Establishment of a dog model for the p53 family pathway and identification of a novel isoform of p21 cyclin-dependent kinase inhibitor

Abstract

Spontaneous tumors in the dog offer a unique opportunity as models to study human cancer etiology and therapy. p53, the most commonly mutated gene in human cancers, is found to be altered in dog cancers. However, little is known about the role of p53 in dog tumorigenesis. Here, we found that upon exposure to DNA damage agents or Mdm2 inhibitor nutlin-3, canine p53 is accumulated and capable of inducing its target genes, MDM2 and p21. We also found that upon DNA damage, canine p53 is accumulated in the nucleus, followed by MDM2 nuclear translocation and increased 53BP1 foci formation. In addition, we found that canine p63 and p73 are up-regulated by DNA damage agents. Furthermore, colony formation assay showed that canine tumor cells are sensitive to DNA damage agents and nutlin-3 in a p53-dependent manner. Surprisingly, canine p21 is expressed as two isoforms. Thus, we generated multiple canine p21 mutants and found that aa 129 to 142 is required, whereas aa 139 is one of the key determinants, for two p21 isoform expression. Finally, we showed that although the full-length human p21 cDNA expresses one polypeptide, aa 139 appears to play a similar role as that in canine p21 for various migration patterns. Taken together, our results indicate that canine p53 family proteins have biological activities similar to human counterparts. These similarities make the dog as an excellent out-bred spontaneous tumor model and the dog can serve as a translation model from bench-top to cage-side and then to bed-side.

Introduction

The anatomical and physiological similarities between dogs and humans have been the basis of using the dog in biomedical research for over a century. Recently, the spontaneously occurring tumors in companion dogs have become an ideal model for cancer research when compared to the murine model owing to the following features: 1) dog cancers occur spontaneously whereas in laboratory mice, cancers are mostly induced by carcinogens and/or genetic manipulations; 2) dog tumors, including melanoma, osteosarcoma, lung carcinoma, head- and-neck carcinoma, and mammary carcinoma, have similar histological appearance and response to conventional chemotherapies as human tumors; 3) humans and companion dogs live in the same environment; and 4) genetically, dogs are closer to humans than mice, including many oncogenes and tumor suppressor genes, which are found to be altered in spontaneous human tumors, are also found to be altered in spontaneous dog tumors.

The p53 tumor suppressor plays a pivotal role in preserving the integrity of the genome and in maintaining normal cell cycle regulation. Mutations of p53 tumor suppressor occur in ∼50% of human cancers and loss of p53 function is known to play a central role in cancer development and progression (1). p53 is expressed at low levels under unperturbed conditions and rapidly stabilized under stress conditions (1). p53 functions as a transcriptional factor to transactivate its downstream targets involved in cell cycle arrest and apoptosis. Thus, p53 and its downstream targets consist of a network, where p53 is a key molecular node in the network. In the late 90s, two p53-related genes, p63 and p73, were identified due to the high degree of sequence similarity, especially in the DNA-binding domain (2, 3). As a result, p63 and p73 can transactivate some p53-responsive genes involved in cell cycle arrest and apoptosis. Thus, these three proteins form the p53 family. Interestingly, the activities of p53 family proteins are not entirely redundant as p53, p63, and p73 knockout mice exhibit distinct phenotypes (4-7), indicating that each p53 family member has its own specific biological functions.

The canine p53 gene shares 87% sequence similarity with that of human p53 according to the NCBI database. Like its human counterpart, canine p53 is found to be mutated in several types of dog tumors, including osteosarcoma (8), mammary tumours (9), and mastocytoma (10). However, little is known about the biological function of canine p53 and its role in dog tumorigenesis and even less is known about the other two-p53 related genes, p63 and p73. In the present study, we have explored the role of p53 and its family members, p63 and p73, in canine normal and tumor cell lines. We found that DNA damage agents, such as camptothecin and doxorubicin, are capable of up-regulating all three p53 family members as well as p53 targets, p21 and MDM2, whereas the MDM2 inhibitor, Nutlin-3, are only capable of up-regulating p53. Immunofluorescence assay showed that canine p53 is accumulated in the nucleus upon DNA damage, which is coupled with MDM2 nuclear translocation and increased 53BP1 foci formation. In addition, we found that canine tumor cells are sensitive to DNA damage agents and nutlin-3 in a p53-dependent manner. Finally, we showed that canine p21 is expressed as two isoforms and the region from aa 129 to 142 is required for two isoform formation.

Results

The expression pattern of p53 in canine cell lines

In order to explore canine p53 expression, five melanoma cell lines (melanoma-12, -23, -36, -50, and -64) and four osteosarcoma cell lines (D17, 48-4, 348617, and 340529) were used. Some of the cell lines have been characterized to be tumorigenic in nude mice (11). In addition, Madin-Darby canine kidney (MDCK) and Cf2Th thymus cell lines were purchased from ATCC for this study. MDCK cell line was derived from a normal canine kidney and known to contain wild-type p53 (12) whereas Cf2Th cell line was derived from a normal canine thymus tissue. These canine cells were mock-treated or treated with camptothecin (CPT), doxorubicin (DOX), or nutlin-3 for 12 h and the levels of dog p53 protein were measured by Western blotting. Both CPT and DOX are DNA damage agents, which can cause DNA double-strand breaks and are known to stabilize p53 in human cells (13) whereas nutlin-3 is an MDM2 inhibitor and prevents MDM2-mediated p53 degradation (14). We found that canine p53 was accumulated in MDCK cells in response to CPT, DOX, or nutlin-3 (Fig. 1A, p53 panel, compare lanes 2-4 with lane 1), consistent with a previous report (12). Likewise, p53 was induced in four melanoma cell lines (melanoma-12, 23, 50, and 64) upon treatment with CPT, DOX, and nutlin-3 (Fig. 1A, p53 panel, compare lanes 6-8, 10-12, and 14-16 with lane 5, 9, and 13, respectively; Fig. 1B, p53 panel, compare lanes 6-8 with lane 5). However, p53 was undetectable in one melanoma cell line (melanoma-36) regardless of treatment (Fig. 1B, p53 panel, lanes 9-12), suggesting that melanoma-36 is likely to be p53-null. To determine whether the stabilized p53 in these cells is transcriptionally active, we measured induction of two well-known p53 targets, p21 and MDM2. p21 is required for p53-mediated G1 arrest (15) whereas MDM2 is an E3 ubiquitin ligase and can target p53 for proteasomal degradation (16, 17). We showed that in response to CPT, DOX, and nutlin-3, both p21 and MDM2 were induced in cells with wild-type p53, including MDCK and four melanoma cell lines, but not in p53-deficient melanoma-36 cell line (Fig. 1A-B, p21 and MDM2 panels). Surprisingly, canine p21 is expressed as two isoforms (Fig. 1A-B, p21 panels), which was addressed in the latter part of this study.

Figure 1. The expression pattern of p53 in canine cell lines.

(A-B) p53 is induced by DNA damage agents or nutlin-3 in MDCK and four melanoma (melanoma-50, -12, -23, and -64) cells but not in melanoma-36 cells. Cells were treated with or without CPT, DOX, or nutlin-3 for 12 h. Cell lysates were collected and subjected to Western blot analysis using antibodies against p53, p21, MDM2, actin, and GAPDH. (C-D) p53 is induced in four osteosarcoma cells (D17, #48-4, 348617, and 340529) but not in Cf2Th cells. Cells were treated as describe in (A) and then analyzed by Western blot analysis using antibodies against p53, p21, MDM2, and GAPDH.

Next, we analyzed induction of p53, p21, and MDM2 in Cf2Th and four osteosarcoma cell lines (D17, 48-4, 348617, and 340529). We found that all osteosarcoma cells showed a marked accumulation of p53 in response to CPT, DOX, and nutlin-3 (Fig. 1C-D, p53 panels, compare lanes 6-8 and 10-12 with lanes 5 and 9, respectively). In addition, p21 was induced (Fig. 1C-D, p21 panels, compare lanes 6-8 and 10-12 with lanes 5 and 9, respectively). However, MDM2 was highly induced in cells treated with nutlin-3 but little if any with CPT or DOX (Fig. 1C-D, MDM2 panels, compare lanes 9 and 12 with 7-8 and 10-11, respectively). Surprisingly, the steady-state level of p53 protein was found to be very high in Cf2Th cells (Fig. 1D, p53 panel, compare lane 1 with lanes 5 and 9). In addition, p53, p21 and Mdm2 were not found to be induced in Cf2Th cells upon treatment with CPT, DOX, or nutlin-3 (Fig. 1D, p53, p21, and MDM2 panels, compare lanes 2-4 with lane 1). Since Cf2Th cell line was derived from a normal canine thymus tissue, it was assumed to be a non-tumor cell line. However, our results suggest that Cf2Th carries a mutant p53 and thus is likely to be a tumor cell line.

The intracellular localization of p53, MDM2, and 53BP1 in MDCK cells

In response to genotoxic stress, human p53 is accumulated in the nucleus to transactivate its downstream targets. Thus, we sought to determine the intracellular localization of canine p53 in response to doxorubicin by immunofluorescence assay. To do so, MDCK cells were mock-treated or treated with doxorubicin for 12 h, and then stained with anti-p53 and DAPI. We found that in the absence of doxorubicin, there was little positive staining of p53 (Fig. 2A, top panel) whereas in the presence of doxorubicin, condensed nuclear staining of p53 was observed (Fig. 2A, bottom panel). In addition, we analyzed the intracellular localization of MDM2. As a p53 target, it is expected that activation of p53 upon DNA damage would lead to increased expression of Mdm2. Indeed, we found that in the absence of DNA damage, Mdm2 was weakly stained and primarily localized in the cytoplasm (Fig. 2B, top panel). However, upon treatment with doxorubicin, the intensity of Mdm2 was markedly increased (Fig. 2B, bottom panel). Most importantly, the vast majority of Mdm2 protein was translocalized from the cytoplasm into the nucleus in response to DNA damage (Fig. 2B). To further test this, we analyzed the localization of 53BP1, which is known to interact with p53 and functions as a modulator of the p53 and DNA damage response pathways (18). We found that in response to DNA damage, a large number of 53BP1 foci were detected in the nucleus as compared to that under the control condition (Fig. 2C), consistent with previous observations in human cells (19).

Figure 2. The intracellular localization of p53, MDM2, and 53BP1 in MDCK cells.

(A) p53 is accumulated in the nucleus of MDCK cells upon doxorubicin treatment. MDCK Cells were mock-treated or treated with doxorubicin (250 ng/ml) for 12 h. Cells were then fixed and stained for p53 and DAPI. The nuclear DAPI stain is blue and p53 stain is green. (B) Mdm2 is translocated from the cytoplasm to the nucleus in response to DNA damage. Cells were treated as described in (A) followed by immunofluorescence assay using anti-Mdm2 and DAPI. (C) 53BP1 foci formation is increased in response to DNA damage. Cells were treated as described in (A) followed by immunofluorescence assay using anti-53BP1 and DAPI.

The expression pattern of p63 and p73 in canine cells

Human p63 and p73 have been found to play a critical role in tumor suppression and normal development (5-7). However, there has been little study of canine p63 and p73. Based on the predicted sequence, canine p63 and p73 shares 99.6% and 81% sequence identity with their human homologues, respectively. Thus, canine p63 and p73 might be recognized by commercial available antibodies against human p63 and p73, respectively. In this regard, canine p63 was detected by Western blotting with 4A4 anti-p63 antibody. 4A4 was raised against amino acids 1-205 in ΔNp63α of human origin and is capable of recognizing both TA and ΔN p63 isoforms in humans. We found that several distinct bands were detected in MDCK cells treated with camptothecin or doxorubicin compared to that in mock-treated MDCK cells (Fig. 3A, p63 panel, compare lanes 2-3 with lane 1), suggesting that these are likely to represent various TA and/or ΔN p63 isoforms. However, while nutlin-3 was able to stabilize p53 in dog cells (Fig. 1), the levels of these p63 proteins were not significantly altered (Fig. 3A, p63 panel, compare lane 4 with 1). We also found that some of the p63 isoforms were increased in melanoma and osteosarcoma cells upon treatment with camptothecin and doxorubicin but not nutlin-3 (Fig. 3A-C), consistent with the observations in human cells (20). Interestingly, in melanoma-36 and D17 cells, one isoform of p63 was found to be highly expressed (Fig. 3B, lanes 9-12; Fig. 3C, lanes 5-8).

Figure 3. The expression pattern of p63 in canine cell lines.

(A-B) Expression of various p63 isoforms in MDCK and melanoma cells mock-treated or treated with CPT, DOX, or nutlin-3 for 12 h. Cell lysates were collected and subjected to Western blot analysis using antibodies against p63, actin, and GAPDH. (C) Expression of various p63 isoforms in MDCK and D17 cells mock-treated or treated with CPT, DOX, or nutlin-3 for 12 h. The experiment was performed as described in (A-B).

Next, we determined the expression pattern of p73 in canine cells using BL906 antibody. BL906 was raised against amino acids 1-62 in human TAp73 and can only detect various TAp73 isoforms. We noticed that based on the expression pattern and molecular mass, the most predominant isoform in canine cells is likely to represent TAp73α (Fig. 4, marked by dot). We also found that the predominant TAp73 isoform was markedly induced in cells upon treatment with camptothecin and doxorubicin, but not nutlin-3 (Fig. 4A-C). In addition, in melanoma-36 and D17 cells, this isoform of TAp73 was found to be highly expressed (Fig. 4B-C). In sum, our results indicate that like their human counterparts, canine p63 and p73 can be up-regulated in cells upon DNA damage but not by Mdm2 inhibitor nutlin-3.

Figure 4. The expression pattern of p73 in canine cell lines.

(A-B) Expression of TAp73 in MDCK and melanoma cells mock-treated or treated with CPT, DOX, or Nutlin-3 for 12 h. Cell lysates were collected and subjected to Western blot analysis with antibodies against p73, actin, and GAPDH. (C) Expression of TAp73 in MDCK and D17 cells mock-treated or treated with CPT, DOX, or nutlin-3 for 12 h. The experiment was performed as described in (A-B).

The sensitivity of canine cells to DNA damage agents and Mdm2 inhibitor Nutlin-3

One of the most important functions of the p53 family proteins is to suppress cell growth (21). Thus, we determined if nutlin-3, which can induce canine p53 expression, has an effect on dog cell proliferation. To test this, cells were mock-treated or treated with 0.25, 0.5, or 1 μM nutlin-3, and then grown for 10 days. Colony formation assay showed that upon nutlin-3 treatment, the ability of the cells to form colonies was markedly inhibited in a dosage-dependent manner in MDCK and four melanoma cell lines (melanoma-12, -23, -50, and -64) (Fig. 5A-B and Supplementary data Fig. 1). It should be noted that in these cell lines, endogenous p53 is likely to be wild-type (Fig. 1A-B). However, melanoma-36 cells, in which endogenous p53 was undetectable and likely to be lost (Fig. 1B), were insensitive to nutlin-3 (Fig. 5C). Next, the sensitivity of canine cells to chemotherapeutic agents, camptothecin and doxorubicin, was similarly examined. We found that both camptothecin and doxorubicin greatly inhibited colony formation in MDCK and all the melanoma cells (Fig. 6A-C and Supplementary Fig 2) although melanoma-36 cells were slightly less sensitive to camptothecin (Fig. 6C). Similarly, the sensitivity of four osteosarcoma and Cf2Th cell lines to nutlin-3, camptothecin, and doxorubicin, was examined by colony formation assay. MDCK cell line was used as a control and found to be sensitive to nutlin-3 (Fig. 7A) as detected above (Fig. 5A). We found that the ability of osteosarcoma cells to form colony was inhibited upon treatment with nutlin-3 in a dose-dependent manner (Fig. 7B-C and Supplementary Fig. 3), consistent with the possibility that these osteosarcoma cell lines carry wild-type p53 (Fig. 1C-D). In addition, three osteosarcoma cell lines (#340529, #48-4, and #348617) were highly sensitive to camptothecin and doxorubicin (Fig. 8C and Supplementary Fig. 4) whereas D17 osteosarcoma cell line was sensitive to doxorubicin but not camptothecin (Fig. 8D). However, Cf2Th cell line was insensitive to nutlin-3 (Fig. 7D), camptothecin (Fig. 8B) and doxorubicin (Fig. 8B), consistent with the notion that Cf2Th carries a mutant p53 (Fig. 1D).

Figure 5. Nutlin-3 inhibits the ability of canine melanoma cells to form colony in p53- and dosage-dependent manners.

2,000 cells per well were seeded in a 6-well plate for 24 h, followed by treatment with or without nutlin-3 at a concentration of 0.25 μM, 0.5 μM, or 1 μM for 10 days. Cells were then stained with crystal violet for colony formation efficiency.

Figure 6. The sensitivity of MDCK and melanoma cells to DNA damage agents was measured by colony formation assay.

2,000 cells per well were seeded in a 6-well plate for 24 h, followed by treatment with or without 10 nM CPT or 18 nM DOX for 10 days. Cells were then stained with crystal violet for colony formation efficiency.

Figure 7. Nutlin-3 inhibits the ability of canine osteosarcoma cells, but not Cf2Th cells, to form colony.

(A-C) MDCK (A) and osteosarcoma (B-C) cells were seeded at 2,000 cells per well in a 6-well plate for 24 h, followed by treatment with or without nutlin-3 at a concentration of 0.25 μM, 0.5 μM, or 1 μM for 10 days. Cells were then stained with crystal violet for colony formation efficiency. (D) Cf2Th cells were seeded at 2,000 cells per well in a 6-well plate for 24 h, followed by treatment with or without 0.5 μM nutlin-3 for 10 days.

Figure 8. The effect of DNA damage on colony formation by MDCK, Cf2Th, and osteosarcoma cell lines.

2,000 cells per well were seeded in a 6-well plate for 24 h, followed by treatment with or without 10 nM CPT or 18 nM DOX for 10 days. Cells were then stained with crystal violet for colony formation efficiency.

Identification of a novel isoform of canine p21 and possibly human p21

As we showed above, canine p21 is expressed as two isoforms (Fig. 1) whereas human p21 is mainly expressed as one protein (15). To address why this occurs, we first sought to clone and express the canine p21 gene. According to the NCBI database, both human and canine p21 genes contain three exons. Nucleotide sequence analysis showed that the first exon is not conserved whereas the second and third exons are highly conserved between humans and dogs. However, the predicted second exon in the dog lacks the first 45 nt along with an ATG initiation codon, compared to that in human p21. In this regard, the canine genomic DNA sequence was searched for additional sequence homologous to the human p21. Indeed, a highly conserved p21 polypeptide was deduced from the dog genomic DNA sequence and showed in Fig. 9A. Like human p21, canine p21 protein also contains 164 aa (Fig. 9A). Canine p21 protein shares 82% amino acid identity to human p21 compared to 76% sequence identity between mouse and human p21 (Fig. 9A). Based on the p21 genomic DNA sequence, a pair of primers was designed to amplify the canine p21-coding sequence by RT-PCR with total RNA purified from MDCK cells. Next, to make sure that the amplified canine p21 cDNA truly represents the canine p21 gene, the cDNA was cloned into pCDNA3 expression vector and then expressed in both human RKO and canine MDCK cells. We found that like endogenous p21 in MDCK cells, two p21 isoforms were expressed by canine p21 cDNA and had the same migration patterns as endogenous dog p21 in both human and canine cells (Fig. 9B, p21 panel, compare lanes 3 and 5 with lane 4). This suggests that the dog p21 gene, but not the cell type, is responsible for p21 expression pattern. We would also like to note that the predominant form of human p21 detected in RKO cells migrated between the two dog p21 isoforms (Fig. 9B, compare lanes 1-2 with lane 3).

Figure 9. Identification of a novel isoform of canine p21.

(A) Sequence similarity among dog, human, and mouse p21 proteins. (B) Over-expression of canine p21 in RKO and MDCK cells. 3 μg of pCDNA3 control vector and canine full length p21-expressing vector were transfected into MDCK or RKO cells for 24 h, and then the level of canine p21 protein was measured by Western blot analysis with anti-p21 antibody. To stabilize endogenous human p21 in RKO cells as a positive control, RKO cells were treated with 250 nM CPT for 12 h. (C) Schematic presentation of canine p21 protein structure and various canine p21 mutants. (D) Over-expression of canine mutant p21(38-164), and full-length canine p21 in Cf2Th cells. 3 μg of control pCDNA3 vector and pCDNA3 vector that expresses canine mutant p21(38-164) or full-length canine p21 were transfected into Cf2Th cells for 24 h, and then the levels of p21 proteins were measured by Western blot analysis with C-19 anti-p21 antibody. (E) Identification of the region in the p21 protein required for two isoform expression. 3 μg of control pCDNA3 vector and various canine mutant p21-expressing vectors as shown in (C) were transfected into Cf2Th cells for 24 h, and then the levels of p21 proteins were measured by Western blot analysis with anti-HA antibody. (F) Schematic presentation of wild-type and various mutated canine and human p21 constructs. (G) A similar migration pattern of p21 proteins between canine and human. Wild-type and various mutated canine p21 constructs were transfected into Cf2TH cells and the levels of the canine p21 proteins were measured with anti-HA antibody (lanes 1-8). Similarly, wild-type and various human p21 constructs were transfected into 293T cells and the levels of human p21 proteins were also measured by anti-HA antibody (lanes 9-16). The lower bands in the middle panel (lanes 9-16) were artificially drawn to indicate that the expressed human p21 proteins are corresponding to the upper band of the dog p21 doublets.

Next, to determine if the two isoforms of dog p21 is due to alternative translation start site, p21(38-164), an N-terminally truncated p21 was cloned, which begins at codon 38, a potential ATG initiation codon (Fig. 9C). Both canine full-length p21 cDNA and canine p21(38-164) were transfected into Cf2Th cells since endogenous p21 in Cf2Th cells was found to be undetectable (Fig. 1D, lanes 1-4). Like the expression pattern in RKO and MDCK cells (Fig. 9B), canine p21 was expressed as two isoforms (Fig. 9D, lane 2). In addition, the N-terminally truncated p21(38-164) was also expressed as two isoforms (Fig. 9D, lane 3), suggesting that expression of two p21 isoforms are not due to the usage of alternative translation start codons. On the other hand, it suggests that the region from aa 1 to 37 is not responsible for expression of two p21 isoforms in the dog.

To further characterize the region in the dog p21 gene responsible for two isoform expression, various deletion mutants of canine p21 cDNAs were generated (Fig. 9C), including p21(1-152), p21(1-142), p21(1-129), p21(1-121), and p21(1-93). To facilitate protein detection, canine p21 was tagged with an HA epitope at its N-terminus (Fig. 9C). We found that two isoforms were expressed from mutant p21(1-152) and p21(1-142), but not p21(1-129), p21(1-121), and p21(1-93) (Fig. 9E). Interestingly, we noticed that the p21 mutant (1-129) migrated between the two isoforms of the mutant p21(1-142) although it is 13 aa less (Fig. 9E, compare lane 4 with 5). These results suggest that the region from aa129-142 is responsible for the expression of two p21 isoforms in the dog. To further delineate the region of aa 129-142, additional deletion mutants of canine p21 cDNAs were generated (Fig. 9F), including p21(1-141), p21(1-140), p21(1-139), p21(1-138), and p21(1-136). We found that all p21 mutants were expressed as two isoforms (Fig. 9G, left panel). However, the expression patterns of the two isoforms were different among these mutants. The difference in migration between the two isoforms expressed by p21(1-152), p21(1-142), p21(1-141), and p21(1-140) were much greater than that expressed by p21(1-139), p21(1-138), and p21(1-136), suggesting that aa 139 is one of the key determinants for two p21 isoform expression. Due to the remarkable sequence similarity between dog and human p21, we examined whether human p21 has similar patterns in migration. To address this, multiple human p21 deletion mutants, which correspond to the canine p21 mutants, were generated (Fig. 9F). We found that although the full-length human p21 cDNA expresses one major polypeptide, aa 139 in human p21 appears to play a similar role as that in canine p21 for various migration patterns. The lower bands in the middle panel (lanes 9-16) were artificially drawn to indicate that the expressed human p21 proteins are corresponding to the upper band of the dog p21 doublets.

Discussion

The expression pattern of the p53 family members in canine cells

The need for more appropriate animal models in cancer research has led researchers to consider companion dogs with spontaneously occurring neoplasms as a valuable and still under used resource. Due to the similarity in genes and physiology between dogs and humans, dogs may act as environmental sentinels, serve as a model to address cancer etiology in humans, and be enrolled in therapeutic trials and thus facilitate clinic applications of potential drugs to humans. Since little is known about the pathogenesis of tumors in companion dogs, we have sought to establish a dog model for the p53 family pathway. We found that canine p53 is up-regulated in response to DNA damage and Mdm2 inhibition, which is accompanied by induction of p53 target genes, p21 and MDM2. We also found that p63 and p73 are expressed in canine cells and induced in response to DNA damage but not Mdm2 inhibition. This suggests that the canine p53 family members, like their human counterparts, can be up-regulated under a stress condition.

The role of the p53 family in growth suppression

Both human and dog melanomas are highly resistant to radiation- and chemo-therapy but low in p53 mutation (22-24). In this study, we found that among the five melanoma cell lines, only one is deficient in p53 whereas the rest of four lines appear to have a functional p53 pathway. Interestingly, the p53-deficient cell line has much higher levels of p63 and p73 compared to other melanoma cell lines we tested. We also found that the four melanoma cell lines containing wild-type p53 are sensitive to DNA-damaging agents and Mdm2 inhibitor nutlin-3. In contrast, the p53-deficient melanoma cell line is insensitive to Mdm2 inhibitor, but still relatively sensitive to DNA-damaging agents, suggesting that inhibition of cell growth in response to DNA damage is likely due to p63 and p73. Thus, the five melanoma lines are categorized into two groups: one with functional wild-type p53 along with low expression levels of p63 and p73 and the other without functional p53 but with high expression levels of p63 and p73. The finding that nutlin-3 can suppress cell growth in canine melanoma and osteosarcoma cells in p53- and dosage-dependent manners leads us to hypothesize that dog spontaneous tumor models should be explored to test the feasibility of targeting the p53 pathway to manage certain cancers in humans, considering that the induced non-spontaneous mouse tumors are not ideal for studying spontaneous human tumors.

The identification of a novel isoform of canine and possibly human p21

In this study, we have found that canine p21 is expressed as two isoforms, both of which can be induced in cells upon treatment of DNA damage agents or nutlin-3. We then mapped the canine p21 gene and found that the region from aa 129 to 142 is required for the expression of two canine p21 isoforms. Interestingly, only three amino acids, that is, aa 129, 133, and 136, are different between dog and human p21 (Fig. 9A). In addition, we showed that aa 139 is one of the key determinants for two p21 isoform expression (Fig. 9G). It is possible that these variations are responsible for the increased expression of the small isoform of p21 in the dog. While the full-length human p21 cDNA expresses one major polypeptide, aa 139 in human p21 appears to play a similar role as that in canine p21 for various migration patterns (Fig. 9G). Thus, future studies to use a hybrid p21 construct from the human and canine p21 genes would be likely to uncover the mechanism by which various p21 isoforms are generated and how the migration patterns for various mutated p21 proteins occur. One possibility is that in both normal and cancer cells, trans-splicing of mRNAs has been shown to generate a novel isoform of a gene (25) and might be responsible for two p21 isoform formation in dog.

Several alternate human p21 transcripts have been identified by us and other groups (26-28). However, these human p21 transcripts encode the same protein except p21B (26). Notably, Tchou et al. reported that a fast migrating C-terminal truncated p21 protein was expressed in A549 lung carcinoma cells upon treatment with phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA), a potent activator of protein kinase C (29). In addition, Poon et al. found that a p21 isoform, named p21Δ, can be induced in several tumor cells by UV-irradiation (30). Interestingly, these p21 isoforms are found to be expressed in the cytoplasm although it is not certain how these isoforms are generated. Indeed, increased cytosolic p21 expression has been found in breast cancer, which is linked to poorer prognosis (31, 32). These observations point out a possibility that the human p21 gene expresses other isoforms, which may have different biochemical and biological functions. We believe that through elucidating how dog p21 is expressed as two isoforms and whether these isoforms have common or distinct functions, we may be able to uncover the mechanism by which various human p21 isoforms are generated and potential common and distinct functions for these human p21 isoforms.

Materials and Methods

Reagents

Antibodies against p53, p21, MDM2, p63, 53BP1, and GAPDH were purchased from Santa Cruz Biotechnology (Santa cruz, CA). Anti-actin and DAPI were purchased from Sigma (St. Louis, MO). BL906 antibody was purchased from Bethyl Laboratories (Montgomery, TX).

Cell culture, Clonogenic assay, and transfection

Canine melanoma and osteosarcoma cell lines have been defined and used as previously described. MDCK (ATCC #CCL-34), D17 (ATCC #CCL-183), Cf2Th (ATCC #CRL-1430), and HEK293T cell lines were purchased from ATCC (Rockville, MD). These cells were maintained in DMEM (Dulbecco/Vogt Modified Eagle’s Minimal Essential Medium) supplemented with 10% fetal bovine serum (FBS) and 1% non-essential amino acid. Lipofectamine 2000 was used for transfection of p21 cDNAs according to the manufacturer’s instructions (Invitrogen). For detection of p53 protein, cells were treated with or without 250 μM camptothecin, 250 ng/ml doxorubincin, or 4 μM nutlin-3 for 12 h. For colony formation assay, 2000 cells per well in triplicate in a 6-well plate were seeded. After 24 h, cells were continuously treated with or without nutlin-3 (0.25 μM, 0.5 μM, or 1.0 μM), 10 nM camptothecin, or 18 nM doxorubicin for 10 days. Cells were then fixed in methanol/glacial acetic acid (7:1), washed with water, and stained with crystal violet (0.2 g/L).

Immunofluorescence Assay

The assay was performed as described (33). Briefly, cells were grown on four-well chamber slides and treated as indicated. Cells were then fixed and incubated with primary antibody overnight, followed by incubation with fluorescein isothiocyanate (FITC)- or Texas red-conjugated secondary antibodies (Jackson ImmunoResearch and Molecular Probes) for 2 h. Cells were also stained with 4′,6-Diamidine-2′-phenylindole dihydrochloride (DAPI) (Sigma) to visualize nuclei. Intracellular localization of proteins was analyzed by immunofluorescence microscopy.

Western blot analysis

The assay was done as described (34). Briefly, cell lysates were made with 2x SDS sample buffer and boiled for 5 min. Proteins were then resolved in an SDS-PAGE gel and transferred to nitrocellulose membranes. Membranes were then subjected to blocking, washing, antibody incubation, and detection by enhanced chemiluminescence.

Plasmids

To clone canine p21 cDNA, total RNA was isolated with Trizol reagent according to the manufacturer’s instructions (Invitrogen) and RT-PCR was performed with the the iScript cDNA synthesis kit (Bio-Rad Labortories) according to the manufacturer’s instructions. The PCR program used for amplification was: (1) 94°C for 5 min, (2) 94°C for 45 sec, (3) 58°C for 45 sec, (4) 72°C for 1 min, and (5) 72°C for 10 min. From steps 2-4, the cycle was repeated 25 times. The primers used for amplifying full-length p21-coding cDNA were: upstream primer 5′-GCC ATG TCG GAG CCG TCC AGG G-3′ and downstream primer 5′-ATC CGA ATT CAG ATT AGG GCT TCC TCT TGG AG-3′. The PCR products were cloned into pGEMT-easy vector for sequencing. To generate canine p21(38-164), a PCR fragment was amplified using the full-length p21-coding cDNA as a template with upstream primer 5′-ATG GCC AGC TGT GTG CAA GA-3 and downstream primer 5′-ATC CGA ATT CAG ATT AGG GCT TCC TCT TGG AG-3′. The PCR product was then cloned into pcDNA3 vector at EcoRI and HindIII sites. To generate HA-tagged full-length p21-coding cDNA, a PCR fragment was amplified using the full-length p21-coding cDNA as a template with upstream primer 5′-ATC CAA GCT TGC CGC CAT GTA CCC ATA CGA TGT TCC AGA TTA CGC TTC GGA GCC GTC CAG GGA CG-3′ and downstream primer 5′-ATC CGA ATT CAG ATT AGG GCT TCC TCT TGG AG-3′. The PCR product was then cloned into pcDNA3 vector at EcoRI and HindIII sites. To generate various HA-tagged p21 mutants, various PCR fragments were amplifed using the same upstream primer for HA-tagged full-length p21-coding cDNA and a specific downstream primer for each mutant. These PCR products were then cloned into pcDNA3 vector at EcoRI and HindIII sites to generate each individual canine p21 mutant. The downstream primer for canine p21(1-152) is 5′- ATC CGA ATT CTT AGT GAT AGA AAT CTG TCA TGC-3′. The downstream primer for canine p21(1-142) is 5′-ATC CGA ATT CTT ACC GTT TTC GGC CCT GAG AGG-3′. The downstream primer for canine p21(1-141) is 5′- ATC CGA ATT CTT ATT TTC GGC CCT GAG AGG TGC C -3′. The downstream primer for canine p21(1-140) is 5′- ATC CGA ATT CTT ATC GGC CCT GAG AGG TGC CAG GCA C -3′. The downstream primer for canine p21(1-139) is 5′- ATC CGA ATT CTT AGC CCT GAG AGG TGC CAG GCA C -3′. The downstream primer for canine p21(1-138) is 5′- ATC CGA ATT CTT ACT GAG AGG TGC CAG GCA CAC-3′. The downstream primer for canine p21(1-136) is 5′- ATC CGA ATT CTT AGG TGC CAG GCA CAC CCG GGG A -3′. The downstream primer for canine p21(1-129) is 5′-ATC CGA ATT CTT ATG CCT CAG GCC GCT CAG GGG-3′. The downstream primer for canine p21(1-121) is 5′-TTA AGG CAG GAG GGT GCA GGT CAG-3′. The downstream primer for canine p21(1-93) is 5′-TTA CTT GCC CCC TCC CAG GT CATC-3′. To generate various human HA-tagged p21 mutants, a similar strategy was used as that for canine p21 mutants except that different primers were used. Specifically, the upstream primer for the full-length human p21 and various p21 mutants is 5′- ATC CAA GCT TGC CGC CAT GTA CCC ATA CGA TGT TCC AGA TTA CGC TCA GAA CCG GCT GGG GAT G-3′. The downstream primer for human p21(1-152) is 5′- ATC CGA ATT CTT AGT GGT AGA AAT CTG TCA TGC TGG TC -3′. The downstream primer for human p21(1-142) is 5′- ATC CGA ATT CTT ACC GTT TTC GAC CCT GAG AGT CTC C -3′. The downstream primer for human p21(1-141) is 5′- ATC CGA ATT CTT ATT TTC GAC CCT GAG AGT CTC CAG G -3′. The downstream primer for human p21(1-140) is 5′- ATC CGA ATT CTT ATC GAC CCT GAG AGT CTC CAG G -3′. The downstream primer for human p21(1-139) is 5′- ATC CGA ATT CTT AAC CCT GAG AGT CTC CAG GTC CAC C -3′. The downstream primer for human p21(1-138) is 5′- ATC CGA ATT CTT ACT GAG AGT CTC CAG GTC CAC C -3′. The downstream primer for human p21(1-136) is 5′- ATC CGA ATT CTTA GTC TCC AGG TCC ACC TGG GGA C -3′.