Abstract
Organ-specific cancers with activated ras oncogenes most often are associated exclusively with only one ras isoform. For example, only H-ras activation is associated with rat mammary cancers. The mechanism underlying this specificity is mostly unknown. We have shown previously that this tissue specificity of Ras isoforms is defined by the Ras protein itself and not by differential gene expression among Ras family members. Here we show that elements in the known domains in the hypervariable region of Ras (amino acids 170–189) interact in part to control this mammary/H-Ras specificity. In addition, these in vivo mammary studies for the first time identify domains in the mostly homologous region of Ras (amino acids 1–169) that strongly influence the oncogenic potency/specificity of H-Ras.
Introduction
Activated ras is associated with ∼40% of human cancers. Most organ-defined human and rodent cancers with ras activation are associated exclusively with only one isoform of the very homologous members of the Ras family. For example, in humans, K-ras activation is associated with pancreatic and colon cancers; H-ras activation is associated with bladder cancers; and N-ras activation is found in myeloid leukemia (1). In rodents, K-ras is associated with colon and lung cancers, whereas H-ras is associated with mammary and skin cancers (2, 3). The molecular basis of this organ-specific ras activation remains unknown. However, we have shown previously that the activated H-ras is ∼10-fold more potent than the activated K-ras in mammary carcinogenesis and that this organ specificity in the rat mammary gland resides in the Ras protein itself and not in differential regulation of ras family gene expression (4).
Ras proteins can be divided into the homologous region (aa3 1–169; >90% homology) and the hypervariable region (aa 170–189; 10–15% homology; Refs. 5, 6). To define the regions of H-Ras that contribute to its organ specificity in mammary carcinogenesis, we first focused on two regions in the hypervariable region that participate in membrane localization (7, 8, 9, 10). The hypervariable region is known to control Ras isoform cellular localization by a mechanism in which H-Ras travels through the Golgi to the lipid rafts of the plasma membrane, whereas K-Ras bypasses the Golgi and is dispersed throughout the plasma membrane (11, 12, 13). These studies next led us to examine the few nonhomologous regions between H-Ras and K-Ras within the homologous region of Ras. We find that both Ras domains with assigned function as well as those with previously undefined function contribute to the specificity/potency of mammary cancer induction.
Materials and Methods
Construction of ras Chimeras.
To generate ras carboxyl domain exchange constructs between H-ras and K-ras, a HindIII restriction site was created by introducing point mutations into codons 170 and 171 of H-ras cDNA, and codon 170 of K-ras. All of the ras chimera constructs used contain the oncogenic activating mutation at codon 12 (G 35 to A). The existing HindIII restriction site at codon 5 of H-ras and K-ras was removed by PCR-based mutagenesis. Exchanging the 3′ ends of H-ras and K-ras cDNAs by HindIII digestion resulted in H-ras and K-ras carboxyl domain exchange constructs (HKK and KHH). After the exchange of regions that code for the carboxyl terminus, the point mutations engineered during cloning were restored to the original sequences of H-ras and K-ras. A 26-bp linker of v-H-ras 5′-untranslated sequence was fused to the 5′ end of codon 1 of the carboxyl domain exchange constructs to match the sequence of the control constructs, JR/H-ras and JR/K-ras (4).
To exchange the cysteine-aliphatic-aliphatic-any residues (CAAX) boxes of the carboxyl-terminal ends between H-ras and K-ras (HHK and KKH), PCR reactions were performed with H-ras cDNA using a reverse primer, caaxR1, containing the carboxyl-terminal sequence of H-ras with the K-ras CAAX box (CVIM; 5′-ACGCGTCGACTCACATGACTATACACTTGCAGCTCATGCA-3′, the K-ras sequence is underlined for all of the oligos) or with K-ras cDNA using a reverse primer, caaxR2, containing the carboxyl-terminal sequence of K-ras with the H-ras CAAX box (CVLS; 5′-ACGCGTCGACTCAGGACAGCACACACCTTGTCCTTGACTT-3′). To generate chimeras containing exchanged secondary localization signals (HKH and KHK), HKK or KHH cDNA was used as a template in PCR with the caaxR2 reverse primer or caaxR1 reverse primer, respectively.
To generate chimeras of the amino-terminal region of HKK (codons 1–169 for H-ras or 1–168 for K-ras), an AatII site was introduced at codon 109 in the constructs HKK and K-ras, and a SpeI site was introduced in K-ras at codon 158. The chimeras were then constructed by exchanging domains. The construct, α-HKK, is identical to HKK except for the AatII site. All of the constructs were sequenced to verify the absence of spurious mutations.
Vector Infusion Model.
All of the ras chimera constructs were subcloned into the BamHI and SalI restriction sites of the retroviral expression vector JR, in which ras is driven by the Moloney murine leukemia virus long terminal repeat (14). The preparation of concentrated retrovirus was performed as described previously (15). Viral stock (15 μl) at 1 × 107 CFU/ml mixed with Polybrene and a tracking dye was infused into each central duct of all 12 mammary glands, which results in mostly single cellular random integrations of ras into the DNA of the infected small proportion (<0.1%) of mammary epithelial cells (14). Virgin Wistar-Furth female rats at 8 weeks of age were used for all of the infusions. Mammary carcinomas were collected at necropsy (10–11 weeks after vector infusion) and used for histopathological and molecular analysis.
RNase Protection Assay.
The RPA III RNase protection assay kit (Ambion, Austin, TX) was used to determine the expression levels of ras from mammary carcinomas induced by retroviral infusion. A retroviral ras-specific DNA probe (182 bp) was generated using primers to the Moloney murine leukemia virus long terminal repeat region for the forward primer (5′-CCTCCATCCGCCCCGTCTC-3′) and codon 13 of H-ras/K-ras (5′-GCCTTCAGCGCCCACCACC-3′) for the reverse primer. 36B4 ribosomal phosphoprotein (120 bp) was used as an internal control probe. Radiolabeled riboprobes were synthesized using the T7 Maxiscript kit (Ambion) and [α-32P]UTP (NEN Life Science Products, Inc., Boston, MA). RNA was prepared from four carcinomas induced by each construct as well as two untreated normal mammary gland samples. The full-length probes were gel purified and hybridized overnight with 10 μg of total RNA extracted from mammary gland or mammary carcinomas using RNAzolB (Tel-Test, Friendswood, TX). After digestion with RNase following the manufacturer’s protocol, the protected fragments were resolved on a 5% denatured polyacrylamide gel, exposed to a phosphor screen (Molecular Dynamics, Sunnyvale, CA), and analyzed using ImageQuant software.
Immunoblotting.
Lysates were prepared from mammary glands and mammary carcinomas using cold lysis buffer [50 mm Tris (pH 8.0), at 4°C, with 150 mm NaCl, 2 mm EDTA, 10 mm Na2HPO4, 10 mm Na4P2O7-10H2O, 5 mm Na3VO4, 1% (v/v) Triton X-100, 0.1% (w/v) SDS, 0.5% (w/v) sodium deoxycholate, 0.25 mg/ml Pefabloc, 100 μg/ml soybean trypsin inhibitor, and 20 μg/ml leupeptin] (16). Protein (10 μg) was loaded onto a 15% PAGE and transferred onto an Immobilon-P membrane (Millipore, Bedford, MA). The membrane was then probed for total Ras protein using mouse anti pan-Ras antibody (Ab-3; Calbiochem, San Diego, CA) and reprobed for α-tubulin using mouse anti α-tubulin antibody (Ab-1; Calbiochem). The blot was developed with a SuperSignal West Pico Chemiluminescent Substrate kit (Pierce, Rockford, IL) and exposed to the film.
Construction of ECFP-Ras Fusion Vectors and Fluorescence Microscopy.
All of the ras constructs were used as templates in PCR to amplify the coding region of ras. Each PCR product was then cloned into pECFP-N1, Enhanced Cyan Fluorescent Protein Vector (Clontech, Palo Alto, CA) using BglII and SalI restriction digestion sites. These vectors were transfected into NIH3T3 cells using LipofectAMINE (Life Technologies, Inc., Rockville, MD). Fluorescence was observed using a laser scanning confocal microscope, consisting of an MRC-1000 laser scan head (Bio-Rad, Hercules, CA) equipped with a 15 mW krypton/argon laser and mounted transversely to an inverted Nikon Diaphot 200 microscope.
Statistical Analysis.
The effect of changes in the various combinations of aa sequences on the number of mammary carcinomas was assessed using generalized linear models assuming Poisson-distributed responses. All of the Ps are based on the χ2 test for significant change in the deviance. All of the data presented in the text, and graphs are statistically fitted means.
Results and Discussion
To test the hypothesis that the hypervariable region (aa 170–189) is a major contributor to the specificity of H-Ras in mammary carcinogenesis, portions of this domain from H-Ras and K-Ras were exchanged (Fig. 1, A and B). When these chimeric ras genes were placed into a retroviral vector and infused into the mammary gland ductal lumen (4, 15), carcinomas rapidly developed (Fig. 1 C). H-Ras induced an average of 6.8 carcinomas/rat, whereas K-Ras induced 0.63 carcinomas/rat. A chimera (KHH) in which the last 20 aa of H-Ras replaced those of K-Ras induced an average of 4.5 carcinomas/rat, 7-fold higher than the average number of carcinomas/rat obtained with K-Ras (P < 0.0001). This result revealed that the membrane localization region of H-Ras is able to greatly enhance the oncogenic potency of K-Ras. Unexpectedly, the number of carcinomas obtained in the reciprocal chimera (HKK), in which the last 20 aa of H-Ras were replaced by those aa of K-Ras (HKK), was 2-fold higher (an average of 13.8 carcinomas/rat) than that obtained using H-Ras (P < 0.0001). This result suggests that the membrane localization domain of Ras alone does not fully determine the organ specificity of activated Ras. If it did, we would expect that the HKK chimeric molecule would produce as few carcinomas as did K-Ras.
The membrane localization domain consists of two signal sequences for membrane targeting. For both H-Ras and K-Ras, the last four aa comprise the CAAX box, which is a region signaling for post-translational isoprenylation (7, 8). Additionally, in H-Ras, cysteines 181 and 184 are palmitoylated, whereas in K-Ras, there is a polylysine track (aa 176–181). These serve as secondary membrane localization signals for H-Ras and K-Ras (9, 10), respectively. Having established the importance of the entire carboxyl membrane localization domain in rat mammary carcinogenesis, we next determined the contribution of each membrane localization sequence (Fig. 1, B and C). Exchanging only the CAAX box between H-Ras and K-Ras had no significant effect on mammary carcinogenesis (KKH versus K-Ras, P = 0.72 and HHK versus H-Ras, P = 0.57 in Fig. 1 C). In contrast, exchanging only the secondary membrane localization signal between H-Ras and K-Ras did significantly affect mammary carcinogenesis. Replacing the polylysine track of K-Ras with the two palmitoylation sites of H-Ras generated an average of 1.7 mammary carcinomas/rat (KHK), 2.7-fold higher than K-Ras (P < 0.0001). Unexpectedly, the reciprocal replacement of the two palmitoylation sites of H-Ras by the K-Ras polylysine track generated an average of 38.9 mammary carcinomas/rat (HKH), which was dramatically higher than both the average of 6.8 mammary carcinomas/rat obtained with H-Ras (P < 0.0001) and the average of 13.8 mammary carcinomas/rat obtained with HKK (P < 0.0001). Thus, the effect of the secondary membrane localization signal was modulated by the specific sequence of the CAAX box, suggesting an interaction between these two regions.
A two-cubed factorial model was fit to the number of carcinomas generated by H-Ras, K-Ras, and Ras chimeras to analyze the interactions between aa 1–169 (homologous region), 170–185 (secondary signal), and 186–189 (CAAX box). The three-way interaction was significant (P < 0.0001). Also, the two-way interactions between the secondary signal and the CAAX box were significant when aa 1–169 were held constant (P < 0.01). This analysis suggested that complex interactions between various domains of Ras help define the organ specificity/potency of Ras in neoplastic transformation.
James et al. (17) suggested that both farnesyl protein transferase and geranylgeranyl protein transferase 1 contain a binding site for the polylysine sequence. In addition, K-Ras shows a 20-fold higher affinity to farnesyl protein transferase than H-Ras in vitro (18). The polylysine track of K-Ras was suggested to contribute to this high affinity to farnesyltransferase. Therefore, it is possible that substitution of the 6 lysines of K-Ras made the construct HKH a better substrate than H-Ras for either one or both isoprenyl protein transferases resulting in more efficient membrane localization and a higher number of carcinomas. Alternatively, it has been suggested that the hypervariable region of Ras may participate in interactions with the cysteine-rich domain of Raf (19, 20, 21). This may allow the individual Ras chimeras to differentially modulate the same effector proteins.
Comparison of all of the chimeras also strongly suggests that the few aa residues, which are variable within the largely homologous region of Ras (aa 1–169), play an important role in carcinogenesis. Up until now there has not been any function assigned to these residues. In all of the cases, constructs with aa 1–169 from H-Ras generated more mammary carcinomas than constructs with the corresponding K-Ras residues (P < 0.0001). This observation led us to generate additional H-Ras and K-Ras chimeras within aa 1–169, which differ from each other at only 13 aa, to additionally define the effects of this region on mammary carcinogenesis.
We divided the homologous region (aa 1–169) into three subregions based on the locations of the limited regions of differences between H-Ras and K-Ras. These chimeras, designated respectively as α, β, γ, and δ-HKK, all have the last 20 aa from K-Ras and various regions of aa 1–169 from either H- or K-Ras (Fig. 2,A). Relative to K-Ras, the chimera α-HKK (identical to HKK; see “Materials and Methods”) demonstrated that the replacement of K-Ras aa 1–168 with that of H-Ras resulted in 16-fold more mammary carcinomas (P < 0.0001; Fig. 2,B), confirming the above findings. β-HKK, in which aa 1–108 of K-Ras are substituted with these aa from H-Ras, generated 3.8-fold more mammary carcinomas than K-Ras (P < 0.0001, Fig. 2,B). Interestingly, all of the aa in the known functional domains within this region (GTP/GDP binding and effector domains) are 100% homologous. By replacing K-Ras residues of both regions 1 and 2 (aa 1–157) with those of H-Ras (γ-HKK), carcinogenesis was enhanced 33.6-fold over K-Ras (P < 0.0001; Fig. 2,B). Again no function has been assigned to these regions containing nonhomologous aa. δ-HKK, which is composed of region 1 (aa 1–108) and 3 (aa 158–169) of H-Ras, was 11-fold more potent than K-Ras (P < 0.0001; Fig. 2 B). Region 3, which contains a limited portion of the recently described linker region (22), tripled the carcinogenic potency of δ-HKK compared with β-HKK. These chimeras demonstrate that the few heterologous regions of Ras within the Ras homologous region (aa 1–169) were each able to modulate mammary carcinogenesis. The data also revealed two-way interactions between regions 2 (aa 109–157) and 3 (aa 158–169; P < 0.0001).
Mammary carcinomas induced by all of the Ras chimeras were analyzed histopathologically based on the criteria of Russo et al. (23). Approximately 50% of carcinomas generated by H-Ras were papillary-cribriform (Fig. 3,A), and 50% were comedo carcinomas with areas of necrosis (Fig. 3,B). In contrast, 100% of analyzed carcinomas induced by K-Ras were papillary-cribriform (Fig. 3,A). The morphology of carcinomas induced by the chimeras containing aa 1–169 from H-Ras resulted in the same distribution of histopathologies as observed in H-Ras-induced carcinomas (Table 1). Similarly, the morphology of carcinomas induced by chimeras containing K-Ras aa 1–168 was equivalent to K-Ras morphology. Carcinomas induced by chimeras within aa 1–169 (α-HKK, β-HKK, γ-HKK, and δ-HKK) were all similar to H-Ras-induced tumors in their histopathological distribution. Thus, not only did the regions of aa sequence modulate the frequency of carcinoma induction, but they also helped dictate the morphological phenotype of the carcinomas that arose.
To test the possibility that our observations could be influenced by unequal expression or stability of ras among the different chimeras, steady-state levels of retrovirally encoded ras mRNA was quantified by an RNase Protection Assay, whereas total Ras protein was evaluated by Western analysis in representative carcinomas (n = 4 for each construct). No systematic correlation was observed for either RNA (r2 = 5.87 × 10−4) or protein and vector carcinogenic potency (Fig. 4).
We next determined if each of the Ras chimeras was properly localized to the plasma membrane through fusion of full-length chimeras to ECFP. ECFP was fused to the amino terminus of each Ras chimera, as well as H-Ras and K-Ras. Transient transfections of NIH3T3 cells with ECFP-H-Ras or ECFP-K-Ras demonstrated that the fusion proteins were correctly localized to the plasma membrane and perinuclear structures as reported previously (12). All of the Ras chimeras fused to ECFP were targeted preferentially to the plasma membrane and perinuclear structures as seen with H-Ras and K-Ras (Fig. 5and data not shown).
In summary, we have shown previously that the organ specificity of H-Ras and K-Ras for the induction of mammary cancer resides within the Ras protein itself (4). H-Ras and K-Ras aa sequences vary not only in the carboxyl terminal hypervariable region (aa 170–189) but also to a much lesser extent in the homologous region (aa 1–169). Here we demonstrated that all regions of nonhomology throughout the entire Ras molecule alter Ras potency in the induction of mammary cancers. Whether these regions of H-Ras and K-Ras, which alter potency in this mammary-specific context, specify absolute organ specificity will require comparison of this H-Ras mammary data with data to be generated by future model development and evaluation in organ-specific K-Ras models, such as in lung, pancreatic, or colon carcinogenesis. Finally, it will be important to define the molecular functions of these previously undefined domains in the heterologous regions of the mostly homologous portion (aa 1–169) of Ras that significantly modulates the induction and histopathology of mammary cancer.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by NIH Grant CA77527.
The abbreviations used are: aa, amino acid; CFU, colony forming unit; ECFP, enhanced cyan fluorescent protein.
Group . | No. carcinomas scored . | % of papillary-cribriform carcinomas . | % of comedo carcinomas . |
---|---|---|---|
H-Ras | 29 | 55 | 45 |
HKK | 44 | 30 | 70 |
HHK | 24 | 37 | 63 |
HKH | 67 | 21 | 79 |
K-Ras | 13 | 100 | 0 |
KHH | 37 | 95 | 5 |
KKH | 12 | 92 | 8 |
KHK | 19 | 89 | 11 |
α-HKK | 6 | 0 | 100 |
β-HKK | 12 | 50 | 50 |
γ-HKK | 9 | 56 | 44 |
δ-HKK | 8 | 25 | 75 |
Group . | No. carcinomas scored . | % of papillary-cribriform carcinomas . | % of comedo carcinomas . |
---|---|---|---|
H-Ras | 29 | 55 | 45 |
HKK | 44 | 30 | 70 |
HHK | 24 | 37 | 63 |
HKH | 67 | 21 | 79 |
K-Ras | 13 | 100 | 0 |
KHH | 37 | 95 | 5 |
KKH | 12 | 92 | 8 |
KHK | 19 | 89 | 11 |
α-HKK | 6 | 0 | 100 |
β-HKK | 12 | 50 | 50 |
γ-HKK | 9 | 56 | 44 |
δ-HKK | 8 | 25 | 75 |
Acknowledgments
We thank P. Watson for valuable discussions and editing, D. McFarlin for reagents, and D. Monson for technical assistance.