Generation of pralatrexate resistant T‐cell lymphoma lines reveals two patterns of acquired drug resistance that is overcome with epigenetic modifiers

Abstract While pralatrexate (PDX) has been successfully developed for the treatment of T‐cell lymphoma, the mechanistic basis for its T‐cell selectivity and acquired resistance remains elusive. In an effort to potentially identify synergistic combinations that might circumnavigate or delay acquired PDX resistance, we generated resistant cells lines over a broad concentration range. PDX‐resistant cell lines H9‐12 and H9‐200 were developed, each exhibiting an IC50 of 35 and over 1000 nM, respectively. These lines were established in vitro from parental H9 cells. Expression analysis of the proteins known to be important determinants of antifolate pharmacology revealed increase expression of dihydrofolate reductase (DHFR) due to gene amplification, and reduced folate carrier1 downregulation, as the putative mechanisms of resistance in H9‐12 and H9‐200 cells. Cross resistance was only seen with methotrexate but not with romidepsin, azacitidine (AZA), decitabine, gemcitabine, doxorubicin, or bortezomib. Resistance to PDX was reversed by pretreatment with hypomethylating agents in a concentration‐dependent fashion. Comparison of gene expression profiles of parental and resistant cell lines confirmed markedly different patterns of gene expression, and identified the dual specificity phosphatase four (DUSP4) as one of the molecular target of PDX activity. Reduced STAT5 phosphorylation following exposure to PDX was observed in the H9 but not in the H9‐12 and H9‐200 cells. These data suggest that combination with hypomethylating agents could be potent, and that DUSP4 and STAT5 could represent putative biomarkers of PDX activity.


| INTRODUCTION
Folate metabolism plays a central role in multiple biological pathways and an unbalanced folate metabolism has been implicated in the pathogenesis of different diseases, including lymphomagenesis. 1 Pralatrexate (PDX) was the first drug approved for patients with relapsed/refractory peripheral T-cell lymphoma (PTCL). Unlike other folates, PDX was designed to have high affinity for the reduced folate carrier (RFC), which internalizes both natural forms of folic acid and antifols. This rapid and efficient internalization leads to relatively high intracellular concentrations. While PDX inhibits many of the same enzymes inhibited by other antifols like methotrexate (MTX), it is clear that its activity in PTCL goes well beyond the inhibition of these pathways. This is underscored by the fact that leucovorin does not mitigate its activity, which is the case for classical antifolates like MTX, and suggests that there are other targets the drug influences in PTCL that account for some component of its efficacy. 2 Although PDX is significantly more cytotoxic when compared to first-generation antifolates, its precise mechanism of action and T-cell malignancy is not completely understood. PDX received regulatory approval for the treatment of relapsed or refractory PTCL in 2009.
Early preclinical and clinical experiences have suggested an extraordinarily high level of activity in T-cell malignancies, with minimal to no activity in patients with B-cell lymphoid neoplasm or carcinomas. [3][4][5] Despite the greater selectivity of PDX against T-cell lymphomas, the overall response rate (ORR) in heavily treated patients is about 30%, though recent data from studies in Asia suggest ORR in the 50%. 6,7 Several lines of data suggest that PDX use earlier in the line of therapy can improve the ORR to about 45% to 50%, and lead to markedly better complete remission rates, and progression free and overall survival. This latter observation suggests that there is cross-resistance with other conventional cytotoxic chemotherapy agents which accumulates over time. 8 Intrinsic resistance to PDX appears to be overcome to some extent through the use of combinations, as preclinical and clinical data evaluating PDX combinations have shown marked synergy with histone deacetylase inhibitors (HDACi) like romidepsin, 9 proteasome inhibitors like bortezomib, 10 and cytidine analogs. 11 For example, while the ORR to single agent PDX and romidespin are about 30% and 25%, respectively, the ORR to PDX plus romidepsin in a phase 1 study is in the 70% to 75% range, with a complete response rate of 29%. 9,12,13 While the mechanistic basis for the PTCL selectivity and acquired resistance to PDX are still not entirely understood, it is clear the T-cell malignancies represent one of those neoplastic diseases characterized by high rates of intrinsic drug resistance, and a penchant for rapid emergence of acquired resistance. Herein, we have developed a series of PDX-resistant cell lines as a strategy to better understand mechanisms of acquired resistance and as a strategy to potential identify companion agents that might overcome mechanisms of intrinsic resistance.

| Materials
All drugs were purchased from Selleckchem (Houston, Texas) and were prepared in 100% DMSO.

| Cell lines and culture
The cutaneous T-cell lymphoma line H9 was obtained from the American Type Culture Collection (Manassas, Virginia). Parental and derived resistant cells were grown in RPMI 1640 medium with 10% fetal bovine serum.

| Cytotoxicity assays
Cytotoxicity was performed on cultured cells using the Cell Titer Glo assay (Promega, Madison, Wisconsin) as previously described. 10 Synergy of the combinations was calculated using Excess Over Bliss methodology. 10 Luminescence was detected using the multimode plate reader GloMax Discover system (Promega, Madison, Wisconsin).
For the pretreatment with AZA and decitabine, cell were exposed for 4 days to increased concentration of the hypomethylating agent (125, 250, and 500 nM for AZA and 6.25, 12.5, and 25 nM for decitabine). The compound was added to the cell culture every 24 hours for a 96-hour period. Then, cells were washed and seeded at concentration of 3 × 10 5 cells/mL and exposed to increased concentration of PDX.

| Methylation specific PCR
Genomic DNA was extracted using a Quick-DNA kit (Zymo Research, Irvine, California) and bisulfite treatment was carried out using a EZ DNA Methylation kit (Zymo Research, Irvine, California) following the manufacturer's instructions. Primers specific for the RFC promoter A and B regions and reaction condition were as previously described. 14

| Gene expression profiling
Total RNA was extracted using RNeasy mini kit (Qiagen, Hilden, Germany) from cells collected after 24 hours incubation with or without drugs (MTX and PDX) and RNA quantitation and quality over assessed by the Agilent Bioanalyzer 2100. RNA libraries prepared from poly-A pull-down enrich mRNAs (Illumina TruSeq RNA prep kit, San Diego, California), were sequenced at the Columbia Genome Center using Illumina HiSeq2500/HiSeq4000. DEseq software, an R package based on a negative binomial distribution that models the number reads from RNA-seq experiments and test for differential expression, was employed to test for differentially expressed genes under various conditions. For visualization, raw counts were normalized sample-wise to reads per million and differential expression was calculated for each cell line as a z-score centered at untreated controls. Hierarchical clustering was calculated by euclidean distance using the hclust function in the stats R package and visualized using the heatmap.2 function within the gplots R package. Unclustered heat maps were also generated with the heatmap.    Figure 4C). However, the RFC mRNA level in the H9-200 exposed to AZA was substantially reduced when compared to H9 and H9-12 mRNA levels.  Figure 6C). A supervised analysis of the genes whose expression is modulated (adjusted P value .05, absolute Log 2-fold change 0.5, Supporting information Tables 1 and 2) by each single agent is shown in Figure 6D. Three patterns of modulation were identified: (a) genes whose expression is modified by both agents, (b) genes whose expression is more substantially modified by MTX than PDX, and (c) genes whose expression is more substantially modified by PDX than MTX. Interestingly, none of the patterns are seen in the two resistant cell lines H9-12 and H9-200 following exposure to MTX or PDX.
The DUSP4 was among the genes whose expression was found upregulated after exposure to MTX or PDX ( Figure 6B,C). As shown in Figure 6E, increase in DUSP4 protein level was also observed, when H9 and MyLa, two CTCL derived cell lines and TLOM1, an adult T-cell leukemia derived cell line, were exposed to PDX. DUSP4 has been showed to regulate STAT5 stability in T cells 28 and suppress CD4 T cell proliferation through regulation of STAT5 phosphorylation and IL-2 signaling. 29 As shown in Figure 6E a decrease in STAT5 phosphorylation was observed in parental H9 but not in H9-12 and H9-200 resistant cells. Interestingly, the decrease in STAT5 phosphorylation correlates with a corresponding increase in DUSP4 expression.

| DISCUSSION
The development of drug resistant cell lines can represent a valuable tool in understanding mechanisms of intrinsic and acquired drug resistance, and can provide insights in mechanism of action. 30,31 Resistance to antifolates has been ascribed to several mechanisms, Although the GEP of parental and resistant cell lines after PDX and MTX exposure revealed perturbation of common pathways, the number and type of differentially expressed genes within the common pathways diverge, suggesting that there are substantial differences between these drugs. The notion that PDX has a different mechanism of action making it noncross resistant to MTX has been supported by clinical data, and the observation that leucovorin does not nullify all of its cytotoxic effects. 5 In this instance, the comparison of GEP analysis of parental and resistant cell lines, exposed to MTX and PDX, could provide mechanistic insights into the differential effects The identification of DUSP4 as one of the genes whose expression is increased in H9, MyLa and TLOM1 cells following exposure to PDX is also an example of how we could use this information to generate hypotheses around mechanism. DUSP4 is a member of the dual specificity protein phosphatatse subfamily and DUSPs have been implicated in T-cell lymphomagenesis. These phosphatases negatively regulate members of the mitogen-activated kinase superfamily involved in cellular proliferation and differentiation.
Reduced expression of DUSP4 has been identified as an adverse feature in different tumor types and cancer derived cell lines [34][35][36][37] and found to be epigenetically silenced in 75% of cases of diffuse B cell lymphoma (DLBCL). Moreover, a lack of DUSP4 was a negative prognostic factor in three independent cohorts of DLBCL patients. 38 An increase in DUSP4 mediated by PDX would theoretically counterbalance the adverse pathogenetic implications of reduced DUSP4.
These findings are further supported by our observation of STAT5 decreased phosphorylation in H9 but not in H9-12 and H9-200 exposed to PDX. DUSP4 has shown to regulates STAT5 phosphorylation, 28 and recent findings show that STAT5 activation drives PTCL, 39 while its inhibition induces apoptosis in PTCL. 40 Pimozide, an FDA-approved neuroleptic agent, shown to be a STAT5 inhibitor, demonstrated a concentration dependent reduction in STAT5 activity and in the number of viable cells in PTCL lines. 40 Further studies are underway to explore whether the reduction in STAT5 phosphorylation is the result of DUSP4 induced expression by PDX.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article.