Presenilin/γ-secretase and α-secretase-like peptidases cleave human MHC Class I proteins

Abstract

HLA (human leucocyte antigen)-A2 is an MHC Class I protein with primary functions in T-cell development and initi-ation of immune cell responses. MHC I proteins also play roles in intercellular adhesion, apoptosis, cell proliferation and neuronal plasticity. By utilizing a sequence comparison analysis, we recently identified HLA-A2 as a potential substrate for the Alzheimer's disease-associated PS1 (presenilin 1)/γ-secretase. α-Secretase-like membrane metalloproteinases are responsible for an initial shedding event, partially mediated by ADAM (a disinteg-rin and metalloproteinase)-10. Accordingly, activation or inhibition of α-secretase-like membrane metalloproteinases directly modulated levels of a 14 kDa HLA-A2 CTF (C-terminal frag-ment) in CHO (Chinese-hamster ovary) cells. To show that the HLA-A2 CTF is subsequently cleaved by PS1/γ-secretase, we re-duced its activity in cell lines stably expressing HLA-A2 and in Jurkat T-cells expressing endogenous MHC I. Treatment with specific PS1/γ-secretase inhibitors or expression of a dominant-negative construct led to a significant accumulation of HLA-A2 CTFs. We also identified the PS1/γ-secretase cleavage product of HLA-A2 CTF, termed HLA-A2 intracellular domain, in cell-free and cell-based experiments. In the absence of proteasome inhibitors, HLA-A2 intracellular domain underwent rapid degrad-ation. These data indicate that MHC I proteins undergo extra-cellular domain cleavage mediated by α-secretases and the cleavage product is subsequently cleaved by PS1/γ-secretase.

INTRODUCTION

Along with at least three additional proteins, namely Aph-1, Nicastrin, and Pen-2, the presenilins [PS1 (presenilin 1) and PS2] form a proteolytic complex termed γ-secretase. γ-Secretase is responsible for cleavage of type I membrane proteins in a process termed regulated intramembrane proteolysis. Prior to regulated intramembrane proteolysis, shedding of ectodomain regions occurs by MMPs (matrix metalloproteinases). Once MMPs such as TNFα (tumour necrosis factor α)-converting enzyme or α-secretases ADAM (a disintegrin and metalloprotein-ase)-10 and -17 cleave their substrates, a catalytic aspartate resi-due in the presenilins allows γ-secretase complexes to cleave within the transmembrane domain. γ-Secretase cleavages liberate ICDs (intracellular domains) and several substrates [Notch-1, APP (amyloid precursor protein), CD44, E-cadherin and Erb-B4] have the ability to transduce nuclear signals and affect gene activation [1–7]. γ-Secretase cleavage within the transmembrane domain does not appear to require sequence specificity; however, one site in particular, the APP-ϵ/Notch S3 site near membrane–cytosol interface, has been used to identify novel substrates [8,9].

MHC I (MHC Class I) proteins are type I membrane proteins found on the cell surface of almost all nucleated cells. They largely form non-covalent heterodimeric complexes with β₂-micro-globulin and present small cytosolic peptides of 9–11 amino acids to T-cells. During T-cell development, MHC I complexes are found on antigen-presenting cells where they play a central role in maturation of CD8⁺ cytotoxic T-lymphocytes. Additionally, when cells are invaded by foreign microbes such as viruses, MHC I complexes signal immune surveillance cells to initiate proper immune system responses. In both cases MHC I complexes are primary components of a highly dynamic immunological synapse where they ligate with TCRs (T-cell receptors) and CD8 molecules [10]. The immunological synapse involves additional proteins for proper stabilization between cells and communication: co-stimulatory molecules (CD28, CD80 and CD86), adhesion proteins [LFA-1 (lymphocyte function-associated antigen-1) and ICAM-1 (intercellular adhesion molecule 1)] and additional large glycoproteins to bridge contacts between cells (CD43, CD44 and CD45) [11]. It is interesting that both CD43 and CD44 have previously been identified as γ-secretase substrates [12,13].

In recent years it has been discovered that MHC I proteins are expressed in neurons throughout the CNS (central nervous system). MHC I expression was found to be regulated by neuronal activity during development, with peak expression occurring in the perinatal period [14,15]. Moreover, it has been shown that neuronal MHC I functions as a mediator of synaptic plasticity and activity-dependent refinement during development of the visual system [14,15]. Neuronal MHC I expression is regulated similarly to immune system MHC I, whereupon exposure to cytokines in-duces up-regulation of MHC I [16]. Additional studies suggest a role for neuronal MHC I during treatments such as axotomy where they may regulate the ability of neurons to maintain synapses [16,17].

The present study identifies a specific protein of the HLA (hu-man leucocyte antigen)-A locus, HLA-A2, as a substrate of both α- and γ-secretases. Cell-free and cell-based experiments show that HLA-A2 first undergoes an α-secretase-like shedding event, followed by a PS1/γ-secretase-mediated event that releases an ICD. Since MHC I proteins are important components of the immune system, the developing CNS and during activity-depend-ent synaptic remodelling, the present study provides new insights into the function of PS1/γ-secretase.

MATERIALS AND METHODS

Plasmids and transfections

An expression construct encoding full-length human HLA-A*0201 heavy chain (HLA-A2) was obtained from Dr Hidde Ploegh (Harvard Medical School). The construct contained a signal peptide from the mouse H-2K^b haplotype that enhances in vitro translation [18]. The primers used for PCR amplification were 5′-ACCATGGTACCGTGCACGCTGCTCCT-3′ and 5′-CACTTTACAAGCTGTGAGAGACACAT-3′. Subsequent cloning and transformation were performed using TOPO cloning vector (pcDNA3.1 containing a C-terminal V5/His tag) in One Shot TOP10 Chem competent Escherichia coli (Invitrogen). Sequenc-ing was later confirmed at Massachusetts General Hospital facil-ities. Effectene (Qiagen) was used for transfecting cell lines. We produced stably transfected CHO (Chinese-hamster ovary) cells as well as B104 rat neuroblastoma cells (Dr David Schubert, The Salk Institute, La Jolla, CA, U.S.A.). Jurkat cell line E6.1 was purchased from A.T.C.C.

Western-blot analysis, immunoprecipitation, antibodies and inhibitors

Cell extracts were prepared by directly lysing cells in a buffer containing 10 mM Tris/HCl (pH 6.8), 1 mM EDTA, 150 mM NaCl, 0.25% Nonidet P40, 1% Triton X-100 and a protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN, U.S.A.), followed by centrifugation at 16000 g (4 °C, 15 min). Samples were quantified using the BCA (bicinchoninic acid) protein assay kit (Pierce). Protein (20–100 μg) was resolved on 4–12% gradient Bis-Tris gels. Immunoprecipitations were done as described in [8]. Primary antibodies V5 (1:5000 dilution, Invitro-gen), anti-HLA (1:250), W6/32 were purchased from Biotrend Chemicals. HC10 antibodies were obtained from Dr Hidde Ploegh, and anti-HLA-A and -B (TA-17) were obtained from Dr Tanigaki Nobuyuki (Department of Immunology, Roswell Park Cancer Institute, Buffalo, NY, U.S.A.). The blots were developed using ECL® (enhanced chemiluminescence) with SuperSignal CL-HRP substrate (Pierce) according to the manufacturer's instructions. The γ-secretase inhibitors DAPT {N-[N-(3,5-difluorophenacetyl)-L-alanyl]-(S)-phenylglycine t-butyl ester} and L-685,458 were obtained from Calbiochem, and WPE31C was a gift from Dr Michael S. Wolfe (Department of Neurology, Brigham and Women's Hospital, Boston, MA, U.S.A.). The α-secretase inhibitor, TAPI-1 (TNFα processing inhibitor-1; IC-2), was from Biomol.

Cell-free generation of HLA-A2 ICD

Membrane preparation and cell-free generation of HLA-A2 proteolytic products were performed as described in [8]. The P2 and P3 fractions were resuspended in Buffer H (20 mM Hepes, 150 nM NaCl, 10% glycerol and 5 mM EDTA, pH 7.4) with pro-tease inhibitors. In vitro cleavage experiments were performed by incubating the membrane factions at 37 °C for 1 h in the presence or absence of indicated amounts of DAPT. After incubation, the soluble and membrane-associated fragments were separated by centrifugation of the reaction mixture at 120000 g for 45 min.

Immunohistochemistry

Cells were treated and fixed in 4% (w/v) paraformaldehyde for 10–20 min at room temperature (25 °C) Cells were permeabilized with Triton X-100, and primary and secondary antibodies were incubated for 1 h at room temperature. Anti-V5 (1:200) stained cells were visualized using confocal microscopy (Olympus).

RESULTS AND DISCUSSION

HLA-A2 forms functional MHC I complexes in CHO and B104 cells

HLA-A2 is one of the most commonly expressed MHC I proteins. To analyse proteolyic processing of HLA-A2, we stably transfected CHO and B104 cells with an HLA-A2 cDNA C-terminally tagged with V5/His. Two commonly used antibodies were chosen to characterize the overexpressed HLA-A2 protein. These anti-bodies distinguish between MHC I complexes containing β₂-microglobulin (W6/32) and MHC I proteins lacking β₂-microglobulin, termed β₂-free MHC I (HC10) [19,20] (Figure 1a). Endogenous MHC I proteins are known to be found in complexes with β₂-microglobulin. β₂-free MHC I arise occasionally during biogenesis of MHC I complexes and after both β₂-microglobulin and peptide dissociate from MHC I at the cell surface [21–23]. β₂-free MHC I occur at low levels in naïve cells, but are known to accumulate during T-cell activation [21]. It was previously reported that expression of HLA-A2 in murine cells leads to easily detectable amounts of W6/32-reactive complexes [24].

Figure 1. Stably expressed HLA-A2 in CHO and B104 cells.

(a) Expression construct of HLA-A2 containing a C-terminal V5/His tag. W6/32 and HC10 antibodies recognize dimerized and free forms of HLA-A2 respectively. (b) Anti-V5 immunostaining of CHO-HLA-A2 cells shows surface/early endosome localization of HLA-A2. (c) Immunoprecipitation (IP) of HLA-A2 from CHO and B104 cell lines stably expressing the protein indicates that HLA-A2 forms hybrid complexes with β₂-microglobulin. Heavy chain of W6/32 antibody is not readily detected because of low concentrations used in the immunoprecipitations (0.2 μg for W6/32; 6.8 μg for HC10).

In stably transfected CHO cells we observed HLA-A2 protein at or near the cell surface by anti-V5 immunostaining (Figure 1b). Using the aforementioned antibodies, W6/32 and HC10, we tested whether HLA-A2 was complexed with β₂-microglobulin in these cells. After immunoprecipitation, most HLA-A2 was found in W6/32-positive complexes in CHO-HLA-A2 cells (Figure 1c). This suggests that MHC I complexes containing β₂-micro-globulin were formed. We also tested the B104 rat neuroblastoma cell line stably expressing HLA-A2. In these cells we found an equal distribution of MHC I complexes and β₂-free MHC I after similar immunoprecipitations (Figure 1c). These results show that in hamster and rat cell lines, expression of HLA-A2 protein can form MHC I complexes containing β₂-microglobulin.

HLA-A2 is processed by α-secretases

Prior to γ-secretase cleavage, all substrates undergo ectodomain shedding where they are cleaved by α-secretase-like MMPs at or near the cell surface. To establish a similar role for MMP activity in the generation of HLA-A2 CTFs (C-terminal fragments), we tested whether the α-secretase inhibitor TAPI-1 could affect HLA-A2 CTF generation in CHO-HLA-A2 stable cells. Stably transfected cells differed from transient HLA-A2 expression in that they showed constitutive HLA-A2 CTF generation at 14 kDa under untreated conditions. Constitutive HLA-A2 CTF produc-tion was significantly reduced by TAPI-1 treatment alone for 6 h (Figure 2a). Conversely, PS1/γ-secretase inhibition by DAPT treatment alone for 6 h increased HLA-A2 CTF levels. However, the increase due to γ-secretase inhibition was abolished during co-treatment of DAPT with TAPI-1 (Figure 2a). These data suggest that α-secretase activity is required for generation of HLA-A2 CTFs prior to γ-secretase cleavage.

Figure 2. HLA-A2 is cleaved by α-secretase-like proteases.

(a) Treatment with TAPI-1 (50 μM for 3 h), an α-secretase inhibitor, attenuates HLA-A2 CTF production compared with untreated cells (CTL). Accumulation of HLA-A2 CTF with the PS1/γ-secretase inhibitor DAPT alone is significantly reduced upon co-treatment with TAPI-1 and DAPT. HLA-A2 is stained with an anti-V5 antibody. Transferrin receptor is shown as control. (b) HLA-A2 is cleaved by the α-secretase ADAM-10. Overexpression of ADAM-10 in B104 rat neuroblastoma cells stably expressing HLA-A2 shows that HLA-A2 CTF accumulation by PS1/γ-secretase inhibition is further enhanced in ADAM-10-expressing cells. HLA-A2 is stained with an anti-V5 antibody. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) is used as a loading control. FL, full-length.

To test whether a specific α-secretase, ADAM-10, contributes to HLA-A2 CTF generation, we used B104 neuroblastoma cells stably expressing HLA-A2. Transient expression of ADAM-10 in these cells significantly increased HLA-A2 CTF generation. DAPT treatment of cells expressing ADAM-10 resulted in in-creased HLA-A2 CTF accumulation when compared with cells with normal ADAM-10 levels (Figure 2b). Together, our results indicate that HLA-A2 CTFs are generated by MMP activity prior to PS1/γ-secretase cleavage and the α-secretase ADAM-10 con-tributes to the extracellular-domain shedding event.

HLA-A2 is cleaved by a PS1-dependent γ-secretase-like activity

Next, we characterized PS1/γ-secretase-mediated processing of HLA-A2. Our sequence comparison search using the APP-ϵ/Notch S3-like domain previously revealed γ-secretase substrates including nectin-1 and a voltage-gated sodium channel subunit (β₂) [8,9]. Using the same method, we now found that HLA-A, -B and -C proteins also contain this site homology (Figure 3a). Western-blot analysis of CHO cells transiently transfected with HLA-A2 showed that the full-length HLA-A2 protein migrated at approx. 44 kDa (Figure 3c). In transiently transfected CHO cells treated for 3 h with PMA (‘TPA’), an activator for α-secretases such as ADAM-family metalloproteases, we did not see a consistent and significant accumulation of HLA-A2 CTFs (Figures 3c–3e). However, co-treatment of PMA with the γ-secretase inhibitor DAPT resulted in the accumulation of a C-terminally tagged HLA-A2 fragment at 14 kDa, suggesting that γ-secretase activity may be responsible for cleaving this fragment. In addition, a 6 h treatment of DAPT alone was sufficient to significantly increase HLA-A2 CTFs (Figure 3c). We confirmed these results using two additional γ-secretase inhibitors, L-685,458 and WPE31C. Each inhibitor independently increased HLA-A2 CTFs, which were augmented upon co-treatment with PMA in 3 h treatments (Figure 3d). We also stably expressed HLA-A2 in B104 rat neuroblastoma cells and found accumulation of HLA-A2 CTFs after PS1/γ-secretase inhibition. This effect was potentiated by co-treatment with PMA (Figure 3e). To further confirm that PS1/γ-secretase activity was responsible for cleaving HLA-A2, we tested whether a similar accumulation of the 14 kDa fragment occurs in cells stably expressing dominant-negative mutations in PS1 (D385A). Figure 3(f) shows that lack of γ-secretase activity in PS1 (D385A) cells resulted in accumulation of the HLA-A2 CTF. These results show that PS1/γ-secretase activity is responsible for cleavage of the HLA-A2 CTF detected at 14 kDa.

Figure 3. HLA-A2 is cleaved by PS1/γ-secretase.

(a) Sequence comparison of the APP-ϵ/Notch-S3 domains in ten known human PS1/γ-secretase-like substrates, and in HLA-A2. (b) Predicted fragment sizes before and after PS1/γ-secretase-mediated cleavage of HLA-A2 CTF. (c) CHO cells expressing HLA-A2 were resolved by SDS/12% PAGE and stained with anti-V5. Upon co-treatment with PMA (TPA) and DAPT, an HLA-A2 CTF accumulates around 14 kDa. (d) Anti-V5 immunostaining shows that HLA-A2 CTFs are increased by two additional γ-secretase inhibitor treatments, L-685,458 and WPE31. Transferrin receptor is shown as loading control. (e) Anti-V5 immunostaining of B104 rat neuroblastoma cells stably expressing HLA-A2 shows accumulation of HLA-A2 CTFs after treatment with PMA or DAPT, and further accumulation in co-treatments. Transferrin receptor is shown as loading control. (f) The dominant-negative form of PS1, D385A, elevates HLA-A2 CTF levels. Expression of PS1(D385A) results in the accumulation of the 14 kDa HLA-A2 CTF (arrowhead). The asterisk indicates an unidentified band that is not seen with PS1/γ-secretase inhibitor treatments. (g) Jurkat T-cells show accumulation of an endogenous MHC I CTF. A 24 h incubation of DAPT leads to accumulation of an approx. 12 kDa MHC I CTF. Immunostaining was with anti-HLA-A and -B (TA-17). Transferrin receptor is shown as a control.

Next we tested whether endogenous MHC I in human Jurkat T-cells exhibited a processing pattern similar to overexpressed HLA-A2. For these experiments we used a C-terminally directed antibody known to immunoprecipitate HLA-A and -B [25]. Endogenous MHC I CTF accumulated at approx. 12 kDa after a 24 h incubation with DAPT (Figure 3g). These data indicate that endogenous MHC I proteins are cleaved in cultured human T-cells.

γ-Secretase activity is required for generation of HLA-A2 ICD in vivo and in vitro

For additional confirmation that γ-secretase cleaves HLA-A2 CTFs, we performed a cell-free γ-secretase cleavage assay on CHO cells stably expressing high levels of HLA-A2. Cell-free assays have previously been successful for detection of ICDs, such as nectin-ICD [8]. CHO-HLA-A2 cell membranes were incubated with or without DAPT for 1 h and treated with various protease and proteasome inhibitors. Control samples, incubated at 0 °C, showed significant amounts of HLA-A2 CTF (Figure 4a). After incubation at 37 °C, a fragment of approx. 12 kDa was generated in cells with γ-secretase activity. This band was not observed in DAPT-treated cell extracts, indicating that the HLA-A2 ICD is generated by γ-secretase activity (Figure 4a).

Figure 4. HLA-A2 ICD is produced under cell-free and in vivo conditions.

(a) Cell-free generation of a 12 kDa HLA-A2 ICD. Partially purified membrane fractions of CHO-HLA-A2 were subjected to a γ-secretase cleavage assay [8]. Membranes were incubated with protease inhibitors at 0 and 37 °C in the presence or absence of DAPT. Samples incubated at 37 °C generate a HLA-A2 ICD fragment. Blots were stained with an anti-V5 antibody. (b) In vivo generation of HLA-A2 ICD in CHO-HLA-A2 cells after incubation with a proteasomal inhibitor lactacystin. HLA-A2-ICD accumulation was blocked upon inhibition of PS1/γ-secretase activity with DAPT and L-685,458 respectively. The asterisk indicates an unidentified band that is not seen with PS1/γ-secretase inhibitor treatments.

To investigate whether HLA-A2 ICD can be detected in vivo, we utilized lactacystin β-lactone to inhibit potential proteasomal degradation of the ICD. CHO-HLA-A2 stable cells were first treated with PMA to activate α-secretases and increase both HLA-A2 CTF production and cleavage. Upon co-incubation with PMA and lactacystin β-lactone for 3 h, an HLA-A2 ICD was detected at approx. 12 kDa. The production of this fragment was blocked when similarly treated cells were also incubated with one of two γ-secretase inhibitors, DAPT and L-685,458 respectively (Figure 4b). Our results indicate that γ-secretase activity is required for generation of the HLA-A2 ICD fragment, which is highly unstable and quickly degraded by the proteasome system in vivo. We also observed an approx. 21 kDa band which was increased with PMA and lactacystin β-lactone treatment for 3 h, previously observed in PS1(D385A) cells transiently expressing HLA-A2. This band did not appear consistently in other experiments; per-haps it is a very unstable dimer of HLA-A2 CTFs or an intermediate degradation product. Additional studies would be required for further characterization.

Our study shows that MHC I proteins, such as HLA-A2, are substrates for both α- and γ-secretases in three different cell lines. Metalloprotease or α-secretase activities mediate ectodomain shedding, producing an HLA-A2 CTF, followed by sequential PS1/γ-secretase cleavage. We found that the PS1/γ-secretase cleavage product, HLA-A2 ICD, undergoes rapid proteasomal degradation upon generation by PS1/γ-secretase. The present study adds MHC I proteins to the growing list of PS1/γ-secretase substrates and provides further insight on PS1/γ-secretase func-tion in various tissues.

MHC I functions as a central protein involved in cell–cell inter-actions between antigen-presenting cells and cytotoxic T-lympho-cytes (CD8+) of the immune system. Interestingly, two additional PS1/γ-secretase substrate proteins, CD43 and CD44, are also involved in cell–cell interactions proximal to MHC I proteins [13,26]. Additionally, a novel function for MHC I in the CNS has been suggested in recent years. Several reports support a role for MHC-dependent signalling in development of the mammalian visual system, specifically activity-dependent remodelling and re-traction of inappropriate synapses [14,27,28]. Mice deficient in surface MHC I complexes show abnormal development of retinal projections in the visual system pathways [15]. In addition, these neurons also show perturbations in long-term potentiation and long-term depression in the adult hippocampus [15]. These studies suggest that PS1/γ-secretase cleavage of MHC I may occur not only in immunological cells but also in neuronal cells during synapse formation.

It is well established that the PS1/γ-secretase cleavage product of Notch, NICD (Notch intracellular domain), translocates to the nucleus to initiate transcription [1]. Recent work has shown that α- and PS1/γ-secretase-mediated cleavages of p75 neurotrophin receptors activate Rho and subsequent signals, resulting in axonal growth inhibition [29]. While MHC I-mediated signalling in the CNS is not known, it may play a role in T-cell activation. MHC I-mediated signalling in T-cells has been studied largely using antibody cross-linking, or ligation, of MHC I molecules. MHC I signalling pathways have been suggested to include induction of tyrosine kinase and PLCγ (phospholipase Cγ) activity as well as JAK (Janus kinase) Tyk2 and Stat-3 activation [33,34]. Further-more, MHC I-mediated signalling can activate ZAP-70 [ζ-chain (TCR)-associated protein kinase of 70 kDa] and Src kinase p56^Lck, leading to induction of apoptosis in Jurkat T-cells [35]. Using a different cell type, it was suggested that MHC I-medi-ated signalling can prevent apoptosis and is mediated by intact p56^Lck activity [36]. These data suggest that MHC I-mediated signalling is involved in both T-cell survival and death. However, previous work on MHC I proteins has revealed that all but four proximal amino acids from the ICD can be deleted with no effect on T-cell activation [30,31]. Similarly, membrane-associated MHC I cytoplasmic domains did not appear to affect cellular func-tions, including cytoskeletal association, aggregation and inter-nalization [32].

Our results suggest that α-secretase and PS1/γ-secretase activities are involved in degradation of MHC I proteins under physiological conditions. β₂-free MHC I molecules have been reported as selectively cleaved and released from the plasma membrane by membrane-bound metalloproteinases, similar to α-secretase-mediated cleavages described in our study [37,38]. PS1/γ-secretase-like activity has also been described in virus-mediated down-regulation of MHC I proteins [39]. As previously suggested, PS1/γ-secretase may function as a proteosomal complex for a group of plasma membrane proteins [40]. Character-ization of novel substrate proteins such as MHC I contributes to a full understanding of PS1/γ-secretase function under physiological and pathological conditions.