In vivo Expression of Signaling Proteins in Reconstituted NK Cells

Abstract

Natural Killer cells are cells of the innate immune system that are important for the recognition and clearance of virally infected cells or tumors. Examination of the development and signaling of these cells has been severely hampered due to an inability to over-express proteins in these cells. We developed a novel technique to generate NK cells in vivo, all of which express a gene of interest. IL2Rγ_c^-/-/Rag2^-/- mice do not develop NK cells due to the lack of IL15 signaling. We infected bone marrow from IL2Rγ_c^-/-/Rag2^-/- mice with a retroviral construct encoding EGFP and IL2Rγ_c connected by an IRES. NK cells selectively developed through expression of IL2Rγ_c and 100% of these NK cells were found to be EGFP⁺. In order to test the utilization of this method to examine the function of biologically relevant proteins, constitutively active PI3K p110γ and p110δ isoforms were over-expressed in this system. Constitutively active p110γ revealed profound effects on NK cell development and function in vivo while p110δ had little effect.

Introduction

Natural Killer cells are cells of the innate immune system that are involved in the recognition and elimination of allogeneic cells and stressed autologous cells such as those that are virally infected or transformed (Di Santo and Vosshenrich, 2006). The study of NK cell signaling has been hampered due to difficulties in stably or transiently transfecting NK cells (Maasho et al., 2004). Primary NK cells are resistant to conventional gene transfection methods. The Amaxa nucleofection system™, which directly transfers DNA into the nucleus of the cell, has overcome some of these problems for human NK cells and cell lines. Human NK cell lines can be transfected to varying levels using this method and then sorted by FACS, however, most NK cell lines do not retain all the characteristics of primary NK cells (Maasho et al., 2004). The Amaxa system does not work for murine NK cells and previous studies attempting to use viral systems to over-express proteins in primary murine NK cells have reported very low infection rates for NK cells (Kume et al., 2003). Therefore, an effective system for the over-expression of WT, constitutively active or dominant negative proteins in primary murine NK cells would greatly enhance our understanding of NK development and signaling.

IL2Rγ_c^-/- mice display reduced levels of B and T cells and they lack NK cells due to the absence of IL15 signaling (Cao et al., 1995). Reconstitution of IL2Rγ_c^-/- bone marrow (BM) with the IL2Rγ_c chain permits development of NK cells (Soudais et al., 2000). Here we describe a novel approach harnessing the selective power of IL2Rγ_c to generate NK cells in vivo from transduced progenitors, all of which express a gene of interest. The development of this technique signifies an important step in the progression of NK cell research. For the first time constitutively active proteins, dominant negative proteins or even shRNA may be expressed in 100% NK cells and used for the examination of development, signaling and function of primary NK cells.

Materials and Methods

Constructs

The murine IL2Rγ_c chain was PCR amplified from IL2Rγ_c pMND (a kind gift from Dr. Fabio Candotti) using the following primers (5′-gagatctagcaccatggtgaaactattattg-3′ and 5′-gaattctcaggcttccggcttcagag-3′). An IRES and EGFP were amplified from IRES-EGFP pMPSV with 5′-gaattccgcccctctccctccc-3′, 5′-gcggccgcgttaacggatggttgtggcaagcttatcatc-3′, 5′-gcggccgctgatcaaccatccatggtgagcaagggc-3′ and 5′-catcgataccgtcgactcgagttg-3′ respectively. IL2Rγ_c was cloned into BglII and EcoRI sites to form IL2Rγ_c pMSCV. IL2Rγ_c pMSCV (digested with EcoRI and ClaI), IRES (EcoRI and NotI) and EGFP (ClaI and NotI) were ligated to create γ_c:IRES:EGFP (γ_c:I:EGFP) pMSCV. γ_c:I:EGFP pMSCV (HpaI and XhoI) and IL2Rγ_c pMND (EcoRI, blunted with Klenow fragment followed by XhoI) were ligated to form γ_c:I:γ_c pMSCV. γ_c:I:γ_c pMSCV (EcoRI and BglII) and EGFP (BamHI and EcoRI) were ligated to create EGFP:I:γ_c pMSCV. Constitutively active p110γ and p110δ attached to Src myristylation sequence pBABEpuro (a kind gift from Dr. Amancio Carnero) (SalI, blunted with Klenow fragment followed by BamHI) and γ_c:I:γ_c pMSCV (EcoRI, blunted with Klenow fragment followed by BglII) were ligated to create myr p110γ:I:γ_c pMSCV and myr p110δ:I:γ_c pMSCV.

Bone Marrow Transduction (BMT) and Transplantation

BM was harvested from C57BL/6 or IL2Rγ_c^-/-/Rag2^-/- mice 4 days after I.P. injection with 150mg/kg of 5-fluorouracil (5-FU) (Sigma) in 0.5ml PBS. Isolated BM cells were cultured in X-VIVO 10 medium supplemented with 5% fetal calf serum (FCS), 2mM L-glutamine, 100ng/ml murine stem cell factor (mSCF), TPO, FLT-3L and 50ng/ml IL-6 for 48 h at 37°C and 5% CO₂ at a concentration of 2-3×10⁶ cells/ml. The retroviral constructs were transfected along with pCL Eco (helper plasmid with gag/pol/env genes) into the Phoenix ecotropic packaging cell line with Fugene 6 (Roche) and cultured for 48 h at 37°C and 5% CO₂. Retronectin (Takara) was coated on non-tissue culture treated 6 well plates at 25μg/well in 2ml PBS overnight prior to addition of filtered supernatant from Phoenix transfections at 37°C and 5% CO₂ for 2-3 h. Following removal of supernatant, bone marrow cells were harvested, resuspended in fresh X-VIVO 10 medium supplemented with 5% FCS, 2mM L-glutamine, 100ng/ml mSCF, mTPO, mFLT-3L and 50ng/ml mIL-6 for 48 h at 37°C and 5% CO₂. These cells were plated onto the viral coated plates at a concentration of 2×10⁶ cells/well. The process was repeated 24 h later, the bone marrow cells were harvested following another 24 h and 2-4×10⁶ cells in 200μl of PBS were injected I.V. into the tail vein of sub-lethally irradiated (300 rads) IL2Rγ_c^-/-/Rag2^-/- mice. Repopulation of the recipients was examined at 3-10 weeks post-BMT by staining peripheral blood with α-NK1.1 and α-CD3. 8-12 weeks post-BMT the mice were sacked and peripheral blood, spleens, livers and bone marrow were stained for FACS analysis as below.

Cell Preparation and FACS

Livers were placed in filter bags and stomached for 2 min, washed and centrifuged through a Percoll gradient (40%:80%) for 20 min. The mononuclear cell band was harvested. Splenocytes were mashed through 100μm nylon cell strainers and bone marrow cells were irrigated from femurs with RPMI. Erythrocytes in spleen, BM and blood were lysed in ACK buffer. Cells from all tissues were washed in PBS, pre-incubated with 2.4G2 in 2% mouse serum to block Fc receptor binding, followed by incubation with Abs specific for NK1.1 (PK136), CD122 (5H4), CD117 (2B8), CD127 (A7R34), CD49b (DX5), 2B4 (244F4), CD27 (LG.7F9), Ly49A/D (12A8), Ly49C/I/F/H (14B11) from eBioscience, or CD3 (145-2C11), CD11b (M1/70), Ly49C/I (5E6) from BD Biosciences or Ly49D (4E5) (McVicar et al., 1998). Cells were fixed with BD Cytofix and analysed on BD FACSort.

IFNγ Measurement

Splenocytes were cultured in IL2 (1000U/ml) in RPMI for 24 h, harvested and plated onto wells pre-coated with 1μg/ml anti-NK1.1 (PK136) or control IgG in complete medium. Sub-optimal concentrations of IL12 (0.1ng/ml) and IL18 (1ng/ml) were added to samples followed by addition of 1μg/ml Brefeldin A. After 6 h of culture at 37°C, cells were harvested and Fc receptors were blocked with 2.4G2 in 2% mouse serum followed by staining with PE-conjugated anti-DX5 and PerCPCy5.5-conjugated anti-CD3 or PE- and PerCPCy5.5-conjugated isotype controls. The cells were washed and fixed and permeabilized using BD Cytofix/Cytoperm solution. Fc receptor was blocked by incubation with 2.4G2 and stained with APC-conjugated anti-IFNγ (XMG1.2) (BD Biosciences). DX5⁺CD3^- cells were gated and analyzed for intracellular IFNγ staining.

Western Blots

Phoenix cell transfectants were lysed with Triton X-100 lysis buffer (1% Triton X-100, 300mM NaCl, 50mM Tris pH7.4, 2mM EDTA, 1μg/ml aprotinin, 1μg/ml leupeptin, 1mM phenylmethylsulfonyl fluoride (PMSF) and 2mM sodium vanadate). Protein levels were normalized using a BCA protein assay, 4× Nupage sample buffer and DTT were added to lysates and heated for 5 min at 95°C. Lysates were separated by SDS-PAGE, transferred to PVDF (Millipore) membrane and analyzed by western blot for anti-p110γ and anti-p110δ (Millipore).

Results and Discussion

We hypothesized that reconstitution of IL2Rγ_c^-/-/Rag2^-/- (alymphoid) BM with the IL2Rγ_c chain and a gene of interest (cargo) transcribed as a single RNA should result in the development of 100% cargo⁺ NK cells. To validate this hypothesis, the IL2Rγ_c chain and EGFP, connected by an internal ribosomal entry site (IRES), were cloned into the retroviral vector pMSCV. As previous publications report conflicting results on the efficiency of an IRES in vitro (Royer et al., 2004; Kozak, 2005), we initially cloned the IL2Rγ_c chain upstream of an IRES and EGFP (γ_c:I:EGFP) into pMSCV and then reversed the order of EGFP and the IL2Rγ_c chain (EGFP:I:γ_c) in a second construct (Figure 1A). Retroviral γ_c:I:EGFP and EGFP:I:γ_c constructs were transfected into Phoenix Ecotrophic packaging cells. Virus was adhered to retronectin coated plates and used to infect cultured IL2Rγ_c^-/-/Rag2^-/- BM progenitors (Supplementary Figure 1).

Figure 1.

Retroviral constructs and characterization of splenic NK cells from WT and EGFP:I:γ_c chimeric mice. (A) Schematic structure of γ_c:I:EGFP, EGFP:I:γ_c, p110γ:I:γ_c and p110δ:I:γ_c pMSCV retroviral constructs from the 5′LTR to the 3′LTR. (B) Representative expression of developmental markers on splenic NK1.1⁺CD3^- cells from WT and EGFP:I:γ_c chimeric mice. (C) Graphs display means±(s.e.m.) with n≥3 for % of CD27 and CD11b expression on NK1.1⁺CD3^- cells. Statistical significance was calculated using the unpaired two tailed Students T test (P<0.05). (D) % expression of Ly49 receptors as in B (n≥3).

Chimeric mice were created by injecting γ_c:I:EGFP, EGFP:I:γ_c or mock infected IL2Rγ_c^-/-/Rag2^-/- BM into sub-lethaly irradiated IL2Rγ_c^-/-/Rag2^-/- mice. Retro-orbital bleeds were performed at 4 and 8 weeks following reconstitution to check for the development of NK cells and the BM, liver and blood were examined for the presence and phenotype of NK cells 8-12 weeks post-reconstitution. Mock infected IL2Rγ_c^-/-/Rag2^-/- BM did not support NK cell development while γ_c:I:EGFP or EGFP:I:γ_c infected BM resulted in robust development of NK1.1⁺CD3^- cells. Surprisingly, in γ_c:I:EGFP mice only 31%±5.2 (n=4) of BM NK cells were EGFP⁺ while EGFP:I:γ_c virus successfully reconstituted 97.2%±0.2 (n=4) EGFP⁺ NK cells. NK cells from γ_c:I:EGFP or EGFP:I:γ_c chimeric mice displayed no substantial differences in NK receptors or a panel of developmental markers including CD27 and CD11b (Hayakawa and Smyth, 2006) (Supplementary Figure 2A-C). To test the functional capabilities of these reconstituted cells we examined their potential to produce IFNγ in response to sub-optimal concentrations of IL12 and IL18 in combination with plate-bound Mouse IgG or anti-NK1.1. Reconstituted NK cells produced little IFN_γ in response to low level IL12/IL18 and this was significantly increased with NK1.1 stimulation (Supplementary Figure 2D). Thus, the EGFP:I:γ_c retroviral construct could be adapted to express constitutively active proteins, dominant negative proteins or shRNAs in primary, functional NK cells in vivo.

We next compared the development and phenotype of NK cells in EGFP:I:γ_c chimeric mice to NK cells reconstituted in IL2Rγ_c^-/-/Rag2^-/- mice from WT BM. No significant developmental differences were found in the BM or blood NK cells of these mice (Supplementary Figure 3A,B & D). Additionally, most developmental markers showed similar expression profiles on NK1.1⁺CD3^- splenocytes (Figure 1B). However, EGFP:I:γ_c and WT chimeric mice displayed differences in CD27 and CD11b expression patterns on splenic and liver NK cells suggesting slightly altered maturation levels (Figure 1C and Supplementary Figure 3C). A significant decrease in the number of cells expressing the activating receptor Ly49D was also observed in splenic EGFP:I:γ_c NK cells compared to WT cells (Figure 1D). These changes may be attributed to the underlying genetic background of alymphoid mice (which can affect expression of Ly49 family members) or alternatively to over-expression of CD132 (IL2Rγ_c chain). We therefore used EGFP:I:γ_c chimeric mice as controls for chimeric mice expressing signaling proteins.

We next examined the ability of the technique to over-express biologically relevant proteins in NK cells. PI3K is a critical enzyme in NK cell responses such as cytokine production and cytotoxicity. Knockout mice suggest that concomitant loss of both p110γ and p110δ PI3K isoforms results in a decreased NK cell population with a highly immature phenotype. However, the roles for the individual p110γ and p110δ isoforms in NK cell development and signaling remain controversial (Kim et al., 2007; Tassi et al., 2007). Tassi et al (2007) showed that p110γ^-/- mice developed a higher percentage of immature NK cells expressing significantly higher levels of CD117, CD127 and lower levels of 2B4 and Ly49C/I while p110δ^-/- mice developed normally. Kim et al (2007) found differences in the levels of various Ly49 receptors and developmental markers in both p110γ^-/- and p110δ^-/- mice. To further examine the differential roles of these lipid kinases in NK cell development and function, we expressed constitutively active p110γ or p110δ in NK precursors. Addition of a myristalation sequence to the PI3K isoforms results in their localization to the membrane thereby creating constitutively active proteins (Link et al., 2005). p110γ or p110δ attached to Src myristalation sequences were cloned in place of EGFP to create constitutively active p110γ/δ:IRES/IL2Rγ_c pMSCV (p110γ/δ:I:γ_c) constructs (Figure 1A) and chimeric mice were created in the manner described above. These large cDNA constructs (≈3kb) also test the limits of our method as they are substantially larger than EGFP.

Figure 2A demonstrates that despite their size these constructs can drive expression of EGFP, p110γ, p110δ and IL2Rγ_c. Over-expression of constitutively active PI3K p110δ results in a smaller population of NK cells than the EGFP chimeric mice, however this is likely due to the larger size of the p110 construct compared to EGFP. Over-expression of constitutively active p110γ in progenitors resulted in a further apparent reduction in the development of NK cells (Figure 2B). The NK cells in p110γ:I:γ_c chimeric mice that developed displayed a slightly less mature profile with fewer CD27^-CD11b⁺ cells than EGFP:I:γ_c and p110δ:I:γ_c chimeric mice in the spleen (Figure 2C), liver and blood (Supplementary Figure 4). However, BM cells from p110γ:I:γ_c chimeric mice had a more mature phenotype with more CD27^-CD11b⁺ NK cells than EGFP:I:γ_c mice (Supplementary Figure 5) suggesting these more mature NK cells may not circulate properly into the periphery. Similar to the results shown by Tassi et al (2007), splenic and BM p110γ NK cells also displayed significantly increased expression of the early developmental markers CD127 (IL7Rα) and CD117 (c-kit) and reduced expression of CD122, DX5 and 2B4 (Figure 2D and Supplementary Figure 5) compared to EGFP:I:γ_c and p110δ:I:γ_c NK cells. Expression of Ly49 receptors was unaffected by PI3K expression (Figure 2E). CD122 should be expressed throughout the lifespan of an NK cell (Di Santo and Vosshenrich, 2006), therefore constitutively active p110γ may partially bypass this checkpoint in development as p110γ:I:γ_c BM only contains 67.2%±1.4 CD122⁺NK cells (Supplementary Figure 5). However, mature NK cells appear to acquire CD122 as liver and blood NK cells display a normal level of CD122 expression (Supplementary Figure 4). Therefore, over-expression of constitutively active p110γ results in an altered NK cell phenotype while expression of p110δ has little effect.

Figure 2.

Characterization of splenic NK cells from p110 chimeric mice. (A) Phoenix cell transfectant lysates were immunoblotted with anti-p110δ/γ (top 2 panels). FACS analysis of EGFP and IL2Rγ_c (lower panel) on Phoenix cells. (B) Representative expression of EGFP and 2B4 on splenic NK cells from EGFP:I:γ_c, p110γ:I:γ_c and p110δ:I:γ_c chimeric mice. (C) Means±s.e.m. (n≥4) for % of CD27/CD11b expression on NK1.1⁺CD3^- cells. Statistical significance was calculated using the one-way ANOVA with Dunnetts two tailed post-test (P<0.05). Mean results with n≥3 for % of (D) developmental markers or (E) Ly49 receptors on NK1.1⁺CD3^- cells. (F) IFNγ production was measured by intracellular staining on the DX5⁺CD3^- population following stimulation with Mouse IgG or anti-NK1.1 (1μg/ml) in the presence of low-level IL12 (0.1ng/ml) and IL18 (1ng/ml).

We next examined the effect of over-expressing constitutively active p110γ/δ on NK cell function. p110γ:I:γ_c NK cells did not produce IFNγ, while p110δ:I:γ_c NK cells produced levels of IFNγ similar to EGFP:I:γ_c cells in response to low level IL12/IL18 and anti-NK1.1 (Figure 2F). The lack of IFNγ production in p110γ:I:γ_c mice may be due to the less mature phenotype of these cells. Tassi et al (2007) found that p110δ^-/- mice produced normal levels of IFNγ in response to anti-NK1.1 while p110γ^-/- mice showed severely reduced levels of IFNγ production in response to NK1.1 crosslinking. Our data together with the previous findings of Tassi et al (2007) indicate that tight regulation of p110γ is required for NK cell development and function while regulation of p110δ levels are not as important for these processes.

In summary, we have adapted an existing technique to achieve the unprecedented generation of NK cell populations, all of which express a gene of interest. We have used this technique to over-express biologically relevant proteins and we have shown that effects on development and function of NK cells can be examined using this novel system.

While the system described here has significant potential, it should be noted that the use of alymphoid BM permits examination of transduced NK development and biology only in the absence of T and B cells. Reconstitution of IL2Rγ_c^-/- (rather than alymphoid) BM would circumvent this problem but reportedly results in NK cell development in only ≈60-90% of recipients (Soudais et al., 2000) while our technique resulted in successful NK cell reconstitution in 100% of mice. Therefore the technique described here provides a powerful system to examine NK cell development and function, however if an intact immune system is required, IL2Rγ_c^-/- marrow could be used with the understanding that NK reconstitution will be limited.

A second apparent limitation of the technique described here is the lower yield of reconstituted NK cells from BM transduced with vectors harboring larger inserts. This issue could make comparison between proteins of different sizes difficult. In order to overcome this concern vectors should encode size matched control proteins such as catalytically inactive enzymes or transcription factors with mutations that ablate DNA binding. This approach should preserve the ability to directly assess the role of a variety of proteins on NK cell development and/or function.

Despite these minor limitations, the technique described here should be useful as is, or when slightly modified, for the expression of any native, constitutively active or dominant negative protein or even shRNA, therefore providing a powerful tool to advance NK cell research in the areas of development, signaling and function.