Dihydrofolate Reductase and Thymidylate Synthase Transgenes Resistant to Methotrexate Interact to Permit Novel Transgene Regulation*

Background: Methotrexate targets dihydrofolate reductase (DHFR) and thymidylate synthase (TYMS) to prevent cancer growth, and increases DHFR expression when applied. Results: TYMS resistant to methotrexate inhibits the methotrexate induced increase in DHFR expression. Conclusion: Thymidine synthesis regulates post-transcriptional expression of DHFR and TYMS. Significance: Methotrexate-resistant DHFR and TYMS can be used to regulate cis and trans transgene in primary T cells. Methotrexate (MTX) is an anti-folate that inhibits de novo purine and thymidine nucleotide synthesis. MTX induces death in rapidly replicating cells and is used in the treatment of multiple cancers. MTX inhibits thymidine synthesis by targeting dihydrofolate reductase (DHFR) and thymidylate synthase (TYMS). The use of MTX to treat cancer also causes bone marrow suppression and inhibits the immune system. This has led to the development of an MTX-resistant DHFR, DHFR L22F, F31S (DHFRFS), to rescue healthy cells. 5-Fluorouracil-resistant TYMS T51S, G52S (TYMSSS) is resistant to MTX and improves MTX resistance of DHFRFS in primary T cells. Here we find that a known mechanism of MTX-induced increase in DHFR expression persists with DHFRFS and cis-expressed transgenes. We also find that TYMSSS expression of cis-expressed transgenes is similarly decreased in an MTX-inducible manner. MTX-inducible changes in DHFRFS and TYMSSS expression changes are lost when both genes are expressed together. In fact, expression of the DHFRFS and TYMSSS cis-expressed transgenes becomes correlated. These findings provide the basis for an unrecognized post-transcriptional mechanism that functionally links expression of DHFR and TYMS. These findings were made in genetically modified primary human T cells and have a clear potential for use in clinical applications where gene expression needs to be regulated by drug or maintained at a specific expression level. We demonstrate a potential application of this system in the controlled expression of systemically toxic cytokine IL-12.


Methotrexate (MTX) is an anti-folate that inhibits de novo purine and thymidine nucleotide synthesis. MTX induces death in rapidly replicating cells and is used in the treatment of multiple cancers. MTX inhibits thymidine synthesis by targeting dihydrofolate reductase (DHFR) and thymidylate synthase (TYMS). The use of MTX to treat cancer also causes bone marrow suppression and inhibits the immune system. This has led to the development of an MTX-resistant DHFR, DHFR L22F, F31S (DHFR FS ), to rescue healthy cells. 5-Fluorouracil-resistant TYMS T51S, G52S (TYMS SS ) is resistant to MTX and improves MTX resistance of DHFR FS in primary T cells. Here we find that a known mechanism of MTX-induced increase in DHFR expression persists with DHFR FS and cis-expressed transgenes.
We also find that TYMS SS expression of cis-expressed transgenes is similarly decreased in an MTX-inducible manner. MTX-inducible changes in DHFR FS and TYMS SS expression changes are lost when both genes are expressed together. In fact, expression of the DHFR FS and TYMS SS cis-expressed transgenes becomes correlated. These findings provide the basis for an unrecognized post-transcriptional mechanism that functionally links expression of DHFR and TYMS. These findings were made in genetically modified primary human T cells and have a clear potential for use in clinical applications where gene expression needs to be regulated by drug or maintained at a specific expression level. We demonstrate a potential application of this system in the controlled expression of systemically toxic cytokine IL-12.
Antifolate drugs have been in use for seven decades in the treatment of cancer (1). MTX 3 is a commonly used agent from this class and inhibits several folate-dependent enzymes including DHFR and TYMS. Inhibition of these proteins adversely affects the de novo synthesis of purine and thymidine nucleotides, which is vital for survival of rapidly replicating cells. MTX has proven valuable for treating rapidly replicating cancers such as leukemia, but the impact of MTX on healthy, replicating tissue leads to dose-limiting toxicities such as bone marrow suppression (2,3). This disadvantage of MTX has led to intense study into mechanisms of resistance (4) and alternative antifolates for the treatment of cancer (3).
Stemming from the discovery of DHFR mutants that conferred MTX resistance, a mutant that weakly binds MTX, DHFR FS , was developed (5). A strategy was formulated where bone marrow cells would be genetically modified ex vivo with DHFR mutants resistant to MTX (6). The use of DHFR FS was later implemented in genetically modified T cells. T cells modified with DHFR FS are desired following bone marrow transplant for relapsed leukemia to overcome immune suppression of MTX, allowing genetically modified T cells to survive and target cancer (7,8). An advantage of DHFR FS is that it can be used to select for transgenes useful for therapeutic efficacy such as tumor targeting proteins such as chimeric antigen receptors (8), suicide genes (9), or imaging genes (10). In an effort to broaden the application of DHFR FS , our group added a human TYMS mutant, discovered in a bacterial screen, resistant to 5-fluorouracil (5-FU) (11). Upon the addition of human mutant TYMS SS with DHFR FS in human T cells, we demonstrated improved resistance of T cells to 5-FU and increased resistance to MTX beyond that of DHFR FS alone, in agreement with a series of experiments performed in murine bone marrow cells where linking of DHFR FS to a mutant of TYMS led to improved resistance to 5-FU and the antifolate pemetrexed (12,13). We suggested that the addition of TYMS SS restores thymidine synthesis in MTX-treated T cells (9).
In working with DHFR and TYMS, it was noted that human DHFR is known to bind and inhibit translation of DHFR mRNA as a post-translational mechanism of regulation. Furthermore, MTX was shown to prevent this binding (14) and lead to an increase in DHFR protein expression. The nucleotide sequence to which DHFR protein binds DHFR mRNA within the coding domain is known (15,16), and the specific amino acids that DHFR protein utilizes to bind DHFR mRNA have been elucidated (17). However, subsequent studies found that a mutation to mRNA binding amino acids of DHFR still resulted in increased DHFR protein expression in the presence of MTX (18). Conversely, mutations to amino acids unrelated to DHFR binding mRNA (17) prevented MTX-inducible expression of DHFR (18). This suggests that another mechanism of posttranscriptional regulation for DHFR is in effect, and findings detailed in this study lead us to propose that a mechanism independent of mRNA binding is affecting MTX-induced DHFR up-regulation. Notably, we demonstrate that this mechanism can be utilized for regulation of transgene expression when transgenes are located in a 3Ј cis arrangement with DHFR FS .
By codon-optimizing DHFR FS to remove the mRNA binding sequence previously identified (15,16), we find that MTX continues to induce an increase in DHFR FS expression independent of the mRNA sequence and that the addition of TYMS SS unexpectedly blunts this increase. This finding suggests a novel method of post-transcriptional DHFR expression that is regulated by the enzymatic action of TYMS SS . In performing these experiments in primary human T cells, we also identified that the application of MTX decreases TYMS SS expression in an MTX-dependent manner, likely due to TYMS binding its own mRNA to repress translation (19). Translational repression of TYMS is lost in the presence of tetrahydrofolate (19), and MTX prevents the synthesis of tetrahydrofolate (3). Consistent with this hypothesis, restoration of tetrahydrofolate synthesis in T cells by expressing DHFR FS along with TYMS SS prevented the MTX-induced expression decreases for TYMS SS . Based on these findings, we were led to discover that expression of DHFR FS and TYMS SS is linked post-transcriptionally and results in correlated expression of DHFR and TYMS within human T cells. Here we present methods to increase, decrease, or link the expression of transgenes based on the use of mutant DHFR FS and TYMS SS in a clinically relevant context. This system will permit new forms of post-translational regulation to be utilized for improved control of gene therapy vectors.

Experimental Procedures
Cells-The Jurkat human T cell line was used for controlled expression experiments. Cell line origin and attributes have been previously discussed (9). T cells were derived from healthy donor-derived peripheral blood mononuclear cells (PBMC) from the MDACC Blood Bank, Houston, TX. PBMC were isolated and cryopreserved until use (9).
Genetic Constructs-A mutant DHFR FS containing the original DHFR mRNA codon sequence with point mutations to L22F,F31S was designed to co-express eGFP C terminus to a T2A amino acid sequence. This construct is referred to as FLAG-DHFR FS -2A-eGFP pSBSO (D FS G). The DHFR FS was also codon-optimized to remove the original codon sequence, with specific attention paid to the coding region where DHFR is known to bind its own mRNA (15,16). This codon-optimized (CoOp) DHFR FS otherwise had the same design as D FS G and is known as FLAG-CoOp DHFR FS -2A-eGFP pSBSO (CoOp D FS G). TYMS SS without codon optimization was co-expressed with a C terminus eGFP or RFP. The constructs are referred to as FLAG-TYMS SS -2A-GFP pSBSO (TS SS G) and FLAG-TYMS SS -2A-RFP pSBSO (TS SS R). A selection-free RFP plasmid NLS-mCherry pSBSO (RFP) was utilized as a control. The construct Myc-ffLuc-NeoR pSBSO (NRF) was electroporated, and G418 (InvivoGen, Ant-gn-1) was used for selection as described below when alternative selection was desired. A construct to select for inducible caspase 9 (iC9) FLAG-DHFR FS -2A-iC9 pSBSO (D FS iC9) was also used. The design and synthesis of these constructs have been described elsewhere (9). The construct FLAG-TYMS SS -2A-IL-12p35-2A-IL-12p40 pSBSO (TS SS IL-12) was synthesized from codon-optimized (GeneArt, Life Technologies) IL-12 p35 and IL-12 p40 transgenes and digested within the 2A regions to ligate IL-12 p35 and IL-12 p40 with a TYMS SS fragment also digested within the 2A region. TS SS G backbone digestion points 5Ј to the start site of TYMS SS and 3Ј to the IL-12p40 stop site ligated the three components into the TS SS G backbone in a four-part ligation. All constructs described integrate into the cellular genome with the aid of Sleeping Beauty transposase (20).
Culture Conditions-Both Jurkat cells and human PBMC were electroporated, maintained, and selected as described previously (9). Briefly, Jurkat cells were electroporated with transgene, and 48 h later, drug was applied for 2 weeks. Surviving cells were rested in complete media for another 3-5 weeks before testing. Human PBMC were electroporated, and the following day, T cells were propagated with activating and propagating cells at a 1:1 ratio including recombinant human IL-2. 48 h after co-culture was initiated, drug was applied to culture for a 2-week period. Propagation of PBMC with activating and propagating cells and IL-2 selectively expanded T cells and was repeated weekly. The T cells surviving drug treatment from days 2 to 14 were co-cultured without drug for another 3 weeks (9, 20, 21). On day 35 of propagation, T cells were stimulated with anti-CD3, anti-CD28, and IL-2 (9). MTX (Hospira, Lake Forest, IL), pemetrexed (Lilly, Indianapolis, IN), and 5-FU (APP Pharmaceuticals, Schaumburg, IL) were attained from the MDACC pharmacy, whereas raltitrexed (Abcam Biochemicals, Cambridge, MA, catalog number Ab142974) was not.
Flow Cytometry-T cells were prepared for flow cytometry by washing once in FACS buffer, and surface-stained for anti-human CD3-APC (BD Pharmingen, 340661, 1:33 dilution), as described previously (21). Intracellular staining for Myc tag Alexa Fluor 488 (MBL International, M047-A48, 1:33 dilution) and anti-human IL-12 p40/p70-PE (BD Pharmingen, 554575, 1:20 dilution) also proceeded as described previously (21). Transgene expression was analyzed on either a BD LSRFortessa or a BD FACSCalibur (BD Biosciences), with the same flow cytometer used consistently within a given experiment. FlowJo v 10.0.5 (TreeStar Inc., Ashland, OR) was used to analyze flow cytometry data. Statistical Analysis-Statistical analysis and graphical representation were performed as before (9). Briefly, non-Gaussian distributions of a multivariate nature were analyzed by one-way ANOVA Kruskal-Wallis test followed by Dunn's multiple comparison test. Univariate tests (experimental versus control) utilized t tests. Two-way ANOVA was performed when appropriate and followed by Šidák's multiple comparison test. The analysis is denoted in context. Pearson's correlation and linear regression analysis were performed where denoted. Statistical significance was designated as ␣ Ͻ 0.05. Means with standard deviations are depicted within each figure

Results and Discussion
Mayer-Kuckuk et al. (22) demonstrated that DHFR linked to a transgene of interest can be used to induce an increased expression of that transgene in vitro and in vivo when MTX is administered. We attempted to recapitulate an increase in the expression of MTX-resistant DHFR FS after the administration of MTX within human T cells, which Skacel et al. (18) has shown in a hamster cell line. As a control for our studies, DHFR FS transgene was codon-optimized to remove any of the mRNA sequence or structure that contributed to mRNA binding (16). It was hypothesized that this modification would increase DHFR FS expression by removing DHFR protein binding to DHFR FS mRNA and prevent the increase of DHFR FS expression previously noted in the presence of MTX. The expression of DHFR FS and CoOp DHFR FS selected for uniform expression in Jurkat T cell line is shown in Fig. 1A. CoOp DHFR FS did not contribute to a significantly higher expression of DHFR FS as indicated by a cis-expressed eGFP, nor did it prevent MTX-induced increases in transgene expression as noted in Fig. 1B. This was unexpected. However, the loss of MTX-induced increase in DHFR FS expression was noted when TYMS SS was co-expressed with DHFR FS , as seen in Fig. 1, A and B (9). The addition of TYMS SS led to an insignificant reduction in the expression of un-optimized DHFR FS in the absence of MTX. The addition of MTX was unable to induce the same increase in DHFR FS expression seen during the sole expression of either DHFR FS version. Thus, TYMS SS is playing a role in the MTX-inducible increase of DHFR FS , and likely through the restoration of thymidine synthesis.
Expression of these transgenes in primary T cells was evaluated to recapitulate the findings of MTX-inducible increases in DHFR FS expression that were prevented by TYMS SS . Expression of DHFR FS , TYMS SS , or [DHFR FS & TYMS SS ] was achieved with stability and purity by selecting from days 2-14 of propagation with the respective drugs MTX, 5-FU, and G418 when the selection vector containing neomycin resistance was included (9). The expression of DHFR FS -linked eGFP and TYMS SS -linked RFP can be noted in Fig. 1C for DHFR FS , TYMS SS , or [DHFR FS & TYMS SS ]. Again it is noted that DHFR FS expression is increased in the presence of MTX (Fig.  1D), but this increase is blunted and no longer significant when TYMS SS is co-expressed with DHFR FS , as in Jurkat cells. Of note, expression of TYMS SS without DHFR FS was successfully achieved in primary T cells by selection with 5-FU and a trans neomycin resistance gene selected by G418 (9). When TYMS SS was tested for inducible changes in the presence of high doses of MTX (Fig. 1D), it was found that TYMS SS -linked RFP decreased significantly. MTX induced a reduction in the expression of TYMS SS that MTX-resistant DHFR FS restored. Although not definitive, these findings could indicate that TYMS SS is being repressed by a lack of 5,10-methylenetetrahydrofolate (5,10-CH 2 THF). This is consistent with the findings of Chu et al. (19) that 5,10-CH 2 THF prevents TYMS-mediated repression of TYMS mRNA translation. Those findings indicate that in a healthy cell where DHFR activity is uninhibited, TYMS is not significantly bound to its mRNA. However, we propose that MTX, which leads to a drop in 5,10-CH 2 THF, is causing TYMS protein to bind TYMS and TYMS SS mRNA, preventing expression (23). It should be noted that TYMS SS is equivalent to the native sequence with the exception of the point mutations (9). Based on findings in Fig. 1, A and B, we propose that DHFR FS expression is also regulated by the synthesis of thymidine. Likewise, based on findings in Fig. 1, C and D, we propose that TYMS SS expression is regulated by the synthesis of tetrahydrofolate (THF). As a derivative of THF is used to make thymidine, a logical conclusion was made that DHFR FS regulates the expression of TYMS SS and TYMS SS regulates the expression of DHFR FS . If this is the case, a correlated expression of DHFR FS and TYMS SS should be noted within individual cells. A correlated expression of DHFR FS and TYMS SS was indeed observed in flow cytometry plots of Jurkat cells in Fig. 1E and primary T cells in Fig. 1F. A control RFP vector co-expressed with DHFR FS , but not modulated by cis expression with TYMS SS , did not appear to have the same co-expression pattern (Fig. 1F). To quantify this observation, Jurkat cells expressing [DHFR FS & TYMS SS ] were treated with antifolates MTX, pemetrexed, and raltitrexed at varying concentrations for 2 weeks. DHFR FSlinked eGFP mean fluorescence intensity (MFI) and TYMS SSlinked RFP MFI for each separate well were then plotted and correlated. The linked expression between DHFR FS and TYMS SS was significant and fit a linear regression (Fig. 1G). These findings support a general mechanism for regulation of DHFR and TYMS, which leads to a linear co-expression of DHFR FS and TYMS SS independent of MTX. This model is shown in Fig. 1H.
Based on the above model in Fig. 1H, it appears that TYMS SS expression will be stabilized by DHFR FS from strong expression changes in the presence of MTX. This was tested in Fig. 2A with primary T cells expressing DHFR FS along with either RFP or TYMS SS linked to RFP by applying increasing doses of MTX. As expected, DHFR FS -linked eGFP was increased by rising concentrations of MTX, and this increase was blunted by the presence of TYMS SS (Fig. 2A, panel I). This conserves the model in Fig. 1H where restoration of thymidine synthesis prevents the MTX-induced increase in DHFR FS . Further conserving the model, RFP linked to TYMS SS did not significantly increase over any concentration of MTX used ( Fig. 2A, panel II). When DHFR FS -linked eGFP increased, so too did the control RFP, and a rise in the expression of RFP alone was not expected. A possible explanation is that this increase was noted above 0.5 M MTX, and DHFR FS alone is only resistant to 0.5 M MTX (9). This suggests that higher doses of MTX begin to select for cells with higher gene content of DHFR FS and associated transgenes. This is supported by Kacherovsky et al. (10) where greater DHFR FS expression was selected by the addition of MTX. Notably, DHFR FS co-expressed with TYMS SS is resistant to concentrations of up to 1 M MTX. This further supports the use of TYMS SS to modulate transgene expression and prevent unwanted selection toward higher gene expression levels of genes expressed in cis or trans with DHFR FS . Our group previously described a construct of DHFR FS cis expressing an inducible suicide gene, iC9 (9). This construct, called D FS iC9, selects for T cells expressing D FS iC9 in the presence of MTX and ablates D FS iC9 ϩ T cells in the presence of drug that activates iC9 to induce apoptosis (24). Based on the above findings, the DHFR FS in D FS iC9 could be used to modulate and potentially ablate the expression of a transgene of interest that is otherwise too toxic to express without regulation. IL-12 is such a transgene. IL-12 is a cytokine capable of inducing a strong immune response against tumor from tumor-specific T cells (25). However, systemic IL-12 is highly toxic and hence of low efficacy (26). Groups seeking to apply IL-12 to a clinical setting have sought inducible systems to overcome the toxicity related to this cytokine where T cells express IL-12 after activation (27). Presented here is an alternative approach where IL-12 is expressed cis to TYMS SS to decrease and stabilize the expression level of IL-12. In Fig. 2B, a flow plot demonstrates the expression of IL-12 cis-expressed with TYMS SS and iC9 cis-expressed with DHFR FS expression. The primary T cells were either left untreated or treated with high doses of MTX for 7 days. This expression pattern appears to indicate that IL-12 can be stably expressed even in toxic doses of MTX. A further analysis of similarly manipulated donors (Fig. 2C) demonstrates the potential of TYMS SS when co-expressed with DHFR FS to stabilize the expression of the potentially toxic transgenes of interest (here IL-12).
As for the endogenous regulatory mechanism proposed in Fig. 1H, there is evidence to support a natural regulatory pathway leading to linked expression of TYMS and DHFR. Regulation of TYMS expression by TYMS binding its own mRNA (19) and regulation of DHFR expression by DHFR binding its own mRNA (14) are both translationally regulated by the presence of folates. Our findings indicate that a DHFR mRNAindependent translational or post-translational mechanism is also involved in directing the co-expression of DHFR and TYMS. In this study, DHFR FS and TYMS SS expression was tracked by fluorescent proteins linked by a 2A ribosomal "slip" site. These linkages express two independent proteins due to slippage of the completed N terminus peptide at the 2A sequence. In this study, the N terminus peptide was DHFR FS and TYMS SS (28). The above findings of MTX-inducible changes suggest that if the fluorescent protein is regulated by either DHFR FS or TYMS SS , then this regulation is occurring before or during translation rather than after the proteins sep-arate post-translationally. Thus, our findings add complexity to the translational regulation of DHFR and TYMS to meet an unknown biological need for combined expression of DHFR and TYMS. Findings by Anderson et al. (29) demonstrated that DHFR and TYMS co-localize as a multi-enzyme complex within the nucleus during replication of DNA. Stabilization and construction of this multi-enzyme complex within the cell may explain the need for tight translational control of DHFR and TYMS.
In conclusion, we have demonstrated an endogenous mechanism of post-transcriptional regulation for DHFR and TYMS that can be co-opted with either DHFR FS or TYMS SS to regulate expression of cis-or trans-expressed transgenes. The purpose of this study is to describe a novel system for transgene regulation in gene therapy. This system is of significant clinical interest as the genes described are human transgenes of low immunogenicity that utilize MTX to regulate transgene expression. MTX is readily available for use in both hospital and clinical settings. As these studies demonstrate that this phenomenon occurs in primary human T cells, this system can readily be used in the field of transgenic T cell immunotherapy to improve the safety and efficacy of T cell therapies used to treat cancer (30). What remains to be determined is how an improved understanding of the regulatory mechanism surrounding DHFR and TYMS expression may lead to a better understanding of developmental disorders and improved chemotherapeutics.