PIM Kinases as Potential Therapeutic Targets in a Subset of Peripheral T Cell Lymphoma Cases

Currently, there is no efficient therapy for patients with peripheral T cell lymphoma (PTCL). The Proviral Integration site of Moloney murine leukemia virus (PIM) kinases are important mediators of cell survival. We aimed to determine the therapeutic value of PIM kinases because they are overexpressed in PTCL patients, T cell lines and primary tumoral T cells. PIM kinases were inhibited genetically (using small interfering and short hairpin RNAs) and pharmacologically (mainly with the pan-PIM inhibitor (PIMi) ETP-39010) in a panel of 8 PTCL cell lines. Effects on cell viability, apoptosis, cell cycle, key proteins and gene expression were evaluated. Individual inhibition of each of the PIM genes did not affect PTCL cell survival, partially because of a compensatory mechanism among the three PIM genes. In contrast, pharmacological inhibition of all PIM kinases strongly induced apoptosis in all PTCL cell lines, without cell cycle arrest, in part through the induction of DNA damage. Therefore, pan-PIMi synergized with Cisplatin. Importantly, pharmacological inhibition of PIM reduced primary tumoral T cell viability without affecting normal T cells ex vivo. Since anaplastic large cell lymphoma (ALK+ ALCL) cell lines were the most sensitive to the pan-PIMi, we tested the simultaneous inhibition of ALK and PIM kinases and found a strong synergistic effect in ALK+ ALCL cell lines. Our findings suggest that PIM kinase inhibition could be of therapeutic value in a subset of PTCL, especially when combined with ALK inhibitors, and might be clinically beneficial in ALK+ ALCL.


Introduction
Peripheral T cell lymphomas (PTCL) are a very aggressive and heterogeneous group of hematological malignancies [1,2]. Very little is known about their molecular biology, and consequently, the search for efficient therapies that would improve the outcome of these patients remains challenging [3]. Several factors are responsible for our limited knowledge, such as the low incidence of PTCL, the heterogeneity of its subtypes and the few representative models (cell lines or mouse models) available. It is worth to note that majority of the available cell lines cover very few PTCL subtypes, and are mostly derived from cutaneous T cell lymphomas, including its two most prevalent forms: Sézary Syndrome and Mycosis Fungoides [4].
Among the different PTCL subtypes, PTCL-Not Otherwise Specified (PTCL-NOS), Angioimmunoblastic T-cell Lymphoma (AITL) and Anaplastic Large Cell Lymphoma (ALCL) are the most frequent ones. Within the ALCL group, there are two subgroups, depending on the presence or absence of the chromosomal translocation t(2; 5) (p23; q35), which involves the ALK and NPM1 genes and leads to the overexpression of the fusion protein NPM-ALK [3]. This is considered to be the main oncogenic force in ALK+ ALCL, because it activates the Jak/ STAT pathway [5,6]. The ALK+ ALCL is the only PTCL subgroup with a relatively good prognosis [7], however, around 40% of ALK+ ALCL patients fail to be cured using standard therapeutic approaches [3]. New drugs, such as the ALK inhibitor Crizotinib, seem to improve the survival in these patients in early clinical trials [8].
Although different histological subtypes of PTCL have been identified, the treatment approach has been essentially based on the application of anthracycline-based combination chemotherapy, resulting in poor outcomes [9]. To date, only 3 agents have been recently approved by the FDA for the treatment of relapsed or refractory PTCL: pralatrexate, romidepsin and brentuximab vedotin [9,10]. Nevertheless, the development of new, efficiently targeted therapies is of great importance to PTCL patients [9,11,12].
The Proviral Integration site of the Moloney murine leukemia virus (PIM) family is an important mediator of cell survival, comprising three ubiquitously expressed serine/threonine kinases (PIM1, PIM2 and PIM3) with a broad range of cellular substrates that promote cell growth, proliferation and drug resistance. They are overexpressed in a number of human cancers and frequently associated with poor prognosis in most hematological malignancies [13]. PIM kinases are typically induced by the activation of transcription factors downstream of growth factors, cytokines and mitogenic stimuli signaling pathways, such as the Jak/STAT and NF-kB [13,14], and are also protected from proteasomal degradation by HSP-70 and HSP-90 [15]. Their activities are mediated through the phosphorylation of a number of proteins, including regulators of transcription (MYC, MYB, RUNX1, RUNX3), cell cycle (p21, p27, CDC25A, CDC25C), protein translation (EIF4E, 4E-BP1), apoptosis (BAD, BCL2, ASK1), signaling intermediates (SOCS1, SOCS3, MAP3K5, mTOR, AKT), and drug resistance proteins (ABC transporters) [13,14,15]. Studies using transgenic mice have shown that PIM kinases cooperated with important genes involved in B-and T-cell lymphomagenesis, such as, c-Myc, BCL6 and E2A-PBX1 [14]. GSEA ranked all significantly altered genes between PTCL and LN according to its correlation with PIM1 or PIM2 expression and displayed them in the red-to-blue bar. Each gene belonging to every pathway was interrogated whether it appeared positively (in the red region of the bar) or negatively (in the blue side) correlated. Using this approach GSEA identified a positive and significant correlation between PIM1 and PIM2 expression and Jak/STAT, NF-kB and IL-2 signaling pathways in the PTCL molecular signature (FDR,0.25). (C) PIM family genes mRNA level was measured by RT-qPCR in eight PTCL cell lines and (D) primary tumoral T cells from 5 Sézary Syndrome patients (SS #1-5), and compared with normal T cells isolated from 3 healthy donors (Control #1-3). The relative RNA amount of PIM has been calculated as a relative quantification, as described in the Methods section (RQ = 2 2DCt ), normalized with non-tumoral cells: RQ in PTCL/RQ in healthy #3. In both settings, PIM1, and especially PIM2, but not PIM3 expression was found to be increased in PTCL. (E) PIM kinase protein basal levels in PTCL cell lines measured by Western blot. PIM1 and PIM2 isoforms are also shown. (F) Distribution of PIM2 protein in a series of tumoral samples from 136 PTCL patients measured by immunohistochemistry. Negative, weakly positive and strongly positive samples were defined by the presence of ,5%, 5-20% and .20% positive cells. (G) Distribution of PIM2 protein in the most common PTCL subtypes measured by immunohistochemistry. doi:10.1371/journal.pone.0112148.g001 On the other hand, triple PIM1-PIM2-PIM3 knockout mice have been reported to be viable, fertile, and just smaller compared with wild type littermates [13,14,15]. Very recently, an abnormal hematopoiesis has been described in these triple-knockout mice [16]. These findings indicate that PIM kinases are important for lymphomagenesis and their absence is well tolerated, suggesting that selective PIM kinase inhibitors might have a low toxicity profile [13]. Based on this, along with some differences in the structural conformation of the ATP-binding pocket in the active site compared with other kinases, PIM kinases have been proposed as promising therapeutic targets for pharmacological inhibition. So far, a number of small molecule inhibitors have been tested in vitro, but clinical data are only available for a handful of them. One of the most promising PIM inhibitors (PIMi) was SGI-1776, a compound with activity against PIM1, PIM2 and PIM3 at nanomolar concentrations [13,14,15], which induced apoptosis at micromolar doses in chronic lymphocytic leukemia [17], mantle cell lymphoma [18], and acute myeloid leukemia [19]. Unfortunately, the phase I clinical trial of this compound was discontinued in November 2010 because of a strong cardiotoxic effect that impaired its further development [15,20].
This study aimed to determine the efficiency of PIM kinase inhibition in PTCL, to explore the molecular response of PTCL cells to pharmacological pan-PIM inhibition and to identify those PTCL subgroups that are more susceptible to PIM inhibition.

Ethics statement
The research was approved by the Hospital Universitario Marqués de Valdecilla ethics committee (Santander, Spain).
All the human samples used in this study have been procured from the Spanish CNIO Biobank, located in the Spanish National Cancer Research Centre (Madrid, Spain) (https://www.cnio.es/ ing/servicios/biobanco/index.asp), and according to the Spanish legal framework regarding written informed consent and sample anonymization.
Additionally, some samples used here were also previously used in [15].

Bioinformatics analysis in the PTCL patient series
The gene expression profiles of frozen tumoral samples from 38 PTCL patients and 6 reactive lymph nodes were compared using microarrays. Briefly, differentially expressed genes between PTCL and reactive lymph nodes were identified using a t-test. Then, Gene Set Enrichment Analysis (GSEA) ranked them according to its correlation with PIM1 or PIM2 expression. More details are provided in the Supplementary Information (Methods S1). All raw microarray data regarding the PTCL patient series are available at the Gene Expression Omnibus under accession number GSE36172.

Cell lines, primary samples and reagents
Eight human PTCL cell lines were used in this study. HH (cutaneous T cell lymphoma) and MJ (HTLV1+ PTCL) were obtained from the American Type Cell Collection (ATCC, Rockville, MD); MyLa (Mycosis Fungoides) and HuT78 (Sézary Syndrome) were obtained from the European Collection of Cell Cultures (ECACC, Salisbury, UK); DERL7 (hepatosplenic gamma-delta T cell lymphoma) and SR786, KARPAS-299 and SU-DHL-1 (ALK+ ALCL) were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). All of them except MJ were cultured in RPMI 1640 medium (IMDM medium for MJ cells) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% penicillin/ streptomycin (all from Life Technologies, Carlsbad, CA) in a humidified atmosphere at 37uC and 5% CO 2 . The DERL7 cell line was supplemented with 20% FBS and 20 ng/ml human IL-2 (PeproTech, Rocky Hill, NJ). Cell lines were previously authenticated by DSMZ (year 2010-2011).
Human primary samples were used to measure the basal PIM1, PIM2 and PIM3 mRNA levels. Tumoral and normal T cells were respectively isolated from the peripheral blood of 5 Sézary Syndrome patients and 3 healthy donors, through negative selection using the RosetteSep kit (StemCell Technologies, Grenoble, France). Sample purity was checked by flow cytometry, and an enrichment of .90% of CD3+ cells was ensured in all samples. Additionally, the PIM inhibitor sensitivity of primary T cells from 8 PTCL patients (5 Sézary Syndrome and 3 Mycosis Fungoides) and 5 healthy donors was tested.
The pan-PIM inhibitors ETP-39010 [21], ETP-47652, ETP-47551 and ETP-46638 were developed by the Experimental Therapeutics Programme of the Spanish National Cancer Research Centre (Madrid, Spain). The chemical structure of these compounds has been published in [22] under publication number WO 2011/080510. (http://www.wipo.int/portal/index.html.en). The ALK inhibitor (ALKi) Crizotinib was obtained from Selleck Chemicals (Houston, TX). All inhibitors were dissolved in DMSO and the stocks were kept at 220uC. They were diluted in culture medium at desired concentrations immediately before use. For the controls DMSO concentration in the medium was lower than 0.2%.

PIM genetic silencing experiments
Transient genetic silencing was performed in PTCL cell lines as follows: HH, SR786, SU-DHL-1 and MyLa cell lines were electroporated with specific small interference RNAs (siRNAs) against PIM1, PIM2 and PIM3 genes, using the Neon Transfection System (Life Technologies) and following the manufacturer's instructions, as previously described [23]. Briefly, cells were incubated without antibiotics overnight and resuspended in R buffer at a density of 500,000 cells/ml. Then, siPIM1 (s10527), siPIM2 (s21749), siPIM3 (HSS140560) and the Non-Template Control (NTC, AM4635) (all from Life Technologies) were added to the cells at several concentrations (25-100 nM). Microporation conditions were set up for each cell line (900 V, 30 ms and 2 pulses for HH, SR786 and SU-DHL-1; 1300 V, 20 ms and 2 pulses for MyLa), aiming for the highest transfection efficiency with the minimum loss of cell viability. Cells were then electroporated under these conditions to allow for the entry of the siRNAs into the cell and 100 ml of the suspension were seeded in 2 ml for 24, 48, 72 and 96 h.
In addition, a stable PIM knockdown was carried out in PTCL cell lines using the MISSION product line from Sigma-Aldrich (St Louis, MO) according to the manufacturer's instructions. The base vector (pLKO.1-puro) contains the Puromycin resistance gene for mammalian cells selection. Thus, sensitivity to Puromycin was first tested in several PTCL cell lines, and optimal concentrations were chosen from a wide range. Then, the optimal amount of lentiviral particles was assessed using the control transduction particles, both the negative (Non-targeting shRNA, SHC016V) and the positive (Turbo-GFP, SHC003V) lentiviruses (Sigma-Aldrich). MyLa was the only used cell line showing high infection efficiency, and therefore, was the best model to test PIM stable knockdown. Briefly, MyLa cells were infected with MISSION lentiviral transduction particles containing specific short-hairpin RNA (shRNA) against PIM1 (SHCLNV-NM_002648), PIM2 (SHCLNV-NM_006875) and PIM3 (SHCLNV-NM_001001852) (all from Sigma-Aldrich) using Polybrene (hexadimethrine bromide) as a transduction enhancer (8 mg/ml). After 24 h post-infection, lentiviral particles were removed and Puromycin (4 mg/ml) was added to culture media. Green fluorescence was checked for 15 days using a Nikon Ti Epi-Fluorescence microscope and the imaging software NIS-Elements (Nikon, Amsterdam, Netherlands) and flow cytometry.

Cell viability assay
For drug cytotoxicity experiments, PTCL cell lines and primary tumoral and normal T cells were seeded in 96-well plates at a density of 10,000 cells per well, and pan-PIM inhibitors, ALKi, Cisplatin or combinations were added at a range of doses for 48 h (for primary cells) and 72 h (for cell lines), using DMSO as control. Cell viability was measured as the intracellular ATP content using CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI), following the manufacturer's instructions.
For drug combination experiments, cells were treated with a wide range of doses and cell viability was measured as explained above. The combination index (CI) was calculated using CalcuSyn software (Biosoft, Ferguson, MO) following the Chou & Talalay method [24], where values of CI,1, <1 and .1 indicate synergism, an additive effect and antagonism, respectively.

Flow cytometry analysis
The distribution of cells during the phases of the cell cycle and induction of apoptosis were evaluated by flow cytometry using propidium iodide (PI, Sigma-Aldrich, St Louis, MO) staining and the APC-Annexin V (Beckton Dickinson, BD, Franklin Lakes, NJ) binding assay, respectively. Data from 10,000 cells were detected on a FACS Calibur flow cytometer (BD) and analyzed using CellQuest Pro software (BD).

RNA extraction and quantitative RT-PCR
Total RNA was extracted and purified using RNeasy Mini-Kit (Qiagen, Valencia, CA) following the manufacturer's instructions in order to check the PIM genes' knockdown efficiency, to understand the molecular response to the pan-PIMi and to measure basal mRNA levels of these genes in PTCL cell lines and primary tumoral and normal T cells.

XRCC2
Hs03044154_m1, XRCC5 Hs00897854_m1, and the endogenous control RN18S1 Hs03928990_g1; all from Life Technologies) in a final volume of 15 ml. PCR amplification was performed using the Applied Biosystems Prism 7900HT Sequence Detection System (Life Technologies) under the following thermal cycler conditions: 2 min at 50uC, 10 min at 95uC and 30 cycles (15 s at 95uC and 1 min at 60uC).
Relative Quantification (RQ) was calculated following the DCt method: RQ = 2 2DCt , where DCt is the difference between the Ct of the gene of interest and the Ct of the endogenous gene control RN18S1. In addition, in knockdown experiments RQ was normalized as RQ = 2 2DDCt , where DDCt is the difference between the DCt in knockdown cells and the DCt in control cells.

Microarray hybridization and data analysis
The molecular response of PTCL cells to the pan-PIMi ETP-39010 was explored through gene expression analysis. DERL7, HuT78, MyLa and SR786 cells were treated with 10 mM pan-PIMi for 0, 2, 4, 6, 10 and 24 h. At each time, DMSO-and pan-PIMi-treated cells were harvested and total RNA extracted as described above, and the quality assessed in a 1% agarose-gel.
Samples were hybridized onto 4644 K microarray slides (Whole Human Genome, Agilent Technologies, Inc., Santa Clara, CA), as described in the Supplementary Information (Methods S1). Short Time-series Expression Miner (STEM) [25] identified differentially expressed genes under each condition, and Gene Ontology (GO) categories were used to recognize functional groups. A False Discovery Rate (FDR) ,0.05 was considered significant. All raw microarray data regarding the molecular response to pan-PIMi in PTCL are available at the Gene Expression Omnibus under accession number GSE42595.

Immunofluorescence and immunohistochemistry
After treatment with 10 mM of the pan-PIMi ETP-39010 for 24 h, cH2A.X was measured in MyLa cells by immunofluorescence. Immunohistochemical staining of PIM2 was carried out in a series of formalin-fixed and paraffin-embedded tumoral samples from 136 PTCL patients (Table S1) in a Bond Max automatic immunostainer (Leica Microsystems, Wetzlar, Germany). Details are provided in the Supplementary Information (Methods S1).

Statistical analysis
Unless otherwise specified, all experiments were done three times and all numerical data were expressed as the average of the values 6 the standard error of the mean. Statistical significance of differences between groups was established by Student's independent samples t-test (SPSS v17.0). p-values ,0.05 were considered significant.

PIM kinases as potential therapeutic targets in PTCL
First, using microarrays, PIM1 and especially PIM2 genes, but not PIM3, were found to be significantly overexpressed (FDR, 0.05) in tumoral samples from 38 PTCL patients compared with 6 reactive lymph nodes ( Figure 1A). PIM1 and PIM2 expression was significantly correlated with Jak/STAT, NF-kB and IL-2 pathways in our PTCL patient series ( Figure 1B), indicating a strong relationship between these pathways and the expression of PIM kinases in PTCL. Furthermore, PIM1 and, again, especially PIM2, but not PIM3 expression was increased in a panel of 8 PTCL cell lines ( Figure 1C) and primary tumoral T cells from 5 Sézary Syndrome patients ( Figure 1D) relative to normal T cells from 3 healthy donors. Similarly, PIM protein levels were also detected by Western blot in all PTCL cell lines, with slight differences in the most expressed PIM2 isoform ( Figure 1E).
Since PIM2 was the most upregulated PIM kinase in PTCL (both patients and cell lines) at the mRNA and protein levels, we explored the expression of the PIM2 protein by immunohistochemistry in a series of 136 PTCL patients. We found that 77% of these samples were positive for PIM2 expression ( Figure 1F and Figure S1), and that the trend was largely maintained in the most common PTCL subtypes, with a slight predominance in the AITL subtype ( Figure 1G). Although our series was limited, a preliminary significant association was found between PIM2 expression and a shorter overall survival only in the ALCL subtype, both ALK+ and ALK2 cases, but not in other PTCL subtypes ( Figure  S2).
These findings suggest that PIM kinases could be of potential therapeutic value in PTCL.

PIM kinase genetic silencing in PTCL
To test this hypothesis, genetic silencing experiments with siRNAs were performed specifically to abolish the expression of PIM1, PIM2 or PIM3 genes in PTCL cell lines. Knockdown efficiency differed with the cell line and time point, varying from around 70 to 95% in the PIM1 or PIM2 mRNA, and lower for PIM3 ( Figure 2A). However, no significant effects on cell survival were observed, either with respect to apoptosis induction ( Figure 2B and Table S2) or cell cycle arrest ( Figure S3). These results indicated that the remaining protein or other untargeted genes were responsible for triggering survival.
As PIM genes belong to the same family and have a high homology in their sequences [15,19,20], they could share functions. This prompted us to measure the mRNA levels of each of the PIM members when one of them was knocked down. Strikingly, we found an upregulation of PIM2 and PIM3 when PIM1 was silenced. Likewise, an increase in PIM1 and PIM3 was observed when PIM2 was inhibited ( Figure 2C). Because the PIM3 knockdown was less efficient than other PIM genes silencing, PIM1 or PIM2 upregulation after PIM3 inhibition was less evident (data not shown). Again, very similar results were found in all cell lines, suggesting the existence of a compensatory mechanism among the PIM genes in PTCL.
This led us to exploit the simultaneous silencing of the 3 PIM genes: since the recommended maximum siRNA concentration is 100 nM and we aimed to inhibit 3 genes at the same time, the concentration of each siRNA was reduced to 33 nM. This meant that the knockdown efficiency was lower than for individual silencing: around 70% at the mRNA level on average for the 3 genes 24 h after the microporation (Figure 3A), and about 50% at the protein level after 48 h ( Figure 3B). Once more, however, cell survival was unaffected, producing no significant induction of apoptosis ( Figure 3C) or cell cycle arrest ( Figure S4) in any cell line under any of the studied conditions. These results could be due to the transient knockdown triggered by siRNAs.
Additionally, a more stable knockdown of PIM genes was approached using lentiviral particles containing shRNAs inserted into the pLKO.1-puro vector. Since infected cells should be selected with Puromycin, first we tested the sensitivity of several PTCL cell lines to this antibiotic. SU-DHL-1 and SR786 cell viability was rapidly impaired in the presence of Puromycin, while HH and MyLa cells showed a greater resistance (data not shown). Then, we checked the infection efficiency using lentiviral particles containing GFP, and observed a high proportion of green MyLa cells, while no green HH cells were found even after 15 days postinfection (data not shown). Thus, MyLa cell line was chosen to be infected with shPIM-lentiviral particles: 74% of cells were infected 8 days after lentiviral addition ( Figure 4A). This efficiency was checked to be as high as possible, because in parallel, non-infected cells were cultured in the presence of Puromycin, and at this time point all these cells were dead ( Figure 4A). However, under these conditions around 30% of PIM-mRNA was still detectable by RT-qPCR (data not shown). Even when the 3 shRNAs (shPIM1+ shPIM2+ shPIM3) were simultaneously added and cells were selected for 15 days, we found a decrease of only 30%, on average, both at the mRNA ( Figure 4B) and protein ( Figure 4C) levels. Again, these conditions did not affect cell survival (Figures 4A and  4D). These results could be due to the incomplete silencing of all 3 PIM kinases, with the remaining active protein still triggering enough survival signaling.

Pharmacological pan-PIM kinase inhibition in PTCL
In order to inhibit the catalytic activity of all PIM kinases more efficiently, the pharmacological pan-PIM inhibitor ETP-39010 [21] was used. We found that this drug reduced cell viability in all PTCL cell lines in the same low micromolar range of IC 50 values ( Figure 5A). This effect was mainly due to a strong dose-and timedependent induction of apoptosis ( Figure 5B and Figure S5A, S5B and S5C), without cell cycle arrest ( Figure S5D). The subG0 population (reflecting dead cells) increased even with shortduration, low-dose treatments, especially in the KARPAS-299, SU-DHL-1 and SR786 cell lines, which are members of the ALCL subtype ( Figure S5D).
These results indicated a direct and strong cytotoxic effect of the pharmacological pan-PIMi on PTCL cell lines.
Additionally, we tested the ex vivo efficiency of the pan-PIMi in PTCL. Primary T cells from 8 PTCL patients (Mycosis Fungoides and Sézary Syndrome) and 5 healthy donors were treated for 48 h, and we observed that tumoral T cell viability was slightly but significantly reduced, while normal T cells remained unaffected ( Figure 5C).

Molecular response of PTCL to the pharmacological pan-PIMi
To confirm that the pharmacological pan-PIMi was really inhibiting PIM kinase activity, we measured the phosphorylation status of 4E-BP1, a well-established substrate of PIM kinases [13,14,15]. A decrease in p4E-BP1 was found in PTCL cell lines after short treatment with pan-PIMi ( Figure 6A). Moreover, taking into account the dramatic proapoptotic effect of this drug, two key proteins involved in apoptosis were also examined: we found that the pan-PIMi induced cleavage and activation of Caspase-3 and decreased the levels of BCL2 ( Figure 6B). These observations support and explain the strong apoptosis induced by the pan-PIMi in PTCL.
To understand the molecular response of PTCL cells to the pharmacological pan-PIMi, 4 PTCL cell lines were treated with 10 mM for varying periods, and changes in gene expression over time were examined. We found 390 genes that were differentially expressed (FDR,0.05) and commonly deregulated in all 4 cell lines upon drug treatment ( Figure 6C). On the basis of GO categories we found that the upregulated genes were those involved in the positive regulation of the cell cycle pathway, which could explain the aforementioned absence of cell cycle arrest, and that the downregulated genes were related to the response to DNA damage, repair and replication, which could be added and enhance the strong apoptosis induced by the pan-PIMi ( Figure 6C). A more detailed list of genes and pathways deregulated in each PTCL cell line treated with the pan-PIMi can be found in the Tables S3 and S4, respectively. To validate this result, the expression of several genes involved in DNA damage repair, such as ERCC8, XRCC2 and XRCC5 ( Figure S6A) was measured by RT-qPCR. We found that treatment with the pan-PIMi downregulated the expression of these genes in a time-and dose-dependent manner ( Figure S6B).
To functionally confirm that pharmacological PIM kinase inhibition impaired the DNA damage repair machinery, we measured the amount and distribution of cH2A.X protein, the classical hallmark for DNA damage [26,27], in MyLa cells. After treatment with the pan-PIMi increases in the amount and formation of cH2A.X foci corresponding to DNA damage foci were observed ( Figure 6D).
These results indicated that the pharmacological pan-PIMi strongly induces DNA damage through the downregulation of genes involved in the DNA repair machinery.
Based on this, we hypothesized that the response to the pan-PIMi could be even improved by a DNA damaging agent, such as Cisplatin. Thus, 4 PTCL cell lines were treated with the drug combination pan-PIMi + Cisplatin for 72 h. In all tested cell lines, a synergistic effect between both drugs (Combination Index ,1) was observed ( Figure 6E), highlighting again the functional link between PIM kinases and DNA repair.

Synergism between PIM and ALK inhibition in ALCL
Since ALCL cell lines were the most sensitive to the pan-PIMi ETP-39010, and PIM2 expression was preliminarily associated with poor prognosis in our limited ALCL series, we decided to explore the therapeutic relevance of the PIM pathway in ALCL, especially in the ALK+ ALCL subtype, because ALK translocation is known to activate STAT3 [5,6], and STAT3 triggers PIM2 expression [13,14]. First, we treated 2 ALK+ ALCL cell lines (KARPAS-299 and SU-DHL-1) and 2 ALK2 PTCL cell lines (MyLa and DERL7) with the ALKi Crizotinib ( Figure 7A) and found, as expected, that the ALK+ cells were about 10 times as sensitive as the ALK2 cells ( Figure 7B). Next, we combined this ALKi with the pan-PIMi ETP-39010 to inhibit the same pathway at two sites, which produced a strong synergistic effect between these drugs only in ALK+ cells, but not in ALK2 cells. Strikingly, 24 h of the combined treatment strongly enhanced apoptosis in KARPAS-299 and SU-DHL-1 cells (Combination Index ,1), while a slight additive effect or even antagonism, was found in MyLa and DERL7 cells (Combination Index .1) ( Figure 7C). The effects driven by the pan-PIMi shown here (24 h, 5 mM) were comparable with those in Figure S5A.
Finally, since the selectivity profile of the pan-PIMi ETP-39010 was not very specific, it was important to rule out the possibility that the effects we observed were due to off-target consequences. To this end, we confirmed the most significant results with newly developed and more specific pan-PIMi: ETP-47652, ETP-47551 and ETP-46638. Of these, ETP-47551 was the compound with the best selective profile (Figures 8A and 8B). We treated our panel of 8 PTCL cell lines with these new pan-PIMi, and found that all cell lines showed the highest sensitivity to the ETP-47551, with IC 50 values comparable to those obtained with ETP-39010 ( Figure 8C). Additionally, we observed that the more specific ETP-47551 reduced cell viability ( Figure 8D) and strongly induced apoptosis in PTCL cell lines in a time-dependent manner ( Figure 8E), similarly to the ETP-39010 compound. Lastly, the synergistic effect between the ALKi and the pan-PIMi ETP-39010 in ALK+ ALCL ( Figure 8F) was confirmed using the combination ALKi plus the specific pan-PIMi ETP-47551 ( Figure 8G). These results could help discard potential off-target effects driven by ETP-39010.

Discussion
We hypothesized that PIM kinase inhibition could be of therapeutic value in PTCL because: 1) PIM kinases have an important role in CD4+ T cell responses [28]; 2) PIM1 and especially PIM2 expression is increased in PTCL patients, cell lines and primary tumoral T cells of Sézary Syndrome patients; 3) they are significantly correlated with survival pathways, such as Jak/STAT, NF-kB and IL-2 signaling; and 4) pharmacological PIM inhibition is effective in other T cell-mediated malignancies, such as T cell acute lymphoblastic lymphoma [29].
PIM family members, especially PIM2, were found to be overexpressed in PTCL, as in many other tumor entities of hematological or epithelial origin, such as chronic lymphocytic leukemia, mantle cell lymphoma, diffuse large B cell lymphoma, acute myeloid leukemia, and prostate, pancreatic, gastric, colon and hepatocellular carcinomas [13,15]. This increased expression was found to be significantly correlated with Jak/STAT, NF-kB and IL-2 signalings in our PTCL patient series, suggesting that these pathways could be the responsible for PIM activation and could contribute to PTCL cell survival.
However, in spite of all of the supporting evidence, our results indicate that individual genetic silencing of PIM1, PIM2 or PIM3 genes does not affect PTCL cell survival, either at the level of apoptosis or the cell cycle. We found that this could be due, at least in part, to a compensatory mechanism among the 3 PIM genes, since PIM1 knockdown was accompanied by the upregulation of both PIM2 and PIM3, and vice versa. These redundant functions have also been described in in vivo models: mice lacking PIM1 had a higher level of PIM2 expression, while those deficient for PIM1 and PIM2 selectively activated PIM3 [14,15]. These overlapping functions can be explained by the substantial homology (50-70%) of the PIM kinases at the amino acid level [15,19,20,30].
These observations suggested that in order to effectively treat PTCL an inhibition of all three PIMs would be required, as described for other hematological malignancies, such as multiple myeloma [31] and acute myeloid leukemia [32].
Unexpectedly, the simultaneous genetic inhibition of PIM1+ PIM2+PIM3 did not affect PTCL cell survival either. Nevertheless, it is important to note that the simultaneous use of 3 siRNAs makes them less efficient than when they are on their own; moreover, these cells are not easily transfected/infected, making the genetic inhibition approach less informative. The lack of effect could be due to incomplete PIM family inhibition (indeed, around 50% of each protein remained after triple-knockdown si/shRNA, in our case), or the dispensable effects that have been described for PIM (the triple-knockout mice were still viable and their mainly described phenotypic characteristics was a markedly reduced body size [13,14,15]). To address this possibility, we adopted a pharmacological inhibition approach to abolish all PIM kinase activity. Although it is conceivable that the different ATP-binding region of PIM kinases compared with other kinases would allow specific PIM inhibitors to develop, in practice this specificity does not seem to be reached, especially because they also inhibit FLT3, PDGFR and KIT [15,19,21]. Some pharmacological inhibitors are available that are selective for one of the PIM kinases [21], but consequently they will not avoid the compensatory mechanism among the other PIM kinases. We used the ETP-39010 compound, which is a pan-PIMi with a low specificity profile [21] for all the functional assays. Nevertheless, inhibition of PIM kinases by this drug was assessed, since it reduced 4E-BP1 phosphorylation, which is a very well established PIM kinase substrate [13,14,15,21], and could be a biomarker for PIM kinase inhibition in PTCL. In addition, the most significant effects observed with ETP-39010 were confirmed with a newly developed and much more selective compound (ETP-47551). Among our most striking findings was the potent cytotoxic effect in all PTCL cell lines upon pan-PIMi treatment, at doses similar to those used with other pan-PIMi, such as SGI-1776 in prostate cancer [33], acute myeloid leukemia [19], chronic lymphocytic leukemia [17] and mantle cell lymphoma [18]. Surprisingly, and in contrast to the findings of the great majority of these studies, apoptosis induction was not accompanied by cell cycle arrest in PTCL cell lines. An increase in the subG0 population was observed even at lower doses or with shorter-duration treatments, highlighting the potent efficiency of this pan-PIMi in PTCL, especially in ALCL cell lines. Moreover, this strong induction of apoptosis was in part due to the cleavage of Caspase-3, the decrease in BCL2 protein levels (as extensively described in [13,14,15,32]), and the enhancement of the DNA damage, since we found that the pan-PIMi downregulated the expression of a number of genes involved in DNA damage repair signaling, leading to the formation of cH2A.X foci, the most well established surrogate biomarker for DNA damage [26,27]. Accordingly, there are several lines of evidence involving PIM kinases in the DNA repair machinery [34,35,36,37].
The cytotoxic effect found in pan-PIMi-treated PTCL cell lines was explored ex vivo in primary T cells from cutaneous T cell lymphoma patients. Interestingly, we found that although the effects on neoplastic cells was not very dramatic, normal T cells from healthy donors were not affected by the pan-PIMi, recalling the limited cytotoxicity observed in SGI-1776-treated normal lymphocytes [19]. These evidences could support the proof of concept that the PIM kinase inhibition strategy might be a preliminary safe therapeutic approach. Moreover, it has been reported that SGI-1776 treatment reduces tumor volume without causing significant changes in body weight [19]. These findings, along with the fact that the triple PIM1+PIM2+PIM3 knockout mice had a mild phenotype [13,14], support the rationale of using pharmacological pan-PIMi as safe antitumoral agents.
A large fraction of the PTCL patients showed increased PIM2 protein expression, regardless of their subtype (although with a slight predominance in AITL, where a PIM2 increased expression has been already reported [38]). Importantly, PIM2 protein levels were significantly correlated with a worse outcome in ALCL patients, as described for the majority of malignancies [13]. It is important to note that in our limited ALCL series (n = 27), this association between PIM2 expression and a worse survival was found taking into account both ALK+ and ALK2 ALCL patients. Although more samples are needed if more statistically significant conclusions are to be drawn, since in our series, the well-known prognostic marker ALK expression is almost significantly associated with outcome (p = 0.08).
This preliminary finding, along with the fact that ALCL cell lines are the most sensitive to the pan-PIMi ETP-39010, led us to hypothesize that the ALK -STAT3 -PIM2 pathway could be important for ALCL survival, at least in ALK+ ALCL, since ALK is a well-known STAT3 activator [5,6], and STAT3 has been extensively described to increase PIM2 expression [13,14]. Thus, we aimed to target this axis at two different points using the ALKi Crizotinib plus pan-PIMi (i.e., the less specific ETP-39010 and the more selective ETP-47551). As expected, the simultaneous inhibition of ALK and PIM strongly affected cell survival in ALK+ ALCL but not in other PTCL cell lines, synergizing the apoptosis induced by each drug alone only in ALK+ ALCL cells. These results could highlight the potential therapeutic usefulness of this pathway in ALK+ ALCL.
Although ALK+ ALCL is the PTCL subtype with the most favorable outcome, frequently relapses have been reported in around 30% of patients treated with primary chemotherapy [39]. Some studies have recently described the efficacy of ALK inhibition in ALCL, both in murine models [40] and in preliminary clinical studies: in one trial 2 ALK+ ALCL patients reported complete remission of the disease within 1 month of treatment with Crizotinib, the response being sustained 5-6 months later [8]. A later clinical trial with 9 ALK+ ALCL patients treated with Crizotinib showed an objective response rate of 100%, a complete remission rate of 100%, a median duration of response of 10 months and 3-year progression-free survival of 63% with a plateau in the curve after 6 months [9]. In fact, Crizotinib has been approved by the FDA for the treatment of ALK+ non-small cell lung cancer [3]. Unfortunately, despite its initially impressive efficacy, resistance to Crizotinib has been found in patients carrying mutations in the fused ALK proteins [9,41]. For this reason, it might be worthwhile exploring drug combinations targeting downstream effectors of the oncogenic-driver ALK translocation. Additionally, PIMi are known to synergize strongly with other antitumoral agents, such as Cisplatin, as demonstrated here in PTCL, the MEK inhibitor UO126 in precursor T cell lymphoblastic leukemia [29], the PI3 K inhibitor GDC-0941 in acute myeloid leukemia [21], the HDAC inhibitor SAHA in classical Hodgkin lymphoma [42], Bendamustine in mantle cell lymphoma and splenic marginal zone lymphoma [43], the BCL2 inhibitor ABT-737 [44] and taxanes [45] in prostate cancer and the multi-kinase inhibitor Sunitinib in renal cell carcinoma [46].
In conclusion, our results suggest that the simultaneous inhibition of all PIM kinases could be an efficient therapeutic strategy in those PTCL with PIM upregulation. This strategy seems to be particularly relevant in the ALK+ ALCL subtype, whereby the increased expression of PIM2 is associated with shorter survival and the combinatory inhibition of ALK and all PIM kinases potently enhanced apoptosis.  Table S3. Some important genes, such as ERCC8, XRCC2 and XRCC5 (highlighted by arrows) were randomly selected to be validated. (B) Validation of microarray data by RT-qPCR. The expression of ERCC8, XRCC2 and XRCC5 genes was confirmed to be reduced in a time-and dose-dependent manner after pan-PIMi treatment in MyLa and SR786 cell lines. RQ, relative quantification, was calculated as described in the Methods section as RQ = 2 2DCt . (TIF)