Characterization of an androgen-responsive, ornithine decarboxylase-related protein in mouse kidney

We have investigated and characterized a novel ornithine decarboxylase (ODC) related protein (ODCrp) also annotated as gm853. ODCrp shows 41% amino acid sequence identity with ODC and 38% with ODC antizyme inhibitor 1 (AZIN1). The Odcrp gene is selectively expressed in the epithelium of proximal tubuli of mouse kidney with higher expression in males than in females. Like Odc in mouse kidney, Odcrp is also androgen responsive with androgen receptor (AR)-binding loci within its regulatory region. ODCrp forms homodimers but does not heterodimerize with ODC. Although ODCrp contains 20 amino acid residues known to be necessary for the catalytic activity of ODC, no decarboxylase activity could be found with ornithine, lysine or arginine as substrates. ODCrp does not function as an AZIN, as it neither binds ODC antizyme 1 (OAZ1) nor prevents OAZ-mediated inactivation and degradation of ODC. ODCrp itself is degraded via ubiquination and mutation of Cys363 (corresponding to Cys360 of ODC) appears to destabilize the protein. Evidence for a function of ODCrp was found in ODC assays on lysates from transfected Cos-7 cells where ODCrp repressed the activity of endogenous ODC while Cys363Ala mutated ODCrp increased the enzymatic activity of endogenous ODC.


Introduction
Polyamines are small ubiquitous aliphatic polycations involved in or are essential for fundamental cellular processes and events ranging from cell growth and proliferation to synthesis, function, and stability of macromolecules. Elevated polyamine levels have been linked to tumorigenesis. Thus, the intracellular polyamine concentration is tightly regulated at the levels of synthesis, catabolism, uptake, and excretion. Ornithine decarboxylase (ODC, EC 1.1.1.17) is the first and rate-limiting enzyme in the polyamine synthesis pathway [1][2][3].

Experimental animals
ICR mice (10-to 12-weeks old) were used. Intact males, orchiectomized males, and female mice were injected subcutaneously with testosterone (T, 1 mg/mouse/day in 0.1 ml mineral oil) or vehicle. Orchiectomized males received T four days after the operation. Gene expression analyses were performed with mice treated with T for 3 days. ChIP assays were performed 2 h after a single T injection. All animal experiments were approved by Finnish Review Board for Animal Experiments and performed according to the guidelines for animal experiments at the University of Helsinki (permit number ESLH-2008-09035/Ym23). The mice were killed by carbon dioxide inhalation and different organs were either snap-frozen in liquid nitrogen for RNA isolation or fixed in 10% buffered formalin (Sigma-Aldrich) and embedded in paraffin for immunohistochemistry.

ChIP and ChIP-sequencing
Minced fresh mouse tissues were cross-linked in 1% formaldehyde (Merck KGaA, Darmstadt, Germany) at room temperature for 20 min. After washing twice with ice-cold PBS, tissues were homogenized in hexylene glycol buffer to isolate a crude nuclear fraction [25]. Sonication (Sonicator 3000, Misonix, Inc., Farmingdale, U.S.A.) was performed in 500 μl of RIPA buffer to yield chromatin fragments of 100-500 bp in size. Immunoprecipitation was carried out with polyclonal anti-androgen receptor (AR) antibody or normal rabbit IgG (sc-2027, Santa Cruz, Dallas, U.S.A.) as previously described [26]. After reverse cross-linking overnight at 65 • C, immunoprecipitated and input DNA was purified using PCR purification kit (Qiagen, Hilden, Germany) and eluted in 100 μl of elution buffer. For ChIP qPCR, 5 μl of ChIP or input DNA was used in each reaction with SYBR Green master mix (Roche, Basel, Switzerland) and specific primers (Odcrp −4 kb: forward primer 5 -AGGGTCAGGATGTTCCTGTG, reverse primer 5 -GAGAGCTTTGGCTCCTGATG; Odcrp +30 kb: forward primer 5 -CAGCCCAGATGCAGAGTTTC, reverse primer 5 -TTCCAGCCTTTGAGTTTGCT). Results from IP samples were normalized to respective input sample, and the results (mean + S.E.M.) for four replicates are shown as percent of input. DNA libraries from ChIP samples were prepared according to Illumina's instructions and sequenced using Illumina Genome Analyzer II. Peak calling Immunohistochemistry Five-micrometer thick sections from formalin fixed and paraffin-embedded kidneys were stained with 1:1200 diluted rabbit anti-ODCrp[A] antibody and with its preimmune serum as control using Vectastain Elite ABC Kit (Vector Laboratories Inc., Burlingame, U.S.A.) according to the manufacturer's instructions essentially as described [29]. Light microscope photographs were taken with an Olympus BX51 microscope (Olympus Optical, Tokyo, Japan) and a Nikon Digital Sight DS-5M camera (Nikon Corporation, Tokyo, Japan) using NIS-Elements F2.30 software (Nikon Corporation). Digital image processing was performed with PhotoScape v3.6.1 (Mooii Tech, Informer Technologies Inc., Los Angeles, U.S.A.).

Cell cultures and transfections
Cos-7 cells were cultured in DMEM medium. ODC-deficient CHO cells (a kind gift from Dr Lo Persson, Lund, Sweden), devoid of endogenous ODC activity, were cultured in RPMI-1640 medium supplemented with putrescine. Both media also contained 10% (v/v) FBS (Gibco, Life Technologies, Thermo Fisher Scientific Inc.), l-glutamine and penicillin and streptomycin. ODC-deficient CHO cells were plated without putrescine 24 h before transfection. Cells were transfected with the desired plasmids using the FuGENE6 transfection reagent (Promega) according to the manufacturer's instructions. In co-transfection experiments, the transfection mix contained equal amounts of both plasmids. Production of the transfected proteins was verified by SDS/PAGE and immunoblotting.

In-vitro translation
Radiolabeled in vitro translated proteins were produced using an In-vitro translation (IVT) kit (TNT Coupled Reticulocyte Lysate System, Promega) according to the manufacturer's instructions with L-[ 35 S]-methionine (PerkinElmer, Waltham, U.S.A.). Samples of the translated proteins and Amersham Rainbow [ 14 C] methylated protein molecular weight marker (Amersham Biosciences, GE Healthcare Life Sciences, Chicago, U.S.A.) were separated by SDS/PAGE (12% gel). The gel was fixed for 30 min in 30% methanol and 10% acetic acid solution and incubated for 1 h in Amplify Fluorographic Reagent (Amersham Biosciences). The gel was vacuum dried (Model 853 Gel Dryer, Bio-Rad, Hercules, U.S.A.) on to filter paper for 2 h at 80 • C and used to expose an X-ray film (FUJI) overnight.

Degradation assay
Protein degradation assay was performed as described previously [23]. Of the in-vitro translated proteins used, only ODC was radiolabeled. As a negative control, IVT lysate without translated proteins was mixed with radiolabeled ODC. The reactions were set up by mixing 1 μl OAZ with 14 μl ODCrp/ 1-13 ODCrp/AZIN1/lysate for 10 min at room temperature. ODC (2 μl) was added to the mixture, which was kept at 4 • C for 5 min, whereafter prewarmed (37 • C) ATP-regenerating buffer (50 mM Tris/HCl, pH 7.5, 5 mM MgCl 2 , 2 mM DTT, 0.5 mM ATP, 10 mM p-creatine, and 5 μg/ml creatine kinase) was added to a total volume of 60 μl. The reactions were incubated at 37 • C and 5 μl samples were taken after 0, 10, 30 min, and 1 h. The samples were immediately mixed with 2× Laemmli sample buffer + 2-mercaptoethanol and separated by SDS/PAGE (12% gel). Radiolabeled ODC was visualized by fluorography.

Immunoprecipitation
In co-precipitation experiments, Cos-7 cells were co-transfected with plasmids producing FLAG-tagged and Myc-tagged proteins. Myc-tagged proteins were immunoprecipitated with 4 μg mouse monoclonal anti-Myc antibody (Sigma-Aldrich) and protein G-agarose (Roche) using an immunoprecipitation kit (Roche) according to the manufacturer's instructions. Alternatively, FLAG-tagged proteins were immunoprecipitated with 4-5 μg mouse monoclonal anti-FLAG antibody (Sigma-Aldrich) and protein G-agarose. Finally, agarose pellets were suspended in 20 μl 2× Laemmli sample buffer + 2-mercaptoethanol, heated at 95 • C for 5 min and separated by SDS/PAGE. Co-precipitated FLAG-tagged proteins were detected by immunoblotting with rabbit polyclonal anti-FLAG antibody (Sigma-Aldrich). Alternatively, co-precipitated Myc-tagged proteins were detected by immunoblotting with rabbit polyclonal anti-Myc antibody (MBL, Medical & Biological Laboratories Co., Ltd, Nagoya, Japan). In the ubiquitination experiments, Cos-7 cells were transfected with plasmids encoding FLAG-tagged proteins that were immunoprecipitated with mouse anti-FLAG antibody and analyzed by immunoblotting with rabbit anti-ubiquitin antibody (DAKO).

Treatment with cycloheximide
Cos-7 cells were transfected with ODCrp-pCDNA3.1 or ODCrp C363A -pCDNA3.1 constructs and treated with 50 μg/ml cycloheximide. Samples were collected after 1, 3, and 5 h of treatment and lysed on ice in 60 μl cold Pawson lysis buffer for 10 min. Supernatants of centrifuged lysates were recovered for SDS/PAGE and analyzed by immunoblotting.

Statistical analysis
The statistical significance was calculated by unpaired Student's t test, and a value of P<0.05 was considered significant. In the figures, the statistical significance of differences between two samples is indicated by a square bracket and an asterisk (*), except for Figure 7, which contains brackets only.
OAZ binds to the 117-140 region of ODC [13]. The residues Gln 119 , Ala 124 , Asn 125 , Gln 129 , Glu 136 , Val 137 , and Met 140 of human ODC are most important for OAZ binding [37]. Lys 141 and Phe 397 also interact with residues of OAZ, while Lys 69 , Lys 92 , and Tyr 323 come in close proximity to the surface of OAZ [19]. Human and mouse ODC proteins differ at only two residues within the 117-140 region, where Asn 125 and Val 137 of human ODC are substituted for Ser 125 and Ile 137 in mouse ODC. In ODCrp, only 14 of the 24 residues of the OAZ-binding motifs are conserved, including only three (Ala 124 , Gln 129 , and Glu 136 ) of the seven most important residues. However, of the other residues interacting with or coming in close proximity to OAZ, all except for Lys 92 are conserved in ODCrp (Figure 1). AZINs have a higher affinity for OAZ than ODC [19]. The higher affinity is mediated by differences at residues 125 and 140, where serine and methionine of ODC are replaced by two lysine residues in AZIN1 [38]. In AZIN2, only the former lysine is conserved, while the latter is changed to alanine. It is unlikely that ODCrp has an AZIN-like affinity for OAZ, as in ODCrp these residues are arginine and isoleucine, respectively. ODC has two PEST sequences [39] that are recognized by the proteasome after OAZ binding to ODC [14]. The PEST sequences consist of regions 293-333 and 423-449, of which the latter is more important for the OAZ-mediated degradation [40]. In ODCrp, the region of the first PEST sequence has undergone many changes and is partially missing, while the latter region is lacking entirely (Figure 1). The mutated OAZ-binding motif suggests that ODCrp does not interact with OAZ, which together with the missing PEST sequences suggest that ODCrp degradation is likely to be initiated by ubiquitination.
The N-and C-terminal differences between ODCrp and ODC may not affect the functions of ODCrp, as those regions are located on the outer surface of the folded protein away from the active sites or dimer interface [41,8]. It is plausible to assume that ODCrp functions as an independent enzyme that does not interact with OAZ. Alternatively, ODCrp could serve as an ODC enhancer by binding ODC to form a catalytically active and/or a degradation-resistant heterodimers.

Tissue expression and androgen responsiveness of Odcrp
According to the UniGene expressed sequence tag (EST) profile (Mm.387701) [42], Odcrp is mainly expressed in kidneys and to a lesser extent in liver. The result of qPCR experiments on RNAs from kidney, liver, brain, lung, spleen, heart, prostate, and testis showed that Odcrp mRNA is almost exclusively expressed in kidney, which is in-line with the EST profile. Although some expression was also found in the brain, the Odcrp mRNA level in the brain was only approximately 3% of that in the kidney. Odcrp mRNA level in the liver was less than 0.03% of that in the kidney (results not shown).
Odc is known to be androgen-inducible in mouse kidney [43][44][45]. qPCR experiments showed that the steady-state Odcrp mRNA level was approximately two-fold higher in male than in female kidneys. The difference in ODCrp expression in male and female kidneys was statistically significant (P<0.05) in a sample size of 17 male and 19 female mice. Treatment of ICR mice with 1 mg T for 3 days increased accumulation of Odcrp mRNA approximately 5.3-fold in male and 4.4-fold in female kidneys (Figure 2).
In vivo ChIP-seq experiments [46] were performed to examine whether there are AR-binding sites adjacent to the Odcrp locus to support the notion that androgen induction of Odcrp mRNA accumulation is a transcriptional event. AR binding in vivo to renal chromatin in the absence of androgen was marginal, while T treatment resulted in loading of AR on to specific sites adjacent to the Odcrp locus. More specifically, there were several AR binding events at +30 kb and −4 kb regions of the Odcrp transcription start site after 2 h of T exposure ( Figure 3A). AR loading on to the two Odcrp regulatory regions was validated by using direct ChIP assays, and the results showed that androgen exposure result in approximately ten-fold enrichment of AR binding in vivo at both loci in murine kidneys ( Figure  3B).
Sections from kidneys of T treated and control male mice were stained with rabbit anti-ODCrp[A] antibody. In kidneys of control mice, ODCrp antigen level was most abundant in the subcapsular area and in the inner part of the cortex closest to the medulla, where the staining was strongest in the epithelial cells of the proximal tubules ( Figure  4A). In kidneys of T-treated mice, ODCrp expression was seen throughout the whole cortex. Similar to ODC and AR expression [46,47], the staining of ODCrp antigen was strongest in the epithelial cells of the proximal tubules ( Figure  4B). These cells also increased in size ( Figure 4B), a known hypertrophic effect of androgens in mouse kidney [48]. Androgen regulation of Odc does not require catalytic ODC activity [44], and subsequent polyamine accumulation is not needed for androgen-induced hypertrophy of mouse kidney [44,49]. The effect of androgen action in mouse The 20 amino acid residues most critical for the catalytic ODC activity are marked in dark gray. The small arrow indicates the residue Cys 363 in ODCrp and Cys 360 in ODC. The antizyme-binding region is marked by a square. The five most important residues within the antizyme-binding region of ODC are marked in black, while the two residues of AZIN1 enhancing its affinity toward antizyme are marked in dark gray. Residues coming into close proximity with the surface of antizyme in the ODC-antizyme complex are marked in gray. Two of these residues also belong to the 20 most critical residues marked in dark gray, in that case, a gray dot (.) was placed above them. The two PEST sequences of ODC are marked with underlining bars, with the most important region marked in by a bolder bar. The alignment shows that ODCrp contains all the residues needed for catalytic activity, while the regions needed for antizyme binding and antizyme-mediated degradation are either mutated or missing. The alignment of the antizyme-binding regions of mouse and human ODC is also shown at the bottom. The two differing residues within the antizyme-binding region of mouse and human ODC are marked in gray.  kidney comprises several hundreds of genes that are up-or down-regulated by androgens [46,50], and Odcrp belongs to the category of up-regulated genes. Of note, genes involved in DNA replication or cell proliferation are not regulated by androgens in mouse kidney as opposed to mouse prostate [46].

Catalytic activity and dimer formation of ODCrp
Dimerization is required for the catalytic ODC activity [6]. To investigate whether ODCrp is also capable of dimerization, Cos-7 cells were co-transfected with two ODCrp cDNA constructs, one producing a FLAG tag and the other a Myc tag. FLAG-tagged proteins were recovered by immunoprecipitation from cell lysates and analyzed by SDS/PAGE and immunoblotting with anti-Myc antibody. ODCrp was found to form dimers similar to ODC ( Figure 5). In addition to native ODCrp, the modified forms 1-13 ODCrp, ODCrp C363A , and 1-13 ODCrp C363A also dimerized. In  We next investigated whether ODCrp catalyzes decarboxylation of ornithine. ODC-deficient CHO cells that are devoid of endogenous ODC activity [51] were transfected with cDNA constructs encoding different ODCrp variants or mouse ODC as positive control. Immunoblotting of the lysates from transfected cells showed protein expression with all cDNA constructs. However, in the ODC assay, only lysates from the cells transfected with ODC cDNA displayed catalytic activity above background (results shown in Supplementary Figure S1), which means that ODCrp itself does not display measurable decarboxylase activity under the conditions used for a conventional ODC assay.
Since ODCrp did not exert catalytic ODC activity, we tested whether ODCrp was able to catalyze decarboxylation of other substrates like lysine and arginine [52]. Constructs with different ODCrp cDNA variants were transfected into Cos-7 cells that are devoid of endogenous arginine and lysine decarboxylase activity. Lysates from transfected Cos-7 cells were assayed for lysine and arginine decarboxylase activity under the same conditions as that for the ODC assay. Since no positive controls were available for lysine and arginine decarboxylase assays, lysates of Cos-7 cells transfected with an ODC cDNA construct were assayed for ODC activity to serve as a positive control. No lysine or arginine decarboxylase activity was detected with the ODCrp constructs used, indicating that the substrate of ODCrp is neither lysine nor arginine (results shown in Supplementary Figures S2 and S3).
Although ODCrp neither catalyzed ornithine decarboxylation nor formed dimer with ODC, it could nevertheless affect endogenous ODC activity by interacting with other endogenous proteins or by acting upon a substrate that is naturally present in the cells. To examine this possibility, we performed an ODC assay with lysates from Cos-7 cells transfected with cDNA constructs for the different ODCrp variants and with mouse ODC cDNA as positive control. Human AZIN1, which should bind the endogenous green monkey OAZ1 (human and green monkey OAZ1 are 99% identical), was also included to mark the highest activity achievable with endogenous ODC. Compared with the vector control, the presence of ODCrp and 1-13 ODCrp had a weak but statistically significant repressing effect  on the total ODC activity. This confirmed that ODCrp itself did not have any intrinsic ODC activity. On the contrary, expression of the mutated and 'inactive' counterparts ODCrp C363A and 1-13 ODCrp C363A had a slight but statistically significant enhancing effect on the total ODC activity (Figure 7). The activities of the lysates with the different ODCrp forms were smaller than with AZIN1 and only approximately one-tenth of that with the positive control ODC (not shown), suggesting that the different ODCrp forms somehow affected the catalytic activity of endogenous ODC. The intact and mutated variants of ODCrp should essentially have the same molecular interactions, with the difference that the mutated forms should not catalyze decarboxylation [33]. If the concentration of this putative substrate is low and/or only a small fraction of it is free and metabolically available, as is the case with polyamines [53], and if both the intact and mutated ODCrp forms bind the substrate but only the intact forms complete the reaction, it could explain why the total ODC activity is repressed in the presence of ODCrp and 1-13 ODCrp and increased in the presence of ODCrp C363A and 1-13 ODCrp C363A . These opposite effects seen with the intact and mutated forms are unlikely to result from differences in protein levels (further elaborated below) or cell proliferation, as immunoblotting revealed the presence of all ODCrp forms at similar total protein concentrations (shown in Supplementary Figure S5).

ODCrp degradation and stability
Unlike ODC, ODCrp is likely to be degraded by ubiquitination as both the OAZ-binding motifs and the two PEST sequences are impaired. To investigate the degradation pathway of ODCrp, FLAG-tagged proteins were recovered by immunoprecipitation from lysates of transfected Cos7-cells and analyzed by SDS/PAGE and immunoblotting with antiubiquitin antibody. ODCrp was found to be ubiquitinated like AZIN1 (Figure 8) [22]. No difference was observed between ODCrp and 1-13 ODCrp.
When verifying the amounts of transfected proteins in the ODC assay samples by SDS/PAGE and immunoblotting, we observed that the levels of ODCrp C363A and 1-13 ODCrp C363A appeared lower than the levels of ODCrp and [1][2][3][4][5][6][7][8][9][10][11][12][13] ODCrp. To examine the reason for this finding, cells transfected with ODCrp or ODCrp C363A constructs were treated with 50 μg/ml cycloheximide, lysed, and the lysates were analyzed by immunoblotting with anti-ODCrp[B]  antibody. The results showed that the mutation of Cys 363 (corresponding to Cys 360 of ODC) to alanine caused destabilization of ODCrp (Figure 9), suggesting that this residue is of importance for the functional role and/or conformation of the protein.

Interaction between ODCrp and antizyme and the effect on ODC degradation
In ODC (and AZIN1), the residues 117-140 constitute the OAZ-binding motif [13]. Despite the fact that the sequence alignment showed that only 14 of these residues are conserved in ODCrp, an interaction between ODCrp and OAZ could still be mediated by residues outside of the OAZ-binding region. We performed a co-immunoprecipitation experiment to investigate whether ODCrp displays AZIN functions by binding to OAZ. Cos-7 cells were co-transfected with constructs producing FLAG-tagged ODCrp or AZIN1 and Myc-tagged OAZ. The Myc-tagged OAZ was recovered by immunoprecipitation from the lysates and analyzed by SDS/PAGE and immunoblotting with anti-FLAG antibody. There was no evidence for direct binding between ODCrp and OAZ, whereas AZIN1 co-precipitated with OAZ ( Figure 6A, last two lanes).
The possibility exists that ODCrp inhibits the OAZ-mediated degradation of ODC without direct binding to OAZ. It is also possible that the interaction between ODCrp and OAZ or between ODCrp and ODC occurs even if it cannot be found by co-immunoprecipitation. To investigate this, we performed degradation assays with in-vitro translated proteins. The results showed that neither ODCrp nor [1][2][3][4][5][6][7][8][9][10][11][12][13] ODCrp influences the OAZ-mediated degradation of ODC under the conditions where AZIN1 blocked ODC degradation ( Figure 10). Thus, since ODCrp does not directly bind OAZ or protect ODC from OAZ-mediated degradation, it is very unlikely that ODCrp displays AZIN functions in vivo.
To further investigate functional interactions between ODC, OAZ, ODCrp, and AZIN1, we performed ODC assays with in-vitro translated proteins. The results showed that: (i) ODCrp did not release ODC from OAZ-mediated inhibition under conditions where AZIN1 displayed such a function, and (ii) the activity of 10 μl IVT ODCrp was very close to that of 10 μl AZIN1 and less than approximately one-third of the activity of 2 μl ODC, suggesting that ODCrp was devoid of measurable intrinsic ODC activity under the conditions where the positive ODC control catalyzed ornithine decarboxylation as expected ( Figure 11). We could not get evidence for any functional or regulatory role of the unique N-terminal extension ODCrp.

Conclusion
ODCrp is an androgen-inducible protein specific to mouse kidney, where its expression is most prominent in epithelial cells of the proximal tubules closest to the medulla. ODCrp is neither an AZIN nor a direct ODC enhancer, as no interaction could be shown between ODCrp, and OAZ or ODC. While ODCrp forms dimers, it does not catalyze decarboxylation of ornithine, lysine, or arginine. Its putative substrate remains to be identified. However, ODCrp displayed a small but significant regulatory influence on the catalytic activity of endogenous ODC. Last, no indication for a functional or regulatory role of the unique N-terminal extension ODCrp was uncovered.