Cells Exposed to Antifolates Show Increased Cellular Levels of Proteins Fused to Dihydrofolate Reductase: a Method to Modulate Gene Expression

Home , Antifolate

Cells exposed to antifolates show increased cellular levels of proteins fused to dihydrofolate reductase: A method to modulate gene expression

Philipp Mayer-Kuckuk*, Debabrata Banerjee*†, Sundeep Malhotra‡, Mikhail Doubrovin§, Marian Iwamoto*, Tim Akhurst¶, Julius Balatoniʈ, William Bornmann**, Ronald Finnʈ, Steven Larson¶, Yuman Fong‡, Juri Gelovani Tjuvajev§, Ronald Blasberg§, and Joseph R. Bertino*††

*Molecular Pharmacology and Therapeutics Program, Departments of ‡Surgery and §Neurooncology, ¶Nuclear Medicine Service, ʈRadiochemistry͞ Cyclotron, and **Preparative Synthesis Chemistry Core Facilities, Memorial Sloan–Kettering Cancer Center, New York, NY 10021

Communicated by Paul A. Marks, Memorial Sloan–Kettering Cancer Center, New York, NY, January 22, 2002 (received for review November 27, 2001) Human cells exposed to antifolates show a rapid increase in the levels quently results in increased translation of DHFR protein (5–7). In of the enzyme dihydrofolate reductase (DHFR). We hypothesized that addition to the described translational regulation of DHFR in this adaptive response mechanism can be used to elevate cellular cancer cells exposed to MTX, increased levels of DHFR also result levels of proteins fused to DHFR. In this study, mouse cells transfected through DHFR gene amplification, a common mechanism of to express a green fluorescent protein-DHFR fusion protein and acquired resistance to this drug (8). In contrast to rapid transla- subsequently exposed to the antifolate trimetrexate (TMTX) showed tional modulation of DHFR, gene amplification occurs in response a specific and time-dependent increase in cellular levels of the fusion to chronic exposure to antifolates, and elevated cellular levels of protein. Next, human HCT-8 and HCT-116 colon cancer cells retrovi- DHFR result from transcription of multiple DHFR gene copies. rally transduced to express a DHFR-herpes simplex virus 1 thymidine Regulation of exogenous gene products in vivo has become kinase (HSV1 TK) fusion protein and treated with the DHFR inhibitor increasingly important with the development of gene therapies TMTX exhibited increased levels of the DHFR-HSV1 TK fusion protein to treat human diseases such as cancer (9, 10). We initiated the and an increase in ganciclovir sensitivity by 250-fold. The level of study presented here with the hypothesis that proteins of interest fusion protein in antifolate-treated human tumor cells was increased can be fused to DHFR and thereby adopt the cellular regulation in response to a 24-h exposure of methotrexate, trimetrexate, as well mechanisms of the DHFR protein, thus regulating exogenous as dihydrofolate. This effect depended on the antifolate concentra- fusion proteins in a nontranscriptional manner by small drug tion and was independent of the fusion-protein mRNA levels, con- molecules. Here, we report intracellular up-regulation of exog- sistent with this increase occurring at a translational level. In a enous DHFR fusion gene products via antifolates in vitro and in xenograft model, nude rats bearing DHFR-HSV1 TK-transduced HCT-8 vivo. Presented data indicate that the up-regulation is specific tumors and treated with TMTX showed, after 24 h, a 2- to 4-fold and translationally controlled. increase of fusion-protein levels in tumor tissue from treated animals compared with controls, as determined by Western blotting. The Materials and Methods fusion-protein increase was imaged with positron-emission tomog- Materials. Tissue culture material was obtained from Corning. raphy, where a substantially enhanced signal of the transduced Enzymes were purchased from NEB or Stratagene. The anti- tumor was detected in animals after antifolate administration. Drug- enhanced green fluorescent protein (EGFP) antibody was from mediated elevation of cellular DHFR-fused proteins is a very useful ␤ method to modulate gene expression in vivo for imaging as well as Roche Molecular Biochemicals, the anti- -actin antibody was from therapeutic purposes. Sigma, and the secondary antibodies were from Santa Cruz Bio- technology. All DNA primers and probes were obtained from Keystone Laboratories (Menlo Park, CA). Antifolates were pur- ihydrofolate reductase (DHFR; E.C. 1.5.1.3.) catalyzes the chased from U.S. Bioscience (TMTX; West Conshohocken, PA) or DNADPH-dependent formation of 5,6,7,8-tetrahydrofolate Immunex (MTX), and ganciclovir (GCV) was from Roche Clinical (THF) from 7,8-dihydrofolate (DHF). The most significant con- Laboratories, Burlington, NC. [124I]2Ј-fluoro-2Ј-deoxy-5-iodoura- sequence of DHFR inhibition by methotrexate (MTX) or trime- cil-␤-D-arabinofuranoside (FIAU) was synthesized by the Prepar- trexate (TMTX) is a decrease of thymidylate biosynthesis by means ͞ 5 10 ative Synthesis Chemistry and Radiochemistry Cyclotron Core of depletion of the N ,N -methylene-THF pool resulting in DNA Facilities at Memorial Sloan–Kettering Cancer Center (MSKCC). synthesis inhibition and cell death. MTX is used to treat acute lymphoblastic leukemia, lymphoma, gastrointestinal cancers, and Tissue Culture. Colon cancer cells (HCT-8, HCT-116) were cul- breast cancer. However, antifolate resistance remains an obstacle to tivated in RPMI medium 1640 supplemented with 100 units͞ml successful cancer treatment. Early studies showed that treatment of penicillin͞100 ␮g/ml streptomycin͞10% dialyzed FBS (dFBS; leukemia patients with MTX increased DHFR activity in blast cells Atlanta Biologicals, Norcross, GA). The AM12 and the NIH 3T3 by 6-fold (1). Subsequently, Hillcoat et al. (2) observed a 12-fold increase in DHFR activity of cultured lymphoblasts after a 24-h exposure to MTX. Importantly, the increase of activity after 24–48 Abbreviations: DHFR, dihydrofolate reductase; MTX, methotrexate; TMTX, trimetrexate; h was not affected by the addition of actinomycin D, but was moi, multiplicity of infection; RT, reverse transcriptase; GCV, ganciclovir; EGFP, enhanced inhibited by the addition of cycloheximide. Later studies (3, 4) green fluorescent protein; PET, positron-emission tomography; FIAU, 2Ј-fluoro-2Ј-deoxy- showed no alteration of DHFR enzyme in human cells either 5-iodouracil-␤-D-arabinofuranoside; HSV1, herpes simplex virus 1; IRES, internal ribosome entry site; TK, thymidine kinase; TS, thymidylate synthase. treated with MTX in vitro or isolated from patients treated with † MTX. Furthermore, mRNA levels were not increased after MTX Present address: Cancer Institute of New Jersey, New Brunswick, NJ 08901. treatment, and the half-life of DHFR in human cells was not ††To whom reprint requests should be addressed at the present address: Cancer Institute of New Jersey, Robert Wood Johnson School of Medicine, 195 Albany Street, New Bruns- affected by the presence or absence of MTX (4). Recent mecha- wick, NJ 08903. E-mail: [email protected]. nistic studies from our laboratory and others indicate that DHFR The publication costs of this article were defrayed in part by page charge payment. This binds to DHFR mRNA in the coding region, and that inhibition of article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. DHFR by MTX releases the enzyme from the mRNA and conse- §1734 solely to indicate this fact.

3400–3405 ͉ PNAS ͉ March 19, 2002 ͉ vol. 99 ͉ no. 6 www.pnas.org͞cgi͞doi͞10.1073͞pnas.062036899 Downloaded by guest on October 2, 2021 cells were cultivated in DMEM containing high glucose (4.5 mg͞ml), penicillin (100 units͞ml), streptomycin (100 ␮g͞ml), and 10% dFBS. Cells were maintained at 37°C in a humidified atmosphere containing 5% CO2 and regularly checked for mycoplasma contamination.

Plasmids. All retroviral vectors used are based on the Moloney murine leukemia virus-based (SFG) plasmid described (11). The coding sequence for a double mutant (Phe-22–Ser-31) DHFR has been described (12). Use of the internal ribosome entry site (IRES) element was reported before (13). Published coding sequences for EGFP (accession no. U55761) and herpes simplex virus 1 (HSV1), thymidine kinase (TK) (accession no. AB009258) were used. Recombinant DNA manipulations were verified by DNA sequencing.

Fig. 1. Schematic representation of the retroviral vectors used. The retroviral Transfections and Retroviral Infections. Transfection. Transfections ϩ vectors code for (A) a DHFR (double mutant Phe-22-Ser-31)-EGFP fusion protein, of parental GP envAM12 cells (14) with EGFP-containing (B) DHFR and EGFP separated by an IRES element, (C) EGFP or (D) a DHFR-HSV1 TK plasmids were carried out by using DOTAP transfectant reagent fusion protein. Transgene expression is controlled by the retroviral 3Ј-long ter- {N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium minal repeats (LTR) and enhanced by a myeloproliferative sarcoma virus (MPSV) methylsulfate; Roche Molecular Biochemicals}. element (). Unique restriction sites are indicated. ␺, packaging signal. Retroviral infection. Retrovirus-producing cell lines were generated by transfection of the DHFR-HSV1 TK plasmid in parental GPϩenvAM12 cells. Transfection was performed three times at cell sities was carried out with an Alpha Imager (Alpha Innotech, densities of 30%, 50%, and 75% using Superfect (Qiagen, Chats- San Leandro, CA) and ALPHA EASE 5.1 software. worth, CA) as described by the manufacturer. Clones of retrovirus- producing cells were selected in 150 nM TMTX. Viral titers were Treatment of Cells with Drugs and Preparation of Lysates. Drug determined against NIH 3T3 cells. Retrovirus-producing AM12V treatment. Compounds used in this study were dissolved in sterile cells were grown in medium without drug to a density of 60–80%; water (TMTX, DHF) or DMSO (cycloheximide, actinomycin the day before the infection, medium was changed. Viral transduc- D). For chronic TMTX treatment, cells were treated in three tion was performed by infecting colon cancer cells plated 48 h cycles with medium containing increasing concentrations of before infection in 10-cm dishes. Four single virus exposures with TMTX (25 nM, 50 nM, and finally 100 nM) for 9 days in each the desired multiplicity of infection (moi), each of 6 h duration, were cycle. For short time drug treatment, cells in log phase were carried out. For each single exposure, fresh AM12V supernatant cultured in 25-ml of medium in 15-cm dishes. was filtered through a 0.45-␮m cellulose acetate filter and supple- Cell lysates. Preparation of cell lysates was performed at 4°C. mented with polybrene (8 ␮g͞ml). Likewise, the 2-h viral trans- Cells were washed with PBS, collected by centrifugation (1,000 ϫ duction was performed in 6-well plates. g, 1 min) and resuspended in TBS buffer containing 200 ␮M PMSF, 1.5 ␮g͞ml aprotinin and 10 mM ␤-mercaptoethanol. After two Standard and Quantitative Reverse Transcriptase (RT)-PCR. Standard sonications (30% output, Vibra Cell, Sonics & Materials, Danbury, PCR. The transgene was amplified by using specific primers (forward CT) for 10 s each cycle, the lysate was centrifuged (14,000 ϫ g,25 primer: 5Ј-ACATTTGCACTGGTAAACTTCATCA-3Ј, reverse min), and the supernatant was immediately used or stored at primer: 5Ј-CTCAAGGAACCTCCACAAGGAG-3Ј), Herculase Ϫ20°C. polymerase (Stratagene) and the GeneAmp 9600 (Perkin–Elmer͞ Tumor lysates. Tumor tissue was pulverized in liquid nitrogen, ϫ Cetus) thermocycler at T1,94°C, 120 sec; T2,63°C, 30 sec; T3,68°C, weighed, suspended in a volume equal to 10 the tumor weight 60 sec (1 cycle); and T1,94°C, 30 sec; T2,63°C, 30 sec; T3,68°C, 60 of SDS-PAGE sample buffer without glycerol and dye, and sec (33 cycles). boiled for 5 min. Lysate was centrifuged at 14,000 ϫ g for 25 min, QRT-PCR. The PCR reaction was performed by using a TaqMan and the supernatant was immediately used or stored at Ϫ20°C. 1,000 RXN Gold with Buffer A kit and the ABIPrism 7700 thermocycler (Applied Biosystems). The temperature profile was Cytotoxicity Assays. Colony formation assay. Cells (1 ϫ 103) were ␮ 95°C, 10 min (1 cycle), and T1,95°C, 15 sec; T2,60°C, 60 sec (42 plated in a volume of 12 ml of medium with or without drug (20 M cycles). Sequences for primers and probes were DHFR-HSV1 TK: GCV) for 12 days in 10-cm dishes. Cells were stained with crystal forward primer, 5Ј-GGAGGAGAAAGGCATTAAGTACAAA- violet solution, and colonies with a size of at least 0.5 mm were 3Ј; probe, 5Ј-6FAM-ATCTGGTTCCGCGTGGATCCCTC- counted. TAMRA-3Ј; reverse primer, 5Ј-CGCAGACGCGTGTTGATG- XTT assay. In each well of a 96-well plate, 1.5 ϫ 103 cells were 3Ј. ␤-actin : forward primer, 5Ј-TGAGCGCGGCTACAGCTT-3Ј; plated in 300 ␮l of medium containing the desired concentration probe, 5Ј-FAM-ACCACCACGGCCGAGCGG-TAMRA-3Ј; re- of drug, and after 5 days, the XTT assay was performed verse primer, 5Ј-TCCTTAATGTCACGCACGATTT-3Ј. according to standard protocols. Absorbance was measured in a microplate reader (EL 340, Bio-Tek, Burlington, VT) at a Western Blotting. Western blotting was done at room temperature wavelength of 450 nm (cut off ϭ 630 nm). with TBS buffer containing 0.1% Tween 20 and 5% dry nonfat milk powder, according to standard protocols and manufacturer Xenograft Tumor Model. All work with animals was carried out instructions. Equal protein loading was controlled by PonceauS according to institutional guidelines under the auspices of an animal staining and Western blotting by using anti-␤-actin antibody. The protocol approved by the Institutional Animal Care and Use enhanced chemilumiscence reagent (Amersham Pharmacia) was Committee at MSKCC. Male athymic (RNU) rats, 6–8 weeks old, used according to the manufacturer’s instructions to detect the were obtained from National Institutes of Health and housed in an

secondary antibody on the blot. If required, blots were stripped MSKCC animal facility. To generate a flank tumor model, rats were SCIENCES two times for 30 min in 200 mM glycine (pH 2.2)͞0.1% SDS͞1% anesthetized with 3 ml͞kg of a 1:2 mixture Ketaset (100 mg͞ml, Fort Tween-20 at room temperature. Measurement of signal inten- Dodge Animal Health, Overland Park, KS) and Aceproject (10 APPLIED BIOLOGICAL

Mayer-Kuckuk et al. PNAS ͉ March 19, 2002 ͉ vol. 99 ͉ no. 6 ͉ 3401 Downloaded by guest on October 2, 2021 Fig. 3. Retroviral transfer and expression of the DHFR-HSV1 TK fusion gene in colon cancer cells. (A) Detection of the fusion gene in transduced HCT-8–116 cells by PCR. The transgene-speciﬁc 860-bp fragment was ampliﬁed from a genomic DNA template. Mock, mock-transduced control cells. (B) Western blot analysis using an anti-DHFR antibody.

Fluorescent images of randomly selected areas of the monolayer were taken with a Zeiss Axiovert S 100 microscope. PET imaging. Each rat received 2 ml of 0.9% NaI 48 h before imaging. The [124I]FIAU (260 ␮Ci per animal; 1 Ci ϭ 37 GBq) was injected under anesthesia into the rat penile vein 24 h before PET scanning. Sterile water was injected i.p. the day of imaging to enhance tracer clearance. Animals were sedated in the PET suite as described above and placed in a specially designed animal holder in the PET camera field. Imaging was performed with a GE Advanced PET tomograph (General Electric). The durations of transmission and emission scans were 7 and 20 min, respectively. Immediately after the imaging session, the animals were killed, and tumor samples were weighed and assayed for [124I]FIAU radioactivity by using a Packard 5500 ␥ spectrometer (Packard). Results Specific and Time-Dependent Increase of DHFR-EGFP Fusion-Protein Fig. 2. Specific increase of DHFR-EGFP fusion-protein levels in mouse cells exposed to TMTX. AM12 cells expressing DHFR-EGFP (A), DHFR-IRES-EGFP (B), or Levels in Mouse Cells Exposed to Trimetrexate. To determine if EGFP (C) were exposed to 1 ␮M TMTX for 24 and 48 h and subsequently analyzed up-regulation of DHFR fusion proteins occurs in cells exposed to for EGFP protein levels by Western blotting. (D) AM12 cells expressing DHFR-EGFP TMTX, we first generated a retroviral vector encoding for a were treated with TMTX for 48 h. After 24 h, samples were exposed to cyclohex- DHFR-EGFP fusion protein (Fig. 1). Transfection of the vector imide (50 ␮g͞ml) or actinomycin D (1 ␮g͞ml). Cellular DHFR-EGFP levels were containing the fusion DHFR-EGFP into mouse cells (AM12) determined by Western blotting using an anti-DHFR antibody. (E) DHFR-EGFP- ␮ resulted in DHFR-EGFP fusion-protein expression (Fig. 2A). expressing AM12 cells were exposed to 1 M TMTX (control, water) for 48 h. After Increased cellular DHFR-EGFP fusion-protein levels were de- 24 (data not shown) and 48 h, cell fluorescence was monitored with fluorescent ␮ microscopy. Pictures were taken with the same settings, first from the TMTX- tected at 24 and 48 h after treatment with 1 M of the antifolate treated cells, then followed by the control cells. trimetrexate (Fig. 2A). We next assessed whether the observed up-regulation of EGFP was specific for the fusion protein. Mouse cells were transfected with a vector encoding either for DHFR and mg͞ml, Vetus Animal Health, distributed By Burns Veterinary EGFP separated by an IRES element or a vector containing only Supply, Farmers Branch, TX) given i.p. and subsequently injected EGFP. These two nonfusion constructs did not result in an increase s.c. with 1 ϫ 106 cells in 100 ␮l of serum-free medium. in EGFP levels in response to TMTX treatment (Fig. 2 B and C). Next, DHFR-EGFP fusion protein-expressing cells were exposed Fluorescent Microscopy and Positron-Emission Tomography (PET) Im- for 48 h to TMTX. After 24 h TMTX exposure, the cells were aging. Fluorescent microscopy. Cells transfected with EGFP- treated with cycloheximide, resulting in a reduction of the trans- expressing vectors were cultured for at least 12 days and subse- lational up-regulation, whereas addition of the transcription inhib- quently sorted for EGFP-positive cells by using a FACS Vantage itor actinomycin D had no effect, as determined by Western blotting SE cell sorter (Becton Dickinson). For fluorescent microscopy, (Fig. 2D). these cells were grown in T25 flasks. Confluent cells were In a subsequent set of experiments, fluorescent microscopy was cultured in fresh media supplemented with or without TMTX. used to visualize directly the effect of TMTX on cellular DHFR-

3402 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.062036899 Mayer-Kuckuk et al. Downloaded by guest on October 2, 2021 exhibit a moderate and 116HT cells exhibit a higher GCV sensitivity and TMTX resistance as compared with the parental cell line (Fig. 4 A and C). This finding correlates with the observed levels of fusion-protein expression in these cells (Fig. 3B). In a similar way, detected fusion-protein expression correlates with the drug cytotoxicity in transduced HCT-8 cells. 8HT cells, but not 8LT, cells showed GCV sensitivity and TMTX resistance (Fig. 4 B and D). In summary, HCT-8 cells showed substantially lower transduction and transgene expression in comparison to HCT-116 cells under comparable transduction conditions and correlated with lower levels of GCV sensitivity and TMTX resistance. When HCT-8 cells were treated chronically with increasing concentrations of TMTX, a 250-fold increase in GCV sensitivity (IC50) was measured (by XTT-assay; Fig. 4B), accompanied by an increased gene copy number (data not shown) and fusion- protein expression in the 8LT-CT cells (Fig. 3B and Fig. 5). Thus, amplification of the fusion gene and increased expression of HSV1 TK occurred with chronic antifolate treatment. In another set of experiments, the retrovirally transduced HCT- 116 and HCT-8 cells were exposed for 24 h to TMTX. Increased DHFR-HSV1 TK levels were observed in all transduced cell lines (Fig. 5). However, the fold increase upon TMTX exposure was cell-line dependent. The 8HT cells showed a moderate increase, Fig. 4. GCV sensitivity and TMTX resistance of parental, transduced, and 116LT cells a higher increase, and 116HT cells an even higher TMTX-exposed colon cancer cells. GCV and trimetrexate sensitivity of parental increase in fusion-protein levels. When the chronically TMTX- and transduced HCT-116 and HCT-8 cells were measured with an XTT assay. treated 8LT-CT cells were allowed to grow in the absence of drug, Cell line abbreviations: 8LT, low-transduced HCT-8 cells; 8HT, high-transduced HCT-8 cells; 8LT-CT, low-transduced and antifolate treated HCT-8 cells; 116LT, a significant induction of the fusion protein occurred after a low-transduced HCT-116 cells; 116HT, high-transduced HCT-116 cells. subsequent new exposure to TMTX. To demonstrate that the fusion protein contained DHFR, an anti-DHFR antibody was used. As expected, an increase of endogenous DHFR levels, as well as the EGFP levels in living cells. Fig. 2E shows that the DHFR-EGFP exogenous DHFR-HSV1 TK in response to TMTX treatment, was fusion protein is functionally active, and fluorescence increased noted. Moreover, the amount of increase of the endogenous DHFR over time in DHFR-EGFP-expressing cells treated with TMTX was comparable to the fusion-protein increase. The observed compared with controls. Also, cells expressing DHFR and EGFP increase in DHFR-HSV1 TK or DHFR levels in transduced cells as single proteins did not show an altered EGFP expression in was similar to the DHFR increase in untransduced, parental response to TMTX treatment (data not shown). HCT-116 and HCT-8 cells. To show dose dependence of DHFR-HSV1 TK induction, Increase of DHFR-HSV1 TK Fusion-Protein Levels in Colon Cancer Cells 8LT-CT cells were exposed to increasing concentrations of TMTX Exposed to Antifolates. To provide further evidence for an antifo- or MTX. Concentration-dependent induction of DHFR-HSV1 TK late-induced DHFR fusion gene up-regulation, we studied human was noted with both antifolates (Fig. 6). Interestingly, increased colon cancer cells retrovirally transduced to express a fusion protein fusion-protein levels also were seen after treatment of 8LT-CT cells of DHFR and HSV1 TK (15). Bulk transduction of HCT-116 and with high concentration of the DHFR substrate dihydrofolate. In HCT-8 cells was carried out with an moi of 30 for 24 h or an moi contrast, treatment of cells with etoposide, a topoisomerase II of 2.5 for 2 h. This resulted in the cell lines 116HT (moi 30, 24 h) inhibitor, did not cause induction of the fusion protein (data not and 116LT (moi 2.5, 2 h) as well as 8HT (moi 30, 24 h) and 8LT (moi shown). 2.5, 2 h). Successful gene transfer and transgene expression in these To confirm that the increase in DHFR was related to transla- cell lines were confirmed by PCR (Fig. 3A) and Western blotting tional regulation, the levels of DHFR-HSV1 TK mRNA before and (Fig. 3B), respectively. The percentage of infected cells was deter- after TMTX, MTX, or DHF treatment were measured. As shown mined by colony formation assay using the GCV sensitivity of in Table 1, no alterations of the fusion-protein mRNA expression DHFR-HSV1 TK-expressing cells as follows: 116HT, very high (no in response to the drug exposure were detected, indicating that the colony formation); 116LT, 80%; 8HT, 53%. Infection of 8LT cells increase was caused by relief of inhibition of DHFR translation by was not detectable with the assay used. Next, we measured GCV these folate compounds and͞or protection of DHFR from degra- and TMTX cytotoxicities of the transduced cells. The 116LT cells dation, consistent with previous studies (4–6). SCIENCES Fig. 5. Increased DHFR-HSV1 TK fusion-protein levels in various transduced and parental colon cancer cells exposed to antifolate for 24 h. Transduced and

parental HCT-8 and HCT-116 cells were exposed to 1 ␮M TMTX or MTX for 24 h. Shown is a Western blot analysis using an anti-DHFR antibody. APPLIED BIOLOGICAL

Mayer-Kuckuk et al. PNAS ͉ March 19, 2002 ͉ vol. 99 ͉ no. 6 ͉ 3403 Downloaded by guest on October 2, 2021 Fig. 7. Transduced tumor xenografts show increased DHFR-HSV1 TK levels after TMTX treatment. Tumors derived from 8LT-CT cells were grown in 18 RNU rats to an average size of 560 mm3 (range 90–1,100 mm3). Subsequently, six animals each were treated with either 10 mg͞kg TMTX for 3 days (■)or100 mg͞kg TMTX (Œ) given in two 50 mg͞kg doses the day before tumor sampling. Six control animals received water (ᮀ). Tumors were analyzed ex vivo for Fig. 6. Dose-dependent increase of DHFR-HSV1 TK fusion-protein levels in cellular DHFR-HSV1 TK levels by immunoblotting with an anti-DHFR antibody colon cancer cells exposed to antifolate. Western blot analysis using an anti-HSV1 and, subsequently, after striping of the membrane with an anti-␤-actin anti- TK antibody to detect the fusion protein. DHFR-HSV1 TK and ␤-actin Western blot body. Western blot signal intensities were measured and the fusion-protein signals were quantiﬁed, and the DHFR-HSV1 TK signal intensity relative to the signal intensity relative to the ␤-actin signal was calculated. These values are ␤-actin signal intensity was calculated. These values are represented by the bar shown; bold numbers indicate mean values including SD. graph.

crease is caused by a relief of translational autoregulation of DHFR DHFR-HSV1 TK Levels Increase in Transduced Tumor Xenografts After by MTX or TMTX (1–7, 17). Here, we show that antifolates are able TMTX Treatment. Encouraged by the described in vitro observations, to increase cellular levels of fusion proteins, e.g., DHFR-EGFP and the in vivo relevance of the DHFR-HSV1 TK induction by TMTX DHFR-HSV1 TK in mouse and colon cancer cell lines, respectively was addressed. Rats bearing flank tumors derived from 8LT-CT (Figs. 2 and 5). Thus, the induction is independent of both the cell cells were treated with 10 mg͞kg TMTX for 3 days or a single high type and the protein fused to DHFR. However, as shown in Fig. 5, dose of 100 mg͞kg. The level of DHFR-HSV1 TK fusion protein the intensity of induction measured after 24 h varied. Importantly, in the tumor tissue of antifolate-treated animals, as well as water- only DHFR or DHFR fusion proteins showed this increase. Cells treated rats, was analyzed. All antifolate-treated animals showed transfected with DHFR-IRES-EGFP showed a barely detectable elevated levels of the fusion protein ranging from 1.5 to 4-fold increase in EGFP levels upon exposure to antifolates (Fig. 2B). compared with controls (Fig. 7), and the mean increase was at least Furthermore, by transfecting cells with SFG vector containing 2-fold. The second set of experiments used rats bearing tumors EGFP and subsequent TMTX treatment of the cells we ruled out derived from 8LT-CT and parental HCT-8 cells and in vivo imaging. a nonspecific effect of the SFG vector backbone (Fig. 2C). In Before the imaging in living rats, animals received three cycles of addition, and in accordance with earlier studies, we observed a the three times daily doses of TMTX. By using PET and the tracer decreased cellular induction of the DHFR-EGFP fusion protein [124I]FIAU (16), an increase in tumor-signal intensity in TMTX- after cycloheximide treatment, whereas actinomycin D had no treated rats as compared with control rats was observed (Fig. 8). effect on the induction (Fig. 2D). No change in DHFR-HSV1 TK Measurements of [124I]FIAU uptake (% dose per g) from seven mRNA levels after exposure to antifolates was detected (Table 1), treated and seven control animals showed a 2.6-fold higher as was observed previously (4). [124I]FIAU accumulation in transduced tumor tissue following TMTX treatment (0.348 Ϯ 0.175% dose per g), as compared with untreated controls (0.132 Ϯ 0.04% dose per g). Discussion The ability to control levels of exogenous, therapeutic proteins in cells is an important step toward progress in molecular therapy. Here, we use a unique drug-response mechanism of the highly regulated enzyme dihydrofolate reductase to increase cellular levels of exogenous proteins both in vitro and in vivo. Thus far, the exact mechanism of this rapid ‘‘induction’’ of DHFR after exposure of cells to antifolates has not been described in detail. However, studies provide evidence that this cellular protein in-

Table 1. Fusion protein mRNA expression Relative expression DHFR-HSV1 Cells TK mRNA

8LT-CT 3.7 Ϯ 1.1 8LT-CT, 1 ␮M TMTX, 24 h 4.7 Ϯ 1.4 Fig. 8. TMTX treatment results in increased FIAU accumulation in transduced 8LT-CT, 1 ␮M MTX, 24 h 3.6 Ϯ 0.8 tumor xenografts in vivo. Shown are digital pictures as well as axial and 8LT-CT, 50 ␮M DHF, 24 h 3.5 Ϯ 0.9 transaxial tumor PET scans obtained from (A) an antifolate-treated and (B) water-treated (control) RNU rat. The inserted scans show the heart as control.

3404 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.062036899 Mayer-Kuckuk et al. Downloaded by guest on October 2, 2021 Interestingly, the other mammalian enzyme known to exhibit using a DHFR-cytidine deaminase (CD) fusion cDNA (22). Iden- induction of the protein by the substrate, cofactor, or inhibitor is tification of the minimal region of DHFR mRNA that is essential thymidylate synthase (TS), a protein catalyzing the NADPH- for antifolate-mediated regulation may improve in vivo regulation dependent conversion of dUMP and THF to dTMP and DHF. of the second gene of interest. Thus, TS, like DHFR, is a component of the cell cycle-dependent Retroviral gene transfer delivery is known to be critical for in vivo and highly regulated de novo dTMP production, which is essential applications. We used a retroviral vector in this study because the for DNA synthesis. Similar to DHFR regulation, TS has been gene transfer method provides specificity for dividing cells but, reported to specifically bind to its own mRNA and hence inhibit its compared with adenoviral delivery, lower transduction efficiencies. own translation (18, 19). A recent study raises the possibility that In contrast to most gene therapy studies, we used bulk transduced binding of FdUMP to TS protects this enzyme from degradation cell lines and not clonal cell lines in this study. Bulk transduction and also could be responsible for the observed increased TS levels may reflect retroviral infection in vivo better than clonal cell after fluoropyrimidine exposure (20). In contrast to DHFR, TS is line-derived tumors. We observed significantly different transduc- larger, functional as a dimer, and demonstrates only a 2- to 4-fold tion efficiencies between HCT-8 and HCT-116 cells, which resulted induction in patients (21). Thus, DHFR appears to be more suitable in distinct levels of fusion-protein expression and cytotoxicity than TS for the generation of inducible functional fusion proteins. characteristics. Both cell lines are human, primary colon cancer cell Previous studies with another DHFR fusion protein (22) and lines but may display different levels of retroviral receptors. Taken the findings presented here indicate that DHFR fusion-protein together, our observations point to a significantly cell type- activities are fully functional (Figs. 2E, 4, and 8), and that the dependent retroviral infection efficiency. Chronic exposure to fusion-protein induction is similar to the induction of the en- increasing concentrations of TMTX resulted in a significantly dogenous DHFR (Fig. 5). We used a double-mutant DHFR (12) increased GCV-sensitivity of the low-transduced 8LT cells along to protect transduced cells from antifolate toxicity. with detectable levels of DHFR-HSV1 TK fusion protein and In this study, we describe a nontranscriptional, fusion protein- increased copies of the corresponding transgene and mRNA. We based regulation of an exogenous gene product by a small molecule. were able to increase further fusion-protein levels by stepwise Alternate strategies to regulate cellular protein levels by controlling exposure up to 800 nM TMTX (data not shown). These observa- gene expression involve transcriptional regulation. The potential of tions were likely caused by a selection for transduced cells. 124 doxycycline-responsive gene expression was demonstrated recently Here, we used HSV1 TK, the tracer [ I]FIAU, and PET in a mouse model of a neurodegenerative disease (23), in which the scanning to image a DHFR-HSV1 TK fusion protein in transduced successful transfer and tightly controlled expression of tyrosine tumors in vivo. Recently, a study described the feasibility of non- hydroxylase in brain grafts of human neural progenitors was invasive imaging of a CD-HSV1 TK fusion protein (25). In this reported. Enzyme expression depended on the presence of doxy- study, we demonstrate increased levels of the DHFR-HSV1 TK cycline. Another system used the immunosuppressive drug rapa- fusion protein in transduced tumors after antifolate treatment. mycin to mediate the reconstitution of a functional transcription Additional studies that evaluate the ability to increase PET signal factor and thus transgene expression. Like the first system, this or fluorescence intensities of tumors transduced with DHFR-HSV1 system uses a single-vector coding for the essential control ele- TK or DHFR-EGFP, respectively, in tissue of living animals by ments, as described (24). Both described transcriptional regulations antifolate treatment will be of interest. A further application of show induction of gene expression of approximately four orders of increased cellular levels of therapeutic polypeptides such as HSV1 magnitude in vitro. However, they require vectors containing rel- TK by induction as DHFR fusion protein might be an increased atively long sequences, which encode for at least three transcrip- cellular response to therapies such as GCV because of higher tional control elements. In contrast, the approach presented in this cellular levels of the therapeutic protein after antifolate treatment. study is based on the fusion of the sequence encoding the protein In summary, we have taken advantage of a cellular, adaptive drug-resistance mechanism, i.e., a post translational, fusion protein- of interest to the 560-bp DHFR sequence. It might not be possible based strategy to regulate the level of exogenous proteins in vitro to express all proteins as functional DHFR fusion proteins; how- and in vivo. The observations presented here should be useful for ever, there is evidence that the relatively small DHFR protein is an application in gene therapy, gene expression, and drug resistance appropriate partner in fusion proteins; e.g., DHFR fusion proteins research. have been found to occur in nature in protozoa such as Plasmodium, which expresses a DHFR-TS fusion protein. Moreover, a number The HSV1 TK antibody was a kind gift of Dr. W. C. Summers (School of artificial fusion proteins have been successfully generated. These of Medicine, Yale University, New Haven, CT). This work was supported fusion proteins were used for protein purification strategies using by National Institutes of Health Grants CA 08010 and CA 61586 (to MTX affinity chromatography or gene therapy approaches to J.R.B.) and CA 86438-02 (to D.B.). The Dr. Mildred Scheel Foundation confer drug resistance to antifolates and to cytosine arabinoside, for Cancer Research generously supported P.M.-K.

1. Bertino, J. R., Donohue, D. R., Garbio, B. W., Silber, R., Alenty, A., Mayer, M. & 16. Gambhir, S. S., Herschman, H. R., Cherry, S. R., Barrio, J. R., Satyamurthy, N., Tokokuni, Huennekens, F. M. (1962) Nature (London) 193, 140–141. T., Phelps, M. E., Larson, S. M., Balatoni, J., Finn, R., et al. (2000) Neoplasia (New York, 2. Hillcoat, B. L., Swett, V. & Bertino, J. R. (1967) Proc. Natl. Acad. Sci. USA 58, 1632–1637. NY) 2, 118–138. 3. Bertino, J. R., Cashmore, A. R. & Hillcoat, B. L. (1970) Cancer Res. 30, 2372–2378. 17. Cowan, K. H., Goldsmith, M. E., Ricciardone, M. D., Levine, R., Rubalcaba, E. & Jolivet, 4. Domin, B. A., Grill, S. P., Bastow, K. F. & Cheng, Y. C. (1982) Mol. Pharmacol. 21, 478–482. J. (1986) Mol. Pharmacol. 30, 69–76. 5. Ercikan-Abali, E. A., Banerjee, D., Waltham, M. C., Skacel, N., Scotto, K. W. & Bertino, 18. Chu, E., Koeller, D. M., Casey, J. L., Drake, J. C., Chabner, B. A., Elwood, P. C., Zinn, S. J. R. (1997) Biochemistry 7, 12317–12322. & Allegra, C. J. (1991) Proc. Natl. Acad. Sci. 88, 8977–8981. 6. Ercikan-Abali, E. A., Banerjee, D., Waltham, M. C., Schnieders, B., Scotto, K. W. & Bertino, 19. Chu, E., Voeller, D., Koeller, D. M., Drake, J. C., Takimoto, C. H., Maley, G. F., Maley, 338, J. R. (1993) Adv. Exp. Med. Biol. 537–540. F. & Allegra, C. J. (1993) Proc. Natl. Acad. Sci. USA 90, 517–521. 7. Chu E., Voeller, D., Jones, K. L., Drake, J. C., Takimoto, C. H., Maley, G. F., Maley, F. & 20. Kitchens, M. E., Forsthoefel, A. M., Rafique, Z., Spencer, H. T. & Berger, F. G. (1999) Allegra, C. J. (1993) Biochemistry 32, 4756–4760. J. Biol. Chem. 274, 12544–12547. 8. Schimke, R. T. (1988) J. Biol. Chem. 263, 5989–5992. 21. Steinberg, S. M., Swain, S. M., Lippmann, M. E., Egan, E. F., Drake, J. C. & Allegra, C. J. 9. Anklesaria, P. (2000) Curr. Opin. Mol. Ther. 2, 426–432. 10. Anderson, W. F. (1998) Nature (London) 392, 25–30. (1989) J. Clin. Oncol. 7, 890–899. 11. Riviere, I., Brose, K. & Mulligan, R. C. (1995) Proc. Natl. Acad. Sci. USA 92, 6733–6737. 22. Sauerbrey, A., McPherson, J. P., Zhao, S.-C., Banerjee, D. & Bertino, J. R. (1999) Hum. 12. Ercikan-Abali, E., Mineishi, S., Tong, Y., Nakahara, S., Waltham, M. C., Banerjee, D., Chen, Gene Ther. 10, 2495–2504. W., Sadelain, M. & Bertino, J. R. (1996) Cancer Res. 56, 4142–4145. 23. Corti, O., Horellou, P., Colin, P., Dumas, S. & Buchet, D. (1999) Nat. Biotechnol. 17, 349–354. 13. Frebourg, T., Sadelain, M., Ng, Y. S., Kassel, J. & Friend, S. H. (1994) Cancer Res. 54, 24. Pollock, R., Issner, R., Zoller, K., Natesan, S., Rivera, V. M. & Clackson T. (2000) Proc. Natl. 878–881. Acad. Sci. USA 97, 13221–13226.

14. Markowitz, D., Goff, S. & Bank, A. (1988) Virology 167, 400–406. 25. Hackman, T., Doubrovin, M., Balatoni, J., Beresten, T., Ponomarev, V., Beattie, B., Finn, R., SCIENCES 15. Moolten, F. L., Wells, J. M., Heyman, R. A. & Evans, R. M. (1990) Hum. Gene Ther. 1, 125–134. Bornmann, W., Blasberg, R. & Gelovani Tjuvajev, J. (2002) Mol. Imaging 1, 36–42. APPLIED BIOLOGICAL

Mayer-Kuckuk et al. PNAS ͉ March 19, 2002 ͉ vol. 99 ͉ no. 6 ͉ 3405 Downloaded by guest on October 2, 2021