of April 5, 2018.This information is current as

DipeptidaseAminodipeptidase, Quiescent Cell Prolineof a Novel Post-Proline-Cleaving Vesicular Localization and Characterization

Mathieu, Henry Lee and Brigitte T. HuberMurali Chiravuri, Fernando Agarraberes, Suzanne L.

http://www.jimmunol.org/content/165/10/5695doi: 10.4049/jimmunol.165.10.5695

2000; 165:5695-5702; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/165/10/5695.full#ref-list-1

, 19 of which you can access for free at: cites 28 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2000 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Vesicular Localization and Characterization of a NovelPost-Proline-Cleaving Aminodipeptidase, Quiescent CellProline Dipeptidase1

Murali Chiravuri,* Fernando Agarraberes, † Suzanne L. Mathieu,* Henry Lee,* andBrigitte T. Huber 2*

A large number of chemokines, cytokines, and signal peptides share a highly conserved X-Pro motif on the N-terminus. Thecleavage of this N-terminal X-Pro dipeptide results in functional alterations of chemokines such as RANTES, stroma-derivedfactor-1, and macrophage-derived chemokine. Until recently, CD26/DPPIV was the only known protease with the ability to cleaveN-terminal X-Pro motifs at neutral pH. We have isolated and cloned a novel serine protease, quiescent cell proline dipeptidase(QPP), with substrate specificity similar to that of CD26/DPPIV. In this paper we show that QPP, like CD26/DPPIV, is synthesizedwith a propeptide and undergoesN-glycosylation. Interestingly, this glycosylation is required for QPP enzymatic activity, but notfor its localization. Unlike the cell surface molecule, CD26/DPPIV, QPP is targeted to intracellular vesicles that are distinct fromlysosomes. Proteinase K treatment of intact vesicles indicates that QPP is located within the vesicles. These vesicles appear to havea secretory component, as QPP is secreted in a functionally active form in response to calcium release. The presence of QPP inthe vesicular compartment suggests that molecules bearing the N-terminal X-Pro motif can be cleaved at multiple sites within andoutside the cell. These results expand the potential site(s) and scope of a process that appears to be an important mechanism ofpost-translational regulation. The Journal of Immunology,2000, 165: 5695–5702.

Proteolytic cleavage of proteins can have profound effectson their stability and function (1–3). There is a strikingconservation of X-Pro motifs on the N termini of many

chemokines, cytokines, and other signal peptides, suggesting a se-lective pressure for this motif in the evolution of these molecules(4). For many of these factors, the significance of the X-Pro motifhas yet to be ascertained. Recently, however, the work of severalgroups has shown that the N-terminal X-Pro motif of the chemo-kines RANTES, stroma-derived factor-1a (SDF1a),3 SDF1b,eotaxin, and macrophage-derived chemokine (MDC) is cleaved bydipeptidyl peptidase IV (DPPIV/CD26), resulting in an alterationof their biological function (2, 3, 5, 6). Thus, it appears that theN-terminal X-Pro motif on signal molecules may serve as a site forpost-translational modification and regulation. The cleavage ofproline-containing peptide bonds on the C terminus can also affectthe function of molecules. Angiotensin, for example, undergoes aloss of vasoactive function following post-proline cleavage of the

C-terminal amino acid by prolylcarboxypeptidase (PCP; angio-tensinase C) (7, 8).

Few proteolytic enzymes that cleave peptide bonds containingproline have been identified. These include exopeptidases such asCD26/DPPIV and PCP (4). We have isolated and cloned a serineprotease, quiescent cell proline dipeptidase (QPP), which has sub-strate specificity very similar to that of CD26/DPPIV (9). Boththese enzymes cleave N-terminal dipeptides when the penultimateamino acid is a proline or, less preferably, an alanine residue (9–11). Recently, we observed that highly specific aminodipeptidaseinhibitors cause cell death in quiescent lymphocytes, but not inactivated lymphocytes (12). QPP is the likely target of these in-hibitors, suggesting that this protease plays an important role inresting cells (12). However, unlike CD26/DPPIV, which is foundon the cell surface (10, 11), QPP was isolated from intracellularfractions (9). Although QPP and CD26/DPPIV share substratespecificity, there is no significant sequence homology betweenthese two proteases, suggesting an evolutionary convergence offunctional activity. QPP does, however, share significant sequencehomology with the lysosomal serine protease PCP (42% sequenceidentity) (7, 9). It is interesting to note that the three post-proline-cleaving proteases, CD26/DPPIV, PCP, and QPP, all have thesame sequential ordering of the active site residues, Ser-Asp-His(4, 9), and are functionally active as homodimers (7, 9, 11).

In this work we show that similar to its functional homologue,CD26/DPPIV, QPP is synthesized with a cleaved pro-peptide.QPP undergoesN-linked glycosylation at the Golgi apparatus,which is required for its proteolytic activity. Unlike CD26/DPPIV,QPP localizes to a post-Golgi vesicular compartment. QPP-con-taining vesicles have similar density to lysosomes, but further frac-tionation shows that QPP-containing vesicles form a distinct com-partment, and unlike mannose-6-phosphate (M6P)-dependentlysosomal targeting (13), QPP localizes to vesicles by a M6P-independent mechanism. QPP is localized in the vesicular lumen,

*Department of Pathology, Program in Immunology, and†Department of Physiology,Sackler School of Graduate Biomedical Sciences, Tufts University School of Medi-cine, Boston, MA 02111

Received for publication March 15, 2000. Accepted for publication August 22, 2000.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby markedadvertisementin accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by National Institutes of Health Research Grants AI36696and AI43469 (to B.T.H.).2 Address correspondence and reprint requests to Dr. Brigitte T. Huber, Departmentof Pathology, Tufts University School of Medicine, 136 Harrison Avenue, Boston,MA 02111. E-mail address: bhuber@opal.tufts.edu3 Abbreviations used in this paper: SDF, stroma-derived factor; QPP, quiescent cellproline dipeptidase; DPPIV, dipeptidyl peptidase IV; DPPII, dipeptidyl peptidase II;PCP, prolylcarboxypeptidase; GFP, green fluorescent protein; M6P, mannose-6-phos-phate; LAMP1, lysosome-associated membrane protein-1; MDC, macrophage-de-rived chemokine, TGN,trans-Golgi network; zGGL-AMC, z-glycine-glycine-leucine-aminomethylcoumarin; Ala-Pro-AFC, alanine-proline-aminofluorocoumarin;HA, hemagglutinin; PNS, postnuclear supernatant; ER, endoplasmic reticulum; Glc-NAc, N-acetylglucosamine; PNGase F, peptide-N-glycosidase F.

Copyright © 2000 by The American Association of Immunologists 0022-1767/00/$02.00

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

as it is not susceptible to proteinase K digestion of intact vesicles.Functionally active QPP can also be detected in cell culture su-pernatant, and its secretion can be induced by cell activation andcalcium mobilization, suggesting that there is a secretory compo-nent to the QPP-containing vesicular compartment. The study ofpost-translational modification of N-terminal X-Pro-containingsignal molecules has been a rapidly expanding field in the last fewyears. The discovery of a novel post-proline-cleaving aminodipep-tidase in a distinct compartment from CD26/DPPIV broadens thescope of post-proline-cleavage as a regulatory mechanism.

Materials and MethodsMaterials

Anti-Myc Ab was purchased from PharMingen (San Diego, CA), and anti-hemagglutinin (anti-HA) Ab was purchased from BabCo (Berkeley, CA).The anti-lysosome-associated membrane protein-1 (anti-LAMP1) mAbH4A3 was obtained from the Developmental Studies Hybridoma Bank(Iowa City, IA), and the anti-calnexin mAb was purchased from StressGen(Victoria, Canada). Rab11 and adaptin-a Abs were purchased from Trans-duction Laboratories (Lexington, KY). The reporter substrates alanine-pro-line-aminofluorocoumarin (Ala-Pro-AFC) andz-glycine-glycine-leucine-aminomethylcoumarin (zGGL-AMC) were purchased from EnzymeSystems Products (Dublin, CA). Tunicamycin, ionomycin, and cyclohex-amide were purchased form Sigma (St. Louis, MO), and peptide-N-glyco-sidase F (PNGase F) was purchased from New England Biolabs(Beverly, MA).

Constructs and transfections

To make the QPP-green fluorescent protein (GFP) construct, enhancedGFP was amplified from the cloning vector pEGFP-N1, obtained formClontech (Palo Alto, CA), and cloned into the pCI-neo expression vectorfrom Promega (Madison, WI). The sense primer incorporated five glycinecodons upstream of GFP. QPP was amplified with primers containing re-striction site linkers and cloned in frame upstream of enhanced GFP. TheQPP-Myc fusion was generated by PCR using the high fidelity DeepVentpolymerase (New England Biolabs), with an antisense primer that codedfor 10 aa comprising the c-Myc epitope tag (EQLLISEEDL). Likewise, aQPP-HA fusion was generated using an antisense primer that coded for 10aa comprising the HA epitope tag (YPYDVPDYA). All constructs werecloned into the pCI-neo expression vector. Transfections were performedusing the calcium phosphate method.

Western blot analysis

Cells (1–23 107) were resuspended in lysis buffer (20 mM HEPES, 1.5mM MgCl2, 2 mM EDTA, 10 mM KCl, 0.1% Nonidet P-40, 5mg/mlantipain, and 5mg/ml leupeptin) for 30 min at 4°C. Lysed cells werecentrifuged at 3,000 rpm on a microcentrifuge for 10 min, and the pelletwas discarded. The postnuclear supernatant was subjected to a 30,0003 gcentrifugation for 30 min. The protein concentration was measured usingthe bicinchoninic acid protein estimation kit (Pierce, IL). For the immu-noblot assay of supernatants, the supernatants of QPP-transfected or vec-tor-transfected (control) cells were centrifuged at 3003 g to remove allcells and subjected to SDS-PAGE analysis.

Enzyme assays

QPP and proteasome enzymatic activities were measured using the fluoro-genic substrates Ala-Pro-AFC (20mM in 50 mM HEPES buffer) andzGGL-AMC (100mM in 50 mM HEPES), respectively, on a Fluoromaxfluorescence plate reader (Molecular Devices, Menlo Park, CA; AFC: ex-citation 390 nm; emission 510 nm; AMC: excitation, 460 nm; emission,510 nm). For theb-hexosaminidase assay, the fluorogenic substrate usedwas 4-methylumbelliferyl-b-D-glucosaminide. Twenty microliters of thesamples obtained from the discontinuous gradient were incubated with 0.5ml of the substrate solution (1 mM 4-methylumbelliferyl-b-D-glucosamin-ide and 0.1 M sodium acetate, pH 4.5) for 5 min. For the continuousgradient fractions, 10ml of the samples were incubated for 20 min. Thereaction was stopped with 1.25 ml of stop solution (0.5 M glycine and 0.5M Na2CO3). The samples were read using an excitation wavelength of 364nm and an emission wavelength of 448 nm.

PNGase F assays

Fifty micrograms of lysate from control or QPP-transfected cells wasboiled for 10 min under denaturing conditions (0.5% SDS and 1% 2-ME).

Following this the samples were brought up to 1% Nonidet P-40 and 0.05M sodium phosphate. PNGase F (500 U) was added and incubated for theindicated times at 37°C. The reaction was terminated by boiling the samplein 23 reducing loading buffer for 5 min, followed by SDS-PAGE andimmunoblot analysis. The controls were treated in the same manner exceptfor the addition of PNGase F.

Confocal microscopy

293T human fibroblasts were transfected with QPP-GFP or QPP-Myc,plated on coverslips, pretreated with fibronectin (100mg/ml), and blockedwith 1% BSA in PBS. These cells were fixed in 4% paraformaldehyde andpermeabilized with 0.1% Triton X in PBS. The primary Ab was incubatedfor 1 h at 4°C, followed by three washes and incubation of the secondaryAb, where applicable, for 1 h at 4°C. The samples were washed and vi-sualized by confocal microscopy.

Biochemical organelle fractionation

The isolation of subcellular organelles on discontinuous density gradientswas performed as previously described (14). In addition, a continuous gra-dient fractionation was performed. Briefly, 13 108 cells were lysed bycavitation or Dounce homogenization (Kontes, Vineland, NJ) in 0.25 Msucrose, and the postnuclear supernatant (PNS) was taken from a low speedcentrifugation. For discontinuous fractionation, a metrizamide/Percoll (Ac-curate Chemical & Scientific Corp., Westbury, NY) gradient was set up,using 35% metrizamide, 17% metrizamide, and 6% Percoll, and the PNSwas layered over it. Following ultracentrifugation at 60,0003 g the fourinterface layers were harvested and analyzed. The orientation of the gra-dient is as follows: lightest fraction5 fraction I, most dense fraction5fraction IV. For the continuous gradient fine fractionation of subcellularorganelles, a metrizamide/Percoll gradient was set up as follows: 1 ml of35% metrizamide was placed on the bottom of the tube, followed by alinear gradient of 5–20% metrizamide, 1 ml of 6% Percoll (Pharmacia,Uppsala, Sweden), and 2.5 ml of PNS. The gradients were centrifuged at160,0003 g for 2 h at 4°C, using a SW41 swinging rotor (Beckman,Fullerton, CA). Gradients were fractionated in 250-ml aliquots. These frac-tions were analyzed as described.

Proteinase K treatment

The QPP-containing vesicular fraction was obtained by discontinuous gra-dient fractionation and either were left untreated or were treated with pro-teinase K. Fractions were incubated in chymostatin (30mM) for 10 min onice. These vesicles either were left unpermeabilized or were permeabilizedwith 1% Triton X. Proteinase K (10mg/ml) and CaCl2 (1 mM) were addedto the samples for 20 min at 0°C. The reaction was terminated by theaddition of AEBSF (100mM).

QPP purification

QPP was purified from Jurkat cells as previously described (9). Briefly,Jurkat S-110 supernatant was dialyzed against 4 l of 50 mM acetic acid,titrated to pH 4.5 with NaOH. The protein sample was clarified by cen-trifugation at 10003 g for 10 min at 4°C. The clarified supernatant wasconcentrated on a Centricon 50 membrane to about 10 ml. The concen-trated sample was loaded onto a 3 ml of HiTrap SP Sepharose column andequilibrated with 50 mM acetate, pH 4.5 (start buffer). The column waswashed with 10 column volumes of start buffer and eluted with a linear0–300 mM NaCl gradient in start buffer. Active fractions were pooled andconcentrated to about 1 ml on a Centricon 50 membrane, then to about 0.2ml on a Microcon 30 membrane. The concentrated material was loadedonto a Superose 12 gel filtration column and equilibrated with 50 mMacetate, pH 4.5, and 150 mM NaCl. Active fractions were pooled and usedas a purified preparation of the activity.

ResultsQPP is an intracellular cytoplasmic protease

QPP was initially purified from the cytoplasmic fraction of JurkatT cells (9). To determine the cellular localization of QPP, confocalmicroscopy was performed on cells expressing a QPP-GFP fusionprotein. 293T human fibroblasts transfected with this chimericconstruct showed a cytoplasmic distribution (Fig. 1A). UnlikeCD26/DPPIV, QPP does not localize to the cell membrane. Thiswas confirmed by biochemical means; namely, surface biotinyla-tion of QPP expressing cells, followed by immunoprecipitation,did not yield any biotinylated QPP molecules (M. Chiravuri et al.,

5696 CHARACTERIZATION OF A CD26/DPPIV-LIKE PROTEASE

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

unpublished observations). To ensure that the distribution of QPP-GFP was not altered by the presence of the GFP portion of theconstruct, immunohistochemical analysis was performed with aQPP-Myc construct (Fig. 1B). Fig. 1Cshows that QPP-GFP andQPP-Myc colocalize, demonstrating that the nature of the tag didnot affect QPP localization.

QPP is synthesized with a cleaved propeptide

Proteins are routed to the secretory pathway by signal peptides thatinsert the protein into the endoplasmic reticulum (ER) membrane.

Primary sequence analysis of QPP using a Kyte and Doolittle hy-drophobicity plot (15) indicates a hydrophobic sequence in theN-terminus of QPP, aa 1–21. To verify this, QPP was purified fromJurkat cells and subjected to N-terminal sequencing. Fig. 1Dshows a comparison of the amino acid sequence of full-length QPPdeduced from its cDNA (9) and the N-terminal sequence of matureQPP purified from Jurkat cells. It can be seen that a 29-aa peptideis cleaved from newly synthesized QPP. The cleaved sequencemay be either a signal peptide alone or may include a signal pep-tide and a short propeptide.

QPP undergoesN-linked glycosylation

SDS-PAGE analysis shows that QPP migrates as a 58-kDa species,even though the predicted m.w. of QPP based on its primary se-quence is 53 kDa. Similar to the lysosomal protease PCP (7), QPPhas six potentialN-glycosylation sites, and glycosylation at thesesites could cause the observed increased in molecular mass. Tu-nicamycin is an inhibitor ofN-acetylglucosamine (GlcNAc) trans-ferase that blocks assembly of the GlcNAc-dolichol complex andthus prevents glycosylation of asparagine residues (16). QPP-transfected fibroblasts were treated with tunicamycin (5mg/ml) forvarious periods of time. Fig. 2Ashows that the upper 58-kDa bandgradually disappeared with increased time of tunicamycin treat-ment. We also treated QPP with PNGase F, an enzyme that cleavesbetween the innermost GlcNAc and asparagine residues of highmannose, hybrid and complex carbohydrates fromN-linked gly-coproteins (17). PNGase F treatment also caused the disappearanceof the 58-kDa band (Fig. 2B). These results indicate that the 58-kDa band represents a processed glycosylated form of QPP, whilethe 53-kDa band is the unprocessed form.

FIGURE 1. Localization of QPP by confocal microscopy in 293T hu-man fibroblasts and N-terminal sequencing of mature QPP. Human 293Thuman fibroblasts were transfected with a QPP-GFP (green;A) and a QPP-Myc (red;B) construct. These cells were permeabilized and analyzed withan anti-Myc Ab.C, Merged computer images, where colocalization regis-ters as yellow/orange.D, QPP is synthesized with a propeptide that iscleaved. Purified QPP was subjected to N-terminal sequencing, and thissequence was compared with the full-length sequence of QPP.

FIGURE 2. QPP isN-glycosylated, andN-glycosylation is required for QPP enzymatic activity. Immunoblot analysis was performed with anti-Myc Ab.A, QPP-Myc (transfected) fibroblasts treated with tunicamycin (5mg/ml) for the indicated times.B, Anti-Myc Western blot of QPP-Myc, either untreatedor treated with PNGase F.C, QPP-Myc fibroblasts treated with cyclohexamide (15mg/ml) for the indicated times.D, Analysis of QPP enzymatic activity(f) and proteasome activity (M) in cells treated with tunicamycin for the indicated times.

5697The Journal of Immunology

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

To further confirm this post-translational modification, we ana-lyzed the effect of blocking new protein synthesis with cyclohex-amide. Cells were transfected with QPP and treated for varyinglengths of time with cyclohexamide. As shown in Fig. 2C, follow-ing the inhibition of new protein synthesis, the ratio of the 53-kDaform of QPP to the 58-kDa form decreases. These results indicatethat the 58-kDa species is the mature molecule.

To determine the functional significance of the glycosylation,we analyzed the effect of glycosylation on QPP enzyme activity.Cells transfected with QPP were treated with tunicamycin for var-ious time periods, and the ability of QPP to cleave the reportersubstrate Ala-Pro-AFC was measured. QPP activity was signifi-cantly decreased following tunicamycin treatment, indicating thatthe 58-kDa species is the active form of QPP, while the unglyco-sylated form lacks QPP activity (Fig. 2D). On the other hand,tunicamycin treatment did not alter the localization of QPP (seeFig. 6 below). As a control, the cleaving activity of the proteasomewas measured in each of the samples. The proteasome is a cyto-solic molecular complex (18) that does not undergo a significantdecrease in enzyme activity following tunicamycin treatment.Consistent with this, no change in proteasome activity was seen inthe tunicamycin-treated samples.

QPP localizes to vesicles with similar density to lysosomes

To localize QPP we performed biochemical organelle fractionationusing both discontinuous and continuous metrizamide/Percoll den-sity gradients. This was conducted on human fibroblasts trans-fected with a QPP-Myc construct as well as on untransfected Ju-rkat T cells. The transfected fibroblasts had identical phenotypeand growth characteristics as untransfected cells. QPP-Myc fibro-blasts were lysed by cavitation, and these lysates were fraction-ated. Four fractions were obtained from the discontinuous gradi-ents and were analyzed using organelle-specific markers. As acytosolic marker we measured the chymotrypsin activity of theproteasome complex, using the reporter substrate zGGL-AMC (19,20). The lysosomal marker used wasb-hexosaminidase activity(14), while LAMP1 was used as a marker for late endosomes andlysosomes (21). The ER marker used was calnexin (22), while thetrans-Golgi network (TGN) marker was Rab11 (23). To detectQPP we performed Western blot analysis and measured its activitywith the QPP reporter substrate Ala-Pro-AFC.

The analysis of the four fractions is shown in Fig. 3. The leastdense fraction (I; soluble) had the most proteasome activity (Fig.3A), but contained very little QPP (Fig. 3,F and G). Fractions

FIGURE 3. Subcellular organelle fractionations of QPP-Myc 293T human fibroblasts using discontinuous gradients. Four fractions were obtained andassayed for proteasome activity as a cytosolic marker (A), immunoblot analysis of the secretory vesicle marker Rab11 (B), immunoblot analysis of the ERmarker calnexin (C), immunoblot analysis of the lysosome/late endosome marker LAMP1 (D), enzyme assay of lysosomalb-hexosaminidase activity (E),QPP enzymatic activity (F), and immunoblot analysis with an anti-Myc Ab (G).

5698 CHARACTERIZATION OF A CD26/DPPIV-LIKE PROTEASE

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

enriched in Rab11 (II; Fig. 3B) and calnexin (IV; Fig. 3C) alsoshowed little QPP activity (Fig. 3F). Lysosomes and late endo-somes were identified in fraction III by the presence of LAMP1(Fig. 3D), and this fraction also had maximal activity of lysosomalb-hexosaminidase activity (Fig. 3E). The fraction containing ly-sosomes (III) had the majority of QPP enzymatic activity (Fig.3F). This was confirmed using Western blot analysis (Fig. 3G).

QPP was first isolated from Jurkat cells where it is constitutivelyexpressed. Therefore, we also analyzed QPP distribution withinthese cells. Jurkat cells were lysed by Dounce homogenization,and organelle fractionation was performed as described above. Thesame cell fractionation was conducted on untransfected Jurkatcells (Fig. 4). The proteasome activity localizes to the soluble frac-tion (I; Fig. 4A), and the majority of Rab11 migrates in fraction II(Fig. 4B), while calnexin localizes to the fraction IV (Fig. 4C). As

in the transfected fibroblasts, LAMP1 was highly enriched in frac-tion III, although some was also found on the less dense fractionII. Most of theb-hexosaminidase activity was found in fraction III(Fig. 4, D andE). The presence of membrane-bound LAMP1 butlow b-hexosaminidase activity in fraction II might indicate that afraction of the lysosomes was broken during Dounce homogeni-zation, causing the release of soluble proteins such asb-hex-osaminidase and decreasing the density of the lysosomes. The ma-jority of QPP Ala-Pro-AFC cleaving activity was found in fractionIII (Fig. 4F). These results suggest that QPP localizes to vesiclesthat are either part of the lysosomal compartment or to vesicles thathave similar density to lysosomes.

QPP is routed to a nonlysosomal vesicular compartment

To further define the localization of QPP within the lysosome-containing fraction (III), we performed colocalization experimentswith QPP and the late endosomal and lysosomal marker, LAMP1(21). QPP-GFP and LAMP1 did not colocalize, as determined byconfocal microscopy (Fig. 5,A–C), suggesting that QPP is routedto alternate vesicles that migrate with the lysosomes under thediscontinuous gradient fractionation conditions.

A higher resolution was obtained by performing continuous gra-dient separations. QPP-expressing fibroblasts were separated into51 fractions and probed for the late endosomal and lysosomalmarker, LAMP1, and QPP (Fig. 5D). These fractions were alsoprobed for adaptin-a, a component of adaptor protein 2 that isfound in clathrin-coated vesicles from the plasma membrane (24,25), and Rab11, a small GTP-binding protein that associates withthe TGN, some secretory vesicles, and recycling endosomes (23).

Fig. 5D shows that the distribution pattern of QPP is differentfrom that of LAMP1. Most QPP is found in relatively more densevesicles, with its peak at fractions 45 and 47. The appearance ofQPP in the least dense soluble fractions is presumably from rup-tured vesicles, perhaps due to the increased centrifugation timerequired for this fractionation method. The LAMP1 expressionpattern has two peaks: one at fraction 25, possibly representing thelate endosomes, and the other at fraction 41, representing the ly-sosomes, according to theb-hexosaminidase marker (Fig. 5E). Thedistribution pattern of Rab11 is significantly different from that ofQPP, as it localizes to the less dense vesicles, in agreement withthe data from the discontinuous gradients (Fig. 3B). Adaptin-a alsoshows a different distribution than QPP, and their respective frac-tions of maximum expression are clearly different. Analyses of thefractions for QPP andb-hexosaminidase enzymatic activities showa distribution in accordance with the Western blot analysis (Fig.5E). The QPP activity reaches a maximum in fraction 47 and isslightly elevated in the first few fractions, where the soluble QPPwould be found. Theb-hexosaminidase enzymatic activity corre-lates with the Western blot profile of lysosomal LAMP1 distribu-tion, with maximum activity at fraction 41 and low activitythroughout the endosomal range. It should be noted that the ele-vated QPP activity in the least dense fractions has no correspond-ing increase inb-hexosaminidase activity, further supporting theidea that, while the intact QPP-containing vesicles and the lyso-somes have similar densities, as was seen with the discontinuousgradients, they are not the same vesicles.

Many lysosomal proteins, particularly lysosomal matrix pro-teins, are targeted to the lysosome through the M6P receptor path-way (13). To determine the effect of glycosylation on the local-ization of QPP, fibroblasts treated or untreated with tunicamycinwere subjected to organelle purification using discontinuous gra-dients. As shown in Fig. 6A, unglycosylated QPP was still foundin the same fraction (III) and did not show increased accumulationin the Golgi apparatus. This shows that glycosylation and the M6P

FIGURE 4. Subcellular organelle fractionation of untransfected CD262

Jurkat T cells. The fractions were analyzed for proteasome activity (A),Rab11 expression (B), expression of the ER marker calnexin (C), immu-noblot analysis of LAMP1 (D), lysosomalb-hexosaminidase activity, andQPP enzymatic activity (F).

5699The Journal of Immunology

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

receptor pathway are not required for the routing of QPP, suggest-ing a nonlysosomal vesicular localization. However, M6P-inde-pendent localization of proteins to lysosomes has also been de-scribed (26).

To confirm that QPP is located within vesicles, QPP-HA fibro-blasts were fractionated using discontinuous gradients, and intactvesicles from fraction III were treated with proteinase K. Thisfraction contained lysosomes/late endosomes and QPP-containingvesicles (Fig. 6B,lane 1). When the intact vesicles were first per-meabilized with detergent and then treated with proteinase K, QPPwas degraded (Fig. 6B,lane 2). However, in the absence of de-tergent, no degradation of QPP by proteinase K was seen (Fig. 6B,lane 3), indicating that QPP is located on the inside of the vesicles.

QPP is secreted in an active form

Given that QPP localizes to a post-Golgi vesicular fraction distinctfrom lysosomes, we analyzed the secretion behavior of QPP. Todetermine whether QPP is secreted, we tested the supernatants ofQPP-transfected fibroblasts and vector-transfected fibroblasts forQPP and observed that QPP was secreted in a functionally activeform (data not shown). Calcium-dependent exocytosis is a welldescribed phenomenon (27, 28). QPP release from Jurkat cells wasmonitored after the addition of the calcium ionophore, ionomycin.Jurkat cells in PBS with 1 mM CaCl2 were treated with ionomycinfor 15 min at 37°C, and QPP activity in the supernatant was mea-sured. As Fig. 7 shows, ionomycin treatment of Jurkat cells causedincreased the secretion of QPP. As a control for lysed cells, theactivity of the cytosolic proteasome complex was measured in thesupernatant. The proteasome activity remained similar for both cellpopulations. These results suggest that QPP secretion can be trig-gered by calcium mobilization.

DiscussionA large number of signal proteins from diverse families such ascytokines, chemokines, and neuropeptides show a remarkable con-servation of an X-Pro motif on the N-terminus (4). Over the lastfew years, it has become apparent that a number of these signalpeptides are shorn of their X-Pro motifs by CD26/DPPIV withprofound consequences to their biological function. These includethe chemokines RANTES, SDF1, MDC, and eotaxin (2, 3, 5, 6).The ubiquity of the X-Pro motif on signal molecules suggests that

FIGURE 6. QPP localizes in an M6P-independent manner and is foundwithin the vesicles.A, QPP-Myc transfected fibroblasts were treated withtunicamycin (5mg/ml) for 24 h and then subjected to organelle fraction-ation using discontinuous gradients. The four fractions were analyzed byanti-Myc immunoblot analysis.B, Proteinase K treatment of QPP-contain-ing vesicles. Fibroblasts transfected with QPP-HA were subjected to or-ganelle fractionation. The fraction containing lysosomes and QPP-express-ing vesicles (III) was 1) left untreated, 2) treated with Triton X andproteinase K, or 3) treated with proteinase K alone. These fractions wereanalyzed using the anti-HA Ab.

FIGURE 5. QPP localizes to nonlysosomal vesicles.A, Confo-cal microscopy of QPP-GFP (green) fibroblasts. These cells werepermeabilized and probed with the anti-LAMP1 (red) Ab (B) and amerged computer image (C) was made.D, Cellular fractionation ofQPP-HA fibroblasts on a continuous density gradient. The gradientwas prepared as described inMaterials and Methods. Immunoblotanalysis was performed of LAMP1, adaptin-a, QPP, and Rab 11,respectively.E, Analysis and comparison of QPP andb-hex-osaminidase enzymatic activities in the various fractions.

5700 CHARACTERIZATION OF A CD26/DPPIV-LIKE PROTEASE

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

this motif may be part of an important post-translational regulatorymechanism in a variety of signal pathways in multiple tissues.

Although there are many potential substrates with N-terminalX-Pro motifs, few known enzymes have the requisite substratespecificity to cleave these motifs (4). Until recently, only the cellsurface molecule CD26/DPPIV was known to have the correctsubstrate specificity at physiologic pH to be able to cleave theX-Pro motif from these molecules. Like CD26/DPPIV, QPPcleaves N-terminal dipeptides when the penultimate amino acid isa proline and, to a lesser extent, an alanine (4, 9–11). The discov-ery of a CD26/DPPIV-like activity intracellularly expands the po-tential realm of post-proline-cleavage and regulation.

QPP was initially isolated and cloned from Jurkat T cells (9).Although functionally homologous to CD26/DPPIV, QPP bears nosequence homology with the former enzyme. It does, however,share homology with the carboxypeptidase, PCP (7, 9). Similarlyto CD26/DPPIV and PCP, QPP is targeted to the ER, undergoescleavage of a propeptide, and isN-glycosylated. The glycosylationof QPP is necessary for its enzymatic activity, either to assume itsnative three-dimensional conformation or for substrate binding.However, the absence of glycosylation does not perturb the intra-cellular localization of QPP. While many of the reported experimentswere conducted on QPP-transfected fibroblasts, overexpressing ofQPP did not alter the phenotype or growth characteristics of the cells.

Unlike CD26/DPPIV, QPP does not localize to the cell surface,but to intracellular vesicles distinct from the TGN. This was con-firmed by the fact that surface biotinylation did not yield any bi-otinylated QPP molecules, as tested biochemically with immuno-precipitations (our unpublished observations). In a discontinuousgradient, QPP-containing vesicles comigrated with the lysosomes;however, in a continuos gradient we were able to show that QPPhas a different distribution from the lysosomal marker LAMP1.QPP-containing vesicles are downstream from the TGN and aredistinct from Rab11-containing vesicles and clathrin-coated endo-cytic vesicles. Interestingly, QPP-containing vesicles have a sim-

ilar density as lysosomes, but under more stringent fractionationcan be seen to be more dense than lysosomes. Also, as seen withconfocal microscopy analysis, QPP does not colocalize withLAMP1, and unlike most soluble lysosomal proteins, QPP local-izes in a M6P-independent manner. However, a glycosylation-in-dependent mechanism of routing to lysosomes has also been de-scribed in the literature (26). Within its vesicular compartment,QPP may come into contact with substrate molecules that are inthe process of being synthesized or secreted. QPP does not appearto localize to clathrin-coated endocytic vesicles, but this does notrule out contact with endocytosed substrate molecules.

Calcium-dependent exocytosis of vesicular proteins is a well-described phenomenon (27–29). QPP secretion occurs via a calci-um-dependent mechanism and can be induced by the calciumionophore ionomycin, although secreted QPP represents only aminor part of the total QPP activity. Preliminary experiments in-dicate that QPP is also secreted in response to TCR-mediated sig-naling (H. Lee et al., manuscript in preparation). Such secretionbehavior would allow the cell to release the aminodipeptidase ac-tivity in response to stimulation. Given that extracellular signalmolecules play a pivotal role during cellular activation, the releaseof aminodipeptidase activity may play an important modulatoryrole. It remains to be seen whether QPP and CD26/DPPIV work inconcert or through distinct pathways.

We recently found that highly specific inhibitors of post-proline-cleaving aminodipeptidases trigger a caspase-dependent apoptoticpathway in quiescent lymphocytes, but not activated lymphocytes(12). These results reinforce the idea that post-proline-cleavingaminodipeptidases play an important role in lymphocyte ho-meostasis. Interestingly, the target of these inhibitors appeared tobe QPP rather than CD26/DPPIV, indicating that there may besome separation of function between these two enzymes that sharesubstrate specificity, but have distinct subcellular localizations. Al-though QPP was isolated from lymphocytes, Northern blot analy-sis indicates that QPP is expressed in multiple tissues (M. Chira-vuri et al., unpublished observations), suggesting a diverse role forthis gene product.

Post-prolyl aminodipeptidase-mediated post-translational regu-lation of signal molecules has been a rapidly expanding field ofstudy in the last few years. The discovery of alternate CD26/DP-PIV-like activities in distinct subcellular locations contributes tothe understanding of what appears to be a complex regulatorymechanism at the post-translational level.

AcknowledgmentsWe thank Dr. J. Fred Dice for helpful discussions and support. We alsothank Nicole D’Avirro, Shurjo Sen, and Sreya Urs for critical reading ofthe manuscript.

References1. Varshavsky, A. 1996. The N-end rule: functions, mysteries, uses.Proc. Natl.

Acad. Sci. USA 93:12142.2. Shioda, T., H. Kato, Y. Ohnishi, K. Tashiro, M. Ikegawa, E. E. Nakayama, H. Hu,

A. Kato, Y. Sakai, H. Liu, et al. 1998. Anti-HIV-1 and chemotactic activities ofhuman stromal cell-derived factor 1a (SDF-1a) and SDF-1b are abolished byCD26/dipeptidyl peptidase IV-mediated cleavage.Proc. Natl. Acad. Sci. USA95:6331.

3. Oravecz, T., M. Pall, G. Roderiquez, M. D. Gorrell, M. Ditto, N. Y. Nguyen,R. Boykins, E. Unsworth, and M. A. Norcross. 1997. Regulation of the receptorspecificity and function of the chemokine RANTES (regulated on activation,normal T cell expressed and secreted) by dipeptidyl peptidase IV (CD26)-medi-ated cleavage.J. Exp. Med. 186:1865.

4. Vanhoof, G., F. Goossens, I. De Meester, D. Hendriks, and S. Scharpe. 1995.Proline motifs in peptides and their biological processing.FASEB J. 9:736.

5. Struyf, S., P. Proost, D. Schols, E. De Clercq, G. Opdenakker, J. P. Lenaerts,M. Detheux, M. Parmentier, I. De Meester, S. Scharpe, and J. Van Damme. 1999.CD26/dipeptidyl-peptidase IV down-regulates the eosinophil chemotactic po-tency, but not the anti-HIV activity of human eotaxin by affecting its interactionwith CC chemokine receptor 3.J. Immunol. 162:4903.

FIGURE 7. QPP secretion is regulated in Jurkat cells. Jurkat cells inPBS with 1 mM CaCl2 were treated with either PBS or 10mM ionomycinfor 15 min at 37°C. The supernatant was then analyzed for QPP activity(f). The amount of enzyme released is expressed as a percentage of thetotal activity in the cells. As a control the chymotrypsin activity of theproteasome complex was analyzed (M).

5701The Journal of Immunology

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

6. Proost, P., S. Struyf, D. Schols, G. Opdenakker, S. Sozzani, P. Allavena,A. Mantovani, K. Augustyns, G. Bal, A. Haemers, et al. 1999. Truncation ofmacrophage-derived chemokine by CD26/dipeptidyl- peptidase IV beyond itspredicted cleavage site affects chemotactic activity and CC chemokine receptor 4interaction.J. Biol. Chem. 274:3988.

7. Tan, F., P. W. Morris, R. A. Skidgel, and E. G. Erdos. 1993. Sequencing andcloning of human prolylcarboxypeptidase (angiotensinase C): similarity to bothserine carboxypeptidase and prolylendopeptidase families. [Published erratumappears in1993 J. Biol. Chem. 268:26032.]J. Biol. Chem. 268:16631.

8. Skidgel, R. A., and E. G. Erdos. 1998. Cellular carboxypeptidases.Immunol. Rev.161:129.

9. Underwood, R., M. Chiravuri, H. Lee, T. Schmitz, A. K. Kabcenell, K. Yardley,and B. T. Huber. 1999. Sequence, purification, and cloning of an intracellularserine protease, quiescent cell proline dipeptidase.J. Biol. Chem. 274:34053.

10. von Bonin, A., J. Huhn, and B. Fleischer. 1998. Dipeptidyl-peptidase IV/CD26on T cells: analysis of an alternative T-cell activation pathway.Immunol. Rev.161:43.

11. Morimoto, C., and S. F. Schlossman. 1998. The structure and function of CD26in the T-cell immune response.Immunol. Rev. 161:55.

12. Chiravuri, M., T. Schmitz, K. Yardley, R. Underwood, Y. Dayal, andB. T. Huber. 1999. A novel apoptotic pathway in quiescent lymphocytes identi-fied by inhibition of a post-proline cleaving aminodipeptidase: a candidate targetprotease, quiescent cell proline dipeptidase.J. Immunol. 163:3092.

13. Kornfeld, S., and I. Mellman. 1989. The biogenesis of lysosomes.Annu. Rev. CellBiol. 5:483.

14. Storrie, B., and E. A. Madden. 1990. Isolation of subcellular organelles.MethodsEnzymol. 182:203.

15. Kyte, J., and R. F. Doolittle. 1982. A simple method for displaying the hydro-pathic character of a protein.J. Mol. Biol. 157:105.

16. Elbein, A. D. 1987. Inhibitors of the biosynthesis and processing ofN-linkedoligosaccharide chains.Annu. Rev. Biochem. 56:497.

17. Panneerselvam, K., J. R. Etchison, and H. H. Freeze. 1997. Human fibroblasts prefermannose over glucose as a source of mannose forN-glycosylation: evidence for thefunctional importance of transported mannose. [Published erratum appears in1997J. Biol. Chem. 272:33444.] J. Biol. Chem. 272:23123.

18. Arrigo, A. P., K. Tanaka, A. L. Goldberg, and W. J. Welch. 1988. Identity of the19S ‘prosome’ particle with the large multifunctional protease complex of mam-malian cells (the proteasome).Nature 331:192.

19. Benham, A. M., and J. J. Neefjes. 1997. Proteasome activity limits the assemblyof MHC class I molecules after IFN-g stimulation.J. Immunol. 159:5896.

20. Figueiredo-Pereira, M. E., W. E. Chen, J. Li, and O. Johdo. 1996. The antitumordrug aclacinomycin A, which inhibits the degradation of ubiquitinated proteins,shows selectivity for the chymotrypsin-like activity of the bovine pituitary 20 Sproteasome [published erratum appears in1996 J. Biol. Chem. 1996 271:23602.]J. Biol. Chem. 271:16455.

21. Fukuda, M. 1991. Lysosomal membrane glycoproteins. Structure, biosynthesis,and intracellular trafficking.J. Biol. Chem. 266:21327.

22. Wada, I., D. Rindress, P. H. Cameron, W. J. Ou, J. J. D. Doherty, D. Louvard,A. W. Bell, D. Dignard, D. Y. Thomas, and J. J. Bergeron. 1991. SSRa andassociated calnexin are major calcium binding proteins of the endoplasmic re-ticulum membrane.J. Biol. Chem. 266:19599.

23. Chen, W., Y. Feng, D. Chen, and A. Wandinger-Ness. 1998. Rab11 is requiredfor trans-Golgi network-to-plasma membrane transport and a preferential targetfor GDP dissociation inhibitor.Mol. Biol. Cell. 9:3241.

24. Laporte, S. A., R. H. Oakley, J. Zhang, J. A. Holt, S. S. Ferguson, M. G. Caron,and L. S. Barak. 1999. Theb2-adrenergic receptor/b-arrestin complex recruitsthe clathrin adaptor AP-2 during endocytosis.Proc. Natl. Acad. Sci. USA 96:3712.

25. Chen, H., S. Fre, V. I. Slepnev, M. R. Capua, K. Takei, M. H. Butler,P. P. Di Fiore, and P. De Camilli. 1998. Epsin is an EH-domain-binding proteinimplicated in clathrin-mediated endocytosis.Nature 394:793.

26. Dice, F. J. 1999.Lysosomal Pathways of Protein Degradation. Landes Bio-science, Georgetown, TX.

27. Lledo, P. M. 1997. Exocytosis in excitable cells: a conserved molecular machin-ery from yeast to neuron.Eur. J. Endocrinol. 137:1.

28. Rodriguez, A., P. Webster, J. Ortego, and N. W. Andrews. 1997. Lysosomesbehave as Ca21-regulated exocytic vesicles in fibroblasts and epithelial cells.J. Cell Biol. 137:93.

29. Blank, P. S., M. S. Cho, S. S. Vogel, D. Kaplan, A. Kang, J. Malley, andJ. Zimmerberg. 1998. Submaximal responses in calcium-triggered exocytosis areexplained by differences in the calcium sensitivity of individual secretory vesi-cles.J. Gen. Physiol. 112:559.

5702 CHARACTERIZATION OF A CD26/DPPIV-LIKE PROTEASE

by guest on April 5, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from