Improved Potency of Hyperactive and Actin-resistant Human DNase I Variants for Treatment of Cystic Fibrosis and Systemic Lupus Erythematosus*

The ability of recombinant human DNase I (DNase I) to degrade DNA to lower molecular weight fragments is the basis for its therapeutic use in cystic fibrosis (CF) patients and its potential use as a treatment for systemic lupus erythematosus (SLE). To increase the potency of human DNase I, we have generated and characterized three classes of mutants: (a) hyperactive variants, which have from one to six additional positively charged residues (+1 to +6) and digest DNA much more efficiently relative to wild type, (b) actin-resistant variants, which are no longer inhibited by G-actin, a potent inhibitor of DNase I, and (c) combination variants that are both hyperactive and actin-resistant. For DNA scission in CF sputum where the DNA concentration and length are large, we measured a ∼20-fold increase in potency relative to wild type for the +3 hyperactive variant Q9R/E13R/N74K or the actin-resistant variant A114F; the hyperactive and actin-resistant combination variant was ∼100-fold more potent than wild type DNase I. For digesting lower concentrations of DNA complexed to anti-DNA antibodies in human serum, we found a maximal enhancement of ∼400-fold over wild type for the +2 variant E13R/N74K. The +3 enzymes have ∼4000-fold enhancement for degrading moderate levels of exogenous DNA spiked into human serum, whereas the +6 enzyme has ∼30,000-fold increased activity for digesting the extremely low levels of endogenous DNA found in serum. The actin resistance property of the combination mutants further enhances the degree of potency in human serum. Thus, the human DNase I variants we have engineered for improved biochemical and pharmacodynamic properties have greater therapeutic potential for treatment of both CF and SLE.

In addition to its fundamental role of storing genetic information, DNA is also an important component in the pathogenesis of several diseases. In patients with cystic fibrosis (CF), 1 a defective CF transmembrane conductance regulator gene results in viscous airway secretions containing high concentrations of DNA, which is derived from leukocytes in response to the persistent bacterial infections associated with this disease (1)(2)(3). Recently, CF patients have been treated clinically with recombinant human deoxyribonuclease I (DNase I), which is inhaled into the airways where it degrades DNA to lower molecular weight fragments, thus reducing the viscoelasticity of CF sputum and improving lung function (4 -6).
The presence of anti-nuclear antibodies and immune complexes containing anti-DNA antigens is the hallmark of systemic lupus erythematosus (SLE), an autoimmune disease that affects many organs such as kidney, skin, joints, and central nervous system, resulting in chronic tissue damage (7,8). Anti-DNA antibodies, in particular those against double-stranded DNA, have been implicated in inducing some of the disease manifestations of SLE, especially nephritis, and are generally present at elevated serum levels in clinically active disease (9). In SLE patients, anti-DNA antibodies are associated with glomeruli in stained kidney sections and are lower than expected in the urine, suggesting retention in the kidneys. DNase I may potentially block the progression of SLE by hydrolysis of the DNA component of membrane-deposited DNA⅐anti-DNA immune complexes, facilitating clearance of the anti-DNA antibodies and reducing glomerular nephritis. Alternatively, DNase I could hydrolyze circulating and/or antibody-complexed DNA, reducing the antigen load and subsequent deposition of immune complexes. In a recent study in NZB/W mice, systemic administration of recombinant murine DNase I was found to delay the progression of SLE and demonstrated beneficial effects on renal function and histopathology (10).
DNase I is an endonuclease that catalyzes the hydrolysis of double-stranded DNA predominantly by a single-stranded nicking mechanism under physiological conditions (11). To improve the biochemical activity and potentially the efficacy of DNase I, we have recently assessed the importance of various residues at the DNA-binding interface and engineered hyperactive variants that utilize a more efficient functional mechanism involving processive nicking of double-stranded DNA (12,13). The hyperactive variants were created by introducing basic amino acids into DNase I at the DNA binding interface to generate attractive interactions with the negatively charged phosphates on the DNA backbone. Based on this principle, we constructed and characterized a series of hyperactive variants having from one to six (ϩ1 to ϩ6) basic residue substitutions * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: Dept. of Protein Engineering, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080. Tel.: 650-225-1166; Fax: 650-225-3734; E-mail: lazarus.bob@ gene.com. 1 The abbreviations used are: CF, cystic fibrosis; DNase I, deoxyribonuclease I; SLE, systemic lupus erythematosus; bp, base pair(s); PCR, polymerase chain reaction, ELISA, enzyme-linked immunosorbant assay. One letter codes are used to represent naturally occurring L-amino acids. When referring to mutants the new amino acid follows the residue number which may be preceded by the wild type amino acid as in 13R or E13R; DNase I variants with multiple mutations are denoted with a slash between the individual mutations such as E13R/N74K/ T205K or 13R/74K/205K. The term ϩ1 variant, ϩ2 variant, . . . , ϩ6 variant refers to human DNase I mutants having one, two, . . . , six additional positively charged residues compared with wild type. (14). We found that the degree of hyperactivity was inversely proportional to both DNA concentration and length as well as salt concentration. Under pseudophysiological metal ion and salt conditions, these variants displayed maximal activities ranging from several-to thousands-fold greater than wild type depending on the concentration and length of DNA. In addition to improving the intrinsic DNA-degrading activity of DNase I, we have also recently engineered actin-resistant DNase I variants that no longer bind to G-actin, a potent inhibitor of DNase I having a K i of ϳ1 nM (15). The actin-resistant variants were 10 -50-fold more potent than wild type in reducing viscoelasticity as determined in sputum compaction assays, implicating G-actin as a significant inhibitor of DNase I in CF sputum. Because the DNA and actin binding regions are distinct (Fig.  1), we reasoned that combining these mutations into one enzyme should result in variants that are both hyperactive and actin-resistant and thus even more potent. In the present work, we have characterized hyperactive, actin-resistant, and combination variants of human DNase I in several biologically relevant assays to determine their relative potency and pharmacodynamics in CF sputum and human serum. These results have significant implications for their therapeutic potential for treatment of both CF and SLE.

MATERIALS AND METHODS
Actin Inhibition Assay-Supercoiled pBR322 plasmid DNA (New England Biolabs) at 29 g/ml was treated with diluted culture media from human 293 cells transfected with wild type or variant DNase I (final DNase I concentration of 1 ng/ml as measured by a human DNase I ELISA (13,15)) preincubated with increasing amounts of G-actin for 24 min in the presence of 25 mM Hepes, pH 7, 100 g/ml bovine serum albumin, 1 mM MgCl 2 , and 2.5 mM CaCl 2 at room temperature. After 18 min of DNA digestion, the reaction mix was quenched with 25 mM EDTA, 6% glycerol, xylene cyanol, and bromphenol blue and loaded directly on a 0.8% agarose gel. The gel was run overnight at ϳ1 V/cm in TBE (90 mM Tris borate and 2 mM EDTA), stained with ethidium bromide, and digitized on a Molecular Dynamics model 575 FluorImager. Mock-transfected media displayed no background activity.
CF Sputum DNA Degradation-DNase I degradation of CF sputum DNA was measured by pulsed-field electrophoresis as described previously (15). The sputum collected from a single patient contained 2.24 mg/ml DNA as measured by a minor modification of the DABA method (16) calibrated to a salmon testes DNA standard. The concentration of total actin in this sputum was ϳ375 g/ml as determined by quantitative Western analysis (data not shown). This sputum was aliquoted, frozen for storage, and thawed immediately prior to treatment with 0.5 volume diluted DNase I culture fluid in isotonic buffer containing 25 mM Hepes-NaOH, pH 7.5, 1 mM MgCl 2 , 2.5 mM CaCl 2 , 1 mg/ml bovine serum albumin, and 150 mM NaCl at room temperature for 20 min. The reaction was quenched by adding 30 mM EDTA and 1.2% SDS, followed by treatment with 1 mg/ml proteinase K at 50°C for 2 h. An equal volume of low melting agarose was added to each sample and maintained at 37°C until loading onto a 1% agarose gel prior to pulsed-field electrophoresis. After the reaction mix plugs solidified in the wells, the gel was run for a 20-h cycle and stained in 0.5 g/ml ethidium bromide. Gels were scanned with a FluorImager, followed by quantitation of the intensities found in the 23,000 bp or greater region. The decrease in intensity in this higher molecular weight region over that of the control (no DNase added) is presented as the percent DNA hydrolyzed. Experiments were performed in duplicate. No detectable activity was found using the mock-transfected media.
[ 33 P]DNA Digestion-The [ 33 P]DNA digestion assay follows the extensive degradation of DNA in isotonic buffer or serum (17); [ 33 P]DNA is an attractive alternative to [ 32 P]DNA because of its longer half-life and lower energy emission. Double-stranded 33 P-labeled M13 DNA was adjusted to a constant specific radioactivity of 0.8 Ci/g by the addition of salmon testes DNA (Sigma). Five l of this mixture at 81 g/ml total DNA concentration was added to 90 l of buffer or human serum followed by addition of 5 l of diluted DNase I culture media at room temperature for 2 h. The reaction was stopped by sequential addition of 100 l of ice-cold 50 mM EDTA and 100 l of ice-cold 20% trichloroacetic acid, followed by centrifugation at 1230 ϫ g for 15 min at 4°C. Fifty l of the supernatant, which contains radiolabeled DNA fragments that are ϳ20 bp or less in size, was counted in a scintillation counter. Again, mock-transfected media had no background activity.
Degradation of Chromatin⅐Anti-DNA Immune Complexes in Serum-Preparation of human anti-DNA IgG from SLE patient sera complexed to human chromatin from normal human blood has been described (17). Briefly, 1 volume of the immune complex added to 8 volumes of normal human serum was digested by 1 volume of diluted DNase I transfection media at 37°C for 2 h; the reaction was quenched with 25 mM EDTA. Actin was present in the preparation of chromatin; the final concentration of total actin in this serum was ϳ13 g/ml as determined by quantitative Western analysis (data not shown). For the gel-based assay, the stopped reaction mix was treated with 2% SDS and 1.0 mg/ml proteinase K (Boehringer Mannheim) overnight at 50°C, followed by electrophoresis on 2.7% MetaPhor agarose (FMC)/0.2% SDS gels in TBE buffer. The gels were stained with 1.0 g/ml ethidium bromide and scanned with the FotoAnalyst system (Fotodyne). For the ELISA-based assay, the quenched reaction mix was diluted and added to a plate coated with anti-histone monoclonal antibody (Boehringer Mannheim), followed by detection with peroxidase-conjugated goat anti-human IgG, Fc-specific (17).
DNase I Digestion of Endogenous DNA in Serum-Extracellular DNA was extracted from normal human serum pool using the QIAamp 96 spin blood kit (Qiagen) with modifications in a 96-well format. The purified endogenous serum DNA was then characterized by quantitative PCR 2 using primers and a dual-labeled probe specific for the human Alu gene. Briefly, the fluorescence of one label is quenched by the other label (18). Degradation of the hybridization probe during PCR amplification produces a decrease in fluorescence quenching that is quantified in an Applied Biosystems model 7700 Sequence Detector. The PCR cycle where a fluorescence increase is detected (threshold cycle) is directly proportional to the concentration of target in the sample (19). In this study, using the Alu gene as the target, human DNA concentration was interpolated from a calibration curve generated with a human chromatin standard.

Actin-resistant and Hyperactive DNase I Variants-Previous
protein engineering of human DNase I at the actin-binding interface revealed several key positions involved in actin binding (15). In particular, single point mutations introducing charged, aliphatic, or aromatic residues at Ala 114 of the mature enzyme resulted in actin-resistant variants that had Ͼ10,000fold reduced affinity for G-actin (Fig. 1). In the present work, we chose to characterize the A114F mutant, which might potentially reduce any immunogenicity, because phenylalanine is found at position 114 in several other naturally occurring human DNase I-like enzymes, at least one of which has been shown to be actin-resistant (20 -22). We have further combined the A114F mutation with ones that make DNase I hyperactive to generate several combination variants (14), including Q9R/ E13R/N74K/A114F and Q9R/E13R/H44K/N74K/T205K/A114F, and tested them in an actin-binding ELISA (15) and a plasmidbased actin inhibition assay (Fig. 2). The DNA hydrolytic activity of wild type DNase I is inhibited by actin at concentrations as low as 3 nM, with complete inhibition observed at ϳ3 M. In contrast, the A114F and combination variants were insensitive to actin at concentrations as high as 3 M (Fig. 2), consistent with their lack of detectable actin binding in the ELISA (data not shown). In the absence of actin, the A114F variant showed similar activity and plasmid DNA digestion patterns as wild type DNase I; the lack of a linearized band in the A114F lane (Fig. 2) is not significant and only reflects a slightly lower amount of enzyme added relative to wild type. Both enzymes cleave DNA via a nicking mechanism as evidence by their low linear to relaxed product ratio (14). In contrast, the products of DNA scission by the combination variants are almost exclusively linear DNA, suggesting a fundamental shift in the functional mechanism. In the absence of NaCl, the overall nicking activity of the combination variants, as measured by the disappearance of the supercoiled substrate ( Fig. 2), are actually lower than wild type presumably because of reduced turnover number as a result of stronger DNA binding. However, in the presence of physiological saline, these variants are indeed more active than wild type, especially at low DNA concentrations and size (14). Consistent with the structural model for the additivity of these mutations (Fig. 1), we find that the combination variants are both actin-resistant and hyperactive.
Activity of DNase I Variants in CF Sputum-To evaluate the therapeutic potential of these three classes of DNase I variants, we first tested them in an ex vivo assay in CF sputum, that contained 2.24 mg/ml DNA and ϳ375 g/ml total actin. The degradation of the endogenous high molecular weight DNA in CF sputum as measured by pulsed-field gel electrophoresis showed that the hyperactive variant Q9R/E13R/N74K and the actin-resistant mutant A114F are each ϳ20-fold more potent than wild type, whereas the combination variant Q9R/E13R/ N74K/A114F is ϳ100-fold more active (Fig. 3).
Degradation of Exogenous DNA in Serum-To assess the ability of DNase I variants to degrade DNA in human serum, we characterized them using 4 g/ml of a mixture of 33 Plabeled M13 plasmid (7000 bp) and salmon DNA as substrate. In isotonic buffer, a DNase I variant with a single basic amino acid replacement, N74K, is 16-fold more active than wild type ( Fig. 4A and Table I). The degree of hyperactivity increases progressively with each additional positive charge up to the ϩ3 variant E13R/N74K/T205K, which is ϳ1300-fold more potent than the native enzyme. Further basic residue substitutions lead to variants with progressively less activity such that the level of hyperactivity for the ϩ6 variant is actually less than that of the ϩ2 variant, but still ϳ100-fold greater than that of the wild type (Fig. 4A). We attribute this decrease in hyperactivity of the ϩ4, ϩ5, and ϩ6 variants to a less efficient turnover of substrate as a result of binding to the DNA too tightly (14).
Although the ranking of hyperactive variants remains the same when the [ 33 P]DNA digestion assay is performed in serum, the differences in activity between wild type DNase I and the ϩ2 to ϩ6 variants are even larger than those found in buffer ( Fig. 4B and Table I). The -fold improvement over wild type for the optimal variant E13R/N74K/T205K is ϳ3800, almost three times that observed in buffer. The greater degree of hyperactivity found in serum is because of a greater degree of inhibition of native DNase I relative to the hyperactive variants. For example, the activity of the wild type nuclease in serum is 8.5-fold lower than in buffer, whereas that of the ϩ3 to ϩ6 variants are less than 3-fold lower (Table I).
In isotonic buffer, the actin-resistant A114F mutation had no effect on the activity of wild type or the hyperactive variant Q9R/E13R/N74K in the [ 33 P]DNA digestion assay ( Fig. 4C and Table I). Both the ϩ3 hyperactive and the ϩ3 combination mutants were ϳ500-fold more potent than either wild type or A114F DNase I. Components in serum inhibited the actinresistant and the ϩ3 combination variants by only 5.7-and 1.4-fold, respectively, which was 33 and 52% lower than the -fold inhibition observed for the wild type and the ϩ3 hyperactive variant, respectively (Table I). The effect of actin resistance on DNA scission activity in serum is minimal when comparing wild type and A114F DNase I; however, a significant enhancement of 2.4-fold for actin resistance was found in the background of the ϩ3 hyperactive variant. These results suggest the presence of relatively low levels of G-actin in human serum that can inhibit activity effectively only at lower DNase I concentrations (Fig. 4D).
Digestion of Chromatin⅐anti-DNA Immune Complexes in Serum-Chromatin DNA prepared from normal white blood cells was incubated with SLE serum-derived IgG in normal human serum to form immune complexes. DNase I degradation of these complexes, which have a final DNA concentration of 120 g/ml and a total actin concentration of ϳ13 g/ml, was assessed by an agarose gel-based assay and an ELISA. As shown in Fig. 5, most of the immune complex DNA is Ͼ1,000 bp in size prior to any DNase I treatment. Pulsed-field electrophoresis with higher molecular weight standards revealed that most of the DNA in the immune complexes was 7000 -23,000 bp in length (data not shown). Reaction with either wild type or the ϩ3 hyperactive variant E13R/N74K/T205K converts the high molecular weight substrates to fragments of ϳ180 and ϳ360 bp in length, consistent with cleavage at the linker region of chromatin, which has a nucleosome repeat length of ϳ190 bp and ϳ150 bp of core region (23); intermediate digestion products are also consistent with cleavage between nucleosomes (Fig. 5). DNA scission by the native enzyme is undetectable up to ϳ3 ng/ml with total digestion occurring at 1 g/ml DNase I concentration. The ϩ3 variant E13R/N74K/T205K is ϳ100-fold more potent than wild type, completely degrading DNA at ϳ10 ng/ml.
Degradation of immune complexes in serum was also determined by ELISA, which yields similar results to those of the gel-based assay. For example, the wild type enzyme has no activity at 3 ng/ml and 100% activity at 1 g/ml (Fig. 6A); the EC 50 for DNA scission determined from the ELISA is 32 ng/ml ( Table I). The EC 50 of the ϩ1 variant N74K is ϳ10-fold lower than wild type, whereas that of ϩ2 variant is 400-fold lower. Further addition of positively charged residues decrease the potency so that the ϩ6 variant is only 35-fold more active than the native DNase I. The actin-resistant variant A114F is 20fold more active than wild type, consistent with the detection of actin in the immune complex preparation. Furthermore, the actin resistance and the hyperactivity in the combination variant Q9R/E13R/N74K/A114F results in the most active enzyme, having ϳ650-fold increased potency relative to wild type ( Fig. 6B; Table I).
Digestion of Endogenous Serum DNA-To directly measure DNase I activities in human serum, we applied a Taqman Alu PCR assay to quantitate DNA concentrations. The use of this ultrasensitive assay is necessitated by the extremely low levels of endogenous serum DNA, which range from 25 to 250 ng/ml 2 ; these values are in good agreement with those determined by counterimmunoelectrophoresis (24, 25). We have targeted the Alu family of repetitive sequences, because it is the most abundant repetitive DNA in human genome, having 300 -500 thousand copies per genome (26). Unlike the two previous serum assays (Figs. 4B and 6A), the level of hyperactivity increased progressively with each additional basic amino acid substitution, with the ϩ6 variant being close to 30,000-fold more potent than wild type for digesting endogenous serum DNA (Fig. 7A). The actin-resistant mutation A114F improved the potency of both wild type DNase I and the two ϩ3 variants by ϳ5-fold ( Fig. 7B; Table I), implicating the presence of G-actin in this serum. DISCUSSION We have assessed the relative potency and therapeutic potential of three classes of human DNase I variants by testing them in several biologically relevant assays. For DNA scission in CF sputum, where the DNA and actin concentrations are relatively high, an increase in potency over wild type of ϳ20fold was found for either the hyperactive variant Q9R/E13R/ N74K or the actin-resistant variant A114F; the combination mutant Q9R/E13R/N74K/A114F was even more potent, resulting in an improvement of ϳ100-fold. The increase in potency by the hyperactive variants was much greater for degrading DNA in human serum, although the degree of hyperactivity was highly dependent upon the assay conditions. For the digestion of endogenous serum DNA present at extremely low concentrations, the degree of hyperactivity climbs with each additional positive charge, culminating with the ϩ6 variant being ϳ30,000-fold more potent than wild type. Actin resistance eliminates any inhibition by endogenous G-actin under these conditions, making the ϩ3 variants even more active.
Hyperactivity as a Function of DNA Concentration-We previously characterized a series of hyperactive DNase I variants having from one to six additional positively charged residues relative to wild type in several in vitro assays (14). The level of hyperactivity was found to be inversely proportional to DNA concentration. For example, the ϩ2 variant E13R/N74K was ϳ20-fold more active than wild type for nicking 10 ng/ml of linear plasmid DNA, but was ϳ3-fold less active when the substrate concentration was elevated to 30 g/ml. Our present ex vivo data in human serum and CF sputum is entirely consistent with the tendency of a greater degree of hyperactivity associated with lower DNA concentration. The DNA concentra- tion in CF sputum is typically very high and is correlated with the severity of pulmonary disease; we have measured DNA concentrations as high as 25 mg/ml in CF sputum (data not shown). For the particular CF sputum sample we tested, which had a DNA concentration of 2.24 mg/ml, the ϩ3 variant Q9R/ E13R/N74K was ϳ20-fold more active than wild type. Variants with four or more additional positively charged residues than wild type are not expected to further enhance the potency for CF sputum DNA digestion, since their degree of hyperactivity is actually lower than the ϩ3 variant in assays containing relatively high DNA concentrations, presumably because of a reduction of the turnover rate as a result of binding to the DNA too tightly (14).
The measured hyperactivities in the three serum-related assays are much greater than those determined in the CF sputum DNA digestion assay as a result of much lower DNA  The number of additional positively charged residues relative to wild type DNase I are indicated in parentheses. The [ 33 P]DNA digestion assay was carried out in isotonic buffer containing 150 mM NaCl or in normal human serum. Values in the first two columns are reported as -fold improvements over wild type which has a buffer EC 50 of 130 Ϯ 3 ng/ml and a serum EC 50 of 1100 Ϯ 64 ng/ml. The third column lists the -fold inhibition observed in serum as compared with buffer. The final two columns tabulate the -fold enhancements over wild type, which has an EC 50 of 32 Ϯ 1 ng/ml for degrading chromatin/immune complex in serum and an EC 80 of 1700 Ϯ 50 ng/ml for digesting endogenous DNA in serum, as determined by the Taqman assay. For the endogenous serum DNA assay, we reported the EC 80 instead of EC 50 , because we could not accurately determine the activity at low variant concentrations (Fig. 7). concentrations used. The endogenous DNA concentration in human serum is extremely low, ranging from 25 to 250 ng/ml; the particular serum pool we tested was ϳ180 ng/ml DNA. Consequently, the degree of hyperactivity in the endogenous serum DNA digestion assay is much greater than that found in the CF sputum DNA digestion assay (Table I); a maximal improvement of 30,000-fold relative to wild type was found for the ϩ6 variant. This correlation between the degree of hyperactivity and the number of engineered positive charges on DNase I was also found in a DNA nicking assay using a 32-bp DNA fragment at 6.3 ng/ml as substrate under isotonic conditions in the presence of Mg 2ϩ and Ca 2ϩ ions (14). The maximal -fold improvement over wild type in the in vitro assay was ϳ6000-fold for the ϩ6 variant. The larger -fold enhancement in the ex vivo versus the in vitro assay could be because of the presence of serum DNA binding proteins (27), which might inhibit the activity of wild type more than that of the hyperactive variants; this is consistent with the higher degree of inhibition seen for wild type in the [ 33 P]DNA digestion assay in serum relative to buffer (Fig. 4). It is possible that these DNA binding proteins could sequester some of the 33 P substrate, accounting for the observation that the activity appears to level off at ϳ60% in serum (Fig. 4, B and D). The DNA concentrations in the chromatin-immune complex degradation assay and the [ 33 P]DNA cleavage assay were 120 and 4 g/ml, respectively, intermediate between the mg/ml range found for the CF sputum DNA digestion assay and the ng/ml range for the endogenous serum DNA digestion assay. Not surprisingly, we also found intermediate maximal enhancements of hundredsto thousands-fold over wild type (Figs. 4B and 6A), again consistent with the inverse correlation of the degree of hyperactivity with the DNA concentration.
Hyperactivity as a Function of DNA Length-Besides being inversely dependent on substrate concentration, the degree of hyperactivity also has an inverse correlation with DNA length (14). For example, the ϩ2 variant is ϳ20-fold more active than wild type when nicking the 4361-bp linear plasmid pBR322, but ϳ200-fold more potent when the substrate was a 32-bp DNA fragment. In the present study, the larger difference between the hyperactive variants and the native DNase I measured in the [ 33 P]DNA cleavage assay as compared with that found in the immune complex assay may be explained by the difference in DNA size. In the former assay, the M13 plasmid used is 7000 bp in length, whereas the chromatin DNA used in the latter assay is mostly 7000 -23,000 bp in size. In addition, the former only measures extensive DNA degradation down to 20 bp or less, whereas the latter is likely following digestion down to the size of nucleosome core, at ϳ150 bp. This association of greater hyperactivity with shorter DNA length supports the proposed processive DNA nicking mechanism attributed to the hyperactive variants (14). As the hyperactive DNase I processively digests DNA, it can slide off the ends of a smaller DNA fragment faster than a longer one, thus elevating the turnover rate. This shift in the functional mechanism of DNA cleavage could also explain the thousands-fold maximal improvements observed in the [ 33 P]DNA digestion assay as compared with Ͻ10-fold maximal enhancements found in the plasmid nicking assay (13), both performed in isotonic buffer and in the presence of relatively similar DNA concentrations and substrate sizes. The difference between these two assays is that although the former follows extensive DNA digestion, the latter measures the rate of the initial DNA nick. Consequently, the expected faster generation of smaller molecular weight DNA fragments by processive DNA nicking with a hyperactive variant versus nonprocessive DNA nicking with native DNase I would result in a greater degree of hyperactivity determined in the [ 33 P]DNA digestion assay.
Therapeutic Potential-Based on our ex vivo experiments, we suggest that the engineered variants of human DNase I could significantly increase the efficacy for the treatment of CF. Since the reduction of viscoelasticity of the CF sputum is likely because of moderate DNA degradation down to kilobase in size, a processive DNA nicking mechanism may not be such a dramatic advantage. However, the additional positively charged residues in the hyperactive variants also result in greater resistance to inhibition by physiological saline and perhaps endogenous DNA binding proteins present in CF sputum. The more or less additive effects of hyperactivity and actin resistance leads to a DNase I variant with ϳ100-fold enhancement in potency relative to wild type, which could prove valuable for CF therapy.
The hyperactive class of DNase I variants also has potential for treatment of SLE, where a more dramatic impact on DNA degradation in serum was found. In both normal and SLE sera, the DNA concentrations are very low 2 . Furthermore, it may be necessary to reduce the DNA size down to less than 10 -20 bp to prevent immune complex formation or destroy existing ones. The hyperactive variants are improved relative to wild type in addressing both of these properties, demonstrating an increase in potency of three to four orders of magnitude. DNase I has shown beneficial effects in a murine model of lupus nephritis, albeit at relatively high doses of 7.5 mg/kg/day (10). The observed increase in potency for the most active variants could overcome any dosage limitation of wild type DNase I and show improved efficacy.
The effect of actin-resistance was observed in the CF sputum and some of the serum assays; however, these results should be taken with appropriate caution because of the inherent difficulty of measuring the precise level of G-actin, which is the form of actin that inhibits DNase I. Complicating factors in a biological medium include the equilibrium between F-and G-actin, the presence of actin-binding proteins, and potential artifacts resulting from either cell lysis or proteolysis of actin during the course of the assays. Nonetheless, in the event that G-actin is present, the actin-resistant variants are indeed more potent, as we found in the CF sputum, immune complex, and endogenous serum DNA assays. In the [ 33 P]DNA digestion assay (Fig. 4B), the effect of actin resistance was somewhat attenuated. We attribute these differences to varying concentrations of G-actin in serum for the abovementioned reasons. We have measured total actin levels as high as 125 g/ml in normal human serum 3 ; values ranging between 30 and 50 g/ml in human plasma have been reported (28). Increased levels of an inhibitor of DNase I in human SLE serum, presumably G-actin, have also been reported (29). Finally, actin concentrations could be much higher locally either in CF sputum or in the glomerulus of lupus patients because of increased cell lysis.
The degree of increased activity for the DNase I variants relative to wild type that we have presented is quite substantial. However, extrapolation of ex vivo data to the in vivo environment is inherently speculative and should be taken with caution; in particular, the immunogenicity profile of our DNase I variants has yet to be investigated but could foreseeably pose some limitations to their applicability in the clinic. Nonetheless, the results presented herein suggest that the DNase I variants we have engineered for improved biochemical and pharmacodynamic properties may have improved clinical benefits as well.