Báo cáo khoa học: Murine serum nucleases – contrasting effects of plasmin and heparin on the activities of DNase1 and DNase1-like 3 (DNase1l3)

Murine serum nucleases – contrasting effects of plasmin and heparin on the activities of DNase1 and DNase1-like 3 (DNase1l3) Markus Napirei, Sebastian Ludwig, Jamal Mezrhab, Thomas Klo¨ ckl and Hans G. Mannherz

Abteilung fu¨ r Anatomie und Embryologie, Medizinische Fakulta¨ t, Ruhr-Universita¨ t Bochum, Germany

Keywords DNase1; DNase1l3; plasminogen system; serum; systemic lupus erythematosus

Correspondence M. Napirei, Abteilung fu¨ r Anatomie und Embryologie, Medizinische Fakulta¨ t, Ruhr-Universita¨ t Bochum, Universita¨ tsstraße 150, D-44801 Bochum, Germany Fax: +49 2343214474 Tel: +49 2343223164 E-mail: markus.napirei@rub.de

(Received 8 November 2008, revised 27 November 2008, accepted 10 December 2008)

doi:10.1111/j.1742-4658.2008.06849.x

DNase1 is regarded as the major serum nuclease; however, a systematic investigation into the presence of additional serum nuclease activities is lacking. We have demonstrated directly that serum contains DNase1-like 3 (DNase1l3) in addition to DNase1 by an improved denaturing SDS-PAGE zymography method and anti-murine DNase1l3 immunoblotting. Using DNA degradation assays, we compared the activities of recombinant mur- ine DNase1 and DNase1l3 (rmDNase1, rmDNase1l3) with the serum of wild-type and DNase1 knockout mice. Serum and rmDNase1 degrade chro- matin effectively only in cooperation with serine proteases, such as plasmin or thrombin, which remove DNA-bound proteins. This can be mimicked by the addition of heparin, which displaces histones from chromatin. In contrast, serum and rmDNase1l3 degrade chromatin without proteolytic help and are directly inhibited by heparin and proteolysis by plasmin. In previous studies, serum DNase1l3 escaped detection because of its sensitiv- ity to proteolysis by plasmin after activation of the plasminogen system in the DNA degradation assays. In contrast, DNase1 is resistant to plasmin, probably as a result of its di-N-glycosylation, which is lacking in DNase1l3. Our data demonstrate that secreted rmDNase1 and murine parotid DNase1 are mixtures of three different di-N-glycosylated molecules containing two high-mannose, two complex N-glycans or one high-mannose and one com- plex N-glycan moiety. In summary, serum contains two nucleases, DNase1 and DNase1l3, which may substitute or cooperate with each other during DNA degradation, providing effective clearance after exposure or release from dying cells.

comparing [1,2]. By tracts

it

Abbreviations ANA, antinuclear autoantibodies; DNase1l3, DNase 1-like 3; DPZ, denaturing SDS-PAGE zymography; EndoH, endoglycosidase H; KO, knockout; NLS, nuclear localization signal; NPZ, native SDS-PAGE zymography; Pai-1, plasminogen activator inhibitor 1; pDNA, plasmid DNA; PNGaseF, peptide N-glycosidase F; rER, rough endoplasmic reticulum; rmDNase1 ⁄ rmDNase1l3, recombinant murine DNase1 ⁄ DNase1l3; rrDNase1l3, recombinant rat DNase1-like 3; SLE, systemic lupus erythematosus; SRED assay, single radial enzyme diffusion assay; TAE, Tris–acetate ⁄ EDTA; TBE, Tris–borate ⁄ EDTA.

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1059

DNase1 (EC 3.1.21.1) is an endonuclease secreted into body fluids by a wide variety of exocrine and endo- crine organs which line the gastrointestinal and uro- serum from genital wild-type (WT) and DNase1 knockout (KO) mice, we have demonstrated previously that is the major serum nuclease [3]. A lack or decrease in serum DNase1 activity is associated with the development of systemic lupus erythematosus (SLE) like antinuclear autoantibodies (ANAs) directed against nucleosomes and their constituents, and immune complex-induced glomerulonephritis in humans and mice [4–6]. Previ- ously, we have reported that, in cooperation with different serine proteases, serum DNase1 degrades the chromatin of necrotic cells [3]. Pure DNase1 hydro- lyses ‘naked’ protein-free DNA with high efficiency,

M. Napirei et al.

Murine serum nucleases

the extracellular functions might be the participation in the clearance of autoantigenic chromatin [21].

but efficient chromatin degradation depends on the proteolysis of DNA-bound proteins [3,7]. Heparin pro- motes chromatin degradation by serum DNase1; how- ever, the underlying mechanism for this activation is still unclear [3,7].

In this work, we demonstrate that murine serum contains two chromatolytic activities with different properties. Serum DNase1l3 degrades chromatin at internucleosomal sites on its own and is inhibited by proteolysis by plasmin. In contrast, serum DNase1 degrades chromatin only in combination with prote- such as plasmin. The plasmin resistance of ases DNase1 might be explained by its di-N-glycosylation, which is absent in DNase1l3. Heparin mimics the effect of proteases on DNase1-induced chromatolysis by displacing histones, whereas it inhibits DNase1l3 by binding. We also describe an improvement of the denaturing SDS-PAGE zymography (DPZ) procedure originally described by Shiokawa et al. [14], which allows the simultaneous detection of both nucleases in serum and tissue samples. This test procedure might also be of clinical value, as reduced serum nuclease activity has been reported in patients with SLE and in lupus-prone mice [21].

Results

Murine serum contains two chromatolytic activities with different properties

study, Freshly prepared serum was collected from C57BL ⁄ 6 WT and DNase1 KO mice and employed in nuclear chromatin digestion assays. We found that all sera derived from DNase1 KO mice contained residual nuclease activity (Fig. 1A). In contrast with our previ- ous studies, we found that chromatin breakdown by the serum of WT mice was not inhibited by the addi- tion of aprotinin [3]. Instead, we found that aprotinin accelerated and equalized the overall nucleolytic activi- ties of sera from both mouse strains, leading to an accumulation of mononucleosomal DNA fragments (Fig. 1A). These data imply that the residual serum nuclease activity found in DNase1 KO mice also occurred in WT mice, and was activated by the addi- tion of aprotinin, thereby masking the inhibitory effect of aprotinin on DNase1 ⁄ plasmin-induced chromatoly- sis as described previously [3]. These results contradict our previous studies, which demonstrated that chroma- tin degradation by the sera of WT mice was com- pletely inhibited by aprotinin [3], and imply that, in the earlier the second nuclease activity of murine serum was not always detectable.

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1060

In accordance with our previous studies, we found that heparin accelerated chromatin degradation by the sera of WT mice [3], whereas it inhibited that catalysed serum nuclease activity found in by the residual Previously, we have observed that the serum of some DNase1 KO mice contains residual nucleolytic activity [7]. In contrast with serum DNase1, this nucleolytic activity efficiently degrades chromatin by internucleos- omal cleavage without proteolytic help, and is inhib- ited by heparin. However, the conditions of occurrence and the identity of this additional serum nuclease have not been clarified to date, although preliminary data suggest that it displays biochemical characteristics of recombinant rat DNase1-like 3 (rrDNase1l3; DNase c, DNase Y, LS-DNase, nhDNase) [7,8]. DNase1l3 belongs to the DNase1 nuclease family, which consists of DNase1 and three further DNase1-like endonucleas- es (DNase1L1, DNase1L2 and DNase1l3) [8]. Both DNase1 and DNase1l3 contain an N-terminal signal peptide for their translocation into the rough endoplas- mic reticulum (rER). Indeed, they have been shown to be localized in the secretory compartment and secreted into the cell culture medium by transfected cells [7]. In contrast with DNase1, DNase1l3 contains two nuclear localization signals (NLSs), which might explain its occurrence in the nucleus of certain cells [9]. This find- ing seems to be important for the proposed role of DNase1l3 in chromatin cleavage during apoptosis, as described for several cell types in vitro [10–15] and in vivo [16,17]. Traditionally, the presence of an NLS implies nuclear accumulation by active transport through the nuclear pores after binding of a specific importin to the NLS. However, experiments employing murine and rat DNase1l3-green fluorescent protein constructs did not show any preferential nuclear locali- zation of the fusion proteins after transfection of NIH- 3T3 cells [7]. Instead, we observed secretion of these nucleases into the medium, which was abrogated after deletion of the N-terminal rER signal peptide. It is therefore conceivable that the NLS of DNase1l3 might only be functional under special conditions, such as, for example, apoptosis, leading to the nuclear import of DNase1l3 after its release from the rER into the cytoplasm. Macrophages of different organs have been shown to express DNase1l3 in vivo [18]. Furthermore, DNase1l3 has been isolated from nuclei of rat thymo- cytes [19] and has been demonstrated to be involved in somatic hypermutation in stimulated B cells [20]. These studies imply that DNase1l3 fulfils intra- and extracel- lular physiological functions in the immune system; however, the role of its presumed NLS in fulfilling the intracellular functions proposed is still unclear. One of

M. Napirei et al.

Murine serum nucleases

A

B

by activation of DNase1 ⁄ plasmin-dependent chromatin breakdown. Employing aprotinin and heparin in par- allel, we found that, in the sera of DNase1 KO mice, the inhibitory effect of heparin blocked the accelerat- ing effect of aprotinin on the residual nuclease activity (Fig. 1A). As heparin inhibits rrDNase1l3, as shown previously [3], we concluded that the residual serum nuclease activity was caused by the presence of a DNase1l3-like nuclease. In WT serum, the accelerating effect of heparin on DNase1 ⁄ plasmin-dependent chro- matin degradation (as described previously [3]) over- rides its inhibitory effect on DNase1l3-like activity and the inhibitory effect of aprotinin on DNase1 ⁄ plasmin- dependent chromatolysis (see Fig. 3).

with opposite activation

C

In summary, these and our previous experiments indicate that murine serum contains two chromatolytic properties: activities DNase1 ⁄ plasmin activity, which is activated by hepa- rin and inhibited by aprotinin as a result of plasmin inhibition, and DNase1l3-like activity, which is inhib- ited by heparin and activated by aprotinin. As aproti- nin is a serine protease inhibitor, it is conceivable that the DNase1l3-like nuclease might be sensitive to prote- olysis or indirectly inhibited by proteolysis of DNA- bound structural proteins.

To test whether the chromatolytic activities were also active in undiluted serum, we added cell nuclei directly into pure serum. The data obtained demon- strated complete chromatin degradation by WT serum in an internucleosomal manner, which was less efficient and did not proceed to completion in serum from DNase1 KO mice (Fig. 1B). However, the addition of aprotinin or plasminogen activator inhibitor 1 (Pai-1) to the sera of DNase1 KO mice completed chromatoly- sis to mononucleosomes and even to oligonucleotides (Fig. 1C). These experiments demonstrate that the DNase1l3-like nuclease of murine serum is sensitive to proteolysis by plasmin or inhibited by proteolysis of DNA-bound structural proteins.

Fig. 1. Murine serum contains two chromatolytic activities with dif- ferent properties. Digestion of nuclear chromatin by serum from WT and DNase1 KO mice. (A) Isolated MCF-7 nuclei were incu- bated with 2.5% (v ⁄ v) serum concentrations for 8 h at 37 (cid:2)C. Apro- tinin equalized the internucleosomal chromatin degradation by sera from both mouse genotypes, whereas heparin inhibited that by serum from DNase1 KO mice, but enhanced that by WT serum. (B) Pure serum with 2–8 h of incubation at 37 (cid:2)C under otherwise identical conditions. Chromatin degradation in the serum of a WT mouse proceeded to completion with ongoing incubation time, whereas it stopped in serum from a DNase1 KO mouse. (C) Pure serum with 2 h of incubation at 37 (cid:2)C. Chromatin degradation in the serum of a DNase1 KO mouse was accelerated by the addition of aprotinin and the specific inhibitor for the activation of the plasminogen system Pai-1.

DNA digestion by murine serum nucleases in comparison with recombinant murine DNase1 (rmDNase1) and rmDNase1l3

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1061

DNase1 KO mice (Fig. 1A). As it is assumed that, in addition to DNase1, the residual serum nuclease activ- ity detectable in DNase1 KO mice also occurs in WT mice, the acceleration of chromatin degradation by WT serum in the presence of heparin must be caused To clarify the effect of heparin and aprotinin on the mode of chromatolysis by serum from WT and DNase1 KO mice in more detail, we investigated their influence on rmDNase1 and rmDNase1l3 in plasmid DNA (pDNA) and chromatin digestion assays. For this purpose, we transiently transfected NIH-3T3 cells with expression vectors for the murine DNase1 and DNase1l3 cDNA, and collected cell culture superna- tants containing the secreted recombinant nucleases.

M. Napirei et al.

Murine serum nucleases

A

rmDNase1 directly, but

B

C

D

Fig. 2. DNA digestion facilitated by rmDNase1 and rmDNase1l3. (A) Effect of increasing amounts of heparin on pDNA digestion by rmDNase1 and rmDNase1l3, employing 0.1 and 2 lL of cell culture supernatants, respectively. Incubation for 30 min at 37 (cid:2)C in 10 mM Tris ⁄ HCl pH 7.0, 2 mM MnCl2 and 2 mM CaCl2. (B–D) Chromatin digestion of isolated MCF-7 nuclei by rmDNase1 and rmDNase1l3: 5 lL of cell culture supernatants were employed for 2 h at 37 (cid:2)C. (B) In contrast with pDNA, nuclear chromatin digestion by rmDN- ase1 was enhanced by heparin, leading to a random DNA cleavage pattern. In accordance with pDNA, chromatin digestion by rmDN- ase1l3 was inhibited by heparin. Aprotinin had no effect on chroma- tin digestion by the two recombinant nucleases. (C) Chromatin digestion by rmDNase1 and rmDNase1l3 in the presence of plas- min or thrombin. Plasmin and thrombin induced internucleosomal chromatin degradation by rmDNase1, whereas rmDNase1 per- formed it alone. Plasmin, but not thrombin, inhibited chromatin deg- radation by rmDNase1l3. Conditions as in (B). (D) Pre-incubation of rmDNase1l3, but not rmDNase1, with plasmin for 30 min at 37 (cid:2)C prior to the addition of MCF-7 cell nuclei inhibited chromatin cleavage. The addition of aprotinin after pre-incubation did not restore chromatin cleavage, demonstrating that the inhibition is caused by proteolysis of rmDNase1l3 by plasmin during the pre-incubation period.

First, we evaluated the effect of heparin in DNA digestion assays. As shown in Fig. 2A, heparin had no stimulating effect on pDNA degradation by rmDN- ase1, but inhibited that by rmDNase1l3 at low concen- trations. These results imply that heparin had no effect on protein-free DNA and did not stimulate the activity of inhibited rmDNase1l3. Employing both recombinant nucleases in chromatin in contrast with digestion assays, we found that, pDNA digestion, chromatin breakdown by pure rmDNase1 was weak in comparison with that by pure rmDNase1l3, which efficiently degraded chromatin in an internucleosomal manner (Fig. 2B). In contrast with pDNA digestion, heparin activated chromatin break- down by rmDNase1, leading to a random DNA cleav- age pattern (DNA smear in the agarose gel; Fig. 2B), as described previously for serum from WT mice [3]. In accordance with the pDNA digestion assay, internu- cleosomal chromatin breakdown by rmDNase1l3 was inhibited by heparin (Fig. 2B). In summary, these data demonstrate that heparin has opposing effects on these nucleases: it enhances chromatin but not pDNA cleav- age by rmDNase1, possibly by inducing an alteration in the chromatin structure itself, and inhibits chroma- tin and pDNA cleavage by rmDNase1l3. This inhibi- tion might be caused by direct binding of heparin to DNase1l3 and ⁄ or an alteration of the chromatin struc- ture (see below).

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1062

Employing aprotinin in pDNA (data not shown) and chromatin digestion (Fig. 2B) assays using both recombinant nucleases, we did not observe any effect on their nucleolytic activities. This result suggests that the apparently stimulating effect of aprotinin on serum DNase1l3-like activity (see Fig. 1A,C) and its inhibit- ing effect on serum DNase1 (as described previously [3] and Fig. 3) are facilitated by the inhibition of serum proteases. Therefore, we repeated the chromatin digestion assays employing both recombinant nucleases in the presence of thrombin or plasmin either (Fig. 2C). As described previously [3,7], we found that thrombin as well as plasmin induced chromatin break- down by pure rmDNase1, leading to internucleosomal chromatin cleavage comparable with that induced by pure rmDNase1l3 alone (Fig. 2C). Plasmin was found to be much more efficient than thrombin (Fig. 2C). The simultaneous addition of aprotinin inhibited the promoting effect of plasmin on chromatin breakdown by rmDNase1, whereas the action of thrombin was only slightly inhibited (Fig. 2C), which is most proba- bly explained by the fact that aprotinin inhibits plasmin with higher specificity than thrombin. These these proteases render results strongly suggest that for nucleolytic regions accessible internucleosomal

M. Napirei et al.

Murine serum nucleases

is degraded by plasmin during the pre-incubation per- iod (Fig. 2D).

Activation of plasminogen depletes the DNase1l3-like activity of murine serum

rmDNase1l3

Our finding that the serine protease inhibitor aprotinin and the inhibitor for the activation of the plasminogen system Pai-1 maintained the chromatolytic activity of diluted and undiluted serum of DNase1 KO mice (Fig. 1A,C) implies that DNase1l3-like nuclease activ- ity is sensitive to proteolysis by plasmin. This is sup- is ported by the observation that inactivated by the addition of plasmin (Fig. 2C,D). Therefore, the inability of serum from DNase1 KO mice to cleave chromatin after prolonged incubation indicates that DNase1l3 is inactivated by (Fig. 1B) plasminogen activation during the nuclear chromatin degradation assay. These data explain why,

Fig. 3. Activation of plasminogen depletes the DNase1l3-like activ- ity of murine serum. Chromatin digestion by serum from WT and DNase1 KO mice [2.5% (v ⁄ v) serum concentration, 8 h of incuba- tion at 37 (cid:2)C]. Top panel: serum was stored at )20 (cid:2)C, thawed to room temperature and analysed directly. Bottom panel: identical sera analysed after 2 weeks of storage at 4 (cid:2)C. The addition of aprotinin and Pai-1 demonstrated the presence of a protease-sensi- tive DNase1l3-like nuclease activity in the serum from the DNase1 KO mouse, which disappeared after thawing and prolonged storage of the serum. The DNase1 ⁄ plasmin-dependent chromatolytic activ- ity, which is inhibited by aprotinin and Pai1, remained in the serum from the WT mouse.

still

in previous experiments, the sera of WT and DNase1 KO mice were depleted in DNase1l3-like nuclease [3]. Indeed, when we sub- jected serum frozen at )20 (cid:2)C to thawing to room temperature and subsequently stored it at 4 (cid:2)C, it lost its DNase1l3-like activity within 2 weeks (Fig. 3). Thus, serum from DNase1 KO mice completely lost its ability to induce chromatolysis, whereas serum from WT mice contained DNase1 ⁄ plasmin- dependent chromatolytic activity, which was inhibited by aprotinin and Pai-1 as described previously [3]. From these data, we conclude that the storage con- ditions are crucial for the maintenance of the serum DNase1l3-like nuclease, whereas DNase1 is much more stable.

Heparin displaces core histones from chromatin and alters nuclear structure

inhibit

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1063

attack by rmDNase1, most probably by proteolysis of histone H1 as shown previously [3]. In contrast, inter- nucleosomal chromatin cleavage by rmDNase1l3 was inhibited by plasmin, but not by thrombin (Fig. 2C). Furthermore, the simultaneous addition of aprotinin restored internucleosomal chromatin cleavage by rmDNase1l3 in the presence of plasmin (Fig. 2C). These data reveal that the proteolysis of histones is apparently not necessary and that their intact nature does not internucleosomal chromatin break- down by rmDNase1l3. These results also indicate that plasmin, but not thrombin, proteolytically attacks and inactivates rmDNase1l3 but not rmDNase1. To dem- onstrate this conclusion more directly, we pre-incu- bated both recombinant nucleases with plasmin for 30 min at 37 (cid:2)C, and subsequently added cell nuclei alone or in combination with aprotinin. We found that pre-incubation of rmDNase1 with plasmin had no effect on its ability to cause internucleosomal chroma- tin cleavage, demonstrating that rmDNase1 is not degraded by plasmin (Fig. 2D). In contrast, pre-incu- bation of rmDNase1l3 with plasmin inhibited subse- quent chromatolysis, demonstrating that rmDNase1l3 In a previous study, we showed that the activation of the plasminogen system leads to proteolysis of histone H1 of necrotic cells when incubated in the presence of murine serum [3]. Proteolysis of histone H1 renders to nucleolytic regions accessible internucleosomal attack by serum DNase1, leading to internucleosomal chromatin breakdown. In addition, we found that hep- arin-promoted chromatin degradation by WT serum was accompanied by a switch in the cleavage pattern from internucleosomal to random. Our experiments using pDNA showed that heparin had no direct effect on rmDNase1. The random cleavage pattern of nuclear chromatin suggests that, in addition to H1, the nucleosomal core histones (histones H2A ⁄ H2B ⁄ H3 and H4) are displaced from chromatin.

M. Napirei et al.

Murine serum nucleases

Fig. 4. Heparin displaces core histones from chromatin and alters nuclear structure. Western blot analysis of assay supernatants con- taining MCF-7 cell nuclei and increasing amounts of heparin using an anti-histone H3 serum that cross-reacted with further core (murine histone H3, 15.4 kDa; histone H2A ⁄ H2B, histones (cid:2) 14 kDa; histone H4, (cid:2) 11.4 kDa).

assay does not allow the identification of the residual nuclease activity by, for example, the estimation of the molecular mass of the nuclease.

the

Therefore, we attempted to establish a DPZ proce- dure for the identification of both serum nucleases employing cell culture supernatants of cells transiently expressing mDNase1 and mDNase1l3. We employed the DPZ procedure of Shiokawa et al. [14], and found that the detection of both nucleases in cell culture supernatants became possible, whereas the method usually performed in our laboratory only allowed the efficient detection of DNase1. The main differences between the two methods, which led to the detection of DNase1l3, are as follows: (a) strict maintenance of the reducing conditions by the presence of 2-mercapto- ethanol during electrophoresis and all further incuba- (washing out SDS from gels, nuclease tion steps refolding and reaction within gels); (b) removal of SDS by heat and not by dissolved milk powder; (c) nuclease refolding and reaction in the absence of milk powder (for details, see Materials and methods). Experiments to optimize the DPZ procedure demonstrated that, in the presence of MnCl2 ⁄ CaCl2 instead of MgCl2 ⁄ CaCl2, detection of rmDNase1l3 was preferentially enhanced (see later). This finding was analysed in more detail using pDNA digestion assays (Fig. 5). Indeed, we found that the pH optimum and nucleolytic activity of both nucleases varied in the presence of either Mg2+ or Mn2+ ions. Thus, the pH optimum of rmDNase1

To address this question in more detail, we incu- bated cell nuclei in the presence of increasing amounts of heparin, and subsequently analysed the supernatants by immunoblotting for the presence of core histones, which might have diffused out of the nuclei. As increasing amounts of heparin led to an expected, enhanced dissociation of nucleosomal core histones from chromatin (Fig. 4). These results support the assumption that enhanced chromatolysis by rmDNase1 and serum DNase1 in the presence of hepa- rin is induced by a transition of protein-complexed (chromatin) to protein-free DNA. Whether this transi- tion is also the cause of the inhibition of chromatolysis serum DNase1l3-like nuclease by rmDNase1l3 or remains speculative. As the hydrolysis of protein-free pDNA by rmDNase1l3 is also inhibited by heparin, it is conceivable that heparin, at least, inhibits DNase1l3 directly, for example by binding to the nuclease (see below).

Fig. 5. Influence of Mn2+ and Mg2+ ions on the activity of rmDN- ase1 and rmDNase1l3. pDNA digestion employing cell culture supernatants containing rmDNase1 (0.1 lL supernatant, 10 min of incubation at 37 (cid:2)C) or rmDNase1l3 (1 lL supernatant, 30 min of incubation at 37 (cid:2)C). Influence of the pH value and ion composition: Assays were performed in 10 mM buffers with different pH values (acetate ⁄ NaOH, Mes ⁄ NaOH or Tris ⁄ HCl) in the presence of either 2 mM MgCl2 ⁄ 2 mM CaCl2 (top panel) or 2 mM MnCl2 ⁄ 2 mM CaCl2 (bottom panel).

Establishing DPZ for the detection of rmDNase1 and rmDNase1l3

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1064

In previous experiments, we were unable to detect nuc- leases other than DNase1 in murine serum by native SDS-PAGE zymography (NPZ) and DPZ or the single radial enzyme diffusion (SRED) assay [7]. Failure of detection of mDNase1l3, in contrast with mDNase1, by NPZ (performed at pH 8.6) is most probably explained by its strong basic pI of 8.7, in contrast with the acidic pI of 4.9 of mDNase1. For the SRED assay, we found that the failure of detection of DNase1l3-like nuclease activity in murine serum was most probably caused by its sensitivity to proteolysis. Thus, freshly prepared sera of DNase1 KO mice loaded onto SRED gels displayed residual nuclease activity, which was inhibited by heparin (data not shown). However, this

M. Napirei et al.

Murine serum nucleases

to the

was in the range pH 6.5–7.5 in the presence of Mg2+, and shifted by one pH unit in the presence of Mn2+ (pH 7.5–8.5). Similarly, the pH optimum of rmDN- ase1l3 in the presence of Mg2+ was shifted from pH 4.5–5.5 to pH 5.5–6.5 by Mn2+. Although the activity of rmDNase1 in the presence of Mn2+ was increased only slightly, rmDNase1l3 displayed strongly enhanced nucleolysis. Furthermore, we found by pDNA digestion assays that increasing concentrations of Tris (approximately half activity in the presence of 80 mm Tris) and NaCl (approximately half activity in the presence of 50 mm NaCl) had a greater inhibitory influence on rmDNase1l3 than on rmDNase1 (no inhibitory influence of Tris and approximately half activity in the presence of 150 mm NaCl) (data not shown).

Detection of DNase1 and DNase1l3 in murine serum and tissues by DPZ

To clarify that the DNase1l3-like nuclease in murine serum is indeed DNase1l3, we investigated, by the improved DPZ procedure (reducing conditions), serum samples and tissue extracts of kidney (high DNase1 content [2]) and spleen (high DNase1l3 content [8]) from WT and DNase1 KO mice (Fig. 6A). We used TET and RIPA as extraction buffers (see Materials and methods), and found that nuclease detection was more efficient using RIPA buffer (Fig. 6A). Detection of DNase1 in kidney samples was verified by its absence in samples of DNase1 KO mice. Furthermore, in spleen and kidney we detected a nuclease signal samples of both mice of approximately 34 kDa, which corresponds estimated molecular mass of 33.1 kDa for mature mDNase1l3 (without the N-termi- nal signal peptide of 25 amino acids in length). Indeed, the expression of DNase c (DNase1l3) in human spleen and kidney has been verified previously by RNA dot blot analysis [8], and by RNA in situ hybrid- ization for LS-DNase (DNase1l3) in Rhesus monkey macrophages of the spleen marginal zones, red pulp and the mesangium of the kidney [18]. Previously, expression of LS-DNase has also been shown for hepatic Kupffer cells [8]. By analysing spleen and liver tissue extracts from WT and DNase1 KO mice, we found that the 34 kDa nuclease detectable in spleen

C

A

B

D

Fig. 6. Detection of DNase1 and DNase1l3 in murine serum and tissues by DPZ. (A–C) Modified DPZ under reducing conditions. (D) Conventional DPZ under non-reducing condi- tions (see Materials and methods). (A) Anal- ysis of spleen and kidney tissue extracts from WT and DNase1 KO mice prepared in either TET or RIPA buffer. In spleen and kid- ney of both mice, a (cid:2) 34 kDa nuclease was detected. DNase1 was only detectable in the kidney extract of the WT mouse and displayed a molecular mass of (cid:2) 37 kDa. (B) The 34 kDa nuclease most probably repre- sents DNase1l3, as it was also detectable in the liver of both mice, co-migrated with rmDNase1l3, displayed a higher activity in the presence of Mn2+ instead of Mg2+, and was inhibited by heparin. (C, D) Murine serum possesses two nucleases, DNase1l3 and DNase1, which co-migrate with rmDN- ase1 and rmDNase1l3, respectively. Human serum also contains DNase1l3; however, hDNase1 is only detectable by DPZ under non-reducing conditions. Again, mDNase1l3 and hDNase1l3, by contrast with mDNase1, are inhibited by heparin.

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1065

M. Napirei et al.

Murine serum nucleases

A

extracts was also present in the liver (Fig. 6B). Fur- thermore, this nuclease co-migrated with rmDNase1l3, displayed an enhanced activity in the presence of Mn2+ in comparison with Mg2+ ions, and was inhib- ited by the addition of heparin (Fig. 6B). From these data, we conclude that the detected nuclease must be DNase1l3.

B

Interestingly,

Employing serum from WT and DNase1 KO mice, we demonstrated that murine serum indeed contains both DNase1, as deduced from its absence in the serum of DNase1 KO mice, and DNase1l3 (Fig. 6C). Again, serum DNase1l3 co-migrated with rmDNase1l3 and, in contrast with DNase1, was inhibited by hepa- rin (Fig. 6C). In addition, we found that human serum also contains DNase1l3 (Fig. 6C,D), which was also inhibited by heparin (Fig. 6C). However, in contrast with mDNase1, detection of hDNase1 by DPZ was only possible under non-reducing conditions, employ- ing the method usually performed in our laboratory (Fig. 6C,D). rmDNase1 and DNase1 present in murine kidney extracts and serum displayed a higher molecular mass of (cid:2) 37 kDa in DPZ, in com- parison with the calculated molecular mass of 29.8 kDa for mature mDNase1 (Fig. 6A–D).

Fig. 7. Immunodetection of DNase1l3 after purification from serum by heparin-Sepharose. (A) DPZ under reducing conditions. Murine DNase1l3 was purified from 1 mL of serum collected from DNase1 KO mice by heparin-Sepharose affinity chromatography. Serum samples (2 lL) taken pre- and post-chromatography reveal the effi- cient binding of DNase1l3 to heparin. Binding remained stable dur- ing two washing steps with 0.2 M NaCl (fractions I and II). Elution (fractions III–VII, 10-fold enrichment in comparison with the original serum) with increasing amounts of NaCl revealed a strong affinity of DNase1l3 to heparin, which could only be effectively dissolved by the addition of 1 M NaCl. (B) DNase1l3 of 0.5 mL of serum col- lected from WT mice was purified by heparin-Sepharose affinity chromatography, and the two halves were employed in DPZ under reducing conditions (top panel) and in immunoblotting (bottom panel) against mDNase1l3, using cell extracts of NIH-3T3 fibro- blasts transiently transfected with mDNase1 or mDNase1l3 as a control.

Immunodetection of DNase1l3 after its purification from serum by heparin-Sepharose

the corresponding NIH-3T3 cell extract and that mDNase1l3 purified from WT serum with high speci- ficity in comparison with mDNase1, which was only detected by DPZ (Fig. 7B).

Murine DNase1 is di-N-glycosylated, whereas murine DNase1l3 is not N-glycosylated

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1066

DPZ demonstrated a higher molecular mass of rmDN- ase1 and DNase1 present in murine serum and kidney in comparison with rmDNase1l3 and DNase1l3 detected in murine serum, spleen, kidney and liver samples (Fig. 6). Murine DNase1l3 migrated at an expected molecular mass of (cid:2) 34 kDa in DPZ, which is consistent with the calculated molecular mass of 33.1 kDa for the mature mDNase1l3, i.e. without its In order to provide further proof that the additional serum nuclease detected by DPZ is indeed DNase1l3, and to evaluate whether its inhibition by heparin is caused by direct binding, we attempted to purify the DNase1l3-like nuclease from the serum of DNase1 KO mice employing heparin-Sepharose affinity chromatog- raphy, and to detect it by immunoblotting using a polyclonal anti-mDNase1l3 serum. This antibody was produced by immunizing rabbits with a fusion protein consisting of glutathione S-transferase and the C-ter- minal 25 amino acid residues of mDNase1l3, which are unique for this nuclease among the members of the DNase1 family. Purification by affinity chromatogra- phy revealed that the DNase1l3-like serum nuclease indeed bound to heparin with high specificity, as revealed by its elution from heparin-Sepharose only at high ionic strength (Fig. 7A). This result indicates that inhibition of this nuclease by heparin is caused by a direct interaction. Next, we purified the DNase1l3-like nuclease from 0.5 mL of WT serum and, after further concentration, equal parts of the sample were used in immunoblotting and DPZ. Cell extracts of NIH-3T3 fibroblasts transiently transfected with mDNase1 or mDNase1l3 were employed as control. We found that the anti-mDNase1l3 serum recognized mDNase1l3 in

M. Napirei et al.

Murine serum nucleases

signal peptide.

A

two potential N-glycosylation sites

In contrast, mDNase1 N-terminal migrated at (cid:2) 37 kDa, although the calculated molecu- lar mass for the mature enzyme without its N-terminal signal peptide is 29.8 kDa. As it has been described that bovine DNase1 displays tissue-specific mono- or di-N-glycosylation of the high mannose or complex type [22], we analysed rmDNase1 and rmDNase1l3 for the presence of N-glycosylation. Murine DNase1 pos- (Asn-X- sesses Ser ⁄ Thr) at Asn18 and Asn106, whereas murine DNase1l3 possesses one potential site at Asn283 (the numbering refers to the amino acid sequence of the mature protein without the N-terminal signal peptide) [7]. However, Asn283 is not conserved between DNase1l3 of mouse, rat and humans [7].

B

Fig. 8. Murine DNase1 is di-N-glycosylated, whereas mDNase1l3 is not. DPZ of rmDNase1 and rmDNase1l3 (A) and murine parotid gland DNase1 (B) treated with EndoH or PNGaseF. Recombinant murine DNase1l3 is not N-glycosylated [bottom panel of (A)]. Recombinant murine DNase1 [top panel of (A)] and parotid gland DNase1 (B) represent a mixture of di-N-glycosylated molecules. Approximately one-half of the molecules possessed two complex N-glycans (resistant to deglycosylation by EndoH; molecular mass (cid:2) 37 kDa); the other half possessed one high-mannose and one complex N-glycan (leading to mono-N-glycosylation after EndoH treatment, molecular mass (cid:2) 35 kDa); a very minor proportion pos- sessed two high-mannose N-glycans, leading to complete de-N-gly- cosylation by EndoH comparable with that by PNGaseF treatment (resulting in a molecular mass of 29.8 kDa as calculated from the sequence of mature mDNase1). Mature mDNase1l3 has a calcu- lated molecular mass of 33.1 kDa, which is consistent with the estimated molecular mass of rmDNase1l3 ((cid:2) 34 kDa) from the zymograms.

((cid:2) 37 kDa) and de-

demonstrate that the We treated both nucleases with endoglycosidase H in part, (EndoH), which cleaves high-mannose and, hybrid N-glycans, or with peptide N-glycosidase F (PNGaseF), which cleaves all forms of N-glycans, and subsequently performed DPZ (Fig. 8). We found that rmDNase1l3 is apparently not N-glycosylated, whereas rmDNase1 is di-N-glycosylated (Fig. 8A). Obviously, secreted rmDNase1 is a mixture of molecules differing in the composition of the two N-glycosylation sites. Approximately half of the molecules possessed one high-mannose and one complex N-glycan [only the high-mannose N-glycan was cleavable by EndoH, lead- ing to migration of EndoH-treated rmDNase1 between ((cid:2) 29 kDa) N-glycosylated di- rmDNase1 at (cid:2) 35 kDa). The other half possessed two complex N-glycans [not cleavable by EndoH, leading to migration of EndoH-treated rmDNase1 at the molecu- lar mass of non-treated rmDNase1 ((cid:2) 37 kDa)]. A very minor proportion possessed two high-mannose N-gly- cans (both cleavable by EndoH, leading to migration of EndoH-treated rmDNase1 at (cid:2) 29 kDa, which is con- sistent with the calculated molecular mass for mature mDNase1) (Fig. 8A). As expected, PNGaseF cleaved both N-glycans, leading to completely de-N-glycosylated rmDNase1 (Fig. 8A). In order to verify that di-N-gly- in vivo, we cosylation of mDNase1 also occurs repeated the experiments with murine parotid gland DNase1 and obtained identical results (Fig. 8B). These data suggest that, after transfection, rmDNase1 is gly- cosylated in a random manner by NIH-3T3 cells, and in vivo by the exocrine cells of the parotid gland, and putative furthermore glycosylation site of DNase1l3 is not recognized.

Discussion

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1067

of DNase1 and DNase1l3. By comparing the proper- ties of rmDNase1 and rmDNase1l3 in the hydrolysis of pDNA and chromatin with those of serum col- lected from WT and DNase1 KO mice, we were able to clarify the identity of the nucleolytic activities of murine serum. Our new experiments prove that mur- ine and human sera contain both DNase1 and In the present work, we continued our previous stud- ies on the characterization of the nucleolytic activities

M. Napirei et al.

Murine serum nucleases

DNase1l3. Our previous studies concentrated on the characterization of the properties of serum DNase1, although we suspected the presence of an additional nuclease with biochemical properties of rrDNase1l3 in murine serum [7].

Properties of DNase1 and DNase1l3 in the hydrolysis of DNA substrates

is (DNase1l3) human DNase c

crucial to be for

Histone degradation by proteases is not necessary for chromatolysis by rmDNase1l3. Recombinant mur- ine DNase1l3 and DNase1l3 from other species are able to induce internucleosomal chromatin degradation on their own [7,23]. Indeed, Mizuta et al. [23] have demonstrated that histone H1 functions as a co-activa- tor of DNase c, leading to the degradation of pDNA and chromatin at physiological ionic strength. They hypothesized that DNase c might compete with his- tone H1 for DNA binding and, after histone H1 dis- placement, will gain access to and hydrolyse chromatin DNA. This competition seems to be conceivable, as rmDNase1l3 (pI 8.7) has an estimated charge of +6.7 at pH 7.0; thus, it is a basic protein, like the histones, at physiological pH values. In contrast, rmDNase1 (pI 4.9) is an acidic protein with a charge of )9.4 at pH 7.0, which might explain the opposite behaviour of the two nucleases despite their structural similarities [7]. Alternatively, it has been proposed that histone H1 binding to internucleosomal regions might generate ssDNA portions, which are preferred targets for cleav- age by DNase1l3 [23]. Indeed, an altered DNA confor- mation seems efficient DNA hydrolysis by rmDNase1l3, as demonstrated by our observation of enhanced cleavage of pDNA and chro- matin in the presence of Mn2+. From these findings, we propose that the activating mode of histone H1 and Mn2+ on DNA hydrolysis by rmDNase1l3 may be caused by their similar influence on DNA confor- mation and not by displacement of histone H1. Our data demonstrate that heparin inhibits

Our data demonstrate that rmDNase1 and rmDN- ase1l3 harbour different properties with regard to DNA substrates. Thus, rmDNase1l3 cleaves protein- free pDNA with a lower efficiency in comparison with rmDNase1, but degrades chromatin more rapidly as a result of preferential cleavage at internucleosomal sites. the presence of Mn2+ Our experiments show that instead of Mg2+, in addition to Ca2+-ions enhances, in particular, DNase1l3 activity over a broad pH range. Previously, Mizuta et al. [23] have reported that recombinant a Ca2+ ⁄ Mg2+-dependent ssDNA nuclease with high activity at low ionic strength. Furthermore, it has been reported that Mn2+, in contrast with Mg2+, has dif- ferent effects on DNA conformation: (a) it leads to toroidal condensates of supercoiled pDNA, resulting in more extensive digestion by S1 nuclease [24]; and (b) it affects the CD spectra, especially of GC-rich native DNA, by binding to the GC pairs in addition to the phosphate groups [25]. Proton displacement by Mn2+ from GC pairs leads to conformational changes of the double helix, which are interpreted as tilting of the bases of locally Mn2+-chelated regions [25]. These data may explain why DNase1l3, in particular, which has been described to have a higher affinity and ⁄ or cleavage activity towards ssDNA, is activated in the presence of Mn2+. The activating effect of Mn2+ was used by us to optimize the detection of DNase1l3 in pDNA and chromatin digestion assays, as well as in DPZ.

that normal physiological

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1068

In accordance with Mizuta et al. [23], we found that high ionic strength (NaCl or Tris) more strongly inhib- ited the activity of rmDNase1l3 than of rmDNase1. in Nevertheless, our experiments demonstrate that, undiluted serum, DNase1l3 is sufficiently active to facilitate chromatolysis at physiological ionic strength and composition. Similarly, the murine serum DNase1 concentration is also sufficient to induce chromatoly- sis, provided that the activation of the plasminogen system occurs or other proteases are present. This dependence may explain previous observations indicat- concentrations of ing DNase1 in human serum are insufficient to degrade DNA [26]. the cleavage of chromatin and protein-free pDNA by rmDNase1l3, whereas it activates chromatolysis and does not influence pDNA digestion by rmDNase1. As heparin is a negatively charged sulfated polysaccharide, a direct interaction with polyanionic DNA can be excluded. Therefore, we propose that heparin binds directly to and inhibits DNase1l3. This assumption is consistent with the fact that DNase1l3 binds to hepa- rin-Sepharose at physiological pH values [27]. Indeed, we were able to verify this interaction by purifying the DNase1l3-like nuclease activity of murine serum through heparin-Sepharose affinity chromatography. Subsequent immunoblotting, employing an antibody generated against a peptide comprising the last 25 C-terminal amino acids of mDNase1l3, demonstrated that the second nucleolytic activity of murine serum is identical to that of DNase1l3. In contrast, DNase1 is negatively charged at physiological pH values and is not inhibited by heparin. Previously, we suspected that the activating effect of heparin on chromatolysis by serum DNase1 might be caused by hyperactivation of the plasminogen system [3]. However, our present data

M. Napirei et al.

Murine serum nucleases

show that, in the absence of the plasminogen system, heparin has the same activating effect on rmDNase1. Therefore, its effect is most probably caused by dis- placement of the positively charged histones from thereby generating protein-free DNA, chromatin, which is more efficiently cleaved by DNase1, leading to a switch from an internucleosomal to a random chromatin cleavage pattern. Indeed, in agreement with Hildebrand et al. [28], we verified the displacement of all histones by heparin. Thus, heparin mimics the effect of proteases on DNase1-induced chromatolysis by gen- erating stretches of protein-free chromosomal DNA.

Serum DNase1 and DNase1l3 might suppress antinuclear autoimmunity

it is conceivable that

that DNase1l3 is

thymus, DNase1l3 was originally isolated from rat implied to be essential for chromatin degradation during apoptosis, and therefore regarded as a purely intracellu- lar endonuclease [29]. However, the present data demon- strate that DNase1l3 is also secreted in vivo into murine and human serum in addition to DNase1. We found that, as a result of the activation of the plasminogen sys- tem in the in vitro chromatin digestion assays, murine serum DNase1l3 is rapidly degraded by plasmin, whereas di-N-glycosylation probably protects DNase1 from proteolysis. These findings explain why, in previ- ous studies, serum DNase1l3 was underestimated or not discovered. Thus, our data demonstrate that both serum nucleases are involved in chromatolysis under physio- logical ion concentrations and composition in vitro.

The specific roles of the two serum nucleases in vivo have been investigated poorly to date. For serum DNase1, it has been shown that it is involved in the degradation of chromatin released from necrotic cells (hepatocytes). Thus, after the induction of hepatocellu- lar necrosis by an overdose of acetaminophen, nucleos- omal chromatin fragments accumulate in the blood of DNase1 KO mice, whereas, in WT mice, these frag- ments disappear as a result of further degradation to oligonucleotides [31]. These data imply that DNase1 present in the serum and ⁄ or liberated from necrotic hepatocytes degrades chromatin to oligonucleotides. Indeed, in vitro data have demonstrated that extracel- lular DNase1 penetrates and accumulates within the nuclei of necrotic cells [3]. Necrotic cells induce inflam- mation, accompanied by an increased permeability of serum blood vessels. Thus, DNase1 diffuses into necrotic tissues. The same can be hypothesized for DNase1l3. Furthermore, it is conceiv- able secreted by macrophages recruited into inflamed tissues. Thus, DNase1l3 may function as the primary chromatolytic activity generat- ing nucleosomal fragments, which are subsequently further degraded by DNase1. Activation of the plas- minogen system at sites of necrosis and inflammation is a well-known phenomenon [32]. However, plasmin activity is tightly controlled by extracellular protease inhibitors to prevent extensive tissue damage. Our in vitro results demonstrate the degradation of mDN- ase1l3 by plasmin; however, it is still unclear whether this is also true for the in vivo situation. The exact con- centrations of released and activated enzymes within inflamed tissues, the order of their appearance and the duration of their activity have not been clarified to date. Further in vivo experiments are necessary to resolve these questions.

In summary, our data reveal, for the first time, that serum contains two nucleases – DNase1 and DNase1l3 – which display different substrate specificities. Both nucleases may complement or substitute each other under certain conditions during chromatin degr- adation. It is hoped that future studies on DNase1, DNase1l3 and double KO mice will provide further insight into the exact function and role of the two nuc- leases in the prevention of antinuclear autoimmunity.

Materials and methods

For cloning of the murine DNase1 and DNase1l3 cDNA, total RNA was isolated from kidney (DNase1) or spleen

Cloning of murine DNase1 and DNase1l3 expression vectors

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1069

Serum nucleases seem to fulfil intra- as well as extra- vascular functions in vivo: (a) clearance of chromatin released into the circulation by dying cells to prevent occlusion of capillaries by DNA clots; (b) clearance of nuclear debris within inflamed tissues to prevent auto- antigen formation; (c) clearance of circulating immune complexes composed of ANA and their DNA-contain- ing antigens to prevent their renal deposition; and (d) clearance of deposited nuclear antigens and immune complexes to suppress from basement membranes inflammation caused by hypersensitivity reactions [4]. All of these functions can be summarized as ‘suppres- sion of antinuclear autoimmunity’. Indeed, it has been shown that many patients with SLE [5,30] and SLE- prone mice [4–6] display a lack or decrease in serum DNase1. The participation of murine DNase1l3 in the suppression of antinuclear autoimmunity has also been postulated. Thus, it has been shown that SLE-prone MRL-lpr and NZB ⁄ W F1 mice possess a homozygous missense mutation within the DNase1l3 gene, resulting in a reduced activity of DNase1l3 secreted by spleno- cytes and bone marrow-derived macrophages [21].

M. Napirei et al.

Murine serum nucleases

primer

the C-terminal

primer

and

cleavage and, finally,

(DNase1l3) of a C57BL ⁄ 6 mouse using the RNeasy Mini Kit (Qiagen, Hilden, Germany). The cDNAs were gener- ated by reverse transcription of 2 lg of RNA employing the Omniscript RT Kit (Qiagen) and oligo(dT) primers (12–18 nucleotides; Sigma-Aldrich, Taufkrichen, Germany). RT-PCR of DNase1 cDNA (Genbank Accession Number NM010061, nucleotides 250–1185) was performed using the N-terminal 5¢-GACTGCTGCAGAATTCTCAG ATTGGCT-3¢ and the C-terminal primer 5¢-GTGGAT GCGGCCGCACCAGAAGCA-3¢ containing EcoRI and NotI restriction sites generated by site-directed mutagenesis. RT-PCR of DNase1l3 cDNA (Genbank Accession Number AF047355, nucleotides 153–1114) was performed using the N-terminal primer 5¢-GAAGTCCCAGGAATTCAAAGA TGT-3¢ 5¢-GCGTGAT ACCCGGGAGCGATTG-3¢ containing EcoRI and SmaI restriction sites generated by site-directed mutagenesis. Both cDNAs were first subcloned blunt-end into the MluNI site of the vector pCAPs using the PCR Cloning Kit (Roche Diagnostics GmbH, Mannheim, Germany), subsequently isolated by EcoRI ⁄ NotI cloned into the EcoRI ⁄ NotI the vector pDsRed-N1 sites of (Clontech, Heidelberg, Germany), from which the cDNA of the red fluorescent protein was eliminated by EcoRI ⁄ NotI cleavage.

tube and allowed to coagulate for 1–2 h at 4 (cid:2)C. Subse- quently, the blood was centrifuged for 10 min at 600 g and the serum was transferred into a fresh tube and stored at )20 (cid:2)C until use. For DPZ, 10 lL of serum were mixed with 90 lL of RIPA buffer [10 mm Tris ⁄ HCl pH 7.2, 150 mm NaCl, 0.1% (w ⁄ v) SDS, 1% (v ⁄ v) Triton X-100, 1% (w ⁄ v) sodium deoxycholate, 5 mm EDTA and 1% (v ⁄ v) protease inhibitor cocktail (Sigma-Aldrich)], incubated for 30 min on ice, subsequently mixed with 100 lL of 2· SDS gel-loading buffer [100 mm Tris ⁄ HCl pH 6.8, 4% (w ⁄ v) SDS, 0.2% (w ⁄ v) bromophenol blue, 20% (v ⁄ v) glycerol and 350 mm 2-mercaptoethanol] and 20 lL of these samples were loaded onto the zymograms after heating the samples to 95 (cid:2)C for 5 min and cooling to room temperature. The organs of the dead mice were removed, snap-frozen in liquid nitrogen and stored at )80 (cid:2)C. Tissue extracts were prepared by homoge- nizing the organs in lysis buffer for 30 s using a rotor-stator (Ultra-Turrax T8 homogenizer; IKA Labortechnik, Staufen, Germany) at maximal power (level 6). Either TET buffer [10 mm Tris ⁄ HCl pH 8.0, 20 mm EDTA, 0.5% (v ⁄ v) Triton X-100 and 1% (v ⁄ v) protease inhibitor cocktail] or RIPA buffer (see above) was used as lysis buffer. Samples were sub- incubated on ice for sequently frozen and thawed twice, 30 min, and the cell debris was sedimented by centrifugation at 21 000 g for 10 min at 4 (cid:2)C. The protein content of the supernatants was determined by the standard Bradford pro- cedure [33], and the samples were adjusted to a concentration of 8 mgÆmL)1 using lysis buffer. For DPZ, 100 lL of the tis- sue samples were mixed with 100 lL of 2· SDS gel-loading buffer, heated for 5 min to 95 (cid:2)C, cooled to room tempera- ture, and 20 lL (80 lg protein) were loaded onto the zymograms.

Cell transfection

reagents

Cultivation of NIH-3T3 fibroblasts (ACC59) was per- formed according to the instructions of the German Collec- tion of Microorganisms (Braunschweig, Germany) in 90% (v ⁄ v) DMEM high-glucose (4.5 gÆL)1) medium containing 10% (v ⁄ v) heat-inactivated fetal bovine serum Gold, 2 mm l-glutamine, 1 mm sodium pyruvate and 1% (v ⁄ v) strepto- mycin ⁄ neomycin (all from PAA Laboratories transfection of GmbH, Pasching, Austria). Transient the cells with expression vectors for murine DNase1 and DNase1l3 cDNA was performed by magnet-assisted trans- fection using the MATra-A reagent (IBA BioTAGnology, Go¨ ttingen, Germany) according to the manufacturer’s instructions. Transient gene expression was allowed for 48 h, and the cell culture medium containing the secreted nucleases was divided into aliquots and stored at )20 (cid:2)C until use.

DPZ

Preparation of serum and tissue samples

WT and DNase1 KO mice of the inbred strain C57BL ⁄ 6 were bred in our animal facility. The mice were allowed free access to standard laboratory chow and water, and kept in a light ⁄ - dark cycle for 12 h. All animal procedures carried out in this work conformed with German Animal Protection Law. Blood was collected from ether-anaesthetized animals by tho- racic bleeding after opening the thorax and setting a cut into the heart. The blood was transferred into a microcentrifuge

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1070

Non-reducing conditions Standard SDS-PAGE gels, according to Laemmli [34], were prepared with 4% (v ⁄ v) collecting gels without DNA and 10% (v ⁄ v) resolving gels containing 200 lgÆmL)1 calf thy- mus DNA (D1501, Sigma-Aldrich). Serum samples were prepared using SDS gel-loading buffer without 2-mercapto- ethanol as described above, and loaded onto the zymo- grams. As a molecular mass marker, the PageRuler(cid:3) Prestained Protein Ladder (MBI Fermentas, St Leon-Rot, Germany) was used. Electrophoresis was carried out at 80 V using Tris ⁄ glycine electrophoresis buffer [25 mm Tris, 192 mm glycine, 0.1% (w ⁄ v) SDS, pH 8.7]. After electro- phoresis, SDS was removed and the proteins were refolded by washing the gels overnight with 5% (w ⁄ v) milk powder dissolved in 150 mL of 10 mm Tris ⁄ HCl pH 7.8, 3 mm CaCl2, 3 mm MgCl2 and 10 mm sodium azide. Subse- quently, nuclease reaction was performed by incubating the gels in the same buffer without milk powder for 24–48 h at 37 (cid:2)C. Nuclease activities were detected as dark unstained areas after staining the gels with 0.5 lgÆmL)1 ethidium

M. Napirei et al.

Murine serum nucleases

bromide and photographing the gels on a UV-light transilluminator.

the samples were mixed berg, Germany). Subsequently, with an equal volume of 2· SDS gel-loading buffer, heated to 95 (cid:2)C, cooled to room temperature, and 20 lL of the samples were subjected to DPZ as described above. For the deglycosylation of murine parotid DNase1, 1 lg of parotid protein was treated in an assay of 50 lL with either 100 U PNGaseF or 100 U EndoH for 15 min or 2 h at 37 (cid:2)C. Five microlitres of the samples were mixed with 100 lL of 1· SDS gel-loading buffer, heated to 95 (cid:2)C, cooled to room temperature and 20 lL of the samples were subjected to DPZ under reducing conditions as described above.

Reducing conditions

1 mm 2-mercaptoethanol

either

and

DPZ under reducing conditions, according to the method described by Shiokawa et al. [14], was modified as described in the Results section and performed as follows. Zymo- grams were prepared as described above for non-reducing DPZ. Samples were prepared in SDS gel-loading buffer containing 2-mercaptoethanol as described above, and loaded onto the zymograms. Electrophoresis was carried out at 80 V using Tris ⁄ glycine electrophoresis buffer and, after electrophoresis, SDS was removed by washing the gels twice with 150 mL of 10 mm Tris ⁄ HCl pH 7.8 and 5 mm 2-mercaptoethanol at 50 (cid:2)C for altogether 1 h. Nuclease refolding was performed by incubating the gels in 150 mL of 10 mm Tris ⁄ HCl pH 7.8 containing 1 mm 2-mercapto- ethanol for 24 h at 37 (cid:2)C. Nuclease reaction was performed by incubating the gels in 150 mL of 10 mm Tris ⁄ HCl 3 mm pH 7.8, MnCl2 ⁄ 3 mm CaCl2 (optimal for DNase1l3 activity) or 3 mm MgCl2 ⁄ 3 mm CaCl2 (optimal for DNase1 activity) for 24–48 h at 37 (cid:2)C. To specifically inhibit DNase1l3 activ- ity, 50 UÆmL)1 of heparin was added to the reaction buffer.

To examine the nucleolytic activities of cell culture superna- tants containing the recombinant nucleases, 0.1 lL (rm- DNase1) or 1 lL (rmDNase1l3) of the supernatants was added to 20 lL of substrate solution (100 ng pDNA dissolved in 10 mm Tris ⁄ HCl pH 7.0, containing 2 mm CaCl2 and 2 mm MgCl2) and incubated at 37 (cid:2)C for 10 min (rmDNase1) or 30 min (rmDNase1l3). Buffer and ion compositions, as well as the addition of additives, were varied as indicated in the Results section and the figure legends. Thereafter, the samples were heated to 65 (cid:2)C for 5 min and subjected to 1% (w ⁄ v) Tris–acetate ⁄ EDTA (TAE)-agarose gel electrophoresis.

pDNA digestion assay

Murine serum DNase1l3 was purified using heparin-Sepha- rose (Amersham Biosciences Europe GmbH, Freiburg, Germany) in either a batch or column procedure. Serum was diluted with 5 vol of binding buffer [20 mm Tris ⁄ HCl pH 7.5, 0.15 m NaCl, 5% (v ⁄ v) glycerol, 0.1 mm EDTA and 1 mm 2-mercaptoethanol] and added to the Sepharose (0.1–1 mL, depending on the batch or column procedure). Subsequently, the Sepharose was washed with 10 vol of washing buffer (20 mm Tris ⁄ HCl pH 7.5, 0.2 m NaCl, 0.1 mm EDTA and 1 mm 2-mercaptoethanol) and the bound proteins were eluted with elution buffer [20 mm Tris ⁄ HCl pH 7.5, 5% (v ⁄ v) glycerol, 0.1 mm EDTA and 1 mm 2-mercaptoethanol] containing different concentra- tions of NaCl (0.3–2 m). Standard elution was performed with the concentration of 1 m NaCl. Subsequently, the sam- ples were desalted and concentrated using Ultracel YM-10 Microcon(cid:4) Centrifugal Filter Devices (Millipore GmbH, Eschborn, Germany). All buffers were supplemented with protease inhibitor cocktail (Sigma-Aldrich).

Isolation of MCF-7 cell nuclei was performed as described previously [2]. The cell nuclei were treated with either cell culture supernatants containing the recombinant nucleases or serum derived from WT or DNase1 KO mice [3]. Five microlitres of the cell culture supernatants were added to 105 cell nuclei diluted in 200 lL of reaction buffer (10 mm Tris ⁄ HCl, 50 mm NaCl, 2 mm MgCl2, 2 mm CaCl2, pH 7.0) and incubated at 37 (cid:2)C for 1–2 h. Serum was either employed at a concentration of 2.5% (v ⁄ v) in an assay described for the cell culture supernatants (various times of incubation at 37 (cid:2)C as indicated in the figure legends) or cell nuclei were directly added to undiluted serum and incubated at 37 (cid:2)C for 2–8 h. Some reaction samples also contained murine Pai-1 (Calbiochem Novabiochem, Schwalbach, Germany), heparin (Liquemin(cid:4); Hoffmann La Roche, Grenzach Wyhlen, Ger- many), thrombin, plasmin or aprotinin (all supplied by Sigma-Aldrich) at the concentrations indicated in the figure legends. Subsequent to the incubation step at 37 (cid:2)C, nuclear DNA was isolated using a QIAamp DNA Blood Mini Kit (Qiagen), and the DNA was analysed by 1.5% (w ⁄ v) Tris– borate ⁄ EDTA (TBE)-agarose gel electrophoresis.

Heparin-Sepharose affinity chromatography Nuclear chromatin digestion assay

Deglycosylation

For the generation of a polyclonal rabbit anti-mDNase1l3 serum, the last 25 amino acids of murine DNase1l3 were

Aliquots of cell culture supernatants (30 lL) containing rmDNase1 or rmDNase1l3 were treated with different amounts of either EndoH or PNGaseF according to the manufacturer’s instructions (New England Biolabs, Heidel-

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1071

Anti-mDNase1l3 serum production

M. Napirei et al.

Murine serum nucleases

transfer onto a poly(vinylidene difluoride) membrane using CAPS buffer [10 mm 3-(cyclohexamine)propane-3-sulfonic acid, 10% (v ⁄ v) methanol, pH 11.0] for semidry electro- transfer. Blotting membranes were blocked as described above and immunodetection was performed using the ECL detection system (Amersham Biosciences Europe GmbH). As primary antibody, the polyclonal rabbit anti-mDNase1l3 serum (see above) was used at a dilution of 1 : 2000 over- night at 4 (cid:2)C. As a secondary antibody, an anti-rabbit IgG conjugated with horseradish peroxidase (Cell Signaling Technologies Inc., Danvers, MA, USA) was employed at a dilution of 1 : 2000 for 1 h at room temperature.

Acknowledgements

cloned in fusion to glutathione S-transferase. Using the vector pDs-mDNase1l3 (see above) as a template and employing the N-terminal (5¢-CAGTTGAGTTTAAGCTA CAGT-3¢) and C-terminal (5¢-GGCTCGAGGATACCTA GGAGC-3¢) primers containing EcoRI and XhoI restriction sites, respectively, generated by site-directed mutagenesis, the mDNase1l3 cDNA sequence (Genbank Accession Num- ber AF047355, nucleotides 1115–1128) was amplified by PCR and cloned into the EcoRI ⁄ XhoI sites of the vector pGEX-4T2 of the GST Gene Fusion System (Amersham Biosciences Europe GmbH). The fusion protein was expressed using Escherichia coli BL21 bacteria and, after harvesting and lysing of it was purified the bacteria, Sepharose 4B (Sigma-Aldrich, employing Glutathione Deisenhofen, Germany) in a batch procedure according to the manufacturer’s instructions. The purified GST- mDNase1l3 fusion protein was dialysed against NaCl ⁄ Pi, and a rabbit was immunized by subcutaneous injection of, first, 200 lg and then twice at 1-month intervals 100 lg protein dissolved in TiterMax Gold(cid:4) (HiSS Diagnostics GmbH, Freiburg, Germany). Blood and serum were col- lected and prepared after 3 months, and employed in immunoblotting.

The authors thank Rana Houmany, Eva Maria Kon- ieczny and Swantje Wulf for excellent technical assis- tance and Dr Dirk Eulitz for providing the murine DNase1 cDNA. This work was supported by a grant from the FoRUM programme of the Ruhr-University Bochum (F505-2006).

References

1 Peitsch MC, Irmler M, French LE & Tschopp J

(1995) Genomic organisation and expression of mouse deoxyribonuclease I. Biochem Biophys Res Commun 207, 62–68.

2 Napirei M, Ricken A, Eulitz D, Knoop H & Mannherz HG (2004) Expression pattern of the deoxyribonucle- ase 1 gene: lessons from the DNase1 knockout mouse. Biochem J 380, 929–937.

3 Napirei M, Wulf S & Mannherz HG (2004) Chromatin breakdown during necrosis by serum DNase1 and the plasminogen system. Arthritis Rheum 50, 1873–1883. 4 Napirei M, Gultekin A, Kloeckl T, Moroy T, Froste- gard J & Mannherz HG (2006) Systemic lupus-erythe- matosus: deoxyribonuclease 1 in necrotic chromatin disposal. Int J Biochem Cell Biol 38, 297–306.

5 Lachmann PJ (2003) Lupus and deoxyribonuclease.

Lupus 12, 202–206.

6 Tsukumo S & Yasutomo K (2004) DNaseI in pathogen- esis of systemic lupus erythematosus. Clin Immunol 113, 14–18.

7 Napirei M, Wulf S, Eulitz D, Mannherz HG & Kloeckl T (2005) Comparative characterization of rat deoxyri- bonuclease 1 (DNase1) and murine deoxyribonucle- ase 1-like 3 (DNase1l3). Biochem J 389, 355–364. 8 Shiokawa D & Tanuma S (2001) Characterization of

human DNase I family endonucleases and activation of DNase gamma during apoptosis. Biochemistry 40, 143– 152.

For the detection of histone displacement from chromatin by heparin, 5 · 105 MCF-7 cell nuclei were incubated in 100 lL of 10 mm Hepes pH 7.0, 50 mm NaCl, 40 mm b-glycerophosphate, 1% (v ⁄ v) protease inhibitor cocktail, 2 mm CaCl2 and 2 mm MgCl2 for 2 h at 37 (cid:2)C in the pres- ence of increasing amounts of heparin. Subsequently, the cell nuclei were sedimented by centrifugation for 15 min at 21 000 g. The supernatants were mixed with 0.25 vol of 5· SDS gel-loading buffer, and 20 lL of the samples were sub- jected to 15% (w ⁄ v) SDS-PAGE, followed by protein trans- fer onto a poly(vinylidene difluoride) membrane using the PerfectBlue(cid:3) Semi-Dry-Electroblotter (Peqlab, Erlangen, Germany) and 48 mm Tris, 39 mm glycine, 0.037% (w ⁄ v) SDS and 20% (v ⁄ v) methanol as transfer buffer. Blotting membranes were blocked with 3% (w ⁄ v) low-fat milk pow- der dissolved in Tris-buffered saline supplemented with 0.05% (v ⁄ v) Tween-20. Immunodetection of histone H3 using the alkaline phosphatase protocol, with 5-bromo-4- chloro-3-indolyl phosphate and nitroblue tetrazolium as detection substrates, was performed according to Sambrook et al. [35]. As primary antibody, a polyclonal goat anti-his- tone H3 serum (sc-8654; Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) was employed at a dilution of 1 : 250 overnight at 4 (cid:2)C. As a secondary antibody, a chicken anti-goat IgG conjugated with alkaline phosphatase (sc-2965; Santa Cruz Biotechnology) was employed at a dilution of 1 : 2000 for 1 h at room temperature.

For anti-mDNase1l3 immunoblotting,

9 Shiokawa D, Shika Y & Tanuma S (2003) Identification of two functional nuclear localization signals in DNase

samples were subjected to 10% (v ⁄ v) SDS-PAGE, followed by protein

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1072

Immunoblotting

M. Napirei et al.

Murine serum nucleases

22 Nishikawa A & Mizuno S (2001) The efficiency of

gamma and their roles in its apoptotic DNase activity. Biochem J 376, 377–381.

N-linked glycosylation of bovine DNase I depends on the Asn-Xaa-Ser ⁄ Thr sequence and the tissue of origin. Biochem J 355, 245–248.

10 Boulares AH & Ren T (2004) Mechanism of acetamino- phen-induced apoptosis in cultured cells: roles of cas- pase-3, DNA fragmentation factor, and the Ca2+ and Mg2+ endonuclease DNAS1l3. Basic Clin Pharmacol Toxicol 94, 19–29.

11 Boulares AH, Zoltoski AJ, Contreras FJ, Yakovlev

23 Mizuta R, Mizuta M, Araki S, Shiokawa D, Tanuma S & Kitamura D (2006) Action of apoptotic endonu- clease DNase gamma on naked DNA and chromatin substrates. Biochem Biophys Res Commun 345, 560– 567.

24 Ma C & Bloomfield VA (1994) Condensation of super- coiled DNA induced by MnCl2. Biophys J 67, 1678– 1681.

25 Luck G & Zimmer C (1972) Conformational aspects

AG, Yoshihara K & Smulson ME (2002) Regulation of DNAS1l3 endonuclease activity by poly(ADP-ribo- syl)ation during etoposide-induced apoptosis. Role of poly(ADP-ribose) polymerase-1 cleavage in endonucle- ase activation. J Biol Chem 277, 372–378.

and reactivity of DNA. Effects of manganese and mag- nesium ions on interaction with DNA. Eur J Biochem 29, 528–536.

12 Liu QY, Lei JX, LeBlanc J, Sodja C, Ly D, Charlebois C, Walker PR, Yamada T, Hirohashi S & Sikorska M (2004) Regulation of DNaseY activity by actinin-alpha4 during apoptosis. Cell Death Differ 11, 645–654.

26 Prince WS, Baker DL, Dodge AH, Ahmed AE, Chest- nut RW & Sinicropi DV (1998) Pharmacodynamics of recombinant human DNase I in serum. Clin Exp Immu- nol 113, 289–296.

27 Umemori K, Nishikawa A & Tanuma S (1998) Purifica-

13 Shiokawa D, Kobayashi T & Tanuma S (2002) Involve- ment of DNase gamma in apoptosis associated with myogenic differentiation of C2C12 cells. J Biol Chem 277, 31031–31037.

tion and characterization of a DNase gamma-like endonuclease from Xenopus laevis liver. Apoptosis 3, 145–153.

14 Shiokawa D, Ohyama H, Yamada T, Takahashi K & Tanuma S (1994) Identification of an endonuclease responsible for apoptosis in rat thymocytes. Eur J Bio- chem 226, 23–30.

15 Nishimura K & Tanuma S (1998) Presence of DNase

gamma-like endonuclease in nuclei of neuronal differen- tiated PC12 cells. Apoptosis 3, 97–103.

28 Hildebrand CE, Gurley LR, Tobey RA & Walters RA (1977) Action of heparin on mammalian nuclei. I. Dif- ferential extraction of histone H1 and cooperative removal of histones from chromatin. Biochim Biophys Acta 477, 295–311.

29 Shiokawa D, Ohyama H, Yamada T & Tanuma S

(1997) Purification and properties of DNase gamma from apoptotic rat thymocytes. Biochem J 326, 675– 681.

16 Higami Y, Tsuchiya T, To K, Chiba T, Yamaza H, Shi- okawa D, Tanuma S & Shimokawa I (2004) Expression of DNase gamma during Fas-independent apoptotic DNA fragmentation in rodent hepatocytes. Cell Tissue Res 316, 403–407.

30 Tew MB, Johnson RW, Reveille JD & Tan FK (2001) A molecular analysis of the low serum deoxyribonucle- ase activity in lupus patients. Arthritis Rheum 44, 2446– 2447.

17 Shiokawa D, Tanaka M, Kimura T, Hashizume K, Ta- kasawa R, Ohyama H, Fujita K, Yamada T & Tanuma S (2000) Characterization of two DNase gamma-specific monoclonal antibodies and the in situ detection of DNase gamma in the nuclei of apoptotic rat thymo- cytes. Biochem Biophys Res Commun 275, 343–349. 18 Baron WF, Pan CQ, Spencer SA, Ryan AM, Lazarus

31 Napirei M, Basnakian AG, Apostolov EO & Mannherz HG (2006) Deoxyribonuclease 1 aggravates acetamino- phen-induced liver necrosis in male CD-1 mice. Hepa- tology 43, 297–305.

RA & Baker KP (1998) Cloning and characterization of an actin-resistant DNase I-like endonuclease secreted by macrophages. Gene 215, 291–301.

19 Tanuma S & Shiokawa D (1994) Multiple forms of

32 Del Rosso M, Fibbi G, Pucci M, Margheri F & Serrati S (2008) The plasminogen activation system in inflam- mation. Front Biosci 13, 4667–4686.

nuclear deoxyribonuclease in rat thymocytes. Biochem Biophys Res Commun 203, 789–797.

33 Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding. Anal Biochem 72, 248–254.

34 Laemmli UK (1970) Cleavage of structural proteins

during the assembly of the head of bacteriophage T4. Nature 227, 680–685.

20 Okamoto M, Okamoto N, Yashiro H, Shiokawa D, Su- naga S, Yoshimori A, Tanuma S & Kitamura D (2005) Involvement of DNase gamma in the resected double- strand DNA breaks in immunoglobulin genes. Biochem Biophys Res Commun 327, 76–83.

35 Sambrook J, Fritsch EF & Maniatis T (1989) Molecu- lar Cloning. A Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

21 Wilber A, O’Connor TP, Lu ML, Karimi A & Schnei- der MC (2003) DNase1l3 deficiency in lupus-prone MRL and NZB ⁄ W F1 mice. Clin Exp Immunol 134, 46–52.

FEBS Journal 276 (2009) 1059–1073 ª 2009 The Authors Journal compilation ª 2009 FEBS

1073