CD44 Splice Variants as Potential Players in Alzheimer’s Disease Pathology

Human tissue, RNA preparation, and RT-PCR analysis

Frozen postmortem human hippocampal tissue samples were obtained from The Netherlands Brain Bank (NBB, Netherlands Institute for Neuroscience, Amsterdam). All samples have been collected from donors for or from whom a written informed consent for brain autopsy and the use of the material and clinical information for research purposes had been obtained by the NBB. All material has been collected and used in compliance with the ethical and legal declaration of the Netherlands Brain Bank. Patients should have met the National Institute on Aging (NIA)-Reagan criteria for AD (Braak stage IV/C or higher with no vascular dementia, frontotemporal dementia, or Lewy body pathology [33]).

Total RNA was extracted from frozen hippocampal samples from AD patients (n = 10) and normal control individuals (n = 10) using EZ RNA kit (Biological Industries, Beit Haemek, Israel) and 1.5 μg of total RNA was reverse transcribed using Superscript II (Invitrogen, Carlsbad, CA). Semi-quantitative PCR was conducted on cDNA samples using oligonucleotides pairs in which one oligonucleotide was derived from variant exons V2-V10 sequences and the other from a 3’ or 5’ constant exon. Amplification products were purified from agarose gel (QIAquick gel extraction kit, Qiagen, Chatsworth, USA) and sequenced using one of the primers used in the PCR reaction. Sequence data was analyzed using BlastN algorithm on human RefSeq and genomic databases (NCBI). For quantitative analysis, qPCR was conducted on Applied Biosystems 7500 Real-Time PCR System using Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen) reagent and the following oligonucleotides (500 nM): CD44S (F: gcgcagatcgatttgaatataacc; R: gggcaaggtgctattgaaagc); CD44V3 (F: gcacttcaggaggttacatcttttaca, R: aaactgaggtgtctgtctctttcatc); CD44V6 (F: agaatccctgctaccatccag, R: gaatgtgtcttggtctccagct); CD44V10 (F: gacagaatccctgctaccaatagg, R: ggaatgtgtcttggtctcctgat); GAPDH (F: ccccttcattgacctcaactac, R: gagtccttccacgataccaaag).

The oligonucleotides used for CD44V6 and CD44V10 amplification are located on variant-constant exon junctions aimed to amplify the cDNA species that include only the targeted variant exon without additional variant exons. Single band products were verified using agarose gel electrophoresis and sequencing. The reaction mixture was first denatured at 95°C for 2 min and then 40 cycles were performed using the following protocol: 94°C for 15 s.; 60°C (or 1°C below lowest primer T_M) for 30 s.; 72°C for 30 s.

Standard curves were performed for each pair of primers, to confirm linearity of the qPCR reaction and in order to transform the Ct values to relative expression values.

qPCR using TaqMan probes

The Following TaqMan sets of primers and probes (5’-3’) were used to quantify the expression of CD44 mRNAs in the various gene expression experiments:

The qPCR reactions were performed in Applied Biosystem (ABI) 7300 Real time PCR. The reaction mixture was denatured at 95°C for 2 min and then 40 cycles were performed: 94°C for 15 s.; 60°C for 60 s.

Cell culture

SK-N-SH human neuroblastoma cell lines were obtained from the ATCC. Cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum, L-glutamine, and 1% penicillin-streptomycin mix (Biological Industries). In some of the experiments SK-N-SH cells were treated with 3 μM retinoic acid (Sigma-Aldrich, St. Louis, MO) for 5 days to allow the differentiation of the cells toward neuronal cells.

Frozen primary embryonic cortical C57/bl mouse neurons were obtained from Lonza (Basel, Switzerland) and maintained according to the manufacturer’s instructions. Primary embryonic cortical E15 Wistar rats neurons were prepared at Neuro-Sys SAS (Gardanne, France) as described by Callizot et al. [34]. Human Aβ_1-42 peptide (rPeptide (Bogart, GA, USA) or Bachem (Torrance, CA, USA) treatment of the cells was done after pre-incubation of the amyloid peptide in serum-free medium for 72 h at 37°C. Each experiment was repeated at least 2 times.

Knockdown of CD44 expression by siRNA

At least 3 siRNA sequences were screened for effective knockdown of relevant transcript using HeLa and B16-F10 for human and mouse-directed siRNA sequences, respectively. Cells were plated at 1.5×10⁵ cells in 24 well plates and transfected the day after with the different siRNA molecules (10 nM, IDT) using Lipofectamine RNAi Max (Invitrogen) according to manufacturer’s instructions. For analysis of siRNA efficiency, total RNA was isolated from the cells 48 h post transfection and reverse transcribed into cDNA using SuperScript II reverse transcriptase (Invitrogen). cDNAs were subjected to semi-quantitative PCRs or TaqMan qPCR using β-Actin as the reference gene. Effective siRNA sequences (sense strand) used for functional analysis are (3’ tt or uu are overhangdinucleiotides):

Cell viability

The XTT (Biological Industries) and Alamar Blue (AbD Serotec, Kidlington, UK) were used as cells viability indicators according to the manual instructions. XTT (1:3 dilution with growth medium) was by the end of each experiment and incubated for 1-2 h. Medium Absorbance was measured at 420 nm. Alamar blue was added at 20 to 40-fold dilution to the growth medium, incubated for 0.5–2 h and the medium fluorescence (excitation/emission at 485/590 nm) was read.

Caspase3 activity assays

Activity and proteolytic processing of caspase 3 was measured by using the EnzChek Caspase3 Assay kit (Life technologies, Carlsbad, USA) according to the manufacture’s guidelines. The basis for the assay was the cleaving of the fluorogenic effector caspase substrate Ac-DEVD-AMC. This substrate yields a bright, blue-fluorescent product (excitation/emission maxima ∼342/441 nm) upon proteolytic cleavage. In brief, the cells were lysed and 150 μg of protein was incubated with 200 μM Ac-DEVD-AMC. The increase in fluorescence was measured after 30 min using a fluorescence plate reader.

Immunostaining for survival of rat cortical neurons

Done by Neuro-sys SAS (Gardanne, France). Rat cortical neurons were treated for 24 h of Aβ_1-42 in the presence of 1 μg/ml CD44V10 polyclonal antibodies (Millipore, Billerica, USA) or CD44S mouse monoclonal antibody (clone Hermes-3, purified from hybridoma culture) or antibody-free control. Cells were fixed in ice-cold solution of 95% ethanol and permeabilized with 0.1% saponin. The cells were incubated with mouse monoclonal antibody against the neuronal-specific protein, microtubule-associated-protein 2 (MAP-2, Sigma-Aldrich) followed by incubation with Alexa Fluor 488 goat anti-mouse IgG (Molecular probe). The immune-labeled cultures were automatically examined with ImageXpress (Molecular device USA) at x20 magnification, 30 fields per well (representing ∼80% of the total surface of the well) from 6 wells were analyzed. The total number of neurons was automatically analyzed using Custom module editor (Molecular Devices, USA).

Immunohistochemical staining of hippocampal sections

Done by Premier Laboratory (Longmont, CO, USA). 4-5 μm sections were prepared from formalin fixed, paraffin embedded hippocampal tissues (Netherlands Brain Bank, Amsterdam, Netherlands). Sections were de-parafinized in Xylene and rehydrated through a series of alcohol treatments. Endogenous HRP activity was blocked with 3% H₂O₂. Antigen retrieval was achieved by proteinase K or by heating. After blocking in Protein Block Serum Free (Daco, Santa Clara, USA), Primary antibodies used were anti CD44S (HCAM, rat clone IM7, Santa Cruz, Dallas, USA), CD44V3 (mouse monoclonal vff-327, abcam, Cambridge, UK), CD44V6, and CD44V10 (both rabbit polyclonal, Millipore, USA). References for the specificity of the primary antibodies are given in the vendors’ webpages (e.g., [35] for IM7) Secondary antibodies used were against the relevant species conjugated to HRP, polymer-enhanced (Envision^®, Dako, Santa Clara, USA), using DAB+ as a substrate which was followed by hematoxylin staining. Negative controls were isotype matched serum controls for the rabbit polyclonal antibodies; it was non-immune rabbit serum at the same protein concentration as the primary antibody, and for the mouse antibodies, it was a multiple isotype control (Dako). For the rat antibody, it was a matched isotype control at the same protein concentration as the primary antibody.

Akt activation

SK-N-SH cells were grown for 24 h and then medium was replaced with serum-free medium and incubated for the indicated time in the presence or absence of anti-CD44V10 rabbit polyclonal antibody (Millipore, USA). Cells were lysed using RIPA lysis buffer (Sigma-Aldrich) supplemented with Complete^TM and PhosSTOP^TM (protease and phosphatase inhibitor cocktails, respectively, Roche diagnostics, Switzerland). Protein concentrations were determined with Pierce^TM BCA protein assay kit (ThermoFisher, USA). Lysate samples were separated on 10% SDS-polyacrylamide gel electrophoresis (PAGE). The gel was blotted and probed with anti-phospho-Akt (Ser473) and pan-Akt antibodies (Cell Signaling, USA). Band intensities were determined using the ImageJ gel analysis program (NIH, USA).

Statistics

All statistical analyses were done using 2-tailed, unequal variance Student’s t test. Data are expressed as mean±SEM. p values less than 0.05 were considered significant.

Expression of CD44V10 and other specific splice variants is enhanced in human AD brain hippocampus

Expression of CD44 transcripts was studied in postmortem hippocampal tissue specimens obtained from 10 AD patients and 10 cognitively and pathologically normal individuals with similar age and gender characteristics (Table 1). CD44S and various CD44 splice variants were initially identified by RT-PCR amplification and sequencing. To amplify the variant-containing transcripts, we used primers derived from all 9 human CD44 variant exons (Fig. 1) with the neighboring 3’ and 5’ constant exons (C5 and C6). Sequence analysis indicated multiple CD44 splice variants in the hippocampus that contained one or more human variant exons V2 to V10 (Fig. 1A and data not shown). Three of these identified transcripts, CD44V3, CD44V6, and CD44V10, containing a single variant exon, were selected for further characterization. Expression analysis by real-time PCR (qPCR) performed on the hippocampal samples indicated that the expression of CD44V3, CD44V6, and CD44V10 variants was significantly higher in AD patients compared to the non-AD controls (Fig. 1B and Table 2). Significant upregulation was also observed when a set of primers that amplify constant exons were used, representing the entire CD44 transcripts population. Since the latter population is predominated by CD44S, we refer to the respective RNA and protein products as CD44S. It is important to mention that the PCR amplification of CD44 splice variants required larger number of PCR cycles compared to CD44S (6–8 cycles), indicating a lower expression level of these variants compared to the standard form.

CD44V10 immunoreactivity is localized to AD hippocampal neurons

Immunocytochemical analyses of CD44s, CD44V3, CD44V6, and CD44V10 derived proteins indicated that these proteins were localized histoanatomically and at the cellular level in hippocampal tissue sections. Preparations stained for CD44S were readily distinguishable due to striking immunoreactivity in cortical senile plaques. The reaction product densely filled neuronal perikarya and extended into proximal dendrites, as well as smaller caliber, isolated neuronal processes. Positively stained astrocytes are likely present, as well. Although neuronal somata outside of the plaque were usually unstained, presumptive neuronal processes were commonly positive, although not abundant, even remote from the plaques (Fig. 2A). Positively stained astrocytes were also detected in a cerebellar section (data not shown).

OPEN IN VIEWER

Fig.2

Localization of CD44 splice variants in postmortem hippocampal tissues of AD and control subjects. Immunohistochemistry (IHC) was done on sections of paraffin embedded hippocampus. A) IHC for CD44S showing plaque immunoreactivity involving neuronal somata, processes and astrocytes. The neurons are densely filled with reaction product (inset). Scale bar: 50 μm. B) IHC for CD44 V6 in hippocampal cortex: AD versus Non-AD. AD sample shows both perinuclear granular and superficial punctuate staining patterns. The small positive puncta are often connected with fine, immunoreactive linear structures, probably pre-terminal neuritis (arrows), more clearly appreciated as such in (C). Non-AD sample is minimally reactive. Inset: higher magnification of reactive neuron. C) IHC for CD44V10 with staining pattern similar CD44V6 (2B). D) IHC for CD44V6 in AD: Punctate staining of neuronal surface membrane in the hippocampal formation (CA) reminiscent of the appearance of axonal boutons terminating along several somata and their apical dendrites. E) In white matter, V3 showed dense cytoplasmic staining in astrocytic soma and processes; the latter were found perivascularly, in a typical distribution. AD cases and controls were not distinguishable.

Splice variants CD44V3, CD44V6, and CD44V10 protein products’ immunoreactivity was evident among neuronal cell bodies, and differed from respective findings for CD44S in several ways. Firstly, none of these variants stained elements of plaques. Secondly, neuronal somata never stained as completely or intensely as with CD44S (Fig. 2B, C). In the many immunoreactive neurons, reaction product was distributed both intracellularly and superficially. The intracellular staining was granular, and at least partially overlapped the perinuclear region of lipofuscin autofluorescence (Fig. 2B, C). The second type of neuronal staining consisted of scattered punctate loci that appeared localized to the cell membrane (Fig. 2D). These puncta were occasionally seen to be connected by fine, short processes, suggesting afferent terminal boutons. The staining of these bouton-like, afferent terminal structures was most clearly demonstrated in a CD44V6 stained AD case. In that sample, apical dendrites and/or the perikaryon of many neurons was distinctly outlined by a linear array of such structures (Fig. 2D). Regarding neuroglial staining, the majority of identifiable astrocytic structures were unstained in cortex, although occasional perivascular staining of glial foot-processes was observed. In contrast, astrocytic white matter staining regularly occurs with CD44S and CD44V3 (Fig. 2E) but not with CD44V6 and CD44V10. All components of oligodendrocytes appeared negative for CD44 immunostaining in all locations.

Since AD is on the spectrum of histopathological aging-associated changes in the brain, it was not surprising that some immunoreactivity was also seen in the non-AD, age matched control cases; with CD44S, even rare plaques were encountered. The primary difference was degree; neuronal staining in control cases was quite minimal and evident in only scattered neurons (Fig. 2B). In contrast, astrocytic white matter staining by CD44S and CD44V3 was somewhat comparable in all cases.

CD44 variants upregulation by Aβ_1-42 exposure

We further tested the expression of CD44 splice variants in mouse primary cortical neurons. Using RT-PCR we validated the expression of CD44S, CD44V3, CD44V6, and CD44V10 in these cells (data not shown). We found that treatment of primary neurons with Aβ_1-42 induced the expression of both CD44S, CD44V6, and CD44V10 at the mRNA level as evaluated by qPCR (Fig. 3A). CD44V6 and CD44V10 were also found to be induced by Aβ_1-42 in the human neuroblastoma cell line SK-N-SH (Fig. 3B). SK-N-SH cell line was selected owing to previously published characterization of CD44 splice variants expression in this cellline [36].

OPEN IN VIEWER

Fig.3

CD44 expression in neuronal cells is induced by Aβ_1-42 peptide. A) Mouse primary neurons were treated with the designated concentrations of Aβ_1-42 peptide for 20 h. The mRNA levels of CD44S, CD44V6, and CD44V10 were evaluated using qPCR normalized to β-actin expression (n = 2, 2 technical repeats). B) SK-N-SH cells were differentiated for 5 days using retinoic acid and then treated with Aβ_1-42 peptide for 48 h. The mRNA levels of CD44V6 and CD44V10 were evaluated using qPCR normalized to β-actin expression (n = 3, 2 technical repeats). Bar graphs show means and standard errors. Significance: ^*p < 0.05, ^**p < 0.01 as determined by unpaired Student’s T test.

CD44V10 knockdown by siRNA and anti CD44V10 antibody are neuroprotective

To test whether CD44 splice variants are playing a functional role in AD-related neuronal cell death, we undertook a loss-of-function approach using exon-specific siRNA molecules. We first identified effective mouse and human siRNA molecules by checking reduction of relevant mRNA following transient transfection into relevant CD44-expressing cell lines (Supplementary Figures 1 and 2). We then tested the effects of the selected and non-relevant siRNA molecules on Aβ_1-42 peptide induced cell death in SK-N-SH and in mouse cortical neurons using XTT or Alamar Blue as measures of viability and caspase 3 activity assay, as a marker of apoptosis (Fig. 4). As expected, incubation with Aβ_1-42 resulted in marked increase in cell death (Fig. 4). Transfection of both cell types with CD44V10 siRNA selectively protected the cells against Aβ_1-42 toxicity as evidenced by higher viability and reduced caspase 3 activation in the Aβ_1-42 treated cells (Fig. 4), compared to non-relevant siRNA (NR). Partial but not significant protection was observed with CD44S siRNAs and no protection with CD44V6 siRNA in SK-N-SH and cortical neurons, respectively. This apparent neuroprotection was found consistently in both SK-N-SH neuroblastoma cells (Fig. 4A) and mouse cortical neurons (Fig. 4B), and was observed with two different V10 siRNA molecules, indicating that the effect is specific.

OPEN IN VIEWER

Fig.4

CD44V10-specific siRNA molecules protect neuronal cells from Aβ_1-42 induced cell death. A) SK-N-SH cells were transfected with 10 nM of siRNA molecules with sequences corresponding to two different CD44V10 (V10) locations, CD44S (S) non-relevant (NR) or left untransfected (NT). After 24 h cells were treated with pre-aggregated 80 μM Aβ_1-42 for an additional 24 h when cell viability was determined with XTT and induction of apoptosis evaluated by fluorescence caspase 3 activity assay (n = 2, 2 technical repeats). B) Mouse cortical neurons were transfected with different siRNA molecules in the presence (or absence) of 20 μM Aβ_1-42 and incubated for 48 h. Alamar blue and caspase 3 activity assay were used for evaluation of viability and induction of apoptosis, respectively (n = 3, 2 technical repeats). Significance: p < 0.05, p < 0.01 as determined by unpaired Student’s T test.

We next tested the effect of commercial polyclonal CD44V10 antibody on Aβ_1-42 toxicity in cortical neurons (Fig. 5). In agreement with the CD44V10 knockdown results, we found the CD44V10 antibody protected rat cortical neurons from Aβ_1-42–induced cell death whereas CD44S specific antibody (Hermes-3) had no effect on the viability of these cells (Fig. 5A). We further explored the effect of the anti CD44V10 antibody on serum starvation induced SK-N-SH cell death and found that there was a significantly greater survival in the presence of the antibody compared to vehicle control (Fig. 5B). No protection against serum starvation induced cell death was seen with CD44S or GFP antibodies.

OPEN IN VIEWER

Fig.5

CD44V10 but not CD44S antibody protects cortical neurons from Aβ_1-42 toxicity. A) Rat cortical neurons (E15) were treated with 20 μM Aβ_1-42 for 24 h in the presence of 1 μg/ml CD44V10, CD44S antibodies, or no-antibody control, followed by immunostaining using MAP-2 antibody (a neuronal-specific protein). The number of MAP2-positive cells were evaluated and used as a measure for neuronal survival (n = 6). B) SK-N-SH cells were treated with serum-free medium for 48 h as an inducer of cell death, in the presence of 1 μg/ml CD44V10, CD44S, or GFP antibodies (n = 4). C) Immunoblot and the corresponding densitometry analysis of SK-N-SH cells following treatment with serum-free medium for 10 or 20 min in the presence or absence of 1 μg/ml CD44V10 antibody (V10 pAb). Detection was done by phosphor-Akt (ser473) and pan-Akt antibodies and densitometry analysis by the ImageJ program.

A potential role of CD44V10 in signal transduction

Accumulating evidence demonstrate that CD44 and its splice variants acts as a signaling hub controlling cell surface receptors, among them receptor tyrosine kinases (RTKs) [37]. Therefore we aimed at identifying downstream signaling cascade that may mediate the anti CD44V10 antibody neuroprotection. The PI3K/Akt pathway is activated by RTKs and is well characterized as an anti-apoptotic pathway which is known to mediate survival signaling in neuronal cells [38]. We found that CD44V10 antibody caused a fast and robust Akt activation upon transfer of SK-N-SH cells to serum-free medium compared to antibody-free control medium as reflected by an increase in the level of Akt phosphorylation on Serine 473 (Fig. 5C). AKT activation was not observed when the cells were treated with CD44S antibody (data not shown).