1 1

The essential role of PIM kinases in sarcoma growth and bone

invasion

Maja Narlik-Grassow1, Carmen Blanco-Aparicio1, Yolanda Cecilia1, Sandra

Peregrina1, Beatriz Garcia-Serelde1, Sandra Muñoz-Galvan2, Marta Cañamero3

and Amancio Carnero2,4*

1) Experimental Therapeutics programme, Spanish National Cancer Research Centre, Madrid, Spain, 2) Instituto de Biomedicina de Sevilla (IBiS), HUVR/CSIC/ Universidad de Sevilla, Sevilla, Spain, 3) Biotechnology programme, Spanish National Cancer Research Centre, Madrid, Spain, 4) Consejo Superior de Investigaciones Cientificas

*) Corresponding author: Amancio Carnero; Instituto de Biomedicina de Sevilla

(IBiS), Campus HUVR, Edificio IBIS. Avda. Manuel Siurot s/n. 41013, Sevilla, Spain

email: acarnero@ibis-sevilla.es; Phone: +3495592111;

2 2

ABSTRACT

PIM kinases are a family of serine/threonine kinases composed of three different

isoforms (PIM1, PIM 2 and PIM 3) that are highly homologous. Their expression is

mediated by the JAK/STAT signalling pathway, providing survival and cell cycle

transition signals. PIM kinases are heavily targeted for anticancer drug discovery.

However, very little is known about the relative contribution of the different

isoforms to the tumorigenesis process in vivo, and how their individual inhibition

might affect tumour growth. Taking advantage of genetically modified mice, we

explored whether the inhibition of specific isoforms is required to prevent

sarcomas induced by 3-methylcholanthrene carcinogenic treatment. We found

that absence of Pim2 and Pim3 greatly reduced sarcoma growth, to a similar

extent to the absence of all 3 isoforms. This model of sarcoma generally

produces bone invasion by the tumour cells. Lack of Pim2 and Pim3 reduced

tumour-induced bone invasion by 70%, which is comparable with the reduction

of tumour-induced bone invasion in the absence of all 3 isoforms. Similar results

were obtained in MEFs derived from these knockout mice, where double Pim2/3

knockout MEFs already showed reduced proliferation and were resistant to

oncogenic transformation by the Ras oncogene. Our data also suggest an

important role of Gsk3β phosphorylation in the process of tumorigenesis.

3 3

INTRODUCTION

The PIM kinases are a family of serine/threonine kinases that regulate cell

proliferation and survival [1, 2] and are highly conserved through evolution in

multicellular organisms. This family of kinases is composed of three different members

(Pim1, Pim2 and Pim3) that are highly homologous at the amino acid level, [3] yet

differ partially in their tissue distribution [4]. Functional redundancy of the three PIM

kinases has been shown in vitro [5, 6] and in vivo [7].

Unlike other serine/threonine kinases, the PIM family members are tightly regulated

at the level of transcription and translation [3]. Their expression is mediated by the

JAK/STAT signalling pathway, which is activated by various cytokines and hormones

[1, 2, 8-10].

The PIM kinase family members are considered oncogenes and have been

implicated in tumorigenesis either alone or functioning synergistically with c-Myc [11,

12]. Initially, elevated levels of PIM1 kinase were discovered in leukemia and

lymphoma tumours [13-15]. However, more recently, PIM1 was found to be

increased in solid tumours, including pancreatic and prostate cancer, squamous

cell carcinoma, gastric carcinoma, colorectal carcinoma, liver carcinoma [16, 17],

and recently liposarcoma [18]. Increased levels of PIM2 kinase have been detected

in various lymphomas as well as in prostate cancer. Several in vitro studies also

connected PIM2 kinase with liver cancer [3]. PIM3 kinase has been found aberrantly

expressed in malignant lesions of endoderm-derived organs, liver and pancreas and

also in Ewing’s sarcoma [3]. Recent studies have correlated PIM1 kinase with

chemoresistance in prostate cancer cells, which is a common occurrence in more

aggressive, hormone-refractory prostate cancers [19, 20]. PIM1 kinase has also

been linked to hypoxia-promoted genetic instability in solid tumours facilitating cell

survival, resulting in tumours with a more aggressive phenotype [21].

Although the PIM kinases have been identified as oncogenes in transgenic models,

by themselves they are only weakly transforming. However, they have been shown

to greatly enhance the ability of c-myc to induce lymphomas [11, 12, 22, 23]. It is

understood that mere overexpression of MYC induces an apoptotic response, which

has to be overcome to permit oncogenesis [11, 14, 15]. PIM kinases (PIM1 and PIM2)

4 4

have been shown to counteract this MYC-induced apoptosis by phosphorylating

MYC, thus increasing its protein stability, and consequently its transcriptional activity

[1]. The overexpression of either PIM1 or PIM2 in factor–dependent haematopoietic

cells makes them resistant to apoptosis induced by interleukin-3 withdrawal [24-26],

probably through phosphorylation and inactivation of the proapoptotic protein BAD

[25, 27]. PIM kinases mediate their physiological activities through phosphorylation of

a wide range of cellular substrates which, due to the functional redundancy of the

PIM kinase family, greatly overlap. Previously described phosphorylation targets

include SOCS-1 [28], runt-related transcription factor 1 (RUNX1) and RUNX3 [29]; cell

cycle regulators such as p21waf1 and p27kip1 CDK inhibitors [30, 31], cdc 25A

phosphatase [32] and the kinase cTAK/MARK3/Par1A, and transcriptional repressors

(HP1), activators (NFATc1 and c-Myb) and coactivators (p100) [33-37]. More

recently, PIM2 has been shown to phosphorylate the ribosomal protein 4E-BP1,

causing its dissociation from eIF-4E, which may affect protein synthesis as eIF-4E is a

rate-limiting factor [28]. Interestingly, several of the mentioned substrates are shared

with AKT kinases, and PIM and AKT often phosphorylate the same residues [1, 2] .

Interestingly, animals deficient in Pim1 are viable and show only a minor defect in

haematopoiesis [38, 39], suggesting that PIM1 contributes to haematopoietic cell

growth and survival, although its role is probably secondary to other signalling

pathways. More importantly, mice lacking expression of Pim1, Pim2 and Pim3 are

viable and fertile, although they show a profound reduction in body size at birth and

throughout postnatal life [7]. In addition, the in vitro response of distinct

haematopoietic cell populations to growth factors is severely impaired. These results

indicate that members of the PIM protein family are important but dispensable

factors for growth factor signalling [7].

Together, these studies suggest that PIM kinases may play a distinct role in tumour

formation in vivo, implying that they may be novel targets for cancer therapy [40].

However, very little is known about the relative contribution of the different isoforms

to the tumorigenesis in vivo, and how their individual inhibition might affect tumour

growth. Taking advantage of genetically modified mice, we explored whether the

5 5

inhibition of specific isoforms is required to prevent tumour growth either in vitro, in

MEFs or in vivo in tumours induced by 3-methylcholantrene carcinogenic treatment.

MATERIALS AND METHODS

Culture Conditions and cell line conservation

All cell lines were grown in DMEM culture medium supplemented with glutamine

(Sigma/Gibco), 10 % FBS (Sigma/Genycell), 40 U/ml penicillin, 40 U/ml streptomycin

and 1 µg/ml fungizone (Gibco/Invitrogen), further referred to as complete medium.

Cells were grown on 10-cm ø plates and split when a confluent monolayer of cells

was observed. Retroviral infection of cells was performed as previously described

[41]. Selection of transfected or infected cells was carried out with 2µg/ml

Puromycin.

Generation of MEFs

Mouse embryonic fibroblasts (MEFs) were generated from mouse embryos (day

12.5) of FVB wild-type mice and genetically modified mice for Pim1, 2 and 3. After

eliminating the head and liver (which were lysed and used for genotyping), the

embryo was mechanically disaggregated and trypsinised for 10 min at 37ºC. The

adherent cells were grown in DMEM (Sigma) supplemented with 10% fetal bovine

serum (FBS) from Sigma and 1% penicillin G sulphate/ streptomycin on 10-cm ø

culture plates. Cells were immortalised using the 3T3-protocol. Once immortalised,

one clone of each genotype was selected for further analysis.

3T3-protocol. Was performed as previously described [42]. Briefly, every 3 days cells

were trypsinized, counted and replated at 1 million live cells per 10cm plate.

Proliferation, growth under low serum conditions, foci formation, saturation density

and clonability assays were performed as previously described in [43, 44].

Maintenance of mouse colonies

All animals were kept in the CNIO animal facility according to the norms for animal

welfare.

6 6

Genotyping of mouse lines

Pim1, 2 and 3 KO mice were genotyped by PCR using the standard PCR protocol for

transgenic mice and PCR protocols provided by M. Narwijn (NKI-Netherlands) for the

triple KO-Pim mice [7]. Subsequently, Pim mutants ( KO) were identified by PCR: the

Pim1 alleles were detected using Pim1 and Neo primers (Pim1For,

5`AAGCACGTGGAGAAGGACCG-3_; Pim1Rev, 5´GACTGTGTCCTT GAGCAGCG-3_;

NeoA, 5`CGTCCTGCAGTTCATTCAGG-3_), the Pim2 alleles with Pim2 and Pgp

primers (Pim2For, 5`CACCGCGTCACGGATAGACG-3_; Pim2Rev,

5`CCCACCTTCCACAGCAGCG-3_; HygroFor, 5_AGCACTCGTCC

GAGGGCAAAGG-3_; Pim2Rev2, 5_-CACGGTGGACCAGCCTAGC-3_), and the

Pim3 alleles with Pim3 and lacZ primers (Pim3Sp1, 5’ CTGGACC

AAATTGCTGCCCAC-3_; Pim3Sp4, 5`GGATCTCTGGTTCAAGTATCC3_; LacZ1

5’CGTCACACTACGTCTG-AACG-3_; LacZ2, 5`CGACCAGATGATCACACTCG-3_).

mRNA expression

We analysed the expression of the Pim1, Pim2 and Pim3 transcripts in tumours by

reverse transcription-PCR (RT-PCR). For this, total RNA was isolated from 3-

Methylcholanthrene-induced sarcoma using TriZol (Gibco BRL), treated with DNase,

and reverse transcribed with random hexamer primers (Promega) and avian

myeloblastosis virus-reverse transcriptase (Roche). The cDNA was amplified in a

single PCR using for Pim1 primer combinations (mPim1fwRT 5’

ATGCTCCTGTCCAAGATCAACTC-3_; and Pim1revRT 5’ CTCCAGGATCAGG

ACGAAACTGT-3_;) for Pim2 (Pim2fwRT 5` ATGTTGACCAAGCCTCTGCAGG-3_; and

Pim2revRT 5´ GCTCAAGGACCAGCATGAAGC-3_;) and for Pim3 (Pim3fwRT 5`

ATGCTGCTGTCCAAGTTCGGCT-3_; and Pim3revRT 5´ GCT

CGAACCAGTCCAGCAAGC-3_). As a control, mouse Gapdh was amplified in the

same PCR using the primers GapdhFw (5` AAGGTCGGTGTGAACGGATT-3_) and

GapdhRev (5’ TTGCTGGGGTGGGTGGTC-3_).

Carcinogenesis induced by 3-Methylcholanthrene

The carcinogen 3-Methylcholanthrene (3MC) was dissolved in sesame oil (Sigma) at

a final concentration of 10 mg/ml. The solution was heated up to 100°C for 1 h and

mixed vigorously in order to completely dissolve the 3MC. Mice of 3-5 months of age

were injected with 1 mg (100 µl) of 3MC once into the right hind leg and later

7 7

examined twice a week. The mice were sacrificed when the tumour reached 1 cm

in diameter.

Necropsy and pathological analysis

Once sacrificed, tissues for histological and molecular analysis were taken. Tissues

meant for molecular analysis were frozen in carbonic ice and later stored at -80ºC.

Tissues for histological analysis were fixed in 10% formalin for 24h, dehydrated and

embedded in paraffin. 2-µm cuts were taken off and dyed with haematoxylin and

eosin or different primary antibodies: KI-67 prediluted (Master Diagnostica -

0003110QD), Caspase 3 1:400 (R & D Systems - AF835), Cyclin D1 1:75 (DAKO -

M3635), p211:300 (Santa Cruz – sc-397-G) and p53 1:200 (CM5p) (Leica - NCL-p53-

CM5p). As secondary antibodies the following were used: Omnimap anti-rabbit

prediluted (Ventana – 760-4311), horseradish peroxidase goat anti-rabbit 1:50 (Dako

– P0448), and biotinylated IgG rabbit anti-goat 1:200 (Dako – E0466).

Statistical data analysis

The computer program GraphPad Prism was used for all statistical analysis. To

determine the statistical significance of tumour growth and bone invasion, one-way

student’s T test analysis was used. To determine the statistical significance of survival

curves the Mantel-Cox test was used.

Analysis of protein expression

The cells were washed twice in 1x PBS, pelleted for 30 seconds at 14000x g and lysed

in lysis buffer (Tris-Hcl pH 7.5 50 mM, NP40 1%, glycerol 10%, NaCl 150mM, complete

protease inhibitor cocktail, 2 mM; Roche). Total protein from frozen mouse tissue

(maintained at -80ºC) was extracted after mechanical tissue disintegration using a

sterile plastic piston in 1ml lysis buffer. After centrifugation, supernantant protein

extracts were aliquoted and stored at -80°C until use.

The amount of protein was determined by Bradford assay using BSA (bovine serum

albumin) as a standard. The appropriate protein quantity was dissolved in Laemli

buffer (Tris-HCL pH 6.8 62.5mM, glycerol 10%, SDS 1%, 2-mercapto ethanol 5%,

bromphenol blue 0.0025%) and the proteins were separated in SDS-PAGE gels (12%)

before they were blotted onto Nitrocellulose Transfer membrane (Whatman -

8 8

Protrans). The proteins were visualised via the Odyssey LI-COR system. The primary

antibodies used were: p53 1:1000 (Cell Signalling - 2524), 4E-BP 1:1000 (Cell Signalling

- 9644), p4E-BP Thr37/46 1:1000 (Cell Signalling - 9459), Gsk3β 1:1000 (BD Transduction -

610202), pGsk3βSer9 1:1000 (Cell signalling – 9336L), p27Kip 1:1000 (BD Pharmingen -

610241), AKT 1:2000 (Cell Signalling – 9272), pAKTSer473 1:1000 (Cell signalling – 4051L),

Erk1/2 1:1000 (Cell Signalling – 4695S), pErk1/2 Thr202/Tyr204 1:1000 (Cell Signalling –

9106L), actin 1:5000 (Sigma – A5441), tubulin 1:10000 (Sigma – T6557). The secondary

antibodies used were goat anti-rabbit Alexa Fluor 680 1:5000 (Invitrogen – A21057)

and donkey anti-mouse IRDye 800CW 1:5000 (Rockland Inc. – 605-731-002).

RESULTS

Effect of PIM loss in vitro.

First, we generated MEFs from double pim2/3 knockout (KO) mice (DKO) or from

mice lacking all three isoforms (TKO), in parallel with isogenic wild-type MEFs. To that

end, we first immortalised MEFs from all genotypes. MEFs followed similar senescence

kinetics and were immortalised with similar frequency irrespective of the genotype

(not shown). Once immortalised, MEFs were grown in media with optimal (10%) or

suboptimal (0.5% and 2%) concentrations of serum (Figure 1A). At 0,5%, all MEFs

showed very slow growth but no apoptosis, irrespective of the Pim isoforms that were

expressed. At 2% of serum we observed clear differences in the growth of MEFs

depending on the genotype. TKO MEFs showed poor growth compared with wild-

type MEFs, while DKO MEFs showed intermediate proliferation. The differences

between DKO and TKO increase in respect to wild-type MEFs at 10% serum

concentration (Fig 1A).

However, these differences were more dramatic in MEFs infected with virus-carrying

oncogenic RasV12 (Fig 1B). To that end, MEFs were infected with viruses carrying

oncogenic RasV12 or empty vectors. MEFs carrying RasV12 were seeded as before in

optimal or suboptimal growth conditions. At low serum doses, all MEFs, irrespective

of their genotype, were unable to grow, showing high apoptosis. However, at 10%

serum, WT MEFs expressing RasV12 grew fast, while TKO and DKO MEFs grew initially

but stopped growth with high apoptosis after the initial rounds of duplication (Fig

1B).

9 9

These data suggest that loss of Pim kinases might rend cells resistant to oncogenic

transformation. To examine this hypothesis we transfected WT, TKO and DKO MEFs

with oncogenic RasV12 and measured the number of foci formed. In this assay,

combined loss of Pim2 and Pim3 is sufficient to block the foci formation induced by

RasV12 (Fig 1C). TKO MEFs grow forming a non-homogeneous layer of cells, making it

difficult to assess the foci formation.

To compare the data among the three genotypes and to explore the functional

relevance of each individual Pim kinase in vitro we decided to perform saturation

density analysis. Immortal non transformed MEFs have contact inhibition stopping

growth once the monolayer is formed. However, oncogenically transformed MEFs

continue growth after the monolayer is formed increasing the density of cells. We

observed that the growth of DKO and specially TKO MEFs is lower than the growth of

WT cells (Fig. 1E, lanes C). To compare the individual role of each kinase isoform in

this assay we expressed ectopically Pim1, Pim2 or Pim3 into wild type MEFs, or MEFs

lacking Pim2 and 3 (DKO) or lacking all three Pim isoforms (TKO). We observed that

the increased expression of one isoform of Pim kinases into Pim containing MEFs (WT

or DKO) does not alter significantly the saturation density of control cells (Fig. 1E,

hatched bars). However, in TKO MEFs, the recovery of Pim activity is sufficient to alter

growth forming monolayer (data not shown) and increase the saturation density to

levels similar to DKO cells (Fig. 1E, hatched barrs). Furthermore, we measured the

response of these cells to oncogenic transformation with Hrasv12. Oncogenic

transformation with Ras, as expected, increases saturation density in WT MEFs, and

the increase in one of the Pim isoforms does not seem to provide any advantage

(Fig 1E, solid bars). In DKO cells expressing only Pim1, Ras does not increases

saturation density, correlating with the results observed in Fig. 1C. However,

expressing Ras in MEFs that recovered the expression of 2 isoforms increase the

saturation density to levels similar to WT cells (Fig. 1E, solid bars). Finally, in TKO MEFs,

the expression of Ras does no increase the saturation density in any of the cultures

expressing only one isoform (Fig.1E, solid bars). Our data indicate that at least the

expression of 2 isoforms is necessary to provide ability to support oncogenic

transformation. Only overexpression of Pim1, in the absence of Pim2 and 3 seems to

provide some benefit in the presence of oncogenic Ras.

10 1

Effect of PIM loss in vivo after carcinogenic treatment.

To evaluate whether Pim kinases play a relevant role in the process of

sarcomagenesis induced by the carcinogen 3-Methylcholanthrene (3MC), we

performed intramuscular injection of 3MC, which produced sarcoma tumours with

an onset of tumour growth at approximately 98 days after inoculation. An in deep

evaluation of the time to tumour first observation showed that wild type tumours

appear in average around 12-20 days earlier than DKO or TKO tumours respectively.

Since this first sight, tumours took an average of 9 days in wild-type mice to reach

the end point of 1.7cm3 ( ca. 10% of mouse weight ), indicating a rapid growth of

these tumours (Fig 2A). However, in DKO it took an average of 21 days to reach the

same size, while in TKO mice the tumours reached this size in 32 days after onset of

tumour growth (Fig 2A). This significant time increase indicated that lack of Pim2 and

Pim3 kinases is sufficient to produce a significant delay in the growth of the

sarcoma. Additional loss of Pim1 only produces an additional marginal delay. This

slow sarcoma growth rate was reflected in the survival of the mice. DKO and TKO

mice treated with 3MC died significantly later than wild-type mice (Fig.2B). This was

due to the slower growth rate of the tumour, since the sarcomas appeared at the

same time in all three genotypes (data not shown).

To confirm the relative contribution of each isoform to sarcomagenesis we analysed

whether the tumour generated by the carcinogenic treatment in each mouse

derived from true KO cells, or whether they maintained the relative expression of the

Pim isoforms. We collected mRNA from all the tumours and evaluated the presence

of individual Pim isoforms' RNA expression (Fig 2C). We observed that the sarcoma

originated in wild-type mice maintained expression of all 3 pim isoforms, while DKO

expressed only pim1. Tumours originated in TKO mice did not express any pim

isoform.

The analysis of the sarcomas from all 3 genotypes showed that 3MC activated p53

in DKO and TKO MEFs to a similar extent to in WT MEFs (Fig 3A), which was expected

from the DNA-damage activity of 3MC. As a consequence, all tumours carried

mutated p53 (Fig 3B). All tumours, independently of the genotype, showed similar

11 1

levels of KI67, as it corresponds to fast-growing tumours but with very heterogeneous

distribution (Fig 3C). This large variability among tumours precluded any conclusion

from this marker. However, the percentage of cyclin D1 positive cells, another

proliferation marker, showed that the number of proliferating cells in wild-type

tumours was higher than in DKO and TKO mice (Fig 3C and data not shown).

Although this increase was not statistically significant due to the variability among

tumours, it showed a clear trend that tumours of DKO and TKO mice were less

proliferative. The reduced levels of cyclin D1 might explain the reduced proliferation

rate of the tumours. Similarly, all tumours contain caspase 3 staining (Fig 3C).

However, as in the case of KI67, the large variability among tumours precluded any

conclusion regarding increased apoptosis.

To study the mechanism implicated in the reduction of growth rate in pim-deficient

tumours, we analysed the relative levels of proteins that are direct or indirect targets

of Pim kinases and might be involved in slowing the cell cycle. Pim kinases

phosphorylate p27kip1, inducing its degradation [31], and might alter the protein

translation through 4EBP-1 [45, 46]. On the other hand, Pim3-induced tumours show

increased levels of cyclin D1 [47] and active Gsk3β alters cyclin D1 stability by

enhancing its degradation [48]. Furthermore, since AKT has been reported to be

situated downstream of Pim in some models [49], we also measured the

phosphorylation status of these proteins (Fig 4).

We did only see minimal differences in 4E-BP1 phosphorylation among the tumours

of the different genotypes, with reductions in the DKO and TKO tumors (Fig 4A and

4B). We observed also slightly lower levels of pErk and p27kip1 in most DKO and all

TKO tumours we analysed (Fig 4A and 4B). However, the analysis of

Gsk3β phosphorylation at ser9 (inactive form) by Western blot showed that this

residue is not phosphorylated in most DKO and all TKO tumours (Fig 4 A and 4B). Only

one out of five DKO tumours showed phosphorylated Gsk3β at ser9, while none of

the analysed TKO tumours showed ser9 Gsk3β phosphorylation. By contrast, AKT

phosphorylation did not show significant variations among tumours (Fig 4a and 4B),

indicating that the effect on Gsk3β is not due to inactive Akt in Pim-null tumours.

Levels of other proteins such as MAP kinases did not show variations in the tumours

from different genotypes either (Fig 4A). To explore whether that effect on ser9

12 1

Gsk3β phosphorylation was a general effect due to Pim absence we analyzed

Gsk3β in MEFs from all three genotypes (Fig 4C). All MEFs showed similar content on

Gsk3β and Gsk3β phosphorylated at Ser9, indicating that loss of Gsk3β

phosphorylation at Ser9 occurs during the tumorigenic process.

Loss of Pim kinases reduces bone invasion by sarcoma

One interesting property in our 3MC-induced sarcomas was the microscopically

observed bone invasion (Figure 5A). Sarcoma cells are able to degrade bone and

invade bone marrow. In over 80% of tumours induced in wild-type animals we

detected bone invasion by these sarcomas (Figure 5B). However, the microscopic

analysis showed that lack of Pim2 and 3 reduced the percentage of sarcomas

invading bone to 31%. Lack of all three isoforms reduced bone invasion to 17%

(Figure 5A), confirming the relevance of these kinases in the process. Furthermore,

we have detected that these tumors have reduced expression of MMP9

metaloproteinase correlating with the decreased bone invasion we observed

(Figure 5C).

DISCUSSION

PIM kinases are an active target for drug discovery research, with most compounds

focused on the PIM1 isoform due to its known implications in tumorigenesis.

However, especially in vivo, little is known regarding the specific contribution of

each isoform to tumorigenesis. Taking advantage of genetically modified mice, we

explored whether the inhibition of specific isoforms is required to prevent sarcomas

induced by 3-methylcholanthrene carcinogenic treatment. We found that absence

of Pim2 and Pim3 greatly reduced sarcoma growth, to a similar extent to the

absence of all 3 isoforms. Similar results were obtained in MEFs derived from these

knockout mice, where double Pim2/3 knockout MEFs already showed reduced

proliferation and were resistant to oncogenic transformation. To block bone invasion

by the tumour, although the absence of all 3 isoforms is necessary to achieve the

maximum effect, depletion of Pim2 and Pim3 seems to be sufficient to achieve a

large reduction of the invasive effect.

13 1

We observed that sarcomas arising from Pim-defective mice grow significantly

slower, correlating with reduced proliferation markers, suggesting reduced

proliferative capacity. It has been shown that PIM1 phosphorylates Cdc25A, thereby

increasing its phosphatase activity and the activity of cyclin D1-associated kinases,

which results in cell cycle progression [32]. The Cdc25C-associated kinase 1 (C-TAK1)

is a potent inhibitor of Cdc25C, a protein that actively promotes cell cycle

progression at the G2/M phase. By phosphorylating C-TAK1, PIM1 inhibits its activity

and eventually advances the cell cycle [50]. Furthermore, PIM1 can phosphorylate

p21waf1, another molecule involved in cell cycle progression, at its threonine residues,

resulting in its relocation to the cytoplasm and enhancing its protein stability [30, 51].

These effects are associated with increased cell proliferation. Thus, it is tempting to

speculate that PIM can promote cell cycle progression and thus eventually

contributes to carcinogenesis by modulating the function of molecules that regulate

cell cycle progression.

Our data showed a reduction of cyclin D1 in Pim-null tumours correlating with a

decrease in the proliferative capabilities of these tumours. Furthermore, we also

observed a partial inhibition of 4E-BP1 phosphorylation and slightly lower levels of

pErk in the DKO and TKO tumours (Fig 4A and 4B). It is possible that the addition of

those mild effects contributed to reduced proliferation, slowing the tumour growth

rate.

However, the most important mechanistic feature observed in our model is the loss

of Gsk3β phosphorylation at Ser9 (inactive form) in most tumours of Pim-depleted

mice, but not in MEFs of the same genotype. Absence of Gsk3β Ser9

phosphorylation will maintain high Gsk3β activity, reducing cell cycle transition and

glucose metabolism [48, 52]. This residue has been shown to be phosphorylated by

Akt [48]. It is possible that Akt activation lies downstream of Pim, as has been

described previously [49]. Therefore, loss of Pim might abolish Akt activation and

subsequent Gsk3β phosphorylation. However, analysis of tumours showed similar

levels of Akt activation in Pim-null tumours, suggesting that Gsk3β Ser9 residue

phosporylation depends on a kinase different from Akt. It is possible that Pim itself

phosphorylates Gsk3β at Ser9 and inactivates the protein, since Akt and Pim share

many protein substrates [1-3]. On the other hand, Gsk3β is also a substrate of PP2A

14 1

[53-55], which has been recently described as a target of Pim kinases [56]. Therefore,

loss of Pim kinases might increase the levels of PP2A, leading to Gsk3β

dephosphorylation. Furthermore, loss of Pim expression might increase levels of PP2A

activity, which may mediate the levels of c-Myc and the phosphorylation of proteins

needed for increased protein synthesis. Both of these changes could have a

significant impact on tumour growth [56]. Whichever it is the mechanism, it occurs

only as a consequence of the tumorigenic process in the absence of Pim, since Pim

null MEFS maintain Gsk3β phosphorylation at Ser9 residue, suggesting that there is no

redundancy in this Gsk3b phosphorylation, and therefore, this protein can be a

good therapeutic target for invasion in Pim dependent tumors.

One of the most interesting pieces of data found in our work is the correlation of Pim

kinases with the process of bone invasion in vivo. Absence of Pim kinases blocks the

process of bone invasion induced by the 3MC-induced sarcoma, and the genes

seem to act in an additive manner, since absence of Pim2 and Pim3 produces only

a partial effect and the absence of all three is necessary to achieve the maximum

effect. In agreement with our data, siRNA interference of PIM1 and PIM2 reduced

PC3 migration in vitro by around 50% while inhibition of all 3 PIM kinases by DHPCC-9

(a specific pan PIM inhibitor) reduced migration of PC3 in vitro by 90% [57]. In

addition, overexpression of any PIM family member has the opposite effects by

enhancing cell motility [57]. Silencing of PIM3 was also reported to reduce

endothelial cell spreading, migration and vascular tube formation, further

supporting that this kinase can stimulate metastatic and/or angiogenic potential of

cancerous cells [58]. In addition, PIM1, but not PIM2, was shown to regulate homing

and migration of bone marrow cells, possibly via phosphorylation-mediated

modification of CXCR4 expression in the cell surface [59]. The substrates and

signalling pathways regulated by PIM kinases, that contribute to enhancing the

motility of adherent cancer cells remain, however, to be elucidated. Recently, the

NFAT transcription factors, identified as PIM targets [35] have been implicated in

tumour cell migration and invasion [60]. Since NFAT is also a target of Gsk3β, it is

tempting to speculate that lack of ser9-Gsk3β phosphorylation in Pim-null tumours

contributes to reduced migration by maintaining low levels of NFAT activation.

15 1

Besides our work, several other features justify the use of PIM as a cancer target; its

overexpression in various types of tumours, its role in the regulation of pathways

considered relevant in cancer, as well as the fact that blocking PIM1 function by

introduction of dominant negative PIM1 sensitises pancreatic cancer cells to

apoptosis induced by glucose deprivation under hypoxia [21]. Moreover, dominant

negative PIM1 reduces tumorigenicity in pancreatic cancer cells and HeLa

xenograft mouse models. These and other data have led to the synthesis of PIM

inhibitors that have recently entered clinical trials.

Inhibitors of PIM are of interest as potential therapeutic agents [5, 19, 45, 61, 62];

therefore, the understanding of the process that can be targeted by PIM inhibition is

of great importance to define the potential activity of these compounds. However,

there is some degree of redundancy among PIM kinases that should be studied

while defining the specificity of the pharmacological compounds. The levels of

expression of the different PIM family members vary among tumour types. PIM1 and

PIM2 are mostly overexpressed in leukemia and lymphomas. PIM1 is also

overexpressed in some solid tumours such as prostate, pancreas and colon. PIM3

overexpression has been observed in melanoma, pancreatic and gastric tumours.

The possibility of overlapping substrates and the potential compensatory nature of

the different PIM family members as demonstrated by the triple knockout mouse

model suggests that the inhibition of all isoforms may be more effective than

targeting individual isoforms. In our genetic model, higher activity is achieved in the

absence of all family members although sarcoma growth is already significantly

reduced in PIM2- and 3-depleted mice.

Since both the PI3K and PIM signalling pathways converge on the regulation of

many signalling proteins that are common for AKT and PIM, it is possible that some

redundancy among these pathways also occurs in tumours. In fact, It has been

described that PIM inhibitors synergise with mTOR of PI3K inhibitors [45, 46]. Dual

treatment with the compounds resulted in decreased phosphorylation of 4E-BP1

and BAD proteins as well as an increase in apoptosis. These data suggest that a

combination of drugs that selectively target PI3K/AKT and PIM may have potential in

treating tumours.

16 1

Ackowledgements

We thank Drs. H. Mikkers, J. Jonkers and A. Berns for the kind gift of the PIM TKO mice.

We would like to acknowledge the excellent technical work of Virginia Álvarez,

Maria del Carmen Arriba, Elvira Gil, María Gómez, Patricia González and Natalia

Matesanz of the Comparative Pathology Core Unit at CNIO. This work was

supported by grants from the Spanish Ministry of Science and Innovation and Feder

Funds (SAF2009-08605), Consejeria de Ciencia e Innovacion and Consejeria de

Salud of the Junta de Andalucia (CTS-6844 and PI-0142). AC’s laboratory is also

funded by a fellowship from the Fundacion Oncologica FERO, supported by the

Fundació Josep Botet. MN-G was funded by a fellowship from the Spanish Ministry of

Science and Innovation.

17 1

REFERENCES

1. Nawijn, M.C. et al. (2011) For better or for worse: the role of Pim oncogenes in tumorigenesis. Nat Rev Cancer, 11. 23-34.

2. Amaravadi, R., Thompson, C.B. (2005) The survival kinases Akt and Pim as potential pharmacological targets. J Clin Invest, 115. 2618-2624.

3. Brault, L. et al. (2010) PIM serine/threonine kinases in the pathogenesis and therapy of hematologic malignancies and solid cancers. Haematologica, 95. 1004-1015.

4. Eichmann, A. et al. (2000) Developmental expression of pim kinases suggests functions also outside of the hematopoietic system. Oncogene,

19. 1215-1224. 5. Bullock, A.N. et al. (2005) Structure and substrate specificity of the Pim-1

kinase. J Biol Chem, 280. 41675-41682. 6. Mukaida, N. et al. (2011) Roles of Pim-3, a novel survival kinase, in

tumorigenesis. Cancer Sci. 7. Mikkers, H. et al. (2004) Mice deficient for all PIM kinases display reduced

body size and impaired responses to hematopoietic growth factors. Mol

Cell Biol, 24. 6104-6115. 8. Dautry, F. et al. (1988) Regulation of pim and myb mRNA accumulation

by interleukin 2 and interleukin 3 in murine hematopoietic cell lines. J Biol

Chem, 263. 17615-17620. 9. Fox, C.J. et al. (2003) The serine/threonine kinase Pim-2 is a

transcriptionally regulated apoptotic inhibitor. Genes Dev, 17. 1841-1854. 10. Xu, D. et al. (2011) The oncogenic kinase Pim-1 is modulated by K-Ras

signaling and mediates transformed growth and radioresistance in human pancreatic ductal adenocarcinoma cells. Carcinogenesis, 32. 488-495.

11. Allen, J.D. et al. (1997) Pim-2 transgene induces lymphoid tumors, exhibiting potent synergy with c-myc. Oncogene, 15. 1133-1141.

12. Allen, J.D., Berns, A. (1996) Complementation tagging of cooperating oncogenes in knockout mice. Semin Cancer Biol, 7. 299-306.

13. Selten, G. et al. (1986) The primary structure of the putative oncogene pim-1 shows extensive homology with protein kinases. Cell, 46. 603-611.

14. Cuypers, H.T. et al. (1986) Assignment of the human homologue of Pim-1, a mouse gene implicated in leukemogenesis, to the pter-q12 region of chromosome 6. Hum Genet, 72. 262-265.

15. Wang, Z. et al. (2001) Pim-1: a serine/threonine kinase with a role in cell survival, proliferation, differentiation and tumorigenesis. J Vet Sci, 2. 167-179.

16. Shah, N. et al. (2008) Potential roles for the PIM1 kinase in human cancer - a molecular and therapeutic appraisal. Eur J Cancer, 44. 2144-2151.

17. Bachmann, M., Moroy, T. (2005) The serine/threonine kinase Pim-1. Int J

Biochem Cell Biol, 37. 726-730. 18. Nga, M.E. et al. (2010) PIM-1 kinase expression in adipocytic neoplasms:

diagnostic and biological implications. Int J Exp Pathol, 91. 34-43. 19. Mumenthaler, S.M. et al. (2009) Pharmacologic inhibition of Pim kinases

alters prostate cancer cell growth and resensitizes chemoresistant cells to taxanes. Mol Cancer Ther, 8. 2882-2893.

18 1

20. Chen, J. et al. (2009) Pim-1 plays a pivotal role in hypoxia-induced chemoresistance. Oncogene, 28. 2581-2592.

21. Chen, J. et al. (2009) Hypoxia-mediated up-regulation of Pim-1 contributes to solid tumor formation. Am J Pathol, 175. 400-411.

22. Breuer, M. et al. (1989) Very high frequency of lymphoma induction by a chemical carcinogen in pim-1 transgenic mice. Nature, 340. 61-63.

23. van Lohuizen, M. et al. (1989) Predisposition to lymphomagenesis in pim-1 transgenic mice: cooperation with c-myc and N-myc in murine leukemia virus-induced tumors. Cell, 56. 673-682.

24. Lilly, M., Kraft, A. (1997) Enforced expression of the Mr 33,000 Pim-1 kinase enhances factor-independent survival and inhibits apoptosis in murine myeloid cells. Cancer Res, 57. 5348-5355.

25. Lilly, M. et al. (1999) The PIM-1 serine kinase prolongs survival and inhibits apoptosis-related mitochondrial dysfunction in part through a bcl-2-dependent pathway. Oncogene, 18. 4022-4031.

26. Domen, J. et al. (1993) Impaired interleukin-3 response in Pim-1-deficient bone marrow-derived mast cells. Blood, 82. 1445-1452.

27. Yan, B. et al. (2003) The PIM-2 kinase phosphorylates BAD on serine 112 and reverses BAD-induced cell death. J Biol Chem, 278. 45358-45367.

28. Paukku, K., Silvennoinen, O. (2004) STATs as critical mediators of signal transduction and transcription: lessons learned from STAT5. Cytokine

Growth Factor Rev, 15. 435-455. 29. Aho, T.L. et al. (2006) Pim-1 kinase phosphorylates RUNX family

transcription factors and enhances their activity. BMC Cell Biol, 7. 21. 30. Wang, Z. et al. (2002) Phosphorylation of the cell cycle inhibitor

p21Cip1/WAF1 by Pim-1 kinase. Biochim Biophys Acta, 1593. 45-55. 31. Morishita, D. et al. (2008) Pim kinases promote cell cycle progression by

phosphorylating and down-regulating p27Kip1 at the transcriptional and posttranscriptional levels. Cancer Res, 68. 5076-5085.

32. Mochizuki, T. et al. (1999) Physical and functional interactions between Pim-1 kinase and Cdc25A phosphatase. Implications for the Pim-1-mediated activation of the c-Myc signaling pathway. J Biol Chem, 274. 18659-18666.

33. Koike, N. et al. (2000) Identification of heterochromatin protein 1 (HP1) as a phosphorylation target by Pim-1 kinase and the effect of phosphorylation on the transcriptional repression function of HP1(1). FEBS

Lett, 467. 17-21. 34. Evans, K.E., Fox, S.W. (2007) Interleukin-10 inhibits osteoclastogenesis by

reducing NFATc1 expression and preventing its translocation to the nucleus. BMC Cell Biol, 8. 4.

35. Rainio, E.M. et al. (2002) Cutting edge: Transcriptional activity of NFATc1 is enhanced by the Pim-1 kinase. J Immunol, 168. 1524-1527.

36. Winn, L.M. et al. (2003) Pim-1 phosphorylates the DNA binding domain of c-Myb. Cell Cycle, 2. 258-262.

37. Ishibashi, Y. et al. (2001) Pim-1 translocates sorting nexin 6/TRAF4-associated factor 2 from cytoplasm to nucleus. FEBS Lett, 506. 33-38.

38. Laird, P.W. et al. (1993) In vivo analysis of Pim-1 deficiency. Nucleic Acids

Res, 21. 4750-4755. 39. Domen, J. et al. (1993) Analysis of Pim-1 function in mutant mice.

Leukemia, 7 Suppl 2. S108-112.

19 1

40. Magnuson, N.S. et al. (2010) Why target PIM1 for cancer diagnosis and treatment? Future Oncol, 6. 1461-1478.

41. Carnero, A. et al. (2000) Loss-of-function genetics in mammalian cells: the p53 tumor suppressor model. Nucleic Acids Res, 28. 2234-2241.

42. Carnero, A., Beach, D.H. (2004) Absence of p21WAF1 cooperates with c-myc in bypassing Ras-induced senescence and enhances oncogenic cooperation. Oncogene, 23. 6006-6011.

43. Guijarro, M.V. et al. (2007) MAP17 enhances the malignant behavior of tumor cells through ROS increase. Carcinogenesis, 28. 2096-2104.

44. Leal, J.F. et al. (2008) Cellular senescence bypass screen identifies new putative tumor suppressor genes. Oncogene, 27. 1961-1970.

45. Blanco-Aparicio, C. et al. (2011) Pim 1 kinase inhibitor ETP-45299 suppresses cellular proliferation and synergizes with PI3K inhibition. Cancer Lett, 300. 145-153.

46. Beharry, Z. et al. (2009) Novel benzylidene-thiazolidine-2,4-diones inhibit Pim protein kinase activity and induce cell cycle arrest in leukemia and prostate cancer cells. Mol Cancer Ther, 8. 1473-1483.

47. Wu, Y. et al. (2010) Accelerated hepatocellular carcinoma development in mice expressing the Pim-3 transgene selectively in the liver. Oncogene, 29. 2228-2237.

48. Carnero, A. (2010) The PKB/AKT pathway in cancer. Curr Pharm Des, 16. 34-44.

49. Acosta, J.C. et al. (2008) Chemokine signaling via the CXCR2 receptor reinforces senescence. Cell, 133. 1006-1018.

50. Bachmann, M. et al. (2004) The oncogenic serine/threonine kinase Pim-1 phosphorylates and inhibits the activity of Cdc25C-associated kinase 1 (C-TAK1): a novel role for Pim-1 at the G2/M cell cycle checkpoint. J Biol

Chem, 279. 48319-48328. 51. Zhang, Y. et al. (2007) Pim-1 kinase-dependent phosphorylation of

p21Cip1/WAF1 regulates its stability and cellular localization in H1299 cells. Mol Cancer Res, 5. 909-922.

52. Vivanco, I., Sawyers, C.L. (2002) The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer, 2. 489-501.

53. Hernandez, F. et al. (2010) Regulation of GSK3 isoforms by phosphatases PP1 and PP2A. Mol Cell Biochem, 344. 211-215.

54. Mora, A. et al. (2002) Lithium blocks the PKB and GSK3 dephosphorylation induced by ceramide through protein phosphatase-2A. Cell Signal, 14. 557-562.

55. Ivaska, J. et al. (2002) Integrin alpha 2 beta 1 promotes activation of protein phosphatase 2A and dephosphorylation of Akt and glycogen synthase kinase 3 beta. Mol Cell Biol, 22. 1352-1359.

56. Chen, W.W. et al. (2005) Pim family kinases enhance tumor growth of prostate cancer cells. Mol Cancer Res, 3. 443-451.

57. Santio, N.M. et al. (2010) Pim-selective inhibitor DHPCC-9 reveals Pim kinases as potent stimulators of cancer cell migration and invasion. Mol

Cancer, 9. 279. 58. Zhang, P. et al. (2009) Pim-3 is expressed in endothelial cells and

promotes vascular tube formation. J Cell Physiol, 220. 82-90. 59. Grundler, R. et al. (2009) Dissection of PIM serine/threonine kinases in

FLT3-ITD-induced leukemogenesis reveals PIM1 as regulator of CXCL12-CXCR4-mediated homing and migration. J Exp Med, 206. 1957-1970.

20 2

60. Mancini, M., Toker, A. (2009) NFAT proteins: emerging roles in cancer progression. Nat Rev Cancer, 9. 810-820.

61. Qian, K.C. et al. (2005) Expression, purification, crystallization and preliminary crystallographic analysis of human Pim-1 kinase. Acta

Crystallogr Sect F Struct Biol Cryst Commun, 61. 96-99. 62. Pogacic, V. et al. (2007) Structural analysis identifies imidazo[1,2-

b]pyridazines as PIM kinase inhibitors with in vitro antileukemic activity. Cancer Res, 67. 6916-6924.

21 2

FIGURE LEGENDS

FIGURE 1: A) Proliferation of immortalised MEFs under different serum

conditions. 5x103 cells of each genotype were grown under different serum conditions (10%, 2% and 0.5%, respectively). Every 48hrs for 10 days a time point was taken. Cells were fixed and the number of cells counted as a relative measure of absorbance (see M&M). B) Proliferation

of immortalised MEFs infected with RasV12 under different serum

conditions. Cells of each genotype were infected with viruses carrying H-RasV12.After antibiotic selection (puromycin) 5x103 cells were seeded and grown under different serum conditions (10%, 2% and 0.5%, respectively). Cells were fixed and the number of cells counted as a relative measure of absorbance (see M&M). C) Foci formation. Cells of the different genotypes were transfected with the plasmid pBabepuro-HRasV12 carrying the RasV12 oncogene and grown for 2 weeks, changing the medium every 2 days. Cells were then stained with methylene blue to visualise the foci (see M&M). D) Activation of MAPK (pMAPK) by

oncogenic RasV12 expression in WT or DKO MEFs. Cells of each genotype were infected with viruses carrying H-RasV12. After antibiotic selection (puromycin) 5x105 cells were lysed according to M&M and total protein extracted. Then, western blot to identify MAPK phosphorylation performed. E) Relative role of Pim isoforms in MEFs saturation density. Cells of each genotype (WT, DKO, TKO) were transfected with the different Pim isoform as indicated, under the same constitutive promoter. After selection the different mass cultures were tranfected with a plasmid carrying H-RasV12. After selection, the different mass cultures were seeded and grown under 10% serum conditions until they reach confluence. Then, the cells were cultured in the same plate for one more week changing media every 2 days. After this time, cells were fixed, stained and the number of cells counted as a relative measure of absorbance (see M&M).

FIGURE 2: Sarcomagenesis in Pim-KO mice after 3-Methylcholanthrene

injection. Mice were injected with 3-Methylcholanthrene (3MC) as described in M&M. Animals were monitored twice per week after onset of tumour growth in the first animal was observed. When tumours reached the final size (1,7cm3) animals were sacrificed for a humane endpoint. A) Days of tumour growth from the first sign of growth onset to

the final tumour size of 1,7cm3. B) Percentage of survival of the treated mice from the injection of 3MC until the humane endpoint. C) Relative

expression of the PIM isoform mRNA in tumours of mice treated with 3MC. RNA was extracted from tumours and a reverse transcriptase PCR was performed to obtain cDNA, which was amplified using specifically designed primers. PCR fragment length was checked on 1, 5% agarose gels.

22 2

FIGURE 3: Characterisation of 3MC-induced sarcoma. A) Activation of

p53 by 3-Methylcholanthrene (3MC). Pre-immortalised MEFs were treated with 10 µM 3MC. Total protein was extracted at each time point indicated in hrs and p53 expression was determined by Western blot. B)

Immunohistochemistry of tumours for p53. All tumours were hybridised with antibodies against p53. Mutant p53 is recognised by strong nuclear staining indicating stabilisation of the p53 protein. C)

Immunohistochemistry of KI67, Caspase 3 and Cyclin D1 in 3MC-induced

sarcoma. Several different fields for each tumour were hybridised with antibodies against the proliferation markers KI67 and Cyclin D1 and apoptotic marker caspase 3. Proliferation markers KI67 and Cyclin D1 show similar high staining in every genotype with a high intra- and intertumoral variability. Apoptosis marker Caspase 3 shows low activation regardless of the tumour’s genotype but also a high variability within each tumour and among different tumours.

FIGURE 4: Western blot analysis of Pim kinase targets. A) Western blot

analysis of Pim kinase targets in tumors. For Western blot analysis of direct or indirect Pim kinase targets, whole cell protein was extracted from freshly resected tumors from mice and levels of different proteins were determined by Western blot. Western blot of the phosphorylated proteins recognises phophorylation at the following residues: p4E-BP at Thr37/46, pGsk3β at Ser9, pAkt at Ser473, pErk1/2 at Thr202/Tyr204. 2 tumors per genotype are shown. B) Quantification of the levels of the different proteins analyzed.

Each graph shows the relative levels of the protein indicated p4E-BP (phosphorylated protein at Thr37/46 related to total p4E-BP), pGsk3β phosphorylated Gsk3β at Ser9 related to total pGsk3β), pAkt( phosphorylated Akt at Ser473 related to total Akt), pErk1/2 (phosphorylated Erk1/2 at Thr202/Tyr204 related to total Erk1/2), p27 (total p27 related to tubulin). Graphs show average and St. Dev. of levels in all tumours analyzed as indicated in text. C) Western blots analysis of pGsk3β at Ser9 in MEFs. Whole cell protein was extracted from MEFs and levels of proteins were determined by Western blot. One representative MEF clone per each genotype is shown.

FIGURE 5: Depletion of Pim kinases reduces bone invasion by 3MC-

induced sarcoma: A) haematoxilyn & eosin staining of sarcoma with bone

invasion. Tumour shown was extracted from wild-type mice. B)

Percentage of mice that showed bone invasion of sarcoma. Significantly more wild-type mice show bone invasion of sarcoma compared with mice depleted of Pim kinases. C) Western blots analysis of MMP9 in tumors.

Whole cell protein was extracted from MEFs and levels of different proteins were determined by Western blot. One representative MEF clone per each genotype is shown. D) Quantification of the levels of MMP9 in tumors. The graph shows the relative levels of the MMP9 protein related to tubulin. Graphs show average and St. Dev. of levels in all tumours analyzed as indicated in text.

23 2

Figure 1

24 2

Figure 2

Figure 3

25 2

Figure 4

Figure 5