Regulated expression and intracellular localization of cystatin F in human U937 cells Carl-Michael Nathanson 1 , Johan Wasse ´ lius 2 , Hanna Wallin 1 and Magnus Abrahamson 1 1 Department of Clinical Chemistry, Institute of Laboratory Medicine, and 2 Department of Ophtalmology, University of Lund, University Hospital, Lund, Sweden Cystatin F is a cysteine peptidase inhibitor recently discov- ered in haematopoietic cells by cDNA cloning. To further investigate the expression, distribution and properties of the native human inhibitor the promyeloid cell line U937 has been studied. The cells expressed relatively large quantities of cystatin F, which was found both secreted and intracellu- larly. The intracellular levels were unusually high for a secreted cystatin ( 25% of the cystatin F in 2- or 4-day culture medium). By contrast, U937 cells contained only 3–4% of the related inhibitor, cystatin C. Cystatin F purified from lysates of U937 cells showed three major forms car- rying two, one or no carbohydrate chains. Immunocyto- chemistry demonstrated a marked cytoplasmic cystatin F staining in a granular pattern. Double staining with a marker for endoplasmic reticulum revealed no colocalization for cystatin F. Analysis of the promoter region of the cystatin F gene (CST7) showed that it, like that of the cystatin C gene (CST3), is devoid of typical TATA- and CAAT-box ele- ments. In contrast to the cystatin C promoter, it does not contain multiple Sp1 binding sites, but has a unique site for C/EBPa, possibly explaining the restricted expression of the cystatin F gene. Cells stimulated with all-trans retinoic acid to differentiate them towards a granulocytic pathway, showed a strong ( 18-fold) down-regulation of intracellu- lar cystatin F and almost abolished secreted levels of the inhibitor. Stimulation with tetradecanoyl phorbol acetate, causing monocytic differentiation, also resulted in down- regulation (two fold to threefold) of cystatin F expression, whereas the cystatin C expression was essentially unaltered in both experiments. The results suggest that cystatin F as an intracellular cysteine peptidase inhibitor with readily regu- lated expression, may be a candidate to control the cysteine peptidase activity known to be essential for antigen presen- tation in different blood cell lineages. Keywords: cysteine protease; cysteine protease inhibitor; cystatin F; cystatin C; expression pattern. Mammalian papain-like (family C1) peptidases, such as cathepsins B, H, L, S and K, can be regulated by natural inhibitors belonging to the cystatin superfamily. These inhibitors are of three major types [1]. Those of type 1, cystatins A and B (also called stefins), are approximately 100 amino-acid long polypeptides lacking signal peptides and disulfide bridges. Type 2 cystatins, synthesized with signal peptides, are approximately 120 amino acids long and contain at least two disulfide bridges. Type 3 cystatins, the kininogens, are larger proteins containing three tandemly repeated type 2-like cystatin domains. Type 2 cystatins identified in higher animals include cystatins C, D, S, SA, and SN [2], but also the recently reported cystatin E/M [3,4] and cystatin F [5–7]. Human cystatin F is synthesized as a 145-amino-acid residue preprotein, with the first 19 residues theoretically constituting a signal peptide. Secretion of a 126-residue mature protein has been verified for recombinant cystatin F produced in insect cells [5]. The mature sequence shows 29– 34% identity when compared to the sequences of other human type 2 cystatins. It has two possible N-glycosylation motifs at positions 36–38 and 88–90 (cystatin C numbering). Recombinant cystatin F expressed in Sf9 insect cells is indeed glycosylated [5]. Compared to other cystatins, cystatin F is an unusually specific inhibitor. Among the family C1 cysteine peptidases studied, cystatin F binds papain and cathepsin L with high affinity (K i 0.1–1 n M ), but does not inhibit cathepsin B (K i > 1000 n M ) [5,6]. Cathepsin L is a peptidase involved in the normal lysosomal turnover of proteins. However, more specialized functions have also been attributed to the enzyme. So has cathepsin L been shown to be involved in the loading of MHC II complex at antigen presentation. In cortical thymic epithelial cells from mice devoid of the cathepsin L gene, CLIP 22 and CLIP 10, two intermediate processing products of the invariant chain accumulate [8]. This indicates that the amount of cysteine peptidase activity in the processing compartment for the invariant chain is crucial for antigen presentation and, moreover, that this activity may be regulated by a cystatin [9]. Besides the inhibitory site for family C1 peptidases, cyst- atin F carries a second site for inhibition of the family C13 enzyme, mammalian legumain or asparginyl endopeptidase Correspondence to M. Abrahamson, Department of Clinical Chemistry, Institute of Laboratory Medicine, University Hospital, S-221 85 Lund, Sweden. Fax: + 46 46 130064, Tel.: + 46 46 173445 E-mail: Magnus.Abrahamson@klinkem.lu.se Abbreviations: ATRA, all-trans retinoic acid; TPA, tetradecanoyl phorbol acetate; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. Note: web page available at http://www.klinkem.lu.se/E/abrahamson Note: nucleotide sequences are available in the DDBJ/EMBL/ GenBank databases under accession nos AJ51067, AJ51068, AJ51069 and AJ51070. (Received 12 June 2002, revised 12 August 2002, accepted 11 September 2002) Eur. J. Biochem. 269, 5502–5511 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03252.x (K i for the pig enzyme, 10 n M ) [10]. Like cathepsin L, mammalian legumain has been implicated in antigen pres- entation, but through processing of the antigen rather than the MHC part of the complex. Manoury et al.[11]studied the processing of tetanus toxin in disrupted lysosomes from an EBV-transformed B-cell line. By using a competitive substrate they demonstrated that the degradation of tetanus toxin was due largely to the activity of human legumain, however, common inhibitors of family C1 cysteine peptid- ases or aspartic proteases did not affect the degradation. Thus, cystatin F has the potential to regulate two different enzyme activities relevant for antigen presentation. In this context, it is intriguing that the cystatin F expression seems restricted to haematopoietic cells [5,6]. By Northern blotting of RNA from human tissues, the highest cystatin F mRNA levels were seen in peripheral blood cells and spleen [5]. At cDNA library Southern blotting of 61 human cell types, cystatin F was observed mainly in resting T-cells, premono- cytic cells, activated dendritic cellsand some natural killer cell clones [6]. This is in agreement with analysis of cystatin F EST clones in a collection of > 650 cDNA libraries, showing that 54 cystatin F clones were present in 20 of the libraries, all of which were derived from immune cells (mainly primary dendritic and T cells) [5]. Further analysis of 10 human immune cell lines showed the highest secretion levels from the premyeloid cell line, U937, low secretion levels from T-cell lines and no secretion from B-cell lines [5]. The present investigation was undertaken to further study cystatin F, through analysis of the expression, distri- bution and properties of the native human inhibitor in U937 cells. MATERIALS AND METHODS In silico cloning and sequencing of the human cystatin F gene A full-length cystatin F sequence from cDNA clone HCUDE60 [5] was run in a FASTA search with the GCG package at the Karolinska Institute Sequence Analysis Computer (KISAC, http://130.237.126.207; Karolinska Institutet, Stockholm, Sweden) to identify a full length human genomic clone containing the cystatin F gene. A 106- kb fragment (AL035661) was acquired which contained the entire gene including the promoter region. To verify the in silico cloned cystatin F gene sequence, the four exons including  100 bp of 5¢-and 3¢ flanking sequences as well as the promoter region were amplified by PCR using primers listed in Table 1. Both strands of the fragments were sequenced with an ABI Prism 310 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) using the Big Dye Terminator Cycle Sequencing Kit (Applied Biosystems) and either of the oligonucleotides used for PCR as sequencing primer. The sequencing reactions were performed according to the manufacturer’s advice. The cystatin F gene sequences determined have been deposited in the GenBank/EMBL database under accession nos AJ51067, AJ51068, AJ51069 and AJ51070. Promoter analysis To study the promoter region of the cystatin F and C genes, the nucleotides )1to)600 gene segments were analysed with the MOTIF search engine (the vertebrate database), Bioinformatics Centre, Institute for Chemical Research, Kyoto University, Japan (http://www.motif.genome.ad.jp/) using a cut-off score of 85. Cell culture The human promyleoid cell line U937 (ATCC no. CRL- 2367) was cultured in RMPI medium (Life Technologies Ltd, Paisley, UK) supplemented with 5% (v/v) foetal calf serum (Life Technologies) and antibiotics (10 UÆmL )1 penicillin and 10 lgÆmL )1 streptomycin; Life Technologies) at 37 °C and an atmosphere with 5% CO 2 , in 75 cm 2 culture flasks (Costar, Cambridge, MA, USA). For differ- entiation experiments, cells (1 · 10 6 mL )1 ) were seeded and cultured in the presence of 1 l M ATRA for 4 days or in the presence of 0.13 l M TPA for 2 days. After incubation for 4 and 2 days, respectively, the cells were set to 10 6 cellsÆmL )1 in fresh medium. ATRA or TPA was added at the same concentrations as above. The cells were incubated for an additional 24-h period. Following the second incubation the cells were counted and separated from the media. One portion of the cells was lysed with 1 mL Triton-X100 in NaCl/P i containing 1 · inhibitor cocktail (final concentra- tions of 5 m M benzamidinium hydrochloride, 15 m M NaN 3 and 10 m M EDTA) per 6 · 10 7 cells, for analysis in ELISA and Western blot experiments. Another portion of the cells was spun down to object glasses and used for immunocyto- chemical staining, and from the rest of the cells RNA was Table 1. Primers used for PCR amplification, sequencing and probe amplification. Standard amplification reactions were accomplished in a Perkin Elmer GeneAmp 2400 thermal cycler (Perkin Elmer, Foster City, CA, USA), using 0.6 l M primers and reagents from Perkin Elmer. The PCR cycles repeated 30 times were (denaturation, 94 °C, 30 s – annealing at specified temperature, 30 s – extension, 72 °C, 60 s). The sequencing reactions were performed in the same instrument using standard reaction conditions. Gene part Upstream primer (5¢-to 3¢) Downstream primer (5¢-to 3¢) Annealing temp. (°C) Seg. size (bp) Upper promoter TGA AGC TGG AAA CCA TCA TTC AAA ACA TTA GCA GGA ATT TTC 52 343 Lower promoter GGA GTT CTG CCA GGG AAC CAC GAC AGG GGA GAA CGC CAC TTA 57 587 Exon 1 GTG CTG CCT GAG AAG GAT TG GAA AGT GCC CTG GGG AAG ACC 62 333 Exon 2 TGA AGG CCC CAC TAA CAT CAG TAT ATC CGC CCT GCT CTC CTA 56 314 Exon 3 GAG GCC CTG CTT CCT AGT GGA TGC GTT AGA GAC GTG GTG ACG 58 289 Exon 4 CCG CAG GGA AAG TCT AAG CTC ACA TCT CTG CTG ATT ATT CAG 55 497 Cystatin F probe CTT CTG CTG CCT GGT CTT GA GCA CTT CAC CCG CTC ACT CGT CA 57 542 Cystatin C probe CGG CGA GTA CAA CAA AGC CA GGA GGT GTG CAT AAG AGG TG 57 320 Ó FEBS 2002 Cystatin F in U937 cells (Eur. J. Biochem. 269) 5503 extracted. The culture media were supplemented with 1· inhibitor cocktail and saved at )20 °C until analysed. Western blotting To 1 mL of cell extract prepared as described above, 10 lL of Cm-papain coupled to Sepharose-4B resin (with capacity to bind  10 lg cystatin) [12] was added. The solution was incubated for 4 days at 4 °C on a rocking table. Following incubation the Sepharose gel from 1 mL of mixture was allowed to sink and the supernatant was discarded. Sample buffer containing SDS and reducing agent was added and proteins were separated electrophoretically on 16% SDS- polyacrylamide gels and the buffer system of Laemmli [13]. The gels were electro-blotted using poly(vinylidene difluo- ride) membranes (Millipore, Bedford, MA, USA), the filters were subsequently incubated with a polyclonal rabbit anti- (human cystatin F) IgG (anti-cysF) [5] and visualized by chemoluminescence (ECL Plus kit; Amersham, Bucking- hamshire, UK). Immunocytochemical staining Approximately 80 000 cells were spun down to object glasses and fixed in 3.7% (v/v) formaldehyde (Merck KGaA, Darmstadt, Germany) in NaCl/P i . Unspecific binding sites were blocked with 0.2% (w/v) bovine serum albumin. Anti- cysF, diluted 1 : 10 000 in NaCl/P i for the absorption and double staining experiments and 1 : 1000 for the regulation experiments, was applied in the presence of 0.1% (w/v) saponin. Following incubation with the primary antibody thecellswherewashedandincubatedwithanOregonGreen labelled goat anti-rabbit IgG secondary antibody (Molecular Probes, Inc., Eugene, OR, USA) for 2 h. Finally, the cells were washed and mounted in PVA-DABCO (2.5% w/v 1.4- diazacyclo [2,2,2]octane, 13% w/v polyvinlyalcohol, 33% w/v glycerol, and 0.13 M Tris/HCl, pH 8). To visualize the cells a Bio-Rad MRC1024UV confocal laser-scanning microscope was used and the images taken were processed using Adobe PHOTOSHOP (Adobe Systems, Mountain View, CA, USA). For double staining of cystatin F and ER the staining was performed on cells cultured for 2 days. Cystatin F was stained as described above and after the last washing step 20 lgÆmL )1 Texas Red conjugated concanavalin A (Molecular Probes) was applied and subsequently incubated for 30 min at room temperature. The cells were then washed and mounted as above. Quantitative protein assays Cell lysates and culture media were analysed by ELISA to determine the concentrations of cystatin F [5] and cystatin C [14,15]. The total protein content in cell lysates was determined with the Coomassie Protein Assay Reagent (Pierce, Rockford, IL, USA) according to the manufac- turer’s Micro Method. RNA extraction and Northern blot RNA was extracted from approximately 2 · 10 7 cells according to Chomczynski and Sacchi [16]. Twenty lgof RNA was separated electrophoretically on 1% agarose gels and transferred to Hybond-N nylon membranes (Amer- sham). The filters were prehybridized in Clontech Express Hybridization Solution (Clontech Laboratories, Palo Alto, CA, USA) for 30 min at 65 °C. A [a- 32 P]dCTP labelled cystatin F specific probe was added to 5 mL hybridization solution to give a specific activity of 10 6 cpm/mL. Following hybridization over night the filters were rinsed in 2 · NaCl/ Cit and washed in 2 · NaCl/Cit and 0.1% SDS for 45 min at room temperature; 0.2 · NaCl/Cit and 0.5% SDS for 45 min at room temperature; 0.2 · NaCl/Cit and 0.5% SDS for 30 min at 50 °C. After exposure the filters were stripped and the procedure was repeated with cystatin C and GAPDH specific probes. The cystatin F specific probe used was a 542-bp PCR fragment amplified from the cystatin F cDNA clone, HCUDE60 (Table 1). The cystatin C specific probe was a 320-bp PCR fragment amplified from the full-length human cystatin C cDNA clone, C6a [17] (Table 1) and the GAPDH specific probe used was a human GAPDH cDNA (Clonetech). Probes were labelled with [a- 32 P]dCTP (Amersham) according to the user manual in the Multi- prime DNA labelling system kit (Amersham). Subcellular fractionation U937 cells (10 7 ) were pelleted and resuspended in 1.5 mL homogenizing buffer (0.25 M sucrose, 10 m M tricine/NaOH, pH 7.4, 1 m M EDTA) and homogenized in a glass homog- eniser. The homogenate was treated in a series of centrif- ugation steps. After each centrifugation step the pellet was redissolved in lysis buffer (0.2% (v/v) Triton-X100 in water) and the supernatant was transferred to the next step. Each lysed pellet formed one fraction. The last supernatant formed the supernatant fraction. The centrifugation steps were as follows: debris fraction: 10 min, 600 g;heavy fraction: 10 min, 2000 g; light fraction: 10 min, 15 000 g. Cystatin C, cystatin F and total protein were measured as above. The activity of b-hexosaminidase was measured according to Hultberg et al.[18]. The heavy and the light fractions contained elevated levels of cystatin F and were therefore selected for an ultracentrifugation experiment. Cells (1.4 · 10 8 ) were pel- leted and treated as above with the exception that the heavy and light fraction pellets were not lysed in lysis buffer but resolved in homogenizing buffer. Optiprep (Axis-Shield PoC AS, Oslo, Norway) was diluted to 20% in Optiprep dilutent (8% sucrose, 1 m M EDTA, 20 m M Tricine/NaOH, pH 7.8). Five mL Optiprep solution was overlaid with 400 lL sample and centrifuged at 180 000 g,3h,4°Cina vertical angle rotor (Vti80, Beckman Instruments, Inc., Palo Alto, CA, USA) in a Beckman L8-M80 ultracentrifuge (Beckman Instruments). Finally the heavy fraction was divided in 9, and the light fraction in 11, subfractions by puncturing the plastic centrifuge tube with a syringe connected to a peristaltic pump and a fraction collector. Materials Oligonucleotides were purchased from DNA Technology A/S (Aarhus, Denmark). U937 cells were a kind gift of U. Gullberg, Department of Haematology, Lund University, Sweden. Recombinant human cystatin C was expressed in an Escherichia coli system and purified as described [19]. 5504 C M. Nathanson et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Recombinant human cystatin F was produced in a bacu- lovirus expression system [5]. If nothing else stated chemicals were from Sigma (St Louis, MO, USA). RESULTS Native cystatin F in human U937 cells Earlier reported characteristics of cystatin F have been based on studies of recombinant protein [5,6]. To extend our knowledge about the inhibitor, we wanted to study the properties and distribution of the native protein from a human cell source. The promyelocytic cell line, U937, was chosen as starting material because it showed the highest secretion of cystatin F among 10 cell lines previously analysed [5]. It was not possible to detect cystatin F directly in culture media or cell lysates by immunoblotting. How- ever, by absorption on Sepharose-bound Cm-papain fol- lowed by Western blot analysis, the cell lysate-derived inhibitor could readily be detected. Cystatin F present in U937 cells shows a four-band pattern at SDS/PAGE (Fig. 1). At comparison with insect cell produced recom- binant cystatin F (Fig. 1, lanes 1 and 2), which appears as one mono- (band B) and one di-glycosylated (band A) species at approx. equal proportions [5], it was evident that native human cystatin F exists in these two glycosylated forms, but also as one most likely unglycosylated species (lane 3, band C), with mobility exactly as recombinant cystatin F after deglycosylation with PNGase F [5]. The antiserum also recognized a weaker additional band (lane 3, band a)withM r approximately 3200 Da higher than the diglycosylated cystatin F species. The nature of this band is presently unknown and is discussed below. In order to study the production of cystatin F compared to the ubiquitous inhibitor, cystatin C, we measured the cystatin contents in cell lysates and culture media of cells grown for 2 and 4 days with cystatin F and cystatin C specific ELISAs. Generally, the total cystatin F amounts were approximately 10 times lower than those of cystatin C (Fig. 2A,B) but the proportional distribution of localization of the two proteins differ. The genes for both cystatins encode a signal peptide and the secreted amounts of the proteins are higher than the intracellular ones. However, while cystatin C displays 24–30 times higher secreted levels (Fig. 2B), cystatin F only shows approximately three times higher extracellular protein levels (Fig. 2A). Thus, a signi- ficant part of the cystatin F produced by U937 cells is retained intracellularly or is taken up by the cells immedi- ately following secretion. Accordingly, the relatively low amount of cystatin F produced by U937 cells reported earlier [5] is an underestimate of the real production of cystatin F in U937 cells. The high portion of intracellular cystatin F led us to fractionate cells to determine the localization of the inhibitor. Initially cells were fractionated in a debris fraction (containing nuclei and cell debris), a heavy and a light fraction containing mitochondria and various cell organ- elles, and a supernatant containing the smallest cellular vesicles. Cystatin F, cystatin C and protein contents in the fractions were measured. The highest cystatin F level was seen in the debris fraction (Fig. 3A) but the ratio between cystatin F and protein (Fig. 3B) revealed that cystatin F was enriched in both the heavy and the light fractions compared to the total (with a factor of 2) and supernatant (with a factor of 20) fractions. Cystatin C levels were below the detection limit in all fractions. α C A B 3 1 2 Fig. 1. Western blotting of native human cystatin F. A cystatin F spe- cific polyclonal antiserum was used for immunoblotting, after cysta- tin F in a U937 cell extract had been partially purified by affinity chromatography on immobilized Cm-papain. Lanes 1 and 2 contain 30 and 10 ng insect cell-derived recombinant cystatin F, respectively. The two bands seen represent cystatin F with one (B) and two (A) carbohydrate side chains [5]. Lane 3 shows cystatin F produced by U937 cells. The mono- and diglycosylated forms (A and B) can be seen but also a nonglycosylated form (C). The band marked a is discussed in the main text. 0 1 2 3 4 5 6 7 8 2 4 B 0 10 20 30 40 50 60 70 80 24 Incubation time (days) Cystatin F (ng/culture) A Cystatin C (ng/culture) Fig. 2. Cystatin F and C produced by U937 cells. (A) Cystatin F was quantified in cell extracts and media by ELISA. In the representative experiment shown (one out of three with the same trend of results), intracellular levels (black bars) of cystatin F were approximately 3-fold lower than the extracellular levels (grey bars) both after 2 and 4 days of incubation. (B) Similarly measured by a specific ELISA, the cystatin C levels were 24- to 30-fold higher extracellularly than intracellularly. The total cystatin C amount was  10-fold higher than the total cystatin F amount. Measurements were performed in duplicate and results are presented as mean values with error bars denoting SD. Ó FEBS 2002 Cystatin F in U937 cells (Eur. J. Biochem. 269) 5505 The heavy and the light cellular fractions were subfract- ionated by ultracentrifugation to further localize cystatin F. We also measured the activity of the mainly lysosomal enzyme, b-hexosaminidase, to assess if cystatin F is located in lysosome-like organelles. In both the heavy (Fig. 4A, subfraction 8) and the light (Fig. 4B, subfraction 9) fractions cystatin F coeluted with the peak of b-hexos- aminidase activity. To further investigate the intracellular localization of cystatin F we stained cells with the polyclonal antiserum and visualized the immunoreaction with a FITC labelled secondary antibody in a confocal microscope. Cystatin F showed a vesicular staining pattern (Fig. 5A), whereas after preincubation with recombinant cystatin F the signal was absorbed completely (Fig. 5B). The latter demonstrates that cystatin F is specifically stained by the antiserum at immunohistochemistry. The vesicular staining of cystatin F agrees with the results from the fractionation experiments and taken together, this strongly indicates that the major portion of intracellular cystatin F is present in smaller vesicles. In mouse fibroblasts overproducing human cystatin C, intracellular cystatin C routed for direct secretion is detectable and it shows costaining with markers for the ER [20]. This prompted us to examine a possible costaining of intracellular cystatin F (Fig. 6A) with Texas Red labelled concanavalin A as an ER-specific marker (Fig. 6B). A colocalization of cystatin F with the ER marker would give a yellow colour at overlays, but no such costaining could be observed (Fig. 6C). Thus, the intracellular cystatin F detected by the antiserum is likely not protein detected on the direct transport route to secretion, but rather protein temporarily stored in granules or found in vesicles following uptake from the medium. Fig. 3. Cystatin F concentrations in main subcellular fractions of U937. Cells were mechanically homogenized and fractioned in four fractions by stepwise centrifugation, as detailed in the Materials and methods section. Cystatin F concentrations were measured by ELISA and protein concentrations by a Coomassie binding assay. (A) The debris fraction showed the highest cystatin F concentration and the super- natant the lowest. (B) The cystatin F/protein ratio is > twofold higher in the heavy and light fractions than in the debris fraction or in total U937 cellular extract and > 20-fold higher than in the supernatant fraction. Measurements were performed in duplicate and results are presentedasmeanvalueswitherrorbarsdenotingSD. Fig. 4. Cystatin F in subfractions of the heavy and light subcellular U937 fractions. Cystatin F concentrations were measured with a specific ELISA, total protein concentration was measured with Coomassie binding assay and b-hexosaminidase activity was measured with a colourimetric assay, as detailed in the Materials and methods section. (A) Subfractions of the heavy subcellular fraction obtained by ultra- centrifugation in Optiprep. The total protein concentration (closed triangles and short dashed line, scale on right y-axis) was elevated in subfractions 3, 7, 8 and 9 with the highest concentration in fraction 8. The concentration of cystatin F (closed rectangles and line, scale on left y-axis) and b-hexosaminidase activity (open circles with long dashed line, scale on left y-axis) were elevated in subfractions 7, 8 and 9 with the highest concentrations in fraction 8. (B) Subfractions of the light subcellular fraction obtained by ultracentrifugation. Cystatin F and b-hexosaminidase were elevated in fraction 9. 5506 C M. Nathanson et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Regulation of cystatin F expression To shed light on the regulation of the cystatin F gene, we examined the sequence of the human gene, CST7.Atan initial search of the High Throughput Genome database at Karolinska Institutet (Stockholm, Sweden) by FASTA with the cDNA for cystatin F as probe, a 106 000 bp contig from chromosome 20 was found. Aligning putative exons from the cDNA showed that the contig contained the whole gene plus a long flanking stretch upstream of the cDNA start position. The human CST7 gene consists of four exons [21] as the mouse counterpart [6]. The three introns, 7711 bp in the found contig (7872 bp in Morita et al. [21]), 1468 bp and 590 bp, respectively, are localized in the same pattern as in the mouse gene. The second and third introns are localized in exactly the same positions, taking amino-acid homology match in account, as those of the other six known human type 2 cystatins, strongly indicating that the genes originate from a common ancestor. To verify the exon/ intron junctions and coding segments, primers were con- structed  100 bp 5¢ and 3¢ of the exons in the flanking intron segments. Direct sequencing of PCR products derived from genomic DNA of a normal Caucasian individual was performed and the sequences were compared with the contig. The obtained sequences showed no differences to the database contig sequence. The exon/ intron junctions were all in agreement with the GT/AG consensus sequence. The 5¢ flanking promoter region of the human cystatin F gene does not have an unusually high GC content (over nucleotides )1to)1000,  50%; nucleotides )1to)3000,  45%) and comparing CpG to GpC gives a ratio of  1/5. By contrast, a ratio of  1, indicating a continuously expressed gene [22], is found in the promoter region of the human cystatin C gene [23]. The first 600 bp of the cystatin C(CST3)andF(CST7) promoter regions (calculated from the starting ATG) differs in some obvious manners (Fig. 7). Cap signals are found in the vicinity of the start ATG codons but further upstream CST7 contains a second one 138 bp 5¢- to an alternative start codon [6]. The a-band in Fig. 1 possibly derives from a transcript initiated by this second initiation site. The promoter of the ubiquitously expressed cystatin C gene, CST3, contains eight Sp1 binding sites while CST7 only has one (at )314). Two GC-box elements can be found in CST3 while CST7 has one at )312. At position )500 a CAATT/enhancer binding protein a (C/EBPa) site is found in CST7, with no counterpart in CST3. Expression of cystatin F in differentiated U937 cells Radomska et al. [24] reported that TPA differentiates U937 cells in a monocytic direction and gives a down-regulation of C/EBPa. ATRA gives in contrast a granulocytic differen- tiation of U937 cells and a subsequent up-regulation of C/ EBPa, so we decided to test if the expression of cystatin F is regulated through differentiation. Stimulation by TPA. U937 cells were grown in the presence of 0.13 l M TPA for 2 days. A 2.5-fold down- regulation of cystatin F both in cell lysates and in culture media could be seen (Fig. 8A). Cystatin C was slightly A B Fig. 5. Immunocytochemical staining of cysta- tin F. U937 cells were spun down and fixed on object glasses. (A) The cells were incubated with a specific polyclonal antiserum against recombinant cystatin F and subsequently hybridized with a secondary FITC labelled antibody against rabbit IgG. A vesicular staining pattern can be seen. (B) After prein- cubation of the primary antibody with an excess of recombinant cystatin F [5] a total absorption of the signal was apparent. Fig. 6. Double staining of cystatin F and ER. (A) Cystatin F was stained with a human cystatin F specific polyclonal antiserum and a FITC labelled secondary antibody (green sig- nal). (B) Following the cystatin F staining, the ER was stained with Texas Red conjugated concanavalin A (red signal). (C) Overlay of images in (A) and (B), where a yellow colour would indicate colocalization. Arrows indicate distinct ÔholesÕ in the ER staining where cyst- atin F stains. Ó FEBS 2002 Cystatin F in U937 cells (Eur. J. Biochem. 269) 5507 down-regulated by TPA, with a factor of  1.6 (Fig. 8B). To investigate whether the regulation was transcriptional or translational we extracted RNA from the cells. The results from Northern blotting of the RNA (Fig. 8C) demonstrate a strong down-regulation of cystatin F expression at the mRNA level while cystatin C and control GAPDH mRNA levels were virtually unchanged. The down-regulation of cystatin F was also visualized through immunostaining of cells grown in presence or absence of TPA. A less pronounced staining could be seen in the TPA stimulated cells than in the unstimulated (Fig. 8D). Thus, TPA markedly down-regulates the cystatin F gene expression. This down-regulation would fit with C/EBPa as one regulator of the cystatin F expression in U937 cells. Stimulation by ATRA. The results were similar but even more pronounced when U937 cells were incubated with ATRA (Fig. 9). The intracellular cystatin F levels measured by ELISA were 18-fold lower after ATRA treatment and 9-fold lower in cell culture media (Fig. 9A). Furthermore Fig. 8. Regulation of cystatin expression in TPA stimulated U937 cells. Cells were incu- bated for 2 days in the presence of 0.13 l M TPA. The results from one representative experiment out of three performed are shown. (A) Cystatin F measurement, showing a 2.5-fold down-regulation upon TPA-stimula- ted differentiation (unstimulated, grey bars; stimulated, black bars). (B) Cystatin C meas- urement, demonstrating a 1.6-fold down- regulation (unstimulated, grey bars; stimula- ted, black bars). (C) Northern blot of RNA derived from unstimulated (lane 1) and sti- mulated (lane2) U937 cells, using specific cDNA probes for cystatin F (top), cystatin C (middle) and GAPDH (bottom). (D) Immu- nostaining of cystatin F in unstimulated (left) and TPA-stimulated (right) cells. Fig. 7. Structure of the cystatin F gene and promoter. The structures of the human cystatin F (CST7)andcystatinC(CST3) genes are illustrated schematically at the top. In the upstream region of the CST7 gene, the ATG (arrowhead) at position 0 is the most probable initiation codon for translation. Another ATG and possible initiation site is found at )66. Both possible mRNAs have corresponding cap signals (diamond). Only one Sp1 element (pyramid) can be found in the )600 to )1 segment whereas in the CST3 promoter 7 elements are found. Among other possible transcription factor sites found at a motif search, a unique C/EBPa element in the cystatin F promoter (star) could be of relevance for expression in U937 and native haematopoietic cells. 5508 C M. Nathanson et al. (Eur. J. Biochem. 269) Ó FEBS 2002 were mRNA signals almost completely lost in the treated cells (Fig. 9C). When comparing the immunostained cells before and after treatment the cystatin F signals were more or less extinct after stimulation (Fig. 9D). Such strong evidence of down-regulation could not be seen for cystatin C. The already low intracellular levels did not change and in the culture media only a 1.3-fold decrease was seen (Fig. 9B). Furthermore was the mRNA expression unchanged after ATRA treatment (Fig. 9C) and, thus, no indication of regulation by differentiation was at hand for cystatin C. In contrast to the TPA experiment, the strong down-regulation of cystatin F expression by ATRA con- tradictthatC/EBPa is a major regulator of cystatin F expression in U937 cells. DISCUSSION Cystatin F is a potent inhibitor of several important cysteine peptidases [5,10], but the physiological role of the inhibitor is still unknown. The immune cell restricted expression of cystatin F indicates that the inhibitor is involved in regu- lation of proteolytic events specific to such cells. Regulation of antigen presentation could be one such event. This agrees with results from cDNA libraries [5,6], demonstrating that dendritic cell subpopulations of haematopoietic cells are among the most abundant cystatin F producers. The in vitro properties of recombinant cystatin F show that it is a potent inhibitor of cathepsin L [5,6], as well as mammalian legumain [10]. These enzymes are known to be involved in invariant chain [8] and antigen [11] processing, respectively. Our present results showing a distinct intracellular localiza- tion of cystatin F are interesting in this context. Although a detailed study of the cystatin F transport route was beyond the scope of the present investigation, the fact that cystatin F is present in significant quantities intracellularly supports a possible intracellular function of cystatin F. Furthermore, the apparent distribution to smaller vesicles/ granules in the U937 cells indicates that cystatin F is at least partially localized to organelles resembling the fused lyso- somes/endosomes where antigen and invariant chain pro- cessing take place in antigen presenting cells [9]. Thus, as shown by our present comparisons with the distribution and regulation of cystatin C, cystatin F appears to be a better candidate for regulation of antigen presentation than cystatin C. Although control of critical cathepsin S activity in mouse dendritic cells [9] has been contributed to cystatin C it is a distinct possibility that cystatin F also is involved in such regulation, possibly in co-operation with cystatin C. It should, however, be stressed that this suggested function of cystatin F is hypothetical. Although most cystatins inhi- biting cathepsin L also inhibit cathepsin S, an inhibitory activity of cystatin F against cathepsin S has not yet been reported. The expression of the cystatin F (CST7)andC(CST3) genes differ considerably, both with respect to the overall levels and with respect to tissue/cell-specific expression pattern. CST3 expression is generally higher than that of CST7, as stressed by our present results for U937 cells. Moreover, CST3 is ubiquitously expressed [23], whereas cystatin F transcripts are found exclusively in immune cells [5,6]. Possible explanations to these differing expression patterns can be found in the comparison of the CST3 and Fig. 9. Regulation of cystatin expression in ATRA stimulated U937 cells. Cells were incu- bated for 4 days in the presence of 1 l M ATRA. The results from one representative experiment out of three performed are shown. (A) Cystatin F showed an 18-fold down- regulation in cell extract and a 9-fold down- regulation in medium (unstimulated, grey bars; stimulated, black bars). (B) The cystatin C concentration was barely detectable but unchanged in cell lysate; a slight down-regu- lation could be seen in culture medium. (C) Northern blot of RNA derived from unstim- ulated (lane 1) and stimulated (lane2) U937 cells, using specific cDNA probes for cysta- tin F (top), cystatin C (middle) and GAPDH (bottom). (D) Immunostaining of cystatin F in unstimulated (left) and ATRA-stimulated (right) cells. Ó FEBS 2002 Cystatin F in U937 cells (Eur. J. Biochem. 269) 5509 CST7 promoters presented here (Fig. 7). The CST3 pro- moter contains elements typical for a house-keeping gene (multiple Sp1 and GC-box elements), has an extremely high GC content and a CpG to GpC ratio close to unity, which agrees with its relatively high and unspecific expression pattern. The CST7 promoter, in contrast, lacks most of these CST3 promoter features. This agrees with the relatively low level of cystatin F expression in those few cells where it is expressed at all, such as U937. The expression of CST7 is readily regulated in contrast to that of CST3, as demonstrated in the present study. This difference must clearly be due to other differences in the two promoters. The unique C/EBPa binding site found in CST7 may at least partially explain the restricted and regulated cystatin F expression. The reasons why cystatin F in U937 cells is transported in a different way than the major type 2 cystatin of mammalian tissues, cystatin C, are unknown and certainly merit further studies. The cystatin F and C genes both encode preproteins with typical signal peptides, and the gene products are readily secreted in mammalian [20] and insect [5] cells, respectively. Cystatin F is an unusual type 2 cystatin in that it is a glycoprotein [5,6], and it is known that many proteins found in intracellular vesicular compartments are targeted through their carbohydrate side chains [25]. The comobility of the cystatin F bands from U937 cells with those of recombinant cystatin F in the Western blot experiment of the present study, strongly suggests that the native human inhibitor is glycosylated as predicted. Native cystatin F appears as three major bands when detected in U937 cell homogenates, which correlate exactly with the mobilities of di-, mono- and unglycosylated forms of recombinant cystatin F previously characterized by enzymatic removal of N-linked carbohydrate and carbohydrate composition analysis [5]. It is also noteworthy that all three forms were partially purified by their affinity to Cm-papain, strongly indicating that the carbohydrate side chains do not affect the enzyme-binding properties of cystatin F and should, thus, not affect the inhibition of papain-like family C1 peptidases. The antiserum we used for detection of intracellular cystatin F detected an additional band at Western blotting (Fig. 1). The nature of this band is unknown, but given the high specificity of the antiserum [5] and the absorption step on Cm-papain used, it is most likely an additional form of cystatin F. The band possibly represents a gene product of the alternative ATG start codon (Fig. 7). If transcription starts at the alternative start site then the protein would gain 41 in-frame N-terminal amino acids with no resemblance to a signal peptide but rather to a transmembrane segment [6]. This theoretical cystatin F variant has a calculated M r of 23 000 Da. A calculation taken from the Western blot (Fig. 1) gives a M r of 22 500 Da for the additional band detected. The possibility that the alternative ATG start codon is partially used in human cells is not contradicted by previous results, as previous recombinant cystatin F studies used constructs lacking the upstream ATG codon [5,6]. An alternative suggestion for a physiological function of cystatin F can be speculated from our present results. The expression of cystatin F in unstimulated U937 cells, but not in more differentiated cells of the same lineage resembles the expression pattern of proteins known to be substitutents of secretory granules of granulocytes, such as cathepsin G [26]. Perhaps cystatin F is involved in inflammatory reactions promoting granulocyte migration and release of granule content to combat exogenous threats. Cystatin F could, e.g. inactivate family C1 target enzymes from bacteria or protozoan parasites such as the virulence factor of Chagas’ disease, cruzipain, in Trypanosoma cruzi infections. Cystatin F inhibits cruzipain as strongly as it inhibits cathepsin L in vitro (M. Abrahamson and J. Scharfstein, unpublished results). Clearly, a more detailed study of the subcellular localization and transport of cystatin F in model U937 cells would be valuable in understanding the role of cystatin F in haematopoietic cells. Further work to localize subfractions of native haematopoetic cells expressing cystatin F should provide additional clues to the function of cystatin F in health and disease. Since acceptance of this manuscript for publication, we have become aware of two recent publications that contain results of relevance for the present work. In a study aiming at characterization of the proteome of lysosomes in U937 cells, cystatin F was found as one of 15 proteins binding to immobilised mannose-6-phosphate receptor [27], indicating that either cystatin F contains mannose-6-phosphate or is co-localized with, and binding to, a mannose-6-phosphate containing protein. This favours our suggestion that intra- cellular cystatin F is localized in lysosomes, rather than being present in secretory granules. In a paper describing changes in general gene expression in mature activated dendritic cells compared to immature dendritic cells, the cystatin F gene was among the top 50 of those up-regulated upon maturation of the cells [28]. This supports our conclusion that cystatin F is readily regulated and present in cells of relevance for antigen presentation and, thus, may be a candidate for control of the cysteine peptidase activity essential for this process. ACKNOWLEDGEMENTS We wish to thank Drs Klaudia Brix (Bonn) and Arne Egesten (Malmo ¨ ) for helpful technical discussions. This study was supported by grants from the Swedish Medical Research Council (no. 09915), the Medical faculty at Lund University, the A. O ¨ sterlund Foundation and the Crafoord Foundation. REFERENCES 1. Rawlings, N.D. & Barrett, A.J. (1990) Evolution of proteins of the cystatin superfamily. J. Mol. Evol. 30, 60–71. 2. Abrahamson, M. (1994) Cystatins. Methods Enzymol. 244, 685– 700. 3. Ni, J., Abrahamson, M., Zhang, M., Fernandez, M.A., Grubb, A. &Su,J.,YuG.L.,Li,Y.,Parmelee,D.,Xing,L.,Coleman,T.A., Gentz, S., Thotakura, R., Nguyen, N., Hesselberg, M. & Gentz, R. (1997) Cystatin E is a novel human cysteine proteinase inhibitor with structural resemblance to family 2 cystatins. J. Biol. Chem. 272, 10853–10858. 4. Sotiropoulou, G., Anisowicz, A. & Sager, R. (1997) Identification, cloning, and characterization of cystatin M, a novel cysteine proteinase inhibitor, down-regulated in breast cancer. J. Biol. Chem. 272, 903–910. 5. Ni, J., Fernandez, M.A., Danielsson, L., Chillakuru, R.A., Zhang, J., Grubb, A., Su, J., Gentz, R. & Abrahamson, M. (1998) Cystatin F is a glycosylated human low molecular weight cysteine proteinase inhibitor. J. Biol. Chem. 273, 24797–24804. 6. Halfon,S.,Ford,J.,Foster,J.,Dowling,L.,Lucian,L.,Sterling, M., Xu, Y., Weiss, M., Ikeda, M., Liggett, D., Helms, A., Caux, 5510 C M. Nathanson et al. (Eur. J. Biochem. 269) Ó FEBS 2002 C., Lebecque, S., Hannum, C., Menon, S., McClanahan, T., Gorman, D. & Zurawski, G. (1998) Leukocystatin, a new Class II cystatin expressed selectively by hematopoietic cells. J. Biol. Chem. 273, 16400–16408. 7. Morita, M., Yoshiuchi, N., Arakawa, H. & Nishimura, S. (1999) CMAP: a novel cystatin-like gene involved in liver metastasis. Cancer Res. 59, 151–158. 8. Nakagawa, T., Roth, W., Wong, P., Nelson, A., Farr, A., Deus- sing, J., Villadangos, J.A., Ploegh, H., Peters, C. & Rudensky, A.Y. (1998) Cathepsin L: critical role in II degradation and CD4 T cell selection in the thymus. Science 280, 450–453. 9. Pierre, P. & Mellman, I. (1998) Developmental regulation of invariant chain proteolysis controls MHC class II trafficking in mouse dendritic cells. Cell 93, 1135–1145. 10. Alvarez-Fernandez, M., Barrett, A.J., Gerhartz, B., Dando, P.M., Ni, J. & Abrahamson, M. (1999) Inhibition of mammalian legu- main by some cystatins is due to a novel second reactive site. J. Biol. Chem. 274, 19195–19203. 11. Manoury, B., Hewitt, E.W., Morrice, N., Dando, P.M., Barrett, A.J. & Watts, C. (1998) An asparaginyl endopeptidase processes a microbial antigen for class II MHC presentation. Nature 396, 695– 699. 12. Abrahamson, M., Barrett, A.J., Salvesen, G. & Grubb, A. (1986) Isolation of six cysteine proteinase inhibitors from human urine. Their physicochemical and enzyme kinetic properties and con- centrations in biological fluids. J. Biol. Chem. 261, 11282–11289. 13. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 14. Olafsson, I., Lofberg, H., Abrahamson, M. & Grubb, A. (1988) Production, characterization and use of monoclonal antibodies against the major extracellular human cysteine proteinase inhibitors cystatin C and kininogen. Scand. J. Clin. Laboratory Invest 48, 573–582. 15. Bjarnadottir, M., Grubb, A. & Olafsson, I. (1995) Promoter- mediated, dexamethasone-induced increase in cystatin C produc- tion by HeLa cells. Scand. J. Clin. Laboratory Invest. 55, 617–623. 16. Chomczynski, P. & Sacchi, N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156–159. 17. Abrahamson, M., Grubb, A., Olafsson, I. & Lundwall, A. (1987) Molecular cloning and sequence analysis of cDNA coding for the precursor of the human cysteine proteinase inhibitor cystatin C. FEBS Lett. 216, 229–233. 18. Hultberg, B., Lindsten, J. & Sjoblad, S. (1976) Molecular forms and activities of glycosidases in cultures of amniotic- fluid cells. Biochem. J. 155, 599–605. 19. Abrahamson, M., Dalboge, H., Olafsson, I., Carlsen, S. & Grubb, A. (1988) Efficient production of native, biologically active human cystatin C by Escherichia coli. FEBS Lett. 236, 14–18. 20. Bjarnadottir, M., Wulff, B.S., Sameni, M., Sloane, B.F., Keppler, D., Grubb, A. & Abrahamson, M. (1998) Intracellular accumu- lation of the amyloidogenic L68Q variant of human cystatin C in NIH/3T3 cells. Mol. Pathol. 51, 317–326. 21. Morita, M., Hara, Y., Tamai, Y., Arakawa, H. & Nishimura, S. (2000) Genomic construct and mapping of the gene for CMAP (leukocystatin/cystatin F, CST7) and identification of a proximal novel gene, BSCv (C20orf3). Genomics 67, 87–91. 22. Bird, A.P. (1986) CpG-rich islands and the function of DNA methylation. Nature 321, 209–213. 23. Abrahamson, M., Olafsson, I., Palsdottir, A., Ulvsback, M., Lundwall, A., Jensson, O. & Grubb, A. (1990) Structure and expression of the human cystatin C gene. Biochem. J. 268, 287–294. 24. Radomska, H.S., Huettner, C.S., Zhang, P., Cheng, T., Scadden, D.T. & Tenen, D.G. (1998) CCAAT/enhancer binding protein alpha is a regulatory switch sufficient for induction of granulocytic development from bipotential myeloid progenitors. Mol. Cell. Biol. 18, 4301–4314. 25. Le Borgne, R. & Hoflack, B. (1998) Protein transport from the secretory to the endocytic pathway in mammalian cells. Biochim. Biophys. Acta 1404, 195–209. 26. Senior, R.M. & Campbell, E.J. (1984) Cathepsin G in human mononuclear phagocytes: comparisons between monocytes and U937 monocyte-like cells. J. Immunol. 132, 2547–2551. 27. Journet, A., Chapel, A., Kieffer, S., Louwagie, M., Luche, S. & Garin, J. (2000) Towards a human repertoire of monocytic lyso- somal proteins. Electrophoresis 21, 3411–3419. 28. Hashimoto, S.I., Suzuki, T., Nagai, S., Yamashita, T., Toyoda, N., & Matsushima, K. (2000) Identification of genes specifically expressed in human activated and mature dendritic cells through serial analysis of gene expression. Blood 96, 2206–2214. Ó FEBS 2002 Cystatin F in U937 cells (Eur. J. Biochem. 269) 5511 . of cystatin F in U937 cells. The high portion of intracellular cystatin F led us to fractionate cells to determine the localization of the inhibitor. Initially. function of cystatin F is hypothetical. Although most cystatins inhi- biting cathepsin L also inhibit cathepsin S, an inhibitory activity of cystatin F