The Israeli blind subterranean mole rat belongs to the super-species Spalax ehrenbergi. This group of species consists of at least 12 allospecies in the Near East. The four allospecies in Israel were intensively studied interdisciplinarily across all organizational levels (8, 9). They adapted to four different climatic regimes: Spalax galili (2n = 52) inhabits the cool–humid Upper Galilee Mountains; Spalax golani (2n = 54), in the cool, semi-humid Golan Heights; Spalax carmeli (2n = 58), in the warm–humid central Israel; and Spalax judaei (2n = 60), in the warm–dry southern regions. Hypoxia and hypercapnia climax in the northern S. galili and S. golani, where high rainfall and floods occur during wintertime (8, 9). The entire life cycle of Spalax is spent underground in subterranean burrows at decidedly low-oxygen tension. Hypoxia climaxes during winter rains and flooding, when gas solubility and permeability are restricted and the animal is occupied in tunnel rebuilding (8, 9). The critical pO₂ (the lowest oxygen pressure supporting life) is lower for Spalax than for any other mammal studied to date (8, 9). We measured 7% O₂ inside the Spalax burrows after rains (unpublished data) and in the laboratory, where it can survive at least 14 h under 3% O_2. Under the same conditions, Rattus dies after 2–4 h. This capacity to withstand hypoxic conditions exceeds that manifested by many high-altitude and diving mammals. Since its origin some 40 million years ago, Spalax has evolved multiple adaptations to underground life that allows it to survive and carry out intense activities in a highly hypoxic environment (8–11), including different structure and function of vascular endothelial growth factor (VEGF) (10), hemoglobin (12), haptoglobin (13), myoglobin (14), erythropoietin, neuroglobin, and cytoglobin (unpublished data).

We speculate that Spalax also developed hypoxia adaptation mechanisms that are analogous to the genetic and epigenetic alterations in p53 acquired by cancer cells during tumor progression. We cloned Spalax p53 and compared the sequence with that of the germ-line p53 sequences of humans and mice, as well as to a p53 tumoral mutation database. Two Spalax-specific substitutions in the p53 DNA-binding domain that are identical to known human tumor-associated mutations, altering arginine (R) to lysine (K) in Spalax, were discovered at positions corresponding to codons 174 and 209 in human. Further comparison to an additional 41 different species and subspecies indicated that only codon 174 is highly conserved. The impact of these alterations on the transcription activation of several p53 target genes, using reporter assays, indicates a bias toward G₁ arrest versus apoptosis.

Animals. Spalax was captured in the field and housed in individual cages in the animal house of the Institute of Evolution, University of Haifa. Individuals used in this study were S. galili (2n = 52) from the Kerem Ben Zimra population, S. golani (2n = 54) from the Quneitra population, S. carmeli (2n = 58) from the Mt. Carmel population, and S. judaei (2n = 60) from the Anza population (9). The Ethics Committee of the University of Haifa approved all experiments.

Cloning Spalax p53. Spalax p53 cDNA was cloned from mRNA prepared from a whole embryo belonging to the S. judaei species (GenBank accession no. AJ7-83406). Isolation was achieved by using RT-PCR, using oligos at the 5′ and 3′ UTRs, synthesized according to published human (GenBank accession no. NM000546) and mouse (GenBank accession no. NM011640) p53 sequences.

To confirm the authenticity of the amino acid sequence observed in the binding domain of Spalax p53, this region was further cloned from spleen mRNA of one adult from each of the four different Israeli Spalax species, using RT-PCR and Spalax-specific oligos flanking the Spalax p53-binding domain.

Plasmids. For transfection studies, we constructed expression plasmids of the human p53, human mutated p53, and Spalax p53. Human wild-type p53 was ligated into a pCMV (cytomegalovirus) construct. Two different human mutated p53 constructs (H179R and 342-stop) were generated by using QuikChange XL site-directed mutagenesis kit following the manufacturer's instructions (Stratagene). Spalax p53 was ligated into a pCDNA3 expression vector. Different luciferase constructs containing the p53 response elements of apaf1 (15), bax, puma, noxa, cycG, p21, pten, and mdm2 (a generous gift of M. Oren, Weizmann Institute of Science, Rehovot, Israel) were used.

Cell Cultures. U2OS (human osteosarcoma, ATCC CRL-5803) and H1299 (human non-small cell lung cancer, ATCC HTB-96) were grown in 10% FCS in DMEM (GIBCO/BRL) or RPMI medium 1640 (Sigma), respectively, supplemented with 2 mM glutamine, 100 μg/ml streptomycin, and 100 units/ml penicillin (GIBCO/BRL) at 37°C in a humidified incubator with 5% CO₂.

Transient Transfection. Transient transfection assays were performed on H1299 and U2OS cell lines, using JetPEI transfection reagent (Polyplus Transfection, Ilkirch, France). Cells were plated at 250,000 cells per well in 12-well plates. One day (24 h) after plating, the medium was replaced by 0.5 ml of serum-free medium and a mixture of the various luciferase reporter constructs containing the human promoters of several p53 target genes (500 ng, pGL2-basic vector, Promega), and p53 plasmids (1 μg, pCDNA3 and pCMV vector) were added for 5 h, after which the medium was changed back to one containing 10% FCS. Dual luciferase reporter assay system (Promega) was used with results normalized to Renilla luciferase assay system (cytomegalovirus vector, Promega). The next day, plates were washed twice with cold PBS, collected, and immediately taken for quantification in a luminometer.

Humanization of the Spalax p53 Plasmids. The exchange of arginine instead of lysine at positions 172 and 207 (positions 174 and 209 in humans) of the wild-type Spalax p53 plasmid was done by using the QuikChange XL site-directed mutagenesis kit following manufacturer instructions (Stratagene). The amino acid changes were introduced into the plasmid by using the following primers: K172R forward, 5′-GAC AGA AGT TGT AAG GCG CTG CCC CCA CC-3′; K172R reverse, 5′-GGT GGG GGC AGC GCC TTA CAA CTT CTG TC-3′; K207R forward, 5′-GCA GAG TAT TTG GAT GAC AGG CAC ACT TTT CGA CAC-3′; K207R reverse, 5′-GTG TCG AAA AGT GTG CCT GTC ATC CAA ATA CTC TGC-3′ (underlines indicate the exchanges made for “humanization” of Spalax p53).

Three-Dimensional Structure Analysis of Spalax p53. The structure of the DNA-binding domain of Spalax p53 was modeled with the homology-modeling program nest (16), using the template structure of DNA-bound human p53 (16) (Protein Data Bank ID code 1TSR). The nest program was also used for modeling a dimer of the DNA-binding domain of Spalax based on the dimer that comprises the free-DNA monomer and bound-DNA monomer seen in the crystal structure of human p53 (17). A second dimer was modeled based on a dimer seen in the mouse p53 crystal structure (18) (Protein Data Bank ID Code 1HU8). The solvent-accessible surface areas were calculated by using the surfv program.¶ Superimpositions of structures were made by using the c-alpha match program (20, 21) and insightii (Accelrys, San Diego). Pictures were made by using molscript (22) and raster 3d (23).

Phylogenetic Tree. A phylogenetic tree was computed by using the neighbor-joining method (24) based on Jukes–Cantor distances.

Multiple Alignment Tools. For comparing the p53 protein of 44 different species a multiple sequence alignment was generated by using multalin (http://prodes.toulouse.inra.fr/multalin/multalin.html; ref. 25). For the phylogenetic tree, a multiple sequence alignment (MSA) of the DNA-binding domain of 36 p53's orthologues was constructed by using clustalw (26).

p53 Mutation database. For p53 mutation database, International Agency for Research on Cancer TP53 database (www.iarc.fr/p53/index.html; version R8, June 2003) was used (27).

Comparing the p53 Sequences of Spalax, Mouse, and Human. Spalax p53 was cloned, sequenced, and compared with the known human and mouse p53 mRNAs and proteins. Spalax p53 mRNA is 1,368 bases long, translated into a 391-aa protein, with 83.7% homology to human and 82.2% to mouse p53 mRNA. At the amino acid level, an identity of 85.4% to human and 81.9% to mouse p53 proteins is observed. In the p53 DNA-binding domain region (Fig. 1), there is a nucleotide sequence homology of 88.1% to humans and 86.1% to mice, whereas amino acid homology is 95.8% for humans and 89.5% for mice.

Fig. 1. Sequence of Spalax p53. Amino acid sequence alignment of Spalax (sp53), mouse (mp53), and human (hp53), with full-length p53 proteins (http://prodes.toulouse.inra.fr/multalin/multalin.html; ref. 25). The DNA-binding domain is shown in a box. Identities (more ...)

Multiple alignment analysis (http://prodes.toulouse.inra.fr/multalin/multalin.html; ref. 25) revealed that, in the DNA-binding region, two amino acids differentiate Spalax from human and mouse sequences (Fig. 1). The bases at positions 515 and 620 from the transcription-starting site of Spalax mRNA sequence are adenosine (A) bases, whereas a guanine (G) is present in the human and mouse. Thus, lysine (Lys, AAG) is coded at positions 172 and 207 in Spalax p53 protein (corresponds to codons 174 and 209 in human p53), and arginine (Arg, AGG/AGA) is coded in the human and mouse p53 proteins.

p53 Target Genes Activation: Human Versus Spalax. One of the major functions of p53 is its ability to bind to and activate several target genes. Because mutant forms of p53 may have different effects, we investigated the ability of Spalax wild-type p53 to activate the transcription of several human target genes. For this analysis, expression plasmids carrying either Spalax or human wild-type p53 coding regions were used. Two mutated human p53 plasmids were used as controls: mut179 (H179R) and Mut342-stop (342R). Spalax homologues of p53 target genes have not been cloned; therefore, several human promoter constructs were used. Transient transfections were performed on U2OS (without induction of endogenous wild-type p53) and H1299 (p53-null) human cell lines. Luciferase assays for eight different human p53 target genes, including apaf1, mdm2, bax, puma, noxa, cycG, p21, and pten, were performed. The targets selected are known to be involved in different pathways, including cell cycle arrest, apoptosis, and stabilization/homeostasis of the p53 protein. The relative transactivation capacities of the different p53 plasmids toward the eight target genes (Fig. 2) are summarized with respect to wild-type human p53 (detailed luciferase assays are shown in Fig. 5, which is published as supporting information on the PNAS web site).

Fig. 2. Relative transactivation of eight different human promoters. Human wild-type p53, mutated p53 (mut179 and mut342), and Spalax p53 were examined relative to an array of p53 target gene promoters. The relative transactivation with respect to human wild-type (more ...)

Overall, four different activation patterns were observed. Apaf1 promoter, although activated by the human wild-type p53, was not responsive to Spalax wild-type p53. Spalax p53 ability to activate the promoters of the proapoptotic genes noxa and puma was lower than that of the human p53 (≤0.5-fold of activation compared with human wild-type, P < 0.015) and similar to human wild-type p53 in regard to bax (0.98-fold of activation, P = 0.051). Increased activation of the promoters of pten and mdm2 by Spalax p53 protein was observed (up to 2-fold of activation compared with the human wild-type p53, P < 0.05), with an even greater impact on cycG and p21 promoters (4- and 6-fold of activation, respectively, P < 0.0015). The mutated human p53 used as negative controls resulted in a partial or complete loss of function in regard to all targets tested.

Humanizing Spalax p53 by Site-Directed Mutagenesis. To test whether the lysines in Spalax p53 protein are responsible for the inability to activate apaf1 promoter, Spalax p53 protein was humanized by using site-directed mutagenesis, changing amino acid lysine (K) to arginine (R) at positions 172 and/or 207 (corresponding to positions 174 and 209 in humans). Cotransfections with human apaf1 promoter luciferase construct, together with the different p53 Spalax plasmids (wild-type and humanized) were performed. Human wild-type and mutated p53 were used as controls. Humanization of Spalax p53 at position 174 resulted in a 3-fold increased activation of the apaf1 promoter (Fig. 3), whereas substituting lysine to arginine at position 209 did not have any effect, giving similar results as the original Spalax wild-type p53 protein. A synergistic effect was observed when both amino acids were altered to the human sequence, giving a 5-fold increased activation compared with the original Spalax wild-type p53 protein and 50% of the activity exhibited by human wild-type p53.

Fig. 3. Spalax wild-type and humanized p53 ability to transactivate Apaf1 promoter, with respect to human wild-type and mutated p53. Results are normalized to Renilla-luciferase and presented as fold-change from Spalax wild type (depicted as 1-fold of activation). (more ...)

Crystallography-Based Models of Spalax, Human, and Mouse p53. The structure of Spalax p53 was predicted, based on the human and mouse p53 models, to understand the role of Spalax-specific arginine-to-lysine amino acid changes at codons 174 (171 in mouse) and 209 (206 in mouse). According to human and mouse p53 crystallography-based models, it is established that p53 generates two types of dimers: one that binds DNA through the DNA-binding domain and a second type that is formed before the DNA contact (17, 18, 28).

In the DNA-bound human p53 model (Fig. 4A), Arg-174 and Arg-209 are not located near the DNA-binding site and thus are not involved in stabilizing the folding of the protein. However, according to the DNA-free human and mouse models (Fig. 4 B and C, respectively), Arg-174 contacts residues in the adjacent monomer, mainly Arg-181 (178 in mouse). This contact between two positively charged residues is stabilized by hydrophobic interactions within the arginine side chains and a nearby negatively charged Glu-180 in humans (177 in mice). In the p53 monomer, Arg-174 and -209 are exposed to the surface. By contrast, in the dimer conformation, the surface area of Arg-174 is buried in the dimer, 58% in humans and 85% in mice, whereas Arg-209 only partially participated in the interface, with 13% buried in the dimer in the mouse and totally exposed in the human. Interestingly, in Spalax, the smaller Lys-174 is incapable of mediating the same interactions as the human/mouse Arg-174/171 (Fig. 4 D and E, respectively); thus, the Spalax-formed dimer is unstable and probably adapts to a different contact area.

Fig. 4. Three-dimensional models of human, mouse, and Spalax proteins. (A) DNA-bound human p53 model. Arg-174 and Arg-209 are not located in or near the DNA-binding site and, thus, have no effect on the protein fold and/or stabilization in the DNA bound state. (more ...)

p53 Mutations in Human Cancer and in Underground-Adapted Spalax. Cells exposed to hypoxic conditions accumulate wild-type p53 protein and succumb to apoptotic cell death (29). In human cancer, wild-type p53 expression is often abrogated by somatic mutations. In fact, >50% of human tumors carry mutations in the p53 gene (1, 30). Hypoxia was suggested as a selection pressure for mutating p53 of solid tumors during conversion into ascite tumors, making the cells less prone to hypoxia-induced apoptosis (29, 31). Spalax is a blind mole rat that spends its life underground. Primarily during winter floods, it is exposed to extreme hypoxic conditions (8, 9). We hypothesized that this genus adapts to the extreme hypoxic condition by a battery of adaptive genes (8, 9) involving mutations in the p53 gene, analogous to p53 mutations acquired during tumor progression. This hypothesis is supported by our previous results (10) indicating that Spalax vascular endothelial growth factor (VEGF) expression is similar to its expression in cancer growths. Similar to VEGF expression pattern in human tumors, Spalax VEGF is constitutively expressed in maximal levels under normoxia, with no further increase under hypoxia. Here, we cloned the Spalax p53 mRNA and used reporter assays to establish that Spalax p53 protein is functionally biased toward G₁ arrest activation and impaired toward apoptosis target genes.

The Spalax p53 Lysine Adaptation Mimics Tumor Evolution. The Spalax-specific modifications observed within the p53 DNA-binding domain replace arginine (R) with lysine (K) in the Spalax p53 protein (corresponding to codons 174 and 209 in humans). Arginine at codon 174 is reported to be affected in 57 different tumors of various types (www.iarc.fr/p53/index.html), with an identical arginine-to-lysine exchange in eight different human cancers, including medulloblastoma (32), squamous cell carcinomas of the head and neck (33), oral cancer (34), breast cancer (35), colon cancer (36), ovary adenocarcinoma (37), lung cancer (38, 39), and liver cancer (40). Moreover, this specific mutation was reported in canine thyroid carcinoma (41). Arginine at codon 209 is mutated in 69 tumors, among which five of these tumors harbor the exact arginine-to-lysine alteration including B cell chronic lymphocytic leukemia (42), esophageal carcinoma (43), skin cancer (44), uterine cervical cancer (45), and colon cancer (46).

Spalax p53 Favors Growth Arrest, but Not Apoptotic Target Genes. A large-scale p53 mutation analysis research published recently (47) concluded that three p53 mutant subtypes exist: mutants with no activity (9.6%), mutants with significantly reduced activity but some residual activity (26.5%), and mutants that retain wild-type activity (63.9%). According to these categories, Spalax p53 protein is a partially inactivated protein version, working in a similar manner to several known p53 mutations, favoring the promoters of cell cycle arrest genes rather than apoptotic targets. Apaf1 promoter, a central-player in the mitochondrial apoptotic pathway, recently described by our group and others (15, 48, 49) as a p53 primary target, was completely inactivated by Spalax p53. Interestingly, a neighboring mutated V173L was also incapable of activating Apaf1 promoter (49). Remarkably, however, the same Spalax p53 protein overactivated targets involved in cell cycle and p53 stabilization/homeostasis, including p21, cycG, pten, and mdm2 (Figs. 2 and 5). Such results were obtained with 143A, 175P, and 181L human p53 mutants, retaining the ability to transactivate p21 reporter but failing to activate apoptotic target genes, including bax (143A and 175P) and IGF-BP3 (143A, 175P and 181L) (50). Furthermore, it was previously reported that several mutations of arginine at codon 175 to leucine (28), isoleucine, or serine (51) ablate apoptosis activity but function as a wild type in regard to cell cycle arrest. This phenomenon was also observed at codon 175 mutated from arginine to lysine, the same amino acid alteration observed within Spalax p53 protein (51). Recently, a mouse model was described (52) in which a p53 R175P mutation was introduced. This mutation, neighboring codon 174, which was studied by us, was also defective in initiating apoptosis while still inducing cell cycle arrest.

Several lines of evidence demonstrate that cell cycle arrest and apoptosis are separate p53 functions, with accumulating reports regarding p53 mutants that are able to induce cell cycle arrest while failing to activate apoptosis. The common concept that might explain observations of such differential activities is that cell cycle arrest targets of p53, such as p21, seem to be more efficiently activated than apoptotic targets (53).

Highlighting Spalax p53 Adaptation by 3D Models. The residues in the p53 DNA-binding domain that either directly contact DNA or closely contact those that do so are the most often mutated (54). As such, arginine at codon 175, neighboring the Arg-174 altered in Spalax, located in the L2 loop near the zinc-binding site of the p53 protein (54–56), is one of the hot spots for mutation in human cancers.

Three-dimensional analyses based on human and mouse p53 protein models were conducted to understand the effect of Spalax-specific amino acid changes. Previous human and mouse p53 crystallography-based models established that p53 generates two types of dimers: one that binds DNA through the DNA-binding domain and a second one, which is a non-DNA-bound dimer, that forms before the DNA contact (17, 18, 28, 57). When p53 is bound to DNA, a Spalax lysine-to-arginine alteration at codons 174 and 209 is not located in or near the DNA-binding site and thus has no effect on the protein fold and stabilization. These characteristics allow the replacement of arginine by lysine in Spalax p53 protein without totally abolishing its function, as confirmed by functionality assays conducted in this work (Fig. 2). The H179R p53 mutation, used as a control in our reporter assays, located in a critical zinc-binding region (17), results in a complete loss of function in regard to all promoters tested (Fig. 2). However, in the non-DNA-bound transition state, although Arg-209 plays a minor role, Arg-174 is essential to protein–protein interactions. We suggest that the Lys-174 in Spalax p53 protein assumes, before DNA binding, a dimer conformation that results in the activation of different p53 target genes when compared with human p53. Moreover, the ability of the same protein to differentially express a variety of responsive elements, as demonstrated in our work and by others, may be explained by divergent affinities of the same non-DNA-bound state of the p53 dimer toward the various target gene DNA strands.

Comparative p53 DNA-Binding Domain Across Animal Species. According to all data presented in our work concerning activation of the Apaf1 promoter, the importance of the arginine-to-lysine alteration at codon 174 in comparison with codon 209 was established. These results were further confirmed when a large-scale multiple alignment analysis was conducted between Spalax and a spectrum of 41 other species besides humans and mice (Fig. 6 and Table 1, which are published as supporting information on the PNAS web site). According to the multiple alignments conducted, the p53 DNA-binding domain is highly conserved among the different species and subspecies examined. In evolution, sequences coding for the most critical structural or functional amino acids are the most highly conserved because of selection pressures against functionality-altered proteins. Previous comparisons between species indicated five highly conserved regions in the p53 protein (58). The arginine at codon 174 is situated in the third conserved region, characterized as a hotspot for missense mutations (59). Furthermore, multiple alignment performed among 44 species indicated that the amino acid at position 209 in human, originally assumed to be Spalax-specific when comparing Spalax with humans and mice (Fig. 1), is not a Spalax-unique change. Actually, there is a vast variability at this nonconserved region, with 10 different amino acid possibilities among the species examined (Fig. 7, which is published as supporting information on the PNAS web site).

Focusing on the region surrounding the amino acid at codon 174 (Fig. 7) revealed that this amino acid is conserved as arginine (R) among 38 of the species, whereas only Spalax and six other species carry a lysine (K) residue at this position. These species include Xenopus (African frog), Monodelphis domestica (opossum), two species of Xiphophorus (tropical fish), Loligo forbesi (squid), and Mya arenaria (softshell clam). All of these species, except the opossum, live in hypoxic aquatic and/or muddy bottoms.

Intriguingly, in the group containing the R174K alteration, four species are commonly used as animal models for melanoma, including M. domestica (South American opossum), M. arenaria (60), and a hybrid of two Xiphophorus species, the helleri and maculatus (61). M. domestica and the Xiphophorus rarely develop melanoma in the wild, but melanoma can be induced by UV irradiation. These species are among the very few animals in which melanoma can be induced by UV alone. The Xiphophorus is also actively used as a genetic model organism for the study of neuroblastoma, retinoblastoma, and other malignancies; however, this system is ultimately limited by its evolutionary distance from mammals (62).

Concerning Xenopus laevis and Loligo forbesi, two aquatic species, the Xenopus is submerged for long periods of time, thus making standard inhalation techniques impractical (63), and Loligo forbesi lives in deep water (250–500 m) (64). According to a phylogenetic tree conducted on the p53 DNA-binding domain of Spalax and 35 other species (Fig. 8, which is published as supporting information on the PNAS web site), Spalax is genetically closer to humans and primates than other species examined. Note that similar results were found for other Spalax proteins, including CLOCK (65), Per3 (66), CRY1 (67), and HLA (8), indicating the old origin of Spalax among rodents (8, 9), closer to the primate junction.

To conclude, p53 controls the expression of multiple target genes, and it is therefore considered to be a master gene of diversity, providing the economic creation of multiple germ-line phenotypes (57). Clearly, as observed by various p53 tumor mutant forms along with Spalax p53 reported by us, a single amino acid change could have diverse effects on the ability of p53 to induce the transcription of different target genes. Examples of specific tumoral p53 mutations acquired in response to various environmental stresses are well documented, including exposure to sunlight with skin cancer and tobacco smoking with lung cancer (19). It might be assumed that various p53 mutations may be related to specific types of cancer, such as in UV-induced skin cancer of mice in which a T122L mutation is frequently noticed (57). As such, it is a challenge to establish a connection between the stressful hypoxic underground environments of Spalax and the specific amino acid alteration observed in this genus.

Since its origin some 40 million years ago, Spalax has evolved adaptations to underground life that allow it to survive and carry out intense activities in a highly hypoxic environment (8, 9). We believe that such adaptive mechanisms for hypoxia, analogous to the genetic and epigenetic alterations acquired by cancer cells during tumor development, are demonstrated in Spalax p53 in the present study. We suggest that Spalax holds a promising model for the study of malignancies, highlighting the importance of a large-scale Spalax-genome project in the near future. This may lead to a better understanding of hypoxia stress tolerance and to the elucidation of malignancy evolution and suppression.

This study is in honor of the late Professor N. N. Vorontsov, a great scientist and distinguished evolutionist. We thank G. Klein, H. Werner, E. Feinstein, and N. Minsky for comments improving the paper. We thank the Arison family for their donation to the Center of DNA Chips in Pediatric Oncology, Chaim Sheba Medical Center. G.R. holds the Djerassi Chair in Oncology at the Sackler School of Medicine. This work was supported by the Ancell Teicher Research Foundation for Genetics and Molecular Evolution.