The Bulk of Autotaxin Activity Is Dispensable for Adult Mouse Life

Autotaxin (ATX, Enpp2) is a secreted lysophospholipase D catalysing the production of lysophosphatidic acid, a pleiotropic growth factor-like lysophospholipid. Increased ATX expression has been detected in a number of chronic inflammatory diseases and different types of cancer, while genetic interventions have proven a role for ATX in disease pathogenesis. Therefore, ATX has emerged as a potential drug target and a large number of ATX inhibitors have been developed exhibiting promising therapeutic potential. However, the embryonic lethality of ATX null mice and the ubiquitous expression of ATX and LPA receptors in adult life question the suitability of ATX as a drug target. Here we show that inducible, ubiquitous genetic deletion of ATX in adult mice, as well as long-term potent pharmacologic inhibition, are well tolerated, alleviating potential toxicity concerns of ATX therapeutic targeting.


Introduction
Autotaxin (ATX, Enpp2) is a secreted lysophospholipase D [1] widely present in biological fluids including blood [2][3][4].ATX is the major enzyme catalysing the production of lysophosphatidic acid (LPA) and ATX heterozygous null mice have half of normal LPA levels [5][6][7].LPA is a bioactive, growth factor-like lysophospholipid promoting a large variety of cellular responses in almost all cell types [2,8], mediated from at least six G-protein coupled LPA receptors (LPARs) that exhibit overlapping specificities and widespread distribution [9,10].
ATX expression and LPA production were shown to be necessary for embryonic development, as ubiquitous genetic deletion of ATX resulted to vascular and neuronal defects leading to lethality [5][6][7].In adult life ATX is widely expressed, and upregulated levels have been reported in various chronic inflammatory diseases and cancer [2,11,12].Conditional genetic deletion of ATX attenuated the development of rheumatoid arthritis and pulmonary fibrosis in mouse models [13,14], suggesting a primary role in disease pathogenesis [15,16].Transgenic mammary ATX overexpression resulted in enhanced rates of spontaneous breast cancer development in aged mice, often preceded by chronic inflammation [17].Therefore, ATX overexpression at chronic inflamed or neoplastic tissues, possibly localized to the cell surface through integrin-binding [18], most likely results to increased local levels of LPA, which in turn exert its pathologic effects at the microenvironment [2].
Consequently, ATX has emerged as a promising therapeutic target in chronic inflammatory diseases and cancer [2] and a large number of ATX inhibitors have been developed [19,20].Moreover, the crystal structure of ATX was recently solved [18,21] enabling rational drug design.In this report we show that inducible, ubiquitous genetic deletion of ATX in adult mice, as well as its potent pharmacologic inhibition, are well tolerated, alleviating potential toxicity concerns of ATX therapeutic targeting.

Materials and Methods Mice
All mice were bred at the animal facilities of the Alexander Fleming Biomedical Sciences Research Center, under specific pathogen-free conditions.Mice were housed at 20-22°C, 55 ±5% humidity, and a 12-h light-dark cycle; water and food were provided ad libitum.Mice were bred and maintained in a C57BL/6 genetic background for more than 10 generations.All experimentation in mice for this project was approved by the Institutional Animal Ethical Committee (IAEC) of Biomedical Sciences Research Center "Alexander Fleming" (#376), as well as the Veterinary service and Fishery Department of the local governmental prefecture (#5365).Age and sex-matched littermate mice were assigned randomly to experimental and control groups.All measures were taken to minimize animal suffering and distress; however, neither anesthetics nor analgesics were used during the protocol, as no invasive or painful techniques were performed.The health status of the mice was monitored once per day and no unexpected deaths were observed.Mice were euthanized at predetermined time-points or when they exhibited a <20% body weight loss.Euthanasia was performed in a CO 2 chamber with gradual filling followed by exsanguination.The generation and genotyping protocols for Enpp2 n/n [5], R26Cre-ER T2 [22,23], and R26R [24] mice have been described previously.

Tissue processing and H&E staining
Fresh mouse tissues were fixed in 10% neutral-buffered formalin overnight at 4°C, processed using a standard overnight protocol, and embedded in paraffin blocks.Sections of 4 μm were stained with Hematoxylin and Eosin (H&E) according to standard procedures and imaged using a Nikon Eclipse E800 microscope (Nikon Corp., Shinagawa-ku, Japan) attached to a Q Imaging EXI Aqua digital camera, using the Q-Capture Pro 7 software.Images were taken and evaluated blindly.

X-gal staining
Freshly isolated mouse tissues were embedded in OCT and frozen in liquid nitrogen.Sections of 6 μm were prepared on a cryotome and fixed in 2% formaldehyde/ 0.2% glutaraldehyde for 10 minutes at 4°C, washed twice in cold PBS /2 mM MgCl 2 for 10 minutes and stained in X-gal staining solution (2 mg/ml X-gal (5-bromo-a-chloro-3-inodyl-β-D-galactopyranoside) in 0.1 M Na phosphate buffer pH 7.3, 0.01% Na deoxycholate, 5 mM K3Fe(CN)6, 5.7 mM K4Fe(CN) 6, 2 mM MgCl 2 , 0.02% NP-40) at 37°C in the dark overnight.The sections were rinsed twice in PBS /2 mM MgCl 2 for 10 minutes at room temperature, counterstained with eosin, and visualized under an Eclipse E800 microscope as in the previous section.

ATX activity assay
ATX / LysoPLD activity was measured using the TOOS activity assay.ATX catalyzes the cleavage of lysophosphatidylcholine to lysophosphatidic acid and choline.The released choline is oxidised by choline oxidase to produce betaine and hydrogen peroxide.The latter is used as the oxidizing agent.In the presence of horseradish peroxidase, H 2 O 2 reacts with TOOS (N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3-methylaniline) and 4-AAP (aminoantipyrene) to form a pink quinoneimine dye which absorbs at 555 nm. 1 x LysoPLD buffer (100 mM Tris-HCl pH 9.0, 500 mM NaCl, 5 mM MgCl 2 , 5 mM CaCl 2 , 60 μM CoCl 2 , 1 mM LPC) was incubated at 37°C for 30 minutes.Plasma samples (100 fold diluted) were incubated with 1 x LysoPLD buffer at 37°C for 4 hours at a final volume of 100 μl in a 96-well plate.At the end of the incubation, a colour mix (0.5 mM 4-AAP, 7.95 U/ml HRP, 0.3 mM TOOS, 2 U/ml choline oxidase in 5 mM MgCl 2 / 50 mM Tris-HCl pH 8.0) was prepared and 100 μl were added to each well.Absorbance (A) was measured at 555 nm every 5 minutes for 20 minutes.For each sample, the absorbance was plotted against time and the slope (dA/min) was calculated for the linear (steady-state) portion of each reaction.ATX activity was calculated according to the following equation: Activity (U/ ml) = (μmol/ min/ ml) = [dA/ min (sample)-dA/ min (blank)] Ã Vt/ (e Ã Vs Ã 0.5) where Vt: total volume of reaction (ml), Vs: volume of sample (ml), e: milimolar extinction coefficient of quinoneimine dye under the assay conditions (e = 32,8 μmol/ cm 2 ) and 0.5: the moles of quinoneimine dye produced by 1 mol of H 2 O 2 .

Western blotting
1 μl of plasma samples were diluted in Laemmli buffer, electrophorised on 8% SDS-polyacrylamide gels and transferred to Protran nitrocellulose membranes (Whatman plc, Maidstone, Kent, UK) using the Trans-Blot SD Semi-Dry Transfer system (Bio-Rad Laboratories, Inc., Hercules, CA, USA).Primary a-ATX Ab incubation (rat monoclonal 4F1, kindly provided by J. Aoki, 1:1000; rabbit polyclonal Cayman 1:250) was performed overnight in 5% (w/v) non-fat milk in TBS-Tween 0.05% (TBST) at 4°C.The membranes were then washed three times with TBST and incubated with HRP-conjugated secondary Abs (a-rat or a-rabbit respectively; 1:2000) for one hour at room temperature.Membranes were washed three times with TBST and antibody-antigen complexes were revealed using ECL chemiluminescent reagent (Pierce/ Thermo Scientific, Rockford, IL, USA).Antibody specificity has been scrutinized previously [13].
Plasma biochemical and haematological analyses were performed with the Abbott Architect 8200 and Cell-Dyn 3700 analysers respectively.

Mouse pharmacokinetics and in vivo inhibition of ATX activity
PF8380 was administered by oral gavage (120 mg/kg) dissolved in vehicle (Hydroxypropyl Cellulose 2% / Tween 80 0.1%) to 8 week old mice.Control mice received only the vehicle.Venus blood was collected at time points up to 12 hours post administration, in tubes containing EDTA at a final concentration of 50 mM.For plasma preparation, whole blood was centrifuged at 2000 g for 20 minutes at 4°C and the supernatant was stored in siliconized tubes at -20°C.Plasma samples were examined for ATX activity and PF8380/ LPA levels using the TOOS activity assay and HPLC-MS/MS, respectively.

Statistical analysis
All data are expressed as means ± SEMs.Statistical analysis was performed using SigmaPlot 11.0 (Systat software Inc., IL, USA).The statistical significance was estimated in pair-wise comparisons with control values using a paired Student's t-test, or a Mann-Whitney test in cases of not normal distributions.Survival curves were calculated using Kaplan-Meier survival analysis; the statistical significance of curves was assessed using Logrank test.P values <0.05 ( Ã ), p<0.01 ( ÃÃ ) and p<0.001 ( ÃÃÃ ) were considered significant.The number of samples (n) and the number of experiments/repetitions (exp) are indicated in each figure legend.
In order to confirm efficiency and tissue specificity of Tmx-induced Cre-mediated recombination, R26Cre-ER T2 mice were crossed with mice carrying a Cre-responsive b-galactosidase (LacZ) reporter allele [24].In this reporter strain (R26R), a lacZ transgene has been introduced into the R26 locus, preceded by a floxed neo expression cassette preventing lacZ expression.Upon Cre recombinase activation, this "stop" neo cassette is removed and lacZ expression can be detected using X-gal staining [24].Intraperitoneal (IP) injections of Tmx (100 mg/kg per day for 10 days) in R26Cre-ER T2 /R26R mice resulted in varying degrees of X-gal staining in almost all tissues examined (S1B Fig), as previously reported for similar strains of mice [26][27][28], suggesting efficient and widespread Tmx-induced, Cre-mediated gene targeting.Moreover, no X-gal staining was detected in the absence of Tmx or the R26Cre-ER T2 allele (S1B Fig) , indicating stringent control of gene expression.
For the inducible inactivation of ATX in mice in vivo, Enpp2 n/n mice [5] were crossed with the R26Cre-ER T2 mice yielding R26Cre-ER T2 /Enpp2 n/n mice, which received different concentrations of Tmx via different routes of administration.Administration of 50 mg/kg Tmx IP per day for 10 days was not sufficient to induce recombination (S1 Table ) and had no effect in overall mouse survival rate (Fig 1A ).Increasing the Tmx dose to 100 mg/kg (IP; per day for 10 days) resulted in accurate recombination of the Enpp2 allele, which could only be detected in the presence of the R26Cre-ER T2 allele (S1 Table ).R26Cre-ER T2 -mediated recombination was also detected in littermate R26Cre-ER T2 /Enpp2 n/n mice not receiving Tmx but housed in the same cages (see below and S1 Table) [30].As genomic PCR is not a quantitative assay, the results from the reporter mice (S1 Fig) suggest minimal levels of leaky recombination.
No statistically significant effect in the overall survival rate of mice was observed (Fig 1B).Comparative histopathology of 44 tissues from R26Cre-ER T2 /Enpp2 n/n mice 2 days post Tmx administration (100 mg/kg IP per day for 10 days) at the onset of the limited observed lethality did not identify any significant findings after deletion of ATX, with the exception of lung and gastrointestinal (GI) tract which had mild to moderate inflammatory changes (S2 Table and S2 Fig) and which were followed through the entire study.However, similar histopathological changes were also observed in all control groups receiving Tmx (S2 Table and S2

The majority of ATX activity is not necessary for adult life
To confirm results utilizing a different and more widely used route of administration, Tmx (180 mg/kg) was administered to R26Cre-ER T2 /Enpp2 n/n and control littermate mice with oral gavage (Per Os; PO) for 4 days, followed by a 3-day arrest and Tmx administration was continued for 2 more days (Fig 1D).Similar to the other dosing regimens reported here, Tmxinduced recombination in all tissues could be detected only in the presence of the R26Cre-ER T2 allele, while its minimal leakiness could be avoided with individually housing mice (S1 Table ).No statistically significant effect in the overall survival rate of Tmx-administered R26Cre-ER T2 /Enpp2 n/n mice was observed in comparison to littermate Tmx-administered R26Cre-ER T2 mice, suggesting that the majority of ATX expression is dispensable for adult life and that the observed lethality could be attributed to Tmx-and Cre-mediated toxicity as previously reported [31].No histopathological changes were detected in the lungs and GI tract of all experimental and control mice 18  Based on limited published expression studies, ATX is thought to be widely expressed in adult mice, with the highest mRNA levels reported in brain, ovary/testis, intestine, kidney, and lung [2].Moreover, ATX has been reported to be secreted from adipocytes [32,33], while adipose specific genetic deletion of ATX resulted in marked decreases of serum LPA [34], suggesting the adipose tissue as a major contributor to serum ATX.Real Time RT-PCR analysis of ATX mRNA expression confirmed the brain and adipose (gonadal white adipose tissue, WAT; brown adipose tissue, BAT) tissues as the highest ATX expression tissues in wt C57BL/6 mice (Fig 2A ), in general agreement with on line expression data repositories (http://biogps.org/#goto = genereport&id=18606, http://www.gtexportal.org/home/gene/ENPP2).Tmx-induced, R26Cre-ER T2 -mediated Enpp2 recombination resulted in >80% attenuation of ATX mRNA levels in all organs examined including brain and adipose tissue ( A 50% reduction in plasma ATX (and LPA) levels in heterozygous ATX KO mice was previously reported to result in no overt phenotype [5][6][7].Our results extend these observations proving that even an 80% reduction in ATX activity and LPA levels in adult mice has no effect on tissue histopathology and overall survival.However, we cannot exclude the possibility that residual total LPA may still be enough to maintain normal body functions, although the reported LPA concentration thresholds to observe any effect in any cell type in vitro is much higher (>>1 μM up to 10 μM) [2].In fact, most reported effects of LPA in vitro were observed at non-physiological concentrations corresponding to pathophysiological situations of increased local concentrations of LPA at inflamed/malignant sites, most likely as a result of increased expression of ATX.Moreover, it should be noted that cellular effects of LPA will depend not only on its local concentration (controlled by the relative levels of both ATX and LPPs), but as well as on the local cell-specific abundance of LPA receptors and their suggested transactivation by other factors [35] or the local concentration of interacting proteins that have been reported to modulate its biological effects [36].Additionally and depending on the concentration, LPA may also elicit cellular responses not only through specific binding to cognate receptors, but as well as through the activation of nuclear receptors [37], through the modulation of macromolecular signalling complexes via direct docking to proteins or the metabolic rewiring of cell metabolism [38].

Potent pharmacological ATX inhibition exhibits no toxic effects
Upregulated expression of ATX has been detected in a number of chronic inflammatory diseases and cancer, while genetic interventions have proven a role of ATX in disease pathogenesis; establishing ATX as a promising potential therapeutic target [2].However, the embryonic lethality of ATX knockout mice poses a concern for the suitability of ATX as a drug target.
Inducible, ubiquitous genetic deletion of ATX indicated that the majority of ATX expression is dispensable for adult life, suggesting potential safety of pharmaceutical targeting.To validate the conclusion, mice were treated with the potent ATX inhibitor 6-(3-(piperazin-1-yl)propanoyl)-benzo[d]oxazol-2(3H)-one (PF8380) [39].PF8380 is small molecule with a long linear and flexible structure, good oral bioavailability, and a reported IC50 of 1.7 nM on natural LPC substrates [39].Oral administration of 30 mg/kg PF8380 reduced inflammatory hyperalgesia in a rat air pouch model, exhibiting >95% reduction of LPA levels in both plasma and

Conclusions
Given the extensive steady state reduction of LPA levels upon inducible genetic deletion of ATX and the pharmacokinetics (PK) and pharmacodynamics (PD) profiles of existing potent ATX inhibitors, it is highly unlikely to achieve experimentally a further, 24 hours-long, reduction in LPA levels by combining genetic and pharmacologic approaches.However, such a hypothetical 100% constant ATX inhibition would not only prove that the entire ATX expression and activity (as opposed to 80% shown here) is dispensable for adult life, but would also indicate the relative contribution to serum LPA levels from alternative synthetic pathways possibly mediated by phospholipase A-type enzymes as previously suggested [42].
As the important role of ATX in chronic inflammation and cancer is emerging, a large number of ATX inhibitors have been developed over the last years [2,20], while several of them have exhibited therapeutic potential: in air-pouch inflammation [39], collagen-induced arthritis [43], bleomycin-induced pulmonary inflammation and fibrosis [14], allergen-induced asthma [44], as well as in the metastasis of melanoma and breast cancer cells [41,[45][46][47][48].Moreover, ATX inhibition has been proposed as a potential adjuvant therapy to known cancer treatments, given its suggested role in chemotherapy and radiotherapy resistance [49,50].Noteworthy, the therapeutic effects of inhibiting ATX in most reported pharmacological studies did not always correlate with the PK/PD profile of the compound, when available, and LPA levels were never suppressed >80% for more than three hours.Although the tissue penetration and retention of the compounds and tissue LPA levels post treatment are largely unknown, it's safe to conclude that even the most potent ATX inhibitors that do exhibit therapeutic potential cannot attenuate LPA levels >80% at a steady state for 24 hours.On the other hand the present study has shown that even a constant 80% reduction of LPA is well tolerated systemically.Therefore, ATX inhibition can exert its therapeutic benefits by inhibiting, most likely, the increased amounts of ATX at inflamed and malignant sites locally, without exhibiting systemic, toxic side effects.
Fig), while observed pathologies did not persist 20 days post Tmx administration (S3A Fig), indicating that the observed transient pathologies in the lungs and GI tract were due to Tmx administration and not ATX ablation.To lower Tmx-induced toxicity experiments were repeated using 5 days of IP injections of 100 mg/kg Tmx with similar results.Enpp2 recombination was readily detected (S1 Table), with no statistically significant effects in mice survival rates in comparison to control mice (Fig 1C) and exhibiting no histopathological changes in lung and GI tract 10 and 20 days post Tmx treatment (S3B Fig).
and 30 days post treatment (S4 Fig).Moreover, no deleterious changes indicative of liver, kidney, or pancreatic function were identified in clinical biochemistry analytes from Tmx-administered R26Cre-ER T2 /Enpp2 n/n mice (S5 Fig).Likewise, no differences were detected in blood cell populations (S6 Fig).Therefore, ATX ablation appears to have no major effects in tissue homeostasis and the hematopoietic system.

Fig 1 .
Fig 1.No effect of Tmx-induced genetic inactivation of ATX in mouse survival rate.Kaplan-Meier survival curves of R26Cre-ER T2 /Enpp2 n/n and control mice administered with different Tmx concentrations (50, 100 and 180 mg/kg) using two different routes of delivery (IP, PO), as indicated and as described in the text.Presented results are cumulative from 1 (A), 3 (B) and 4 (C and D) experiments.No statistical significant differences were observed, as assessed with the Logrank test.Overlapping curves are indicated by consecutive symbols.doi:10.1371/journal.pone.0143083.g001 Fig 2B), as quantified with Real-Time RT-PCR.As a consequence, ATX protein was barely detectable in the plasma of Tmx-administered R26Cre-ER T2 /Enpp2 n/n mice (Fig 2C and S7 Fig), resulting to an 80% decrease of plasma ATX activity levels (Fig 2D), as quantified with the TOOS assay on natural LPC substrates.ATX activity levels were followed by a similar decrease in LPA levels (Fig 2E) despite the unaffected levels of possible ATX substrates (Fig 2F), as quantified by HPLC-MS/MS.

Fig 2 .
Fig 2. Tmx-induced (180 mg/kg PO) R26Cre-ER T2 -mediated genetic ablation of ATX results in diminished ATX levels in tissues and plasma.(A) Real-Time RT-PCR analysis of relative ATX mRNA expression levels in different tissues normalized to the expression levels of B2M.(n = 5-10; exp = 2; with the exception of BAT/WAT n = 4, exp = 1).(B) Real-Time RT-PCR analysis of ATX mRNA expression levels, normalized to the expression levels of B2M, in different tissues following Tmx-induced genetic ablation of the Enpp2 gene.(n = 5-10; exp = 2; with the exception of BAT/WAT n = 4, exp = 1).(C) Section from a western blot for ATX (4F1 Ab) in the plasma of the indicated mice.The full images, together with a coomassie brilliant blue staining of the same samples as loading control, and an alternate blot with a commercial antibody can be found at S7 Fig (D) Plasma ATX activity in the plasma of the indicated

Fig 3 .
Fig 3. PF8380 inhibits ATX activity in vitro and in vivo.(A) Dose response ATX activity inhibition curve under various PF8380 concentrations as measured with the TOOS assay using 50 μM of 16:0 LPC as substrate.(B) Plasma PF8380 pharmacokinetic profile following administration of 120 mg/kg PF8380 for 3 weeks, by oral gavage, in female mice.(C) No effect in mouse body weight following treatment with PF8380 as above.(D) Percent (%) residual plasma ATX activity measured in the presence of 1 mM 16:0 LPC with the TOOS assay at various time points post PF8380 administration as above.(E) Percent (%) residual plasma total LPA levels and (F) concentration of different LPA species at different time points post treatment with PF8380 as above.The cumulative values of two independent experiments (n = 4-7) are presented as means (± SEMs).doi:10.1371/journal.pone.0143083.g003