R248Q mutation of p53 alters molecular trafficking and targeted drug responses in high-grade serous ovarian carcinoma

Background p53 mutations are detected in up to 96% of high-grade serous ovarian carcinoma (HGSOC). Moreover, mutant p53 may cause oncogenic gain-of-function phenotypes in sustained activation of epidermal growth factor receptor (EGFR) signaling. In this study, we investigated whether p53 mutation could affect EGFR-related signaling and the corresponding therapeutic strategies in HGSOC.Methods For this study, we selected the p53 R248Q mutant (p53 R248Q ), which has the highest mutation frequency in ovarian cancer. Since we have previously demonstrated that the combined inhibition of EGFR and MDM2-p53 pathways by gefitinib and JNJ-26854165 exerts a strong synergistic lethal effect on p53-mutated HGSOC cells, the gefitinib and JNJ-26854165 responses in different p53 status cells were evaluated by cell viability and target protein expression assays.Results We showed that the phosphorylation of AKT, a critical molecule of EGFR downstream signaling, increased significantly when p53 R248Q was transiently overexpressed. Immunocytochemistry analysis further showed that upon p53 R248Q overexpression, several AKT-dependent mediators translocate in unique patterns within the cell. Subsequent analysis revealed that under the combined inhibition of gefitinib and JNJ-26854165, the cytonuclear trafficking of EGFR and MDM2 is disrupted. Moreover, when we further compared gefitinib and JNJ-26854165 responses in different p53 status cells, we found that the sensitivity to the single- or combined-inhibition treatments is also altered in p53 R248Q -overexpressing cells.Conclusions Our findings suggest that the R248Q mutation of p53 causes significant changes in signaling transduction, molecular trafficking and drug responses, which help to advance our understanding of p53 and to improve therapeutic strategies for HGSOC.


Abstract
Background p53 mutations are detected in up to 96% of high-grade serous ovarian carcinoma (HGSOC). Moreover, mutant p53 may cause oncogenic gain-of-function phenotypes in sustained activation of epidermal growth factor receptor (EGFR) signaling.
In this study, we investigated whether p53 mutation could affect EGFR-related signaling and the corresponding therapeutic strategies in HGSOC.Methods For this study, we selected the p53 R248Q mutant (p53 R248Q ), which has the highest mutation frequency in ovarian cancer. Since we have previously demonstrated that the combined inhibition of EGFR and MDM2-p53 pathways by gefitinib and JNJ-26854165 exerts a strong synergistic lethal effect on p53-mutated HGSOC cells, the gefitinib and JNJ-26854165 responses in different p53 status cells were evaluated by cell viability and target protein expression assays.Results We showed that the phosphorylation of AKT, a critical molecule of EGFR downstream signaling, increased significantly when p53 R248Q was transiently overexpressed. Immunocytochemistry analysis further showed that upon p53 R248Q overexpression, several AKT-dependent mediators translocate in unique patterns within the cell. Subsequent analysis revealed that under the combined inhibition of gefitinib and JNJ-26854165, the cytonuclear trafficking of EGFR and MDM2 is disrupted. Moreover, when we further compared gefitinib and JNJ-26854165 responses in different p53 status cells, we found that the sensitivity to the single-or combined-inhibition treatments is also altered in p53 R248Q -overexpressing cells.Conclusions Our findings suggest that the R248Q mutation of p53 causes significant changes in signaling transduction, molecular trafficking and drug responses, which help to advance our understanding of p53 and to improve therapeutic strategies for HGSOC.

Background 3
Epithelial ovarian cancer is the most prevalent (~90%) type of ovarian carcinoma [1] and could be further divided into subtypes based on histopathology, gene alterations, and prognosis [2]. Among the five main subtypes, high-grade serous ovarian carcinoma (HGSOC) accounts for the highest proportion (~70%) [3]. HGSOC is known to have a high risk of recurrence and platinum-based chemotherapy drug resistance due to genetic changes in tumor suppressor p53 encoded by TP53 [4]. As one would expect, ovarian cancer has the highest rate of somatic TP53 mutation among all cancers [5]. In fact, TP53 is mutated in up to 96% of HGSOC, with arginine 248 (R248) being the most common mutated amino acid [6].
p53 also acts as a transcription factor, with most mutations being located in its DNAbinding domain (DBD) [7,8]. In addition, it has been demonstrated that mutant p53 not only loses its activities but also displays oncogenic gain-of-function, such as accelerating tumor progression and acquiring drug resistance [9]. Since mutant p53 is known to amplify receptor tyrosine kinase (RTK) signaling [10][11][12][13][14][15], the association between p53 R248Q and RTK-regulated signaling in HGSOC could be important for cancer progression. Among the RTK downstream signaling pathways, several studies have revealed that p53 R248Q promotes cell migration and proliferation through AKT signaling [16,17]. However, how the R248Q mutation of p53 affects AKT-dependent carcinogenesis has not been elucidated.
Excessive activation of AKT signaling can mediate a variety of cellular processes, including cell cycle dysregulation, proliferation, and drug resistance, all of which are considered hallmarks of cancer [18]. In addition, it has been elucidated that activated AKT translocates to various organelles, where AKT phosphorylates substrates or interacts with other factors to regulate a complex network of processes [19]. For instance, AKT 4 phosphorylates MDM2 on Serine 166 and Serine 186 to initiate MDM2's nuclear translocation. MDM2 then acts to promote p53 ubiquitination for degradation and reduce p53 transcriptional activities [20]. Similarly, phosphorylation of EGFR's Serine 229 by AKT also promotes nuclear translocation of EGFR [21], which acts as a co-transcription factor of multiple genes regulating cell proliferation and tumor angiogenesis, resulting in enhanced drug resistance and poor patient outcomes [22]. It should be noted that whether the R248Q mutation of p53 affects AKT-dependent molecular trafficking has not been determined.
The taxane-and platinum-based chemoresistance in ovarian cancer is known to be associated with its high mortality rates [ . These findings led us to explore whether the R248Q mutation of p53 would affect the drug efficacy and the underlying mechanisms by which EGFR and 5 MDM2-p53 interact to produce a synergistic effect.
In this study, we first investigated how the R248Q mutation of p53 alters the characteristics of HGSOC. We then explored how the R248Q mutation of p53 regulates the drug responses and synergistic effect under the combined inhibition of EGFR and MDM2.

Cell culture
The serous ovarian cancer cell line used in this study was OVCAR3 (p53 R248Q mutation).
The TP53 status was annotated according to the International Agency for Research on Cancer (IARC) TP53 database (http://p53.iarc.fr/). The cells were cultured at 37°C with 5% CO 2 in RPMI-1640 medium (Gibco) supplemented with 10% fetal bovine serum.

Gene knockdown and overexpression
siRNAs for p53 and control siRNA were synthesized by Genepharma. Cells were transfected with siRNA at a final concentration of 25 nM using lipofectamine (Invitrogen) following the manufacturer's protocols. The p53 R248Q targeting plasmid was constructed in house. Cells were transfected with plasmid at a final concentration of 1 μg/mL using jetPRIME (Polyplus) following the manufacturer's protocols.

Western blot analysis
6 Cell lysates were extracted using RIPA lysis buffer (Roche). The products were then quantitated with a BCA protein assay kit (Pierce). Proteins were separated using 8% or 12% SDS-PAGE and then transferred to PVDF membranes (Millipore). Transferred membranes were immersed in 3% bovine serum albumin (BSA) dissolved in 1X TBS with 0.05% Tween-20 for 1 hour at room temperature. The PVDF membranes were incubated with different primary antibodies diluted in blocking solution overnight at 4°C and then incubated with secondary antibodies (GeneTex) diluted in 5% nonfat milk/TBST at room temperature for 2 hours before signal detection. Finally, the signal was enhanced by Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare). The anti-pEGFR, pAKT, AKT, pMAPK, MAPK, and p21 antibodies were obtained from Cell Signaling Technology. The anti-EGFR, p53 and GAPDH antibodies were obtained from GeneTex.

Immunocytochemistry (ICC)
OVCAR3 cells were seeded on sterile glass coverslips prior to the experiments. Cells were fixed with 4% paraformaldehyde/PBS for 10 minutes and permeabilized with 0.1% Triton X-100/PBS for another 10 minutes. Then, the cells were blocked with 3% BSA/PBS for 30 minutes. The primary antibodies were then added directly into blocking buffer and incubated overnight at 4°C. Afterward, the secondary antibodies, which were anti-rabbit IgG-DyLight488 (Jackson) and anti-mouse IgG-DyLight649 (Jackson), were added and incubated for 2 hours at 37°C. The nucleus and actin cytoskeleton were stained with Hoechst 33342 (Invitrogen) and phalloidin (Invitrogen), respectively. The images of stained cells were captured by confocal microscopy (LSM510 Meta, Zeiss). The anti-EGFR (green signal), AKT, and FOXO3a antibodies were obtained from Cell Signaling Technology.
The anti-EGFR (purple signal) and MDM2 antibodies were obtained from GeneTex. The antimutant p53 antibody was obtained from Merck.

Cell viability assay
OVCAR3 cells were seeded in 96-well plates at a density of 5 × 10 3 cells/well. Cells were transfected with empty plasmid only (1 μg/mL) or with the p53 R248Q construct (1 μg/mL) for 18 hours and then treated with drugs of different concentrations for 48 hours.
Following the manufacturer's protocols, 10 μL Cell Counting Kit-8 (Dojindo) cell proliferation reagent was added to 96-well plates and incubated at 37°C for 1 to 2 hours.
The optical densities to reflect cell viability were determined by absorbance wavelength at 450 nm.

Analysis of combination index
The combination index (CI) values were calculated by CompuSyn calculation software as described by Chou and Talalay [31]. First, the cell viability assay data were converted into the fraction affected (Fa) under specified drug treatment conditions. Next, the Fa values of single agents or drug combinations were uploaded into CompuSyn and generated the CI values, the Fa-CI plots and the isobolograms as instructed by the user manual.

Statistics
Data were processed by GraphPad Prism 6 and expressed as the means ± SEM (standard error of the mean). Statistical significance was determined by p values < 0.05 using Student's t-test.

R248Q mutation of p53 amplified p-AKT signaling in ovarian cancer cells
We first investigated whether p53 R248Q affected EGFR (a member of the RTK family) and its downstream signaling in OVCAR3 cells (HGSOC cell line with p53 R248Q ). Since EGFR is aberrantly overexpressed and activated in ovarian cancer [32, 33], we stimulated prestarved cancer cells with epidermal growth factor (EGF) to activate the EGFR signaling.
Both activated PI3K/AKT and MAPK/ERK pathways under EGF stimulation have been implicated in various cancers. Therefore, we examined these EGFR downstream markers under the condition of p53 siRNA knockdown. We found that p-AKT and p-MAPK decreased in OVCAR3 cells after p53 R248Q knockdown ( Supplementary Fig. S1A). Using an overexpression approach, we found that p-AKT was significantly upregulated after augmented expression of p53 R248Q in OVCAR3 cells ( Supplementary Fig. S1B). Taken together, these data indicated that the R248Q mutation of p53 amplified p-AKT signaling in OVCAR3 cells.

p53 R248Q overexpression and EGF stimulation resulted in similar cytonuclear trafficking of AKT, EGFR, MDM2, and FOXO3a
AKT signaling is known to mediate intracellular trafficking of various receptors and regulatory proteins. We thus investigated how the R248Q mutation of p53 alters the intracellular molecular trafficking of EGFR, MDM2, and FOXO3a in association with AKT ( Fig. 1).
We found that, after either EGF stimulation or p53 R248Q overexpression, most of the AKT translocated into the nucleus completely. However, when p53 R248Q overexpression was combined with EGF stimulation, some AKT remained in the cytoplasm (Fig. 1A).
Consistent with past studies, our data implied that AKT enters the nucleus in response to growth factors to exert regulatory activities [34-36]; thus, we continued to investigate how AKT translocation correlates with the intracellular trafficking of EGFR and MDM2. Similar but different from the aforementioned findings, after EGF stimulation or p53 R248Q overexpression, both EGFR and MDM2 appeared to converge and translocate to the periphery of the cell nucleus ( Fig. 1B and 1C). In comparison to the control group, we also observed some EGFR still localized on the cell membrane, while most of the MDM2 translocated to the periphery of the cell nucleus. Interestingly, when p53 R248Q overexpression was combined with EGF stimulation, the EGFR located on the cell membrane and the MDM2 clustering around the nucleus were not detected, and most of the MDM2 translocated into the cell nucleus.
On the other hand, it has been reported that FOXO3a, a member of the forkhead box O (FoxO) transcription factor families and a tumor suppressor, accelerates its nuclear export when it is phosphorylated by nuclear AKT [37]. Therefore, we speculated that this tumorigenic marker is affected by p53 R248Q . Based on our data, we found that FOXO3a, which mostly localized in the nucleus in the control group, distributed uniformly in the cytoplasm under the following conditions: after EGF stimulation, with p53 R248Q overexpression, and p53 R248Q overexpression combined with EGF stimulation (Fig. 1D).
These findings suggest that the R248Q mutation of p53 promotes tumorigenesis-related AKT signaling by affecting molecular intracellular trafficking.

Combined blockade by gefitinib and JNJ attenuated EGFR and MDM2 cytonuclear trafficking
The comparable translocation patterns of EGFR and MDM2 implied that the molecular cytonuclear trafficking may relate to the synergistic effect of combined EGFR and MDM2 inhibition that has been reported in our previous study [30]. In this present study, we again used gefitinib (EGFR tyrosine kinase inhibitor) and JNJ-26854165 (referred to as JNJ; MDM2 E3 ubiquitin ligase domain inhibitor) to determine whether such combined inhibition would alter the intracellular localization patterns of EGFR and MDM2.
By immunocytochemistry staining analysis, we first reconfirmed that both EGFR and MDM2 would translocate from the cytoplasm to the nucleus in accordance with each other over time after EGF stimulation ( Supplementary Fig. S2). Next, we found that such nuclear convergence of EGFR was attenuated under treatment with gefitinib, JNJ, or both.
Interestingly, the nuclear convergence of MDM2 was only disrupted under the combination treatment of JNJ and gefitinib ( Fig. 2A and 2B). These results indicate that the combined inhibition of gefitinib and JNJ could prevent the converging nuclear translocation of EGFR and MDM2, which might regulate further signaling and oncogenic activity.

R248Q mutation of p53 increased the sensitivity of EGFR and MDM2 inhibitors
To further explore the impact of p53 R248Q on the cellular response to combination therapy for HGSOC, we evaluated the anti-cancer efficacy of gefitinib and JNJ in OVCAR3 cells with different p53 statuses. For a more intuitive comparison, we summarized the data as Fig. 3 to highlight the significant reduction in effective drug concentration under combined inhibition. A synergistic effect was observed whether cells were pre-transfected with the p53 R248Q plasmid or not (Fig. 3A, 3B). Interestingly, gefitinib and JNJ alone were more effective when OVCAR3 cells were pre-transfected with the p53 R248Q plasmid (Fig. 3C, 3D).

As shown in Supplementary
However, the combined inhibition effect remained similar whether OVCAR3 cells were pretransfected with the p53 R248Q plasmid or not (Fig. 3E). Taken together, these results suggest that the R248Q mutation of p53 increases the sensitivity of OVCAR3 cells to EGFR and MDM2 inhibition to various degrees.

R248Q mutation of p53 decreased the synergistic lethal effect of gefitinib and JNJ
To better evaluate the synergistic effects exerted by combined inhibition in different p53 statuses, we adapted the well-known Chou and Talalay's combination index (CI) method to perform in-depth analysis. It should be noted that a CI value less than 1 is defined as drug synergism [31,38].
The CI analysis revealed that in the absence of overexpression of p53 R248Q , the area below the CI=1 line was larger than that of the p53 R248Q overexpression group ( Fig. 4A and 4B). We also generated isobolograms to quantify the drug synergism at different effective doses (EDs). In comparison to the p53 R248Q overexpression group, the symbols (colored as specified) were farther from the corresponding colored lines, and the CI values were also lower in the absence of overexpression of p53 R248Q (Supplementary Table 2 In OVCAR3 cells overexpressing p53 R248Q , the drug combination displayed slight synergism (CI=0.87) at ED 50 , but changed to moderate (CI=0.72) and enhanced synergism (CI=0.67) at ED 75 and ED 90 . Interestingly, OVCAR3 cells without p53 R248Q overexpression showed enhanced synergism (CI=0.65) at lower doses (ED 50 ). This finding implied that a higher dose of JNJ and gefitinib (ED 90 ) was required to reach synergism in OVCAR3 cells overexpressing p53 R248Q . In contrast, lower dose of drug combination (i.e. ED 50 ) exerted synergism in OVCAR3 cells without p53 R248Q overexpression. Such data supports that overexpression of p53 R248Q may attenuate the synergistic lethal effect of JNJ and gefitinib.

MAPK and p21 regulated the effects of single and combined treatment of gefitinib and JNJ
To provide more clues about the regulatory effect of p53 R248Q on the observed differential sensitivity to JNJ and gefitinib, we analyzed the expression profiles of MAPK and p21 (a p53 target gene associated with cell cycle and apoptosis pathways) under different conditions of JNJ and gefitinib concentration and combination.
As expected, the enhanced expression of p53 was detected, which served to validate the overexpression of p53 R248Q after transfection. Next, the results revealed that p-MAPK was significantly reduced by gefitinib in a dose-dependent manner (Fig. 5A). Moreover, at a dose of 10 μM, gefitinib exerted a stronger MAPK-reduction effect in p53 R248Qoverexpressing cells compared to the cells without overexpression (Supplementary Fig.   S3A). We also found that the expression level of p21 was significantly increased by JNJ in a dosage-dependent manner. Upon examination and comparison of the two DMSO control groups, we found that the expression level of p21 was lower in p53 R248Q -overexpressing cells (Fig. 5B). However, with increasing JNJ dose, p21 increased significantly (approximately 17-fold) in p53 R248Q -overexpressing cells compared to the cells without overexpression (approximately 4-fold) (Supplementary Fig. S3B). Strikingly, the effects were reversed when JNJ was combined with gefitinib ( Fig. 5C, Supplementary Fig. S3C,   S3D).
These results imply that in the presence of p53 R248Q overexpression, the augmented sensitivity to JNJ or gefitinib and the weakened synergistic effect of combined treatment might be associated with the alternating expression of MAPK and p21.

Discussion And Conclusions
In this study, we provide new evidence to highlight the effects of the R248Q mutation of p53 on cell signaling regulation and the response to EGFR/MDM2 targeted therapies in HGSOC.
Mutant p53 is generally considered to accelerate carcinogenesis by carrying out a dominant-negative effect on wild-type p53 and by manifesting gain-of function activities [39]. In this study, our findings iterated that the phosphorylation of AKT would be amplified in OVCAR3 cells via p53 R248Q overexpression ( Supplementary Fig. S1). For the effects of enhanced p-AKT signaling in HGSOC, we first speculated that the subcellular localization of AKT, as well as AKT-dependent molecular trafficking, may be altered under the influence of p53 R248Q . This hypothesis is supported by the evidences that nuclear AKT induced by various growth factors and stimuli has been shown to promote tumorigenesis by controlling cell cycle progression, promoting cell survival and DNA repair, and counteracting apoptosis [19]. In this study, we have further examined the trafficking patterns of AKT, EGFR, MDM2, and FOXO3a upon EGF stimulation and p53 R248Q overexpression (Fig. 1). Our data reveals that these molecules' trafficking patterns under p53 R248Q overexpression exert a comparable effect to that of EGF stimulation. Based on our findings, we propose a hypothetical model of molecular trafficking in response to EGF stimulation and p53 R248Q overexpression (Fig. 6A).
In this modeled scheme, the R248Q mutation of p53 could amplify the phosphorylation of AKT, similar to the stimulation of EGF. The resulting p-AKT would further phosphorylate EGFR and MDM2 to facilitate their converging nuclear translocation. Furthermore, nuclear AKT would further phosphorylate FOXO3a to dispel it from the nucleus. Surprisingly, the combined p53 R248Q overexpression and EGF stimulation did not lead to augmentation of nuclear AKT. In contrast, such combined treatment could promote more EGFR and MDM2 to translocate toward or into the nucleus. To the best of our knowledge, such phenomena have not been reported previously. We speculated that the AKT phosphorylation induced by EGFR activation and p53 R248Q results in antagonism, while the duel phosphorylation induced by nuclear AKT and EGF on EGFR leads to synergism. The difference in the translocation pattern of MDM2 may arise from the differential signaling of nuclear AKT and cytoplasmic AKT. The underlying mechanisms shall be further explored in future studies.
To view this study in a broader context, evidences accumulated over the past 20 years have helped to characterize the function of nuclear AKT, which phosphorylates various transcription factors, including members of the FOXO transcription factor families [40]. The phosphorylated FOXO3a would be expelled from the nucleus and weakened its transcriptional capacity. As a result, the expression level of its target gene, such as p21, was reduced [41] (Fig. 5). Nuclear EGFR has been shown to act as a co-transcription factor alongside STAT3, E2F1 and STAT5 to initiate the transcription of several cell cycle regulatory genes, such as Cyclin D1, Aurora A, and Myc [22]. Nuclear MDM2 mainly diminishes cellular levels of p53 and inhibits its transcriptional activity [20,42,43]. Therefore, the nuclear localization of AKT, EGFR, and MDM2 and the exclusion of nuclear FOXO3a have been linked with enhanced drug resistance, poor prognosis, and unfavorable overall survival in various cancers [22,[44][45][46][47].
To date, the combination of paclitaxel-and platinum-based drugs is a standard first-line treatment for ovarian cancer. However, chemoresistance to these drugs may lead to an unfavorable 5-year survival rate in advanced patients [48]. To overcome drug resistance, we suggested a novel combination therapy strategy with EGFR and MDM2 inhibition in our previous study [30]. In the present study, we further identified that the disruption of cytonuclear trafficking by the dual inhibition of EGFR and MDM2 might contribute to the synergistic effect of gefitinib and JNJ (Fig. 2). In addition, we demonstrated that the R248Q mutation of p53 enhanced the efficacy of gefitinib and JNJ alone, but reduced the synergistic effect of their combination ( Fig. 3 and 4). Our data further validated this observation by showing the differential effect of p-MAPK inhibition and the differential effect of p21 induction under different gefitinib and JNJ treatment conditions (Fig. 5). These findings suggested that the degree of synergism of gefitinib and JNJ depends not only on the cytonuclear trafficking of signaling mediators but also on the mutation of p53.
Based on our findings, the proposed model (Fig. 6B) underlines the mechanistic roles of p-MAPK and p21 under the treatment of gefitinib and JNJ, either in the absence or presence of p53 R248Q overexpression. It should be noted that MDM2 is known to ubiquitinate the pro-apoptotic transcription factor FOXO3a via activated MAPK signaling [49]. As expected, gefitinib inhibited its downstream molecule p-MAPK in the absence of p53 R248Q overexpression. The reduction of p-MAPK and upregulation of p21 was controlled by EGFR inhibition and the negative feedback of MDM2. On the other hand, with p53 R248Q overexpression, gefitinib may cooperate with p53 R248Q to reduce p-MAPK. Our model further explains how the reduction of p-MAPK is primarily influenced by p53 R248Q and secondarily influenced by MDM2. Hence, the relevant cell signaling would be disrupted by JNJ, while gefitinib in the presence of p53 R248Q overexpression could not exert its expected effects on p-MAPK and p21 (Fig. 6B). This model has the potential to revise the therapeutic approach for cancer treatment, which suggests the use of gefitinib or JNJ alone if the R248Q mutation of p53 is detected in HGSOC patients; if the R248Q mutation of p53 is not detected, the combination of gefitinib and JNJ should be recommended. While additional work is needed to confirm these predictions, we foresee that further understanding of the relationship between p53 mutation and the altered regulatory

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
Not applicable.

Consent for publication
Not applicable. Before confocal imaging, we stimulated the cells in the presence or absence of EGF (50 ng/mL) for 15 minutes. That is, four conditions were analyzed: the "control" group, in which the cells were serum-starved only; the "+EGF" group, in which the cells were treated with EGF after serum starvation; the "p53R248Q overexpression" group, in which the cells were transfected with p53R248Q before serum starvation; and the "p53R248Q overexpression +EGF" group, in which the cells were transfected with p53R248Q, serum-starved, and then treated with EGF.
Since the cells were only transiently transfected with p53R248Q, we    Symbols (ED90 is marked as green triangles; ED75 is marked as red squares; ED50 is marked as blue circles) above the lines, on the lines, and below the lines represent antagonistic, additive, and synergistic effects, respectively. When the symbols are located below and distant from the lines, it represents strong synergism.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.