Latexin Is Down-Regulated in Hematopoietic Malignancies and Restoration of Expression Inhibits Lymphoma Growth

Latexin is a negative regulator of hematopoietic stem cell number in mice. Its dysregulated expression in other tumors led us to hypothesize that latexin may have tumor suppressor properties in hematological malignancies. We found that latexin was down-regulated in a variety of leukemia and lymphoma cell lines as well as in CD34+ cells from the blood and marrow of patients with these malignancies. 5-aza-2′-deoxycytodine treatment and bisulfite sequencing revealed hypermethylation of latexin promoter in tumor cells. Retrovirus-mediated latexin overexpression in A20 mouse lymphoma cells inhibited their in vitro growth by 16 fold and in vivo tumor volume by 2 fold. Latexin caused growth inhibition of lymphoma cells by significantly increasing apoptosis through the down-regulation of anti-apoptotic genes Bcl-2 and Pim-2. The molecular mechanism underlying latexin-mediated tumor inhibition was not through its canonical carboxypeptidase inhibitor activity. These results are consistent with a tumor suppressor role for latexin and suggest that latexin may have clinical efficacy in the treatment of malignancies.


Introduction
Both solid and liquid tumors are now understood to originate from the malignant transformation of resident adult stem and progenitor cells. [1,2] Nowhere is this paradigm better established than in leukemia, yet events causing neoplastic conversion remain poorly understood. [3] We identified latexin (Lxn) as a novel, homeostatic regulator of the size of the hematopoietic stem cell population in mice. [4] The stem cell pool size was inversely related to quantitative Lxn expression at both the levels of transcript and protein. We found that population size was influenced by Lxn in a stem cell-autonomous manner, and acting through the concerted mechanisms of self-renewal and apoptosis, which were decreased and increased, respectively, by Lxn abundance. [4] Thus, Lxn acts as a brake on the expansion of stem cell population. Unrestrained stem cell expansion carries with it the risk of mutations, genomic instability and carcinogenesis. We therefore hypothesized that Lxn expression patterns in stem and progenitor cells may act as a tumor suppressor by inhibiting stem and progenitor cell proliferation, influencing the crucial steps toward malignancy.
Lxn was primarily studied in the nervous system, involving the specification of cortical brain regions during development [5,6,7,8,9], as well as the speed of nerve transmission in adult peripheral nervous systems [10]. Lxn was found to be involved in the inflammatory response in macrophages owing to its paired cystatin-like domains [11]. It was detected in mast cells associated with a unique type of intracellular granule distinct from histaminecontaining secretory granules and lysosomes [12]. Lxn has about 30% sequence homology, but much greater structural homology, with tazarotene-induced gene 1 (TIG1) (or retinoic acid receptor responder 1, RARRES1), whose expression was down-regulated in an extensive list of tumor types in humans. [11,13,14,15,16]. Lxn and TIG1 are closely linked genetically and may represent members of a family of functionally related genes. Moreover, Lxn was reported to be a TNF-responsive gene in human papillovirus-infected keratinocytes, suggesting that it may contribute to TNF-mediated suppression of cervical cancer development [17]. A recent report revealed that Lxn was down-regulated in patients with gastric carcinomas, and overexpression or knockdown of Lxn inhibited or stimulated tumor growth respectively [18]. Decreased or absent Lxn expression was observed in several human leukemia and lymphoma cell lines [19], as well as in malignant melanoma patients. [20] However, there is still lack of direct evidence for the cause-effect relationship between Lxn and hematopoietic malignancy.
Gene silencing, especially of tumor suppressors, and inappropriate gene activation, especially of oncogenes, are common themes in carcinogenesis. [21] Both often occur through aberrant DNA methylation that is accentuated during aging. [22,23,24,25,26,27] A CpG-enriched region was identified in Lxn promoter and its hypermethylation was found in a variety of human gastric carcinoma cell lines. [18] Promoter hypermethylation of TIG1 was also associated with its silencing in tumor cells. [13,16] However, it is not known whether Lxn expression is regulated by promoter methylation in hematopoietic malignancy.
Molecular mechanisms underlying the role of Lxn in regulating hematopoiesis has not been determined yet. Lxn protein binds to carboxypeptidase A (Cpa) and inhibits its enzymatic activity, indicating that Lxn regulates protein degradation and metabolism. [19] Lxn was highly expressed in mast cells and its expression was further enhanced by LPS, indicating its potential function in inflammation. [11] Proteomic analysis on Lxn knock-out hematopoietic cells revealed that Lxn deletion reduced the abundance of multiple cellular proteins, especially those involving cell-stroma interaction, such as N-cadherin, Tie2, and Roundabout 4. [19] Ectopic expression of Lxn in gastric cancer cell lines led to the differential expressions of several cancer-related genes, including Maspin, WFDC1, SLPI, S100P, and PDGEFB [18], although none of them are overlapped with previously mentioned proteomic results. A recent study revealed ATP/GTP binding protein-like 2 (AGBL2) as a novel binding partner of TIG1. The interaction between TIG1 and AGBL2 regulates the microtubule tyrosination cycle, which is implicated in tumorigenesis, stem cell differentiation and development. [28] Lxn thus may have the similar mechanisms as TIG1 to regulate hematopoiesis.
Here, we provide evidence obtained from a variety of lymphoma and leukemia cell lines, as well as from primary cells from patients with these diseases, that Lxn expression is almost universally absent or significantly reduced from that of normal stem and progenitor cells. Moreover, treatment with a demethylating agent at least partially restores Lxn expression in a variety of tumor cell lines, and methylation level of the CpG island in the Lxn promoter region is inversely associated with its expression. Perhaps most importantly, we show that when Lxn expression was re-initiated ectopically in two lymphoma cell lines using a retroviral expression vector, their growth, both in vitro and in vivo, was significantly blunted. We further show that Lxn inhibits lymphoma cell growth by significantly increasing apoptosis through the down-regulation of anti-apoptotic genes and that its anti-tumor activity is mediated via mechanisms unique from its canonical inhibition of carboxypeptidase A. These results demonstrate that Lxn plays a functional role in tumor cell growth and introduces an unexplored pathway potentially important to cancer treatment in patients.

Loss of Lxn Expression in Malignant Cells
We first determined Lxn mRNA abundance in tumor cell lines by quantitative real-time PCR (Fig. 1A). Compared with normal primitive hematopoietic cells, Lxn mRNA expression was completely absent in a majority of tested leukemic lines, including K562, Molt4, CRF-CEM, J45.01, Jurkat and U937, and was significantly diminished in HL-60, KG-1 and SupB15 cell lines.
LXN protein expression was also tested in all samples using Western blotting and nearly identical results were obtained in tested cell lines (Fig. 1B). Our results are consistent with those previously reported [19] and include additional leukemia and lymphoma cell lines.

Aberrant Promoter Hypermethylation of Lxn in Hematopoietic Malignancy
To assess whether the loss or decrease of Lxn expression in malignant cell lines resulted from promoter hypermethylation, methylation of the 59 CpG island of Lxn gene surrounding its transcriptional start site was determined by genomic bisulfite sequencing. The CpG island spans from within the canonical 59 promoter (2208 nt) to the transcription start site (+1 nt), and extends through the entire first exon (+44 nt). There are 15 CpG dinucleotides within this region ( Fig. 2A). The CpG island identified here is not exactly same as the one reported previously, this may be due to the different criteria we used for defining this region. Figure 2B depicts the quantitative variation in methylation for each CpG site among nine tumor cell lines. Almost complete methylation was seen in J45.01, U937, Jurkat, Molt4 and CCRF-CEM (.90%) lines, which commensurately showed an absence of Lxn expression. Scatteredly methylated CpG sites were found in K562, KG-1 and SupB15 lines, which were linked to the weak expression of Lxn. Surprisingly, although HL-60 had very low Lxn expression, we found that none of the CpG sites were methylated. This is probably due to other epigenetic mechanisms, such as histone deacetylation of latexin promoter in HL 60 cells.

Reactivation of Lxn Expression with Demethylating Reagent Treatment
To test the hypothesis that Lxn promoter hypermethylation might be involved in the loss of expression in leukemia and lymphoma cell lines, we studied the effect of 5-aza-29-deoxycytidine, a DNA demethylating reagent, on Lxn expression (Fig. 2C). After treatment with 2 uM 5-aza-29-deoxycytidine for 4 days, Lxn gene expression was reactivated in cell lines completely lacking Lxn expression prior to treatment (K562, Molt4, CCRF-CEM, J45.01, Jurkat and U937), and was significantly up-regulated in HL-60, KG-1 and SupB15 lines. The up-regulation of Lxn by 5-aza-29deoxycytidine in HL60 cells, which shows no hypermethylation of CpG sties (Fig. 2B), might be because the demethylation of other DNA sequences induces chromatin remodeling, thus exposing Lxn promoter for the transcriptional initiation.

Suppression of Growth of Mouse Lymphoma Cell Lines in vitro and in vivo Following Ectopic Lxn Expression
To determine a correlative or causative relationship between Lxn expression and tumor development, we next asked whether or not the re-initiation of Lxn expression affected the growth rate of malignant cells in vitro and in vivo. To that end, we ectopically expressed Lxn, using a retroviral expression vector containing green fluorescent protein (GFP) marker, in the mouse BALB/cderived A20 B lymphoma cell line, which lacks Lxn expression (Fig. 3C). The controls are A20 cells either un-infected or infected with empty vector. The GFP positive A20 cells were purified by flow cytometric cell sorting, and their growth was determined in vitro and in vivo. Fig. 3 shows the potent effects of Lxn expression on inhibiting tumor growth in vitro and in vivo. Fig. 3A shows that cultures of A20 cells infected with the Lxn expression vector contained only about half the number of cells at day 3 as compared to A20 cells infected with the control (empty) vector or uninfected control cells (A20 control). The growth suppression by Lxn overexpression was exponentially amplified during subsequent days of culture and by day 20 nearly 16 fold less tumor cells were present in Lxnoverexpressing group than in control cultures (right panel). Fig. 3B shows that the fraction of GFP+ cells remained at 90-100% throughout the 20 days of culture in both the Lxn vector-or control vector-infected cells. Fig. 3C shows that at day 20 of culture neither uninfected A20 cells nor A20 cells infected with the control (empty) vector expressed detectable Lxn protein, whereas in the Western blot a strong Lxn band was evident in lysate of cells infected with the Lxn vector (upper band). LXN protein level at day 0 of culture in these cells is nearly identical to that at day 20 (data not shown). In a separate series of experiments, we have found a similar growth reduction in vitro when Lxn was ectopically expressed in WEHI231 lymphoma cells (data not shown). Fig. 3D shows a compilation of two independent experiments in which ectopic Lxn expression suppressed the growth of A20 cell in vivo. 100,000 GFP+ A20 cells infected with either the Lxn expression vector or the GFP only control vector, or uninfected A20 cells, were injected subcutaneously into the flanks of BALB/c mice. Beginning at day 12 when tumors were first palpable, the Lxn-expressing cells caused significantly smaller tumors (filled circles; P,0.005). By day 21 when the experiments were terminated, the tumors in the Lxn vector-injected group averaged only 40% of the volume of tumors in the other two control groups (P,0.05, n = 12 in each group). At day 21, we determined the fraction of tumor cells expressing GFP and Fig. 3F shows that virtually all of the tumor cells in the Lxn and control vector groups were GFP+. More to the point, Western blots confirmed strong expression of Lxn as in the cells analyzed after 20 days in culture (Fig. 3C). Thus, the reduction in tumor growth was due to durable Lxn expression in the tumor cells themselves.
To explore the effects of cell dose on tumor size and latency before being palpable, we ejected graded doses of Lxn-expressing or control A20 cells (Fig. 3E) to BALB/c recipient (n = 4/group). At the 5,000 and 25,000 inoculum sizes, the Lxn-expressing cells not only resulted in more impressive suppression of the size of tumors than with the 100,000 cell inoculum in Fig. 3D, but resulted in delayed onset of measurable-sized tumors. At day 21, the reduction in tumor size caused by ectopic Lxn expression was 83% and 63% at the 5,000 and 25,000 cell doses, respectively. Host animals were necropsied for evidence of gross metastases to spleen, thymus and liver. No evident tumors were found in any of the treatment groups. Similarly, flow cytometry detected no GFP+ cells in these anatomical sites (data not shown).

Lxn Inhibits Tumor Cell Growth through Increasing Apoptosis but not via its Canonical Function
As we previously reported in our normal hematopoietic stem cell studies, high Lxn expression is associated with increased apoptosis and decreased proliferation. To determine if these mechanisms were applied to our results with tumor cells, we next measured these two parameters in vector-infected A20 cells throughout 21 days of culture. As shown in Fig. 4A, flow cytometric analysis of cells stained with BrdU and 7AAD allows for the discrimination of cell subsets that are apoptotic (A), necrotic (N) or reside in G0/G1, S and G2/M phase of cell cycle. By day 5 of culture, 10 fold more Lxn-expressing tumor cells were undergoing apoptosis than control cells (10.6% vs 1%). NO significant differences in the proliferation and the numbers of necrotic cells were observed between Lxn-overexpressing and control cells (data not shown). These results point to apoptosis as the major cellular mechanism in Lxn-mediated tumor suppression. We next investigated the molecular mechanisms underlying Lxninduced apoptosis. We performed apoptosis pathway-specific PCR arrays in A20 cells infected with either empty or Lxn-expressing vectors. We found that 17 out of 84 apoptosis-related genes were differentially expressed in Lxn-overexpressing cells, and majority of them (15 genes) were down-regulated (data not shown). We specifically examined several well-known apoptotic genes, such as Bcl-2, Pim-2, Bcl-xl, Bax, and Bad, and plotted their expression in Fig. 4B. Two anti-apoptotic genes, Bcl-2 and Pim-2, were downregulated by at least 3-fold in Lxn-overexpressing cells (left panel) as compared to the control, whereas pro-apoptotic genes, including Bax and Bad, did not show any change in their mRNA levels. Western blots of these proteins observed the similar reduction in Bcl-2 and Pim-2 (all three isoforms) but not in Bclxl, Bax and Bad in Lxn-overexpressing A20 cells (right panel),  which is consistent with PCR array result. Thus our data indicate that Lxn-induced apoptosis in A20 cells is mainly through the down-regulation of anti-apoptotic regulators, such as Bcl-2 and Pim-2.
A well-known function of Lxn is its role as the sole carboxypeptidase A (CPA) inhibitor in mammalian cells. There are six Cpa genes (1)(2)(3)(4)(5)(6) in this family. In order to test whether or not the suppressive effect of ectopic Lxn expression on the growth of A20 and WEHI231 cells was due to its canonical inhibitory activity, we carried out the following experiments. We first determined which Cpa genes are enriched in hematopoietic stem/progenitor cells. By searching our gene expression profile of hematopoietic stem/ progenitor cells (unpublished data), we found that Cpa3 is the only gene highly expressed in hematopoietic cells (Table S1). Next, we tested Cpa3 expression in A20 cells that were uninfected, or infected with empty or Lxn expression vectors, and found that it was not expressed in any type of these cells ( Figure S1). Last, we treated A20 and WEHI cells with potato carboxypeptidase A inhibitor (PCPI), a 39 amino acid protein which strongly inhibits mammalian CPA, at a series of concentrations ranging from 5 to 60 ug/ml of culture medium. As seen in Fig. 4C, none of the concentrations had any effect on the growth patterns of the tumor lines, despite the continuous presence of PCPI for 12 days of culture. To rule out the possibility that PCPI failed to inhibit tumor growth because it did not enter the cells, we used fluorescein isothyiocyanate (FITC) to label PCPI in cells. Fig. 4D shows A20 cells following a 30 min incubation with labeled PCPI (30 ug/ml of culture medium). The three-color micrograph shows PCPI in green (FITC), the nucleus in blue (DAPI), and the B220 lymphoid cell surface marker in red (phycoerythrin, PE). It is apparent that PCPI is plentiful in the cytosol but is not found in the nucleus. The doses of PCPI chosen for the above experiments were taken directly from a study in which the inhibitor was shown to inhibit the growth of pancreatic adenocarcinoma cells by directly interfering with the epidermal growth factor signaling pathway [29]. Maximal growth inhibition was achieved at 30-50 ug/ml. Thus, the mechanisms by which Lxn regulates both stem cell population size and lymphoma growth inhibition reside in a novel pathway involving apoptosis.

Down-regulation of Lxn in Primary Leukemia and Lymphoma Cells
Based on the strong evidence for the tumor suppressive function of Lxn in leukemia and lymphoma cell lines, we asked whether or not Lxn expression is altered in primary cells derived from patients with malignancies. We isolated stem/progenitor-enriched CD34+ cells from bone marrow and blood cells in lymphoma and leukemia patients, as well as in normal donors. The patients included those with acute myeloid leukemia (AML), T cell prolymphocytic leukemia (T-PLL), plasma cell leukemia (PCL), acute T cell lymphoma (ATLL) and acute lymphoid leukemia (ALL, preB phenotype). The normal CD34+ cell samples were derived from cord blood (CB) and young (31 and 35 years) and old (85 and 97 years) adults. Lxn mRNA expression, quantified by real-time PCR, was decreased by at least two-thirds in primary malignant CD34+ cells (Fig. 5A). Quantification of LXN in CD34+ cells of all human normal and leukemic samples compiled to date is plotted in Fig. 5B. The result shows a significant decrease of LXN protein expression in malignant cells (P = 0.03) even though Lxn expression at the protein level is more variable than that at the mRNA level. Two additional old samples in the normal control group contribute to this variation, which is consistent with the report that the number and functionality of HSCs in old individuals, perhaps stem cell regulatory gene(s), showed more dramatic variations as compared to their young counterparts [30]. In summary, these data indicate that Lxn dysregulation may be involved in human hematological malignancies. In the future study, it will be interesting to examine the expression of two antiapoptotic genes, Bcl-2 and Pim-2, in patient samples to see whether or not they are up-regulated.

Discussion
Recently we found that Lxn is a natural regulator of hematopoietic stem cell population in mice by influencing selfrenewal and apoptosis [4]. Lxn expression in stem cells is inversely proportional to the size of the population in mice and thus acts as a negative regulator through cell-intrinsic mechanisms. The observation that low Lxn expression in hematopoietic cells was associated with increased replication led us to hypothesize that Lxn was down-regulated in malignancies with high proliferative rates and/or low apoptosis.
Here, we show that Lxn is either not expressed or is strongly down-regulated in a variety of leukemias and lymphomas. As a result of these findings and close linkage structurally and genetically with TIG1, we proposed that Lxn may similarly act as a tumor suppressor. Along with this proposal, there is expectation that manipulation of Lxn levels in malignant cells would alter their growth characteristics. Accordingly, two lymphoma cell lines in which Lxn expression was absent were infected with a Lxn expression vector. In one of the most significant findings of this study, we found that re-initiation of Lxn expression dramatically reduced their growth in vitro and in vivo. These results demonstrate that the level of Lxn expression is functionally related to both normal and malignant cell growth and, if the findings are true for human tumors, may provide an avenue for therapeutic intervention. Surprisingly, Lxn knock-out mice did not development leukemia nor lymphoma even though their hematopoietic stem cell pool was over-expanded and blast cells were detected in the peripheral blood [19]. These results suggest that Lxn may influence the crucial, step in carcinogenesis [31]. It thus will be of great interest to introduce additional carcinogenetic ''hits'', such as oncogenes or alkylating agents, to Lxn knockout stem cells to determine whether Lxn deletion will accelerate tumor development. In addition, we observed that re-initiation of Lxn expression did not completely suppress tumor cell growth, suggesting that it would at best be a part of a multi-drug regimen.
We investigated the possible mechanisms by which ectopic Lxn expression may suppress growth of lymphoma cells and found increased apoptosis to be the major mechanism causing inhibition. Consistent with our study of normal hematopoietic cells, our recent findings indicate that high Lxn content is associated with more apoptotic cells, thus fewer stem cells [4]. Apoptosis is controlled by the balance of both positive (pro-apoptotic) and negative (anti-apoptotic) signals. We found that Lxn-induced apoptosis is mainly through the down-regulation of anti-apoptotic genes, such as Bcl-2 and Pim-2. Decreased level of Bcl-2 allows other pro-apoptotic proteins such as BAX and BAK to aggregate, regulation of Lxn expression by 5-aza-29-deoxycytidine. Lxn mRNA was measured by quantitative real-time PCR in leukemic and lymphoma cell lines exposed with or without 2 uM 5-aza-29-deoxycytidine for 4 days. Expression levels were normalized to endogenous control, Gapdh. Results shown are mean6 SD of 12 replicates from 3 independent experiments. doi:10.1371/journal.pone.0044979.g002 inducing cytochrome C release and caspase activation, and apoptosis. Dysregulation of Bcl-2 is implicated in a variety of hematopoietic malignancies. [32] Three Pim-2 protein isoforms are active kinases which phosphorylate pro-apoptotic protein BAD and dissociate its binding from Bcl-xL, resulting in the inhibition of Bad-induced cell death. [33,34] Enhanced expression of Pim-2 was detected in multiple B-cell lymphoma types [35], T-cell lymphoblastic leukemia/lymphoma [36], and acute myeloid leukemia [37]. Thus our studies identify Lxn as a novel regulator of the apoptotic pathway, in which it suppresses tumor survival through the down-regulation of Bcl-2 and Pim-2 anti-apoptotic factors.
The well-known function of Lxn is the sole carboxypeptidase A inhibitor in mammals. [12] Therefore, it was plausible that other carboxypeptidase A inhibitors might mimic Lxn's anti-tumor effects. A CPA inhibitor from potato tubers has been shown to be an effective inhibitor of mammalian CPA and was therefore chosen to test its anti-tumor properties. We found that it freely entered A20 lymphoma cells but did not have suppressive growth effects via apoptosis, irrespective of dose. Since CPAs are not expressed in A20 cells, we conclude that non-canonical functions of Lxn were responsible, yet unknown intracellular functions of Lxn account for its anti-tumor effects. In addition, it is unlikely that tumor suppressive effect of Lxn is through the regulation of ABGL2-mediated microtubule tyrosination cycle, like TIG1, because we did not detect the expression of ABGL2 in hematopoietic cells (data not shown). These results are in line with findings of other investigations in which gene expression or protein abundance regulated by Lxn were not related to carboxypeptidase inhibition [18,19]. Thus, the action of mode of Lxn in regulating normal and malignant hematopoiesis is not through its canonical CPA inhibitor activity. Future studies aimed at identifying direct targets, or downstream signaling pathways, of Lxn will be critical for our understanding of its function in normal and malignant hematopoiesis.
Addressing the issue of why and how Lxn expression is downregulated in tumor cells, we found that CpG dinucleotides in regulatory regions of the Lxn gene were methylated. Most human leukemia and lymphoma cell lines tested showed strong patterns of methylation involving most of the 15 CpG dinucleotides in Lxn promoter. This, in turn, led to a loss of gene expression that was at least partially reversible. When cell lines were treated with demethylating reagent, 5-aza-29-deoxycytidine, Lxn expression was re-initiated or up-regulated. Thus, hypermethylation of the CpG island in the Lxn promoter may contribute to silencing or downregulation of Lxn expression in leukemia and lymphoma cells. Ongoing studies will focus on the genetic and epigenetic regulation of Lxn transcription in cancer stem cells, with promise for improved targeted approaches for cancer prevention, diagnosis, and therapy.

Animals
Young 8-to 12-week old female C57BL/6 (B6) and 7-week old female BALB/c mice were purchased from the Jackson Laboratories (Bar Harbor, ME). Mice were kept in the animal facilities of the University of Kentucky under pathogen-free conditions according to NIH-mandated guidelines for animal welfare. They were fed with acidified water and food ad libitum. All animal work in this study was approved by Institutional Animal Care and Use committee (IACUC) at the University of Kentucky (Identification number: 2010-0753).

Isolation of CD34+ Cells
Primary AML cells were obtained from the peripheral blood and bone marrow of patients at the Markey Cancer Center and Northwestern University. Normal bone marrow was obtained as discarded material following pathologic analysis, surgical marrow harvest, or from the National Disease Research Interchange (NDRI). CB was obtained from patients at the University of Kentucky Obstetrics Department or from the NDRI. All tissues were obtained with the approval of the respective institutional review boards and appropriate informed consent (confirmation number is 11-0315-F3R from University of Kentucky). Some samples were collected form Northwestern University and transferred to University of Illinois at Urbana-Champaign with approval of Material Transfer Agreement (MTA). We processed 9 normal and 9 leukemia and lymphoma samples, the age and disease type of each sample was described in the figure 5 legends. Marrow and blood cells were depleted of erythrocytes by suspending in 150 mM NH4Cl plus 10 mM NaHCO3 for 5 minutes, followed by 2 washes with phosphate-buffered saline (PBS). Blood cells were subjected to Ficoll-Paque (Pharmacia Biotech, Piscataway, NJ) density gradient separation to isolate In vitro growth inhibitory effects of Lxn overexpression on A20 cells. A20 cells uninfected (A20 control) or infected with empty (vector control) or Lxn-containing vector (Lxn vector) were cultured for 20 days and counted by the trypan blue exclusion method. The absolute cell numbers at indicated time-points (X axis) were represented as log scale (Y axis) and were shown in left panel. The relative tumor growth in Lxn-overexpression A20 cells to controls was shown in the right panel. One representative experiment out of 3 independent ones is shown. (B) Proportion of GFP+ A20 cells infected with empty (light grey column) and Lxn-containing vector (black column) throughout 20 days of culture. Flow cytometric analysis was performed to detect GFP signal. (C) Western blot analysis with LXN antibody in total protein lysate from A20 cells at day 20 culture showing durable LXN overexpression. (D) In vivo tumor formation and growth with injection of 100,000 A20 cells. A20 cells were subcutaneously injected and tumor growth was monitored and measured by all three dimensions of tumors (Y axis) on days indicated on X axis. Shown is the compilation of 2 independent experiments with 6 SD (n = 12). The statistical significance is represented by * (p,0.005) and ** (p,0.05). (E) In vivo tumor formation and growth with injection of graded doses of A20 cells. Graded doses of A20 cells (X axis) were subcutaneously injected and tumor growth was monitored and measured in all three tumor dimensions (Y axis) on days indicated on Z axis. Unfilled columns represent A20 cells infected with empty vector and filled columns represent cells with Lxn-containing vector. A representative experiment is shown (n = 4/group). (F) The fraction of GFP-expressing A20 cells in tumors biopsied at 21 days. Single tumor cell suspensions were made on day 21 post injection and subject to flow cytometric analysis for GFP-expressing cells. doi:10.1371/journal.pone.0044979.g003 the mononuclear white blood cell compartment. Resulting leukocytes from marrow or blood were then used for immunoaffinity selection and flow cytometric sorting. For CD34+ cell selection, the Miltenyi immunoaffinity device (VarioMACS) was used according to the manufacturer's instructions (Miltenyi Biotech, Auburn, CA), and further purified by immune-staining with anti-CD34 antibodies (Pharmingen, San Diego, CA) and sorting in a triple-laser FACSVantage flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, CA). In some cases, leukocytes were cryopreserved at a concentration of 5610 7 cells/mL in freezing medium consisting of Iscoves modified Dulbecco medium (IMDM), 40% fetal bovine serum (FBS), and 10% dimethyl sulfoxide (DMSO).

Quantitative Real-time PCR
To measure the expression of Lxn in leukemic cells, quantitative real-time PCR was performed. Identical numbers (200,000) of cells were used for total RNA extraction using RNeasy Mini kit (QIAGEN, Valencia, CA) according to the manufacture's instruction. Isolated total RNA was reverse transcribed into cDNA using random hexamers in a TaqManH reverse transcription solution (PN N8080234) and stored at 220uC. In real-time PCR reactions, primer and probe mix for LXN (human and mouse) were purchased from Applied Biosystems (Foster city, CA, USA). TaqManH human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was served as an endogenous control to normalize LXN expression. PCR reactions were set up according to manufacturer's instructions using TaqManH universal PCR master mix (PN 4304437). Analyses of gene expression were performed in Figure 4. Ectopic Lxn expression increases apoptosis of A20 cells. (A) Flow cytometric analysis of cell cycle and apoptosis in cultured A20 cells. A20 cells without (A20 control) or with empty (vector control) or Lxn-containing vector (Lxn vector) were cultured for 21 days. BrdU and 7AAD staining and flow cytometric analysis was used for detecting different phases of the cell cycle (G0/G1, S, G2/M) and apoptotic (A) and necrotic (N) population. A representative flow cytometric file from Day 5 of culture is shown. (B): Down-regulation of anti-apoptotic genes by Lxn overexpression. Apoptosis PCR arrays were performed on A20 cells infected with empty or Lxn-expressing vectors. Left panel shows the fold changes in the mRNA expression of several selected genes (as indicated in X axis) in Lxn-overexpressing cells compared to the control. Western blot confirms the decreased expression of Bcl-2 and two isoforms of Pim-2 at the protein level in Lxn-overexpressing cells (arrowhead). The other apoptosis-related genes, such as Bcl-xl, Bax and Bad, did not show significant difference in mRNA and protein expression. (C) The growth curve of A20 cells treated with graded doses of potato carboxypeptidase inhibitor (PCPI). No significant difference is detected between the control and any of the doses of PCPI. (D) Internalization of PCPI to cytosol of A20 cells. A20 cells were cultured with fluorescein isothiocyanate (FITC) labeled PCPI. Shown is a three-color micrograph with PCPI in green (FITC), the B220 lymphoid cell surface marker in red (phycoerythrin, PE), the nucleus in blue (DAPI), and merged image of red and green as indicated. doi:10.1371/journal.pone.0044979.g004 single reporter assays in an ABI PRISM 7700 sequence detection system (PE Biosystems, Foster city, CA, USA). RT 2 Profiler Apoptosis PCR Array (PAMM-012Z, QIAGEN) was performed according to the manufacture's instructions. cDNAs was prepared from A20 cells infected with empty or lxn expression vectors with RT 2 First Strand cDNA Kit (C-03, QIAGEN) and added to RT 2 qPCR master mix (PA-012). The mixture was aliquoted across the PCR array which contains 84 apoptosis-related genes. Gene expression and quantification was performed in an ABI PRISM 7700 sequence detection system, and data was analyzed with using web-based software (http:// pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php).
The genes with more than 2-fold change in expression and showing statistically significant (p,0.05) were chosen for candidate gene.

Western Blots
Cell samples were lysed at a concentration of 2610 7 cells/ml in a protein lysis buffer containing: 10 mM Tris pH 7.5, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 5 mM ZnCl2 and 1% Triton X-100, 2.8 ug/ml aprotinin (Sigma-Aldrich; St. Louis, MO), 1 mM phenylmethylsulfonyl fluoride (Sigma), 1 mM sodium vanadate (Na3VO4) 1 ug/ml pepstatin, and 1 mg/ml leupeptin (Oncogene Research, MA, USA). Lysate was incubated on ice for 30 min, and then centrifuged at 15,0006g for 10 minutes to remove debris. The resulting supernatant was then aliquoted and stored at 280uC. For Western blot, protein lysates were thawed and mixed with running buffer and a reducing agent (Novex, San Diego, CA, USA, per manufacturer's instructions) and heated at 95uC for 5 minutes. Samples were then analyzed by denaturing PAGE (Novex, 10% bis-Tris gel) using the equivalent of 4610 5 cells per lane. Following electrophoresis, samples were electro-transferred onto immunobilon-P membranes (Millipore, Bedford, MA, USA), which were subsequently blocked and probed with polyclonal rabbit anti-LXN Ig-G antibody at a 1:3000 dilution. This antibody was generated from the Lxn-specific amino acid sequence CKHNSRLPKEG-QAE at the carboxyl terminus, and was produced by Bethyl Laboratories, Inc. (Montgomery, TX). The antibody for detection of human LXN was purchased from Abcam Inc. (Cambridge, MA) and used at 1:2000 dilution. The antibodies for apoptotic proteins, Bcl-2, Bcl-xl, Bax, Bad were provide by Dr. Bondada. The antibody for Pim-2 was purchased from eBioscience (San Diego, CA). Primary antibodies were detected using alkaline phosphatase-conjugated secondary antibodies (Santa Cruz Biotechnology) and electro-chemifluorescent (ECF) reagent (Pharmacia Biotech) according to the manufacturer's instructions. Blots were visualized using a Molecular Dynamics STORM 860 system and Imagequant Software. Following the detection and quantification of anti-LXN antibody, immunobilon-P membrane was sequentially stripped in 40% methanol and the buffer containing 100 mm ß-mercaptoethanol, 2% sodium dodecyl sulfate and 62.4 mM Tris-HCl to remove ECF reaction product and antibodies, respectively. The stripped membrane was re-probed with anti-actin antibody (Sigma) at 1:500,000 dilution and detected as described previously.

Genomic Bisulfite Sequencing
To investigate the methylation pattern of Lxn promoter, CpG island analysis in the upstream sequence of Lxn open reading frame. Nucleotide sequence of Lxn in upstream region (-1000 bp) and the first 3 exons (+373 bp) was obtained from Ensembl database (www.ensembl.org) with ID number ENSG00000079257. CpG island search using CpG island searcher website (http://www.uscnorris.com/cpgislands2/cpg. aspx) showed a 252 bp region (-208 bp to +44 bp) in upstream of Lxn sequence enriched for CpG repeats. The criteria of 5 CpG island is: GC content .50%, ratio of CpG to GpC .0.6 and 200 bp of minimum length. Genomic DNAs were isolated using AquaPure Genomic DNA kit (Bio-Rad, Hercules, CA) and modified by sodium bisulfite using EpiTectH Bisulfite kit (QIA-GEN, Valencia, CA). For the Lxn promoter methylation study, we designed primers that could amplify a 423 bp fragment in the upstream region of Lxn containing CpG island. The forward primer sequence is 59 GTTGGTGTTTGATAAGTATGTGG 39, and the reverse primer sequence is 59 TTTAACCTTCTA-CACCTCAAACAC 39. The annealing temperatures for primers were 52uC for 2 minutes. Hot-start PCR with a total cycle number of 30 was used in all PCR amplifications. Denaturation and extension cycles were maintained for 95uC, 30 seconds and 72uC, 1 minutes respectively. The amplified fragments were cloned into the pCR2.1-TOPO vector using TOPO TA Cloning Kit (Invitrogen Carlsbad, CA) and sequenced (MWG Technology) (n$ 3 clones for cell line and n$ 8 clones for primary cells).

5-aza-29-deoxycytidine Treatment
To examine the correlation of promoter hypermethylation and Lxn gene expression, the leukemia cell line, shown to have a lack or decrease of Lxn expression, was subject to 5-aza-29-deoxycytidine treatment. Cells were plated with 2 uM 5-aza-29-deoxycytidine (Sigma-Aldrich; St. Louis, MO) and incubated for 4 days. The medium and the drug were replaced every 24 hours (hrs.) and cells were harvested for RNA and DNA extraction 4 days after treatment.

Infection of WEHI231 and A20 Cells with Lxn Expression Vector
Cloning of the mouse Lxn gene into Sfbeta 91 retroviral vector and production of viral supernatant were performed exactly as described previously [4]. WEHI-231 and A20 cells were infected by 10 ml viral supernatant at a density of 1610 6 cells per 10 cm plate along with 4 mg/ml of polybrene for 48 hours. The infected cells (GFP+ cells) were sorted and expanded in culture medium. The expanded GFP+ population, if not used immediately, was cryopreserved at a concentration of 1610 7 cells/mL in freezing medium consisting of 80% fetal bovine serum, and 20% dimethyl sulfoxide (DMSO).

Measurement of Growth of Retrovirally-transduced Tumor Cells
Sorted GFP+ A20 cells over expressing Lxn or Sfbeta 91 empty vector were counted on a hemacytometer using trypan blue dye exclusion and 500,000 cells were seeded into 25 cm 2 tissue culture flask in 4 mls media. Cells were incubated in a humidified atmosphere of 5% CO2 in air at 37uC and subsequently counted on days 3, 8, 12, 16 and 20. At each time point, cells were split and maintained at a concentration of 500,000 cells per 4 ml media. The cumulative cell number was calculated from the cell counts and the dilutions made at each culture split. FACS analysis was also performed at each time-point to measure the percentage of GFP+ cells. For the in vivo measurement of tumor cell growth, various numbers (5,000; 25,000; 50,000 and 100,000) of GFP+ A20 cells over expressing Lxn or Sfbeta 91 empty vector were injected in a 50 ml bolus subcutaneously in the shaved flank of BALB/cJ mice given 3.0 Gy of gamma radiation 4 hrs. prior. Lymphomas were detectable by palpation 10-12 days postinjection and all three dimensions of tumors were measured blind with calipers on days 12, 14, 16, 19, and 21. The same individual made the measurements from day-to-day without knowing the treatment regimen the mice received. At day 21, host mice were euthanized, the lymphomas were excised, and single cell suspensions were made of each to determine the fraction of tumor cells expressing GFP.

Cell Cycle and Apoptotic Analysis
The culture of GFP+ A20 cells was maintained as described above. At each time-point, cell cycle analysis was measured by BrdU labeling using BrdU Flow Kit (Pharmingen, San Diego, CA) according to the manufacturer's instruction. 10 ul of BrdU solution (1 mM) was added to 1610 6 cells in 1 ml culture medium and incubated for 1 hour. The cells were fixed and permeabilized by Cytofix/Cytoperm Buffer and treated with 30 ug DNase for 1 hour at 37uC. After washing with Perm/Wash buffer, cells were stained with PE-conjugated anti-BrdU antibody for 20 minutes at room temperature, washed and 20 ul of 7-AAD was added. The cells were analyzed by flow cytometry on Facscan (Becton Dickinson Immunocytometry Systems, San Jose, CA).

Immunohistochemstry of A20 Cells Treated with Potato Carboxypeptidase Inhibitor (PCPI)
PCI was purchased from Sigma-Aldrich Co. (St. Louis, MO). FITC labeling of PCI was performed by using FLUOROTAG TM FITC CONJUGATION KIT (Sigma-Aldrich) according to manufacturer's instruction, and was used to treat A20 cells for fluorescence internalization assays. Cells are cultured on 22622 mm microscope cover glasses (Fisher Scientific Co., Pittsburgh, PA) in media as described above. Cells were fixed onto cover glasses with 1:1 methanol:acetic acid and washed 3 times by PBS. Cells were incubated with FITC-conjugated PCI at a concentration of 30 ug/ml at 37uC for 30 minutes, washed and stained with phycoerythrin (PE)-conjugated B220 and DAPI. The cover glass coated with A20 cell monolayer was flipped immediately and sealed onto glass slide. The image was taken with a Zeiss Axiovert-200 microscope using a high-resolution Zeiss digital camera (Carl Zeiss Inc., Thornwood, NY).

Culture of A20 Cells with Potato Carboxypeptidase Inhibitor
To determine the effects of PCPI on A20 cell growth, A20 cells were seeded at a density of 6610 5 /well in 6-well plates, cultured overnight before the addition of 0, 5, 15, 30 or 60 ug/ml PCI. The cells were fed every 2 days with fresh medium (as above) containing the respective concentration of PCI, and viable cells were counted on a hemacytometer using trypan blue dye exclusion. Cells were split according to cell population size to maintain a cell concentration of 2-5610 6 /ml and cultures were maintained for 12 days. The cumulative cell number was calculated from the cell counts and the dilutions made at each culture split.

Statistical Analysis
Data were analyzed by either student t-test assuming unequal variance with P,0.05 (two-tail), or a one-way ANOVA. Figure S1 Cpa3 is not expressed in A20 cells. Real-time PCR was performed on A20 cells that were either uninfected (A20 control) or infected with empty (vector control) or Lxn expression vector (Lxn vector) to quantify Cpa3 mRNA expression. The amplification plots for Lxn (A), Cpa3 (B) and Gapdh (C) transcript show that Lxn is highly expressed in A20 cells infected with Lxn expression vector whereas A20 and vector controls have very low expression levels, consistent with the results in Fig. 3c. Cpa3 transcript is not amplified in all types of cells, indicating it is not expressed in A20 cells. These figures show the amplification plots of four individual biological replicates (n = 4) for each gene.

(TIF)
Table S1 Carboxypeptidase A3 (Cpa3) is highly expressed in stem/progenitor cells. Expression level of CPAs were measured by microarray on a bone marrow population null for cell markers characteristic of lineage-specific differentiated blood cells (Lin-negative), and positive for the Sca-1 and c-Kit cell markers (LSK) cells, enriched for hematopoietic stem/progenitor cells in mouse. The isolation of LSK cells as well as microarray analysis was performed as previously described [38] except the array platform is mouse Genome 430 2.0 Array (Affymetrix). Three biological samples were assayed, and the mean expression value (mean_y_B6) for each gene in the table is the average of three readings. Cpa3 (highlighted in gray) is the only Cpa gene that is highly enriched in LSK cells. Gapdh is the endogenous control. (PDF)