to cover, as closely as possible, the same region as the immunogen used to generate each antibody (Table 2).
Figure 1A shows localization of immunoreactivity for APP using Calbiochem antibody #171610 (Table 1) in AD peripheral blood leukocytes. This antibody will detect the soluble C-terminal fragments CTFγ CTFα and CTFβ as well as full length of isoforms APP695, APP751 and APP770. There is both diffuse and punctate immunoreactivity in the cytoplasm of most peripheral blood leukocytes. In addition, there appears to be punctate immunoreactivity in some leukocyte nuclei. It is, however, not possible to determine in these light level images whether this immunoreactivity was within the nucleus or above/below it.
To better localize APP immunoreactivity in PBL we examined a different AD sample with confocal microscopy using the same Calbiochem antibody as in the epifluorescence experiment. Figure 1B shows a similar pattern of immunoreactivity as in Figure 1A with diffuse and punctate immunoreactivity in the cytoplasm of most PBL and the apparent colocalization of APP within the nucleus, demonstrated by the yellow labeling.
Since the apparent immunoreactivity in the nucleus at the epifluorescence and confocal microscopic level could only suggest but not prove nuclear localization, we undertook an immuno-electron microscopic examination of APP in AD peripheral blood leukocytes in order to establish whether or not the immunoreactivity at the light level within leukocyte nuclei was indeed within nuclei. We used the same Calbiochem antibody as in our immunohistochemistry experiments. Figure 2A shows ring-like immuno gold particles within the nucleus. Figure 2B shows immuno gold reactivity in association with a cytoplasmic organelle that probably corresponds to the cytoplasmic punctate immunoreactivity seen at the light level. We were not able to definitively identify the structure with which this cytoplasmic immunoreactivity is associated.
In order to characterize the nature of the leukocyte immunoreactivity seen at the light and e.m. levels, we conducted Western blotting of total protein extracted from peripheral blood leukocytes. Since immunoreactivity of APP peptides has also been described in brain (e.g. Cao and Sudhof, 2001; Gouras et al., 2005; Mohamed and Posse de Chaves, 2011), we included brain samples in our Western blots.
Figures 3A and 3B show that Western blots of brain and blood samples, utilizing antibodies directed at the C-terminus of APP, produce different patterns depending on the specific antibody used. The Calbiochem antibody suggests a basic similarity in APP fragments in both brain and blood, although there are some differences between these two tissue types at higher molecular weights. The intense band at approximately 80 kDa found in both brain and blood is undetermined. However, the landmark study on APP by Selkoe et al., (1988), demonstrated strong, specific staining in immunoblots of human cortical homogenates at ~72 kDa from their antibody derived from the C-terminus of APP. Also, in monocytes and lymphocytes, Jung et al., (1999) found a reactive band at ~82 kDa using an antibody to the C-terminus of APP. Finally, it has been shown that caspase mediated cleavage in the C-terminus of APP produces a polypeptide of ~85 kDa (Gervais et al., 1999; Zhao et al., 2003). These studies demonstrate it may not be unexpected to see immunoreactivity at ~75–85 kDa with an antibody directed against the C-terminus of APP, just as we saw, but further examination is needed to determine its identity or why it was undetectable with the Sigma antibody. The higher molecular weight (~110 kDa) doublet is believed to be the glycosylated form of APP695 and the upper band (~120) is APP770 (Buxbaum et al., 1998).
On the other hand, the Sigma antibody shows major differences in both amount and isoforms of APP in brain and in PBL that suggest differences in processing of APP in these two tissue types. Although the higher molecular weight pattern is the same with both Sigma and Calbiochem antibodies, the undefined ~80 kDa band was virtually undetectable in brain or blood with the Sigma antibody. The band seen at ~60 kDa is believed to be a degradation product. The faint, somewhat diffuse bands below the 17.2 kDa marker are consistent with the expected sizes for CTFα (~9 kDa) and CTFβ (~11 kDa).
Figure 3 demonstrates that there are different patterns of immunoreactivity in brain and in peripheral blood leukocytes from control, AD and PD cases (the latter in Figure 3B), which may be attributed to differential posttranslational processing. In order to investigate tissue differences in APP processing, we examined expression of selected proteins involved in processing APP by Western blotting of brain and PBL protein extracts (Figure 4). We also examined expression of these genes at the transcript level by quantitative PCR (Figure 5). The Western data show that some of the APP processing proteins examined were detected in brain but appear undetectable in protein extracted from AD leukocytes. These were: ADAM10, nicastrin and BACE2 (the latter is, however, present in PD leukocytes). Other enzymes that were detected in protein extracted from both brain and AD leukocytes, but showed different patterns in leukocytes, are neprilysin, PS1 and BACE1, which was undetected in PD leukocytes.
These data are consistent with the suggestion that APP is differentially processed in brain and blood cells in AD.
In order to determine whether the differential appearance of APP processing enzymes in leukocytes and brain was a function of posttranslational processing or differential expression at the transcript level, we quantified transcript expression by qPCR. These data show that message levels of ADAM10, BACE2, nicastrin and PS1 are similar in both brain and leukocytes. BACE1 message was much reduced in leukocytes compared to brain, while neprilysin message was much increased in leukocytes compared to brain (Figure 5).
Our data demonstrate APP immunoreactivity at light and electron microscopic levels in both cytoplasm and nucleus of peripheral blood leukocytes. The apparent nuclear localization of some APP peptide(s) seen at the light level in fluorescence and confocal microscopy was confirmed by immunoelectron microscopy which showed gold particles labeled with an antibody targeting the C-terminus of APP as a ring like structure. We suggest that these structures may correspond to the “spherical nuclear spots” described by von Rotz et al. (2004) in both HEK293 and SH-SY5Y cells co-expressing AICD, Fe65 and Tip60.
Apparently similar localization have been previously described in the cytoplasm and nucleus of neurons (e.g. Cao and Sudhof, 2001; Muresan and Muresan, 2004; Ohkawara et al., 2011). However, our Western blotting data reveal that brain and PBL contain different APP fragments, suggesting that immunohistochemical demonstration of similar locations be viewed with caution, depending on properties of specific antibodies used. We further demonstrate differential expression of APP processing enzymes in brain and in PBL which is consistent with the appearance of differential APP isoforms in these two tissues. This differential of processing enzymes suggests, but does not prove, that the different fragments of APP in brain and in blood cells may be related to differences between these two tissues in posttranslational processing attributable to differential presence of APP processing enzymes in these two tissues. An additional contributor to the difference between brain and leukocyte APP products may also be derived from differential splice isoforms of APP mRNA in leukocytes and neurons (e.g. Sandbrink et al., 1996; Konig et al., 1992). Whether the differential processing we have described here results in different APP fragments localizing to the nucleus in brain and in leukocytes is not clarified by our data. If this were the case, it could have significant consequences for gene expression profiles in these two tissues.
Although there have been studies of specific subclasses of PBL in AD, particularly lymphocytes (e.g. Sultana et al., 2011; Stieler et al., 2012), we did not distinguish among subclasses. However, we argue our data are not the product of any small subclass of leukocytes because the vast majority of cells were immunopositive (Fig. 1) and the Western blot signals were strong (Figs. 3 and and4).4). In addition, we did not examine the entire family of molecules that contribute to posttranslational processing of APP. However, the molecules we did examine (Fig 4), in concert with the Western blotting of APP (Figure 3), serve to establish the principle of differential posttranslational processing of APP in two different tissues.
The discordance between our Western blot and our qPCR data is notable, with Western blots showing undetectable expression of ADAM 10, BACE 2 and nicastrin in PBL relative to brain, while qPCR data show approximately similar message expression of these molecules in brain and PBL. The apparent increased signal for neprilysin in leukocytes relative to brain seen in Figure 4 is consistent with increased signal for neprilysin relative to brain seen in quantitative PCR data (Figure 5). However, although our Western blots show decreased signal for ADAM10, BACE1, BACE2 and nicastrin in leukocytes compared to the signal in brain, our quantitative PCR data only show reduced message for BACE1 in leukocytes compared to brain. ADAM10, BACE2 and nicastrin show virtually equivalent levels of message expression in leukocytes and brain. This raises the question of whether our Western and qPCR data suggest differential stability of these enzymes in brain and in leukocytes, as has been shown for the differential half life of interferon regulatory factor 7 (IRF-7) in different cell types (e.g. Prakash and Levy, 2006). Additionally, it has long been recognized that message and protein expression need not always be in a 1:1 stoichiometric relationship (e.g. Gygi et al., 1999). We conclude that the apparent quantitative discrepancies between our Western data and our qPCR data may be related to a number of factors including differential protein stability as a function of cellular environment, enzyme activity, antibody sensitivity, regulation of translation, etc.
Although the major point of the data presented here was to compare APP processing in brain and in blood, it may be useful to comment that there were no detectable qualitative differences between AD and control leukocytes in expression of APP processing enzymes in Western blotting (Figure 4). Preliminary qRT-PCR data show similar results with a slight increase of BACE2 expression (data not shown). We also draw attention to the differences between the one PD case and AD and control cases in expression of BACE1, BACE2 and neprilysin seen in the Western blots.
Our data showing differential processing of APP in brain and in peripheral blood leukocytes emphasizes the potential variety of sources of APP products in the fluid compartment of blood. The quantitative contributions of differentially processed APP from vascular cells, including leukocytes, and that due to passage of blood through the brain in the composition of the fluid compartments of blood is unknown. From the point of view of blood compartments as providing biomarkers this is irrevelevant. If a biomarker works, that is sufficient. However, the extent to which cellular or fluid compartments of blood may be a reflection of events in the brain must be considered with caution.
This research was supported by NIH AGO-36400 (PC) and the State of Arizona and Arizona Department of Health Services, ADHS12-010553 (PC). We thank the brain bank and the Alzheimer’s Disease Center at the University of Rochester for brain and clinical blood samples. From Banner Sun Health Research Institute, we also thank Dr. Thomas Beach1, Lucia Sue, and their staff for tissue samples and post mortem evaluations, as well as the staff of Dr. Marwan Sabbagh for clinical blood samples.
1The Brain and Body Donation Program is supported by the National Institute of Neurological Disorders and Stroke (U24 NS072026 National Brain and Tissue Resource for Parkinson’s Disease and Related Disorders), the National Institute on Aging (P30 AG19610 Arizona Alzheimer’s Disease Core Center), the Arizona Department of Health Services (contract 211002, Arizona Alzheimer’s Research Center), the Arizona Biomedical Research Commission (contracts 4001, 0011, 05-901 and 1001 to the Arizona Parkinson’s Disease Consortium) and the Michael J. Fox Foundation for Parkinson’s Research.
5 Conflict of interest
To the best of their knowledge, the authors do not have an actual or potential conflict of interest with regard to the present research.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.