Stable Gene Silencing in Human Monocytic Cell Lines Using Lentiviral-delivered Small Interference RNA

Studying mononuclear phagocyte cell biology through genetic manipulation by non-viral transfection methods has been challenging due to the dual problems of low transfection efficiency and the difficulty in obtaining stable transfection. To overcome this problem, we developed a system for mediating RNA interference in monocytic cells. The p110α isoform of phosphoinositide 3-kinases (PI3Ks) was silenced using a lentiviral vector expressing short hairpin RNA (shRNA). This resulted in the generation of stable THP-1 and U-937 monocytic cell lines deficient in p110α. Notably, p110α was silenced without affecting levels of either the other class IA PI3K catalytic subunits p110β and p110δ, or the p85α regulatory subunit. The role of p110α in mediating cell adherence was examined. Monocyte adherence induced in response to either lipopolysaccharide (LPS) or 1α,25-dihydroxycholecalciferol (D3) was blocked by the PI3K inhibitor LY294002. However, although adherence induced in response to D3 was sensitive to silencing of p110α, LPS-induced adherence was not. Expression of the monocyte differentiation marker CD11b was also induced by D3 in a PI3K-dependent manner and gene silencing using shRNA showed that p110α was also required for this effect. Taken together, these findings demonstrate that LPS and D3 use distinct isoforms of class IA PI3K to induce functional responses and that lentiviral-mediated delivery of shRNA is a powerful approach to study monocyte biology.

causation, and developing novel therapeutics, there continues to be a great deal of interest in understanding how the responses of these cells are regulated. However, study of monocyte and macrophage biology through genetic manipulation by non-viral transfection methods has been challenging (reviewed in Ref. 1). Methods involving cationic lipid and liposome-mediated delivery of DNA or physical methods such as electroporation result in low transfection efficiency in monocytic cells, loss of viability, and the difficulty of obtaining stable transfection (2,3). An approach that has been met with greater success in monocytic cell lines is viral-mediated transduction. Although not all viruses can transduce monocytic cells efficiently, lentiviruses have been shown to do so at Ͼ90% efficiency (4 -6).
RNA interference (RNAi) 1 is a sequence-specific post-transcriptional gene silencing mechanism initiated by the introduction of double-stranded RNA (dsRNA) into target cells (7). RNAi is a natural regulatory mechanism that occurs in many organisms, including plants, Caenorhabditis elegans, Drosophila, and mammalian cells (reviewed in Ref. 7). The RNAi pathway begins by processing dsRNA into short (Ͻ30 bp) dsRNA duplexes called small interference RNA (siRNA) by a host RNase Dicer. The siRNA then becomes incorporated into a multicomponent nuclease complex called the RNA-induced silencing complex (RISC). RISC then uses the siRNA sequence as a guide to recognize cognate mRNAs for degradation.
Delivery of siRNAs into mammalian cells by transfection of siRNA or DNA vectors expressing short hairpin RNA (shRNA) has been shown to mediate RNAi successfully (8 -11). Transfection of siRNA is transient lasting only for a week or so (12), although DNA-based vectors may last longer with drug selection (10). In contrast, viral vectors have also been used to deliver siRNA successfully, and these methods tend to provide more stable gene silencing (13)(14)(15). Here we report that human monocytic cell lines can be effectively transduced using a lentiviral vector to stably silence an endogenous lipid kinase.
The phosphoinositide 3-kinases (PI3Ks) constitute a family of at least eight different lipid kinases that phosphorylate the FIG. 1. Construction of a lentiviral vector for transduction of shRNA into target cells. A, to produce the recombinant lentiviral vectors, the packaging cell line HEK 293T was co-transfected by the vector plasmid (pHR-U6-shRNA), helper plasmid (pCMV⌬R8.2), and envelope plasmid (pMD.G). The general strategy in the production of lentiviral vector-delivered siRNA is to segregate the trans-acting sequences that encode for viral proteins from the cis-acting sequences (regions recognized by viral proteins) involved in the transfer of vector sequences encoding the shRNA hydroxyl group of the inositol ring of phosphoinositides at the 3Ј position. The phosphoinositide (PI) metabolites produced as a result are known to be involved in regulating a multitude of cellular events such as mitogenic responses, differentiation, apoptosis, cytoskeletal organization, membrane traffic along the exocytic and endocytic pathways (reviewed in Ref. 16), and various other aspects of monocyte function (17)(18)(19). A considerable amount of research has led to the conclusion that this diversity of cellular control is differentially mediated by distinct PI3K isoforms (20 -22). In vitro, class I PI3K isoforms phosphorylate phosphatidylinositol (PtdIns), PtdIns 4-phosphate, and PtdIns 4,5-bisphosphate. Class I PI3K is further divided into two subclasses (I A and I B ), both of which are known to be activated by cell surface receptors. Mammalian class I A PI3Ks are heterodimers consisting of a regulatory subunit (p85␣, p85␤, p55, or other splice variants) and a p110 (␣, ␤, or ␦ isoforms) catalytic subunit (23)(24)(25). Through their Src-homology 2 domain-containing p85 subunits, class I A PI3K are recruited to and are activated by either cell surface receptors with intrinsic protein-tyrosine kinase activity or receptors coupled to Src-like protein-tyrosine kinases.
1␣,25-Dihydroxycholecalciferol (D 3 ) is a biologically active from of vitamin D and plays an important role in numerous cellular and physiological processes such as calcium homeostasis and regulates cells of the hematopoietic system (reviewed in Refs. 26 -28). For example, D 3 induces maturation markers such as CD11b and CD14 in monocytic cell lines such as THP-1, U-937, and HL-60 (18, 29 -31). In previous work from this laboratory, D 3 was observed to activate PI3K, and PI3K activity was shown to be required for the induction of CD11b and CD14 expression by D 3 (18). D 3 also induces adherence in cells of the human promonocytic cell line THP-1 (29,32,33), although it is not known whether this involves PI3K. Bacterial lipopolysaccharide (LPS) is also known to enhance adherence of leukocytes in vitro (34,35), and it also activates PI3K in monocytic cells (17,36). We and others have previously shown that LPS induces adherence in THP-1 cells (37)(38)(39)(40) and that this is PI3K-dependent (37). However, the roles of individual PI3K isoforms in this phenotype have not been defined.
To provide a basis for studying the roles of specific PI3K isoforms in regulating the function of cells of the mononuclear phagocyte series, the objective of the present study was to create stable monocytic cell lines that express shRNA targeting a specific PI3K isoform. Using this approach, we examined the role of the p110␣ isoform of class I A PI3K in mediating responses to LPS and D 3 .
Lentivirus Packaging-The packaging cell line 293T HEK (5 ϫ 10 6 ) was plated on poly-L-lysine-coated 100-mm tissue culture plates (Corning) and transfected the following day. 10 g of the transducing vector pHR (pHR-U6, pHR-p110␣1, or pHR-p110␣3), 7.5 g of the packaging vector pCMV⌬R8.2, and 2.5 g of the VSV envelope pMD.G were co-transfected by LipofectAMINE 2000 (Invitrogen), according to the manufacturer's instructions. The medium was changed the next day, and cells were cultured for another 24 h. Conditioned medium was then collected and cleared of debris by low speed centrifugation (2,500 ϫ g for 5 min) filtered through a 0.45-m filter and stored at Ϫ70°C. This collection was repeated daily for three more days, and media from the 4 days were pooled and ultracentrifuged at 100,000 ϫ g, at 4°C for 2 h. The pellet was resuspended in 500 l of medium (overnight on a nutator (reviewed in Ref. 44). The vector plasmid contained a U6 promoter-driven shRNA coding sequence, followed by a CMV-driven reporter, enhanced green fluorescent protein (EGFP). The shRNA nucleotide sequence shown is not specific and is only intended to illustrate a generic shRNA. These elements were flanked by long terminal repeats (LTRs) and also contained cis-acting sequences that allowed the vector RNA to be packaged and, subsequently, to be reverse transcribed and integrated in the target cell. The packaging sequence (⌿) was only present in the vector plasmid and not in the other two plasmids. Following transfection, the plasmids pCMV⌬R8.2 and pMD.G were transcribed by CMV promoters, and they provided the viral structural proteins in trans. These included viral integrase, protease, reverse transcriptase, capsid and matrix proteins, and vesicular stomatitis virus G protein. Together, these proteins act to trans-complement the vector by assembling the progeny viral particles, which are limited to a single round of infection. Vector proteins were produced as well, so transfected cells were GFP-positive. Conditioned medium was then harvested, concentrated by ultracentrifugation, and stored at Ϫ70°C. B, transduction of target cells was done at a multiplicity of infection of 10:1. Virions attach at the cell surface via VSV G proteins, fuse with the cell membrane, and release the viral core. Reverse transcription and uncoating of the viral core occurs in the cytoplasm. The dsDNA is then transported into the nucleus where it integrates randomly into the target cell genome. Following integration, the U6 and CMV promoters transcribe their respective genes, and this results in shRNA production and mRNA for the GFP reporter. The shRNA and mRNAs are exported to the cytoplasm. GFP is then translated, and the shRNA is processed by Dicer and then incorporated into the RNA-induced silencing complex (RISC) (reviewed in Ref. 14). RISC then targets and degrades cognate mRNAs. at 4°C), and aliquots were stored at Ϫ70°C. Viral stocks were assayed for the p24 core antigen using the Vironostika HIV-1 Antigen ELISA kit (bioMérieux, Inc., Durham, NC) according to the manufacturer's instructions.
Titration of the Lentiviral Vectors-1 ϫ 10 5 293T cells were plated in each well of a six-well plate. On the following day, cells from three wells were removed with cell dissociation solution (Sigma) and counted to determine the average number of cells at the time of titration. Three dilutions (1/50,000, 1/5,000, and 1/500) of the concentrated viral stocks were used to transduce the cells. Transduction of 293T cells was done in the presence of transduction adjuvant Polybrene (8 g/ml). The medium was changed 24 h after transduction. Cells were removed 48 h posttransduction and analyzed by flow cytometry for GFP expression. The calculation used to determine the titer was as follows: transduction units/ml ϭ (average cell number at the time of transduction ϫ % of GFP-positive cells)/100 ϫ dilution factor.
Transduction of Target Cells-1 ϫ 10 5 THP-1 or U-937 cells were seeded in each well of a 12-well plate in 500 l of complete media and transduced by lentiviral vectors at a multiplicity of infection of 10:1. Transduction was carried out in the presence of Polybrene (8 g/ml). Transduced cells were analyzed by flow cytometry after 6 days. For transduction of 293T HEK cells, 1 ϫ 10 5 cells were seeded in each well of a 12-well plate and transduced the next day.
Western Blot Analysis-Cells were washed once with phosphatebuffered saline and lysed in boiling lysis buffer (1% SDS, 50 mM Tris, pH7.4, 0.15 M NaCl, 1 mM NaF, 10 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, 1 mM EDTA) for 5 min and passed through a 27-gauge needle. Lysates were cleared by centrifugation at 12,000 ϫ g for 1 min, and protein concentration was determined using a Bio-Rad DC protein assay. Equal amounts of protein were separated by 7.5% SDS-PAGE before transfer to nitrocellulose membranes. Membranes were blocked with 5% skim milk or 3% bovine serum albumin in TBST (10 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Tween 20) for 1 h at room temperature depending on the antibody. Primary and secondary antibodies were used according to manufacturer's instructions, followed by detection with enhanced chemiluminescence technique (Amersham Biosciences, Piscataway, NJ).
Adherence Assay-An adherence assay was performed as reported previously with some modifications (37). Briefly, 96-well flat bottomed culture plates were filled with 5 ϫ 10 4 THP-1 cells in 200 l of culture medium. Cells were either treated or not with 25 M LY294002 for 30 min at 37°C, followed by LPS (12.5-500 ng/ml), D 3 (100 nM), or PMA (20 ng/ml). All treatments were done in triplicate. LPS was opsonized with 50% normal human serum for 30 min at 37°C prior to use. Plates were incubated for 24 h (for LPS treatment) or 48 h (D 3 treatment) at 37°C, 5% CO 2 . Non-adherent cells were removed by three washes with 200 l of warm culture medium (37°C). The remaining cells were then fixed with 2% paraformaldehyde/phosphate-buffered saline, for 15 min at 37°C. Fixed cells were then washed once and stained with 0.05% crystal violet in 20% methanol for 10 min at room temperature. Dye was removed, and the wells were rinsed with water three times. The plates were allowed to dry at room temperature. Cell-associated dye was eluted in 100 l per well of 100% methanol, and absorbance was measured at 570 nm in a microtiter plate reader (Bio-Rad Laboratories, Hercules, CA). Absorbance of adherent cells were normalized to PMAtreated cells.
Dual Luciferase Assay-2 ϫ 10 5 293T HEK cells were plated in each well of a 12-well culture plate and transfected the following day with 0.05 g of pGL3-SV40 and 0.02 g of pRL-TK reporter plasmids (Promega Corp., Madison, WI) using LipofectAMINE 2000 (Invitrogen). Cells were lysed 48 h later and assayed for luciferase and Renilla activity using Dual Luciferase Assay reagent (Promega). Activities are reported as ratios of Photinus pyralis GL3 luciferase (Pp-luc) to Renilla reniformis RL luciferase (Rr-luc).
Flow Cytometry Analysis-1 ϫ 10 6 cells were washed once with binding buffer (Hanks' balanced salt solution, 1% fetal calf serum, and 0.1% NaN 3 ), and then stained with anti-CD11b-RPE-conjugated antibody, or isotype-matched control-RPE-conjugated antibody for 20 min at room temperature. Cells were then washed once with binding buffer and resuspended in 1 ml of binding buffer containing 1.85% paraformaldehyde. 10,000 cells were analyzed using a BD Biosciences FACSCalibur flow cytometer. Data were acquired using BD CellQuest software and analyzed using Summit V3.1 software (Cytomation Inc., Fort Collins, CO).
Statistical Analysis-One-way ANOVA was performed on each group and followed by Tukey test for multiple comparisons. A p value of Ͻ0.05 was considered significant. All statistics and graphs were performed using Prism software, version 3.0 (GraphPad, San Diego, CA).

Construction of VSV-pseudotyped Lentiviral Vector Express-
ing shRNA-To deliver siRNA into monocytic cell lines, we developed a lentiviral-based vector that expresses a short-hairpin RNA (shRNA) (Fig. 1, A and B). Vesicular stomatitis viruspseudotyped lentiviral vectors have been shown previously to be able to transduce GFP into monocytic cell lines (4). The VSV glycoprotein G is a substitute for the lentiviral gp120/gp41 as the viral coat protein, because it broadens the range of cell types that can be infected by the virus and also helps to stabilize the virion, yielding higher titers of the virus (41). Furthermore, by stabilizing the virion, VSV G protein also allows the viral particles to be concentrated by ultracentrifugation, thereby providing higher titers for transduction (41,43). The synthesized sense and antisense oligonucleotides encoding the shRNA (Table I) were annealed and then ligated into the BamH1/BseRI site of pSHAG-1 downstream of the RNA polymerase III-specific U6 promoter (not shown) (9). The U6 promoter and the sequence encoding shRNA were then subcloned into the viral transducing vector (pHR-U6-shRNA) using Gateway cloning. The resulting plasmid was then used for transient transfection of the packaging cell line HEK 293T (Fig. 1A). Transfection of separate plasmids encoding viral structural genes (gag-pol gene products and accessory proteins on pCMV⌬R8.2, and VSV G glycoprotein on pMD.G) and nonstructural sequences (packaging sequence ⌿, Rev Responsive Element, and LTRs on pHR-U6-shRNA), ensures that progeny virus and the target cell will not contain any genes that encode viral proteins (44). Recombinant viruses were released into the medium and collected every 24 h for 4 days. Viral particles were concentrated by ultracentrifugation. Verification of viral production was done by performing p24 antigen ELISA on concentrated viral stocks. All viral samples gave high p24 levels (Ͼ Ͼ160 pg/ml). Viral stocks were titered on 293T HEK cells, which became GFP-positive when transduced. We routinely obtained 1 ϫ 10 8 to 1 ϫ 10 9 transducing units/ml.
High Efficiency Transduction of Human Monocytic Cell Lines-THP-1 and U-937 cells were either mock transduced or transduced with HR-U6, HR-p110␣1, or HR-p110␣3 viral stocks at a multiplicity of infection at 10:1. The viral vector RNA, after being reverse transcribed into dsDNA in the cytosol, entered the nucleus and randomly integrated into the genome, thus generating a stable cell line (Fig. 1B). Transduced cells expressed GFP and shRNA (GFP only in HR-U6 vector). Six days post-transduction, cells were analyzed for GFP expression using flow cytometry ( Fig. 2A). The lentiviral vectors were able to consistently transduce Ͼ90% of the target cells.
Stable and Specific Silencing of Class IA PI3K p110␣ Isoform-shRNA specific to PI3K p110␣ mRNA was used to induce RNA silencing by a mechanism involving the RISC (Fig.  1B). Transduced cells were expanded and examined by Western blotting. Fig. 2 (B and C) shows that transduction of THP-1 and U-937 with the HR-p110␣3 viral vectors resulted in nearly complete elimination of PI3K p110␣ isoform expression. In contrast, transduction of cells with either lentiviral vector HR-p110␣1 expressing the ␣1 shRNA sequence, U6-promoter con-trol virus, or mock transduction did not affect p110␣ protein levels. This result was specific in that levels of other Class I A PI3K catalytic subunits p110␤ and p110␦, or the p85␣ regulatory subunits were not affected. The stability of p110␣ silencing was confirmed by Western blotting of cells stored in liquid nitrogen for at least 8 months, and cells that had been in continuous culture for more than 6 weeks.
Stable and Specific Silencing of Luciferase Activity in HEK 293T Cells-To verify that the hairpin construct from HR-p110␣3 does not interfere with the transcription of reporter plasmids, we transduced HEK 293T cells with either HR-p110␣3 or HR-Ff1, which produces a shRNA targeting GL3 firefly luciferase. HEK 293T cells were chosen, because they are much more receptive to transfection than are monocytic cell lines. Western blot analysis demonstrated that the level of p110␣ in HEK 293T cells transduced with HR-p110␣3 viruses was similar to that observed in similarly treated THP-1 and U-937 cells, and as expected HR-Ff1 virus had no effect on p110␣ expression (data not shown). Transduced cells were then co-transfected with firefly luciferase (pGL3-SV40) and Renilla luciferase (pRL-TK) plasmids, the latter serving as an internal control. Forty-eight hours post-transfection, cells were analyzed using dual luciferase assay (Fig. 3). Cell transduction with HR-Ff1 reduced firefly luciferase expression by more than 80%, whereas HR-p110␣3 had no effect compared with mock transduced cells. These results indicate that in cells expressing shRNA targeting p110␣, nonspecific interference with the function of a reporter plasmid is not a problem. The data also show that lentiviral-mediated RNAi is capable of silencing exogenous genes driven by a strong promoter.
Monocyte Adherence Induced by D 3 , but Not LPS Is Dependent on p110␣-We have previously shown that LPS-induced adherence of THP-1 cells is dependent upon PI3K (37). However, use of pharmacological inhibitors did not permit determination of whether this involved class I A PI3K and, if so, which p110 isoform (37). To address this question, we examined LPSinduced adherence in THP-1 cells rendered p110␣-deficient using shRNA (Fig. 4A). Consistent with previous findings that PMA-induced monocyte adherence is resistant to PI3K inhibitors (37), p110␣-deficient cells showed normal PMA-induced adherence (data not shown). Using PMA-treated cells as a control, adherence induced by LPS was determined as the percentage of PMA-induced adherence by a colorimetric based adherence assay as described under "Experimental Proce-dures." Fig. 4A shows that, whereas PI3K inhibitor LY294002 significantly reduced LPS-induced adherence in all transduced cells, silencing of p110␣ did not affect LPS-induced adherence, and this was true over a range of concentrations of LPS (12.5-500 ng/ml). LPS was opsonized by human serum prior to use, however, the presence of human serum alone did not induce adherence (data not shown). Similar to LPS, LY294002 inhibited D 3 -induced adherence in all types of transduced cells (Fig.  4B). In contrast to LPS, however, monocyte adherence induced by D 3 was found to be attenuated in p110␣-deficient cells. Moreover, this inhibitory effect was comparable to that observed in LY294002-treated cells. D 3 -induced CD11b Expression Is Attenuated in p110␣-deficient Cells-We have previously shown that CD11b induction by D 3 in THP-1 cells is sensitive to inhibition by PI3K inhibitors (18). To determine whether this is p110␣-dependent, we incubated cells with LY294002 or not, followed by D 3 for 72 h after which they were collected and stained with anti-CD11b RPE-conjugated antibody or isotype matched control RPE-conjugated antibody. After washing and resuspension in binding buffer cells were analyzed by flow cytometry. Viable cells were gated and analyzed for RPE signals. For transduced cells, only GFP-positive cells were analyzed. As shown in Fig. 5, D 3induced CD11b expression was reduced significantly in p110␣deficient cells when compared with control cells (p Ͻ 0.05). Mean fluorescence intensity (MFI) index is the ratio of (MFI of D 3 -treated samples stained with anti-CD11b antibody Ϫ MFI of isotype-matched control antibody)/(MFI of anti-CD11b antibody-stained, untreated samples Ϫ MFI isotype-matched control antibody). Therefore, an MFI index of 1 indicates that there was no induction of CD11b by D 3 . To calculate % reduction of D 3 -induced CD11b expression in p110␣-deficient cells relative to D 3 -treated control cells, the following calculation was used: % reduction ϭ [(MFI index of control cells Ϫ MFI index of HR-p110␣3 cells)/(MFI index of control cells Ϫ 1)] ϫ 100. The % reduction of D 3 -induced CD11b expression in p110␣-deficient cells relative to controls cells was ϳ59 -63%, similar to that observed in LY294002-treated cells. DISCUSSION A major obstacle in studying monocyte cell biology has been the resistance of these cells to genetic manipulation, particularly when using non-viral methods. In this report, we describe a strategy for stable gene silencing in monocytic cells. Using a VSV-pseudotyped lentiviral vector, monocytic cell lines that stably expressed shRNAs targeting an endogenous gene were generated resulting in silencing of the p110␣ isoform of PI3K. Historically, lentiviral vectors have been shown to be superior to non-viral methods such as cationic lipid-mediated delivery of DNA vectors to monocytic cells because of their much higher transduction efficiency and longer period of transgene expression. Although the mechanism underlying the resistance of monocytic cells to DNA transfection is not known precisely, it has been proposed that much of the exogenous DNA enters the cell via endocytosis resulting in degradation of the DNA by abundant lysosomal nucleases (1). VSV-pseudotyped lentiviral vectors obviate this problem, because the viral core containing the genetic elements of interest is delivered directly into the cytosol after the viral envelope fuses with the plasma membrane of the target cell (44). Another physical approach for transfection that has not been very successful in monocytic cells is electroporation. Poor success here has been related to low viability of cells after electroporation, typically below 10 -22% survival (2,3).
Among the various viral-based approaches, lentiviral-based vectors seem to be the most promising for transduction of monocytic cells. Onco-retroviruses are similar to lentiviruses, but the latter have a more complex genome and, consequently, a more complex replication cycle (reviewed in Ref. 44). One advantage of lentiviral vectors over onco-retroviral vectors lies in their ability to transduce both proliferating and non-proliferating cells, such as liver, muscle, retina, and neurons (45)(46)(47). This has been attributed to the presence of nuclear localization signal sequences present in lentiviral gene products (48), which are absent from onco-retroviral vectors. Interestingly, onco-retroviruses transduce monocytic cell lines THP-1, U-937, and HL-60 at lower efficiencies (1-31%) (5, 6), even though these are proliferating cells. Furthermore, compared with onco-retroviruses, lentiviral vectors are also much less susceptible to transcriptional silencing of the viral transgene, an event that may result from methylation of foreign DNA in the vicinity of the promoter, as well as by integration of the viral elements into condensed chromatin regions (1,49). Taken together, all of the above differences make lentiviruses potentially superior vectors for the delivery of siRNAs into monocytic cells.
Other viruses such as adenoviruses and adeno-associated viruses have also been used to transduce monocytic cells. Adenoviruses do not integrate into the host genome and as a result are not useful for long term expression of the exogenous sequences (1). Adeno-associated viruses, in contrast, do integrate into the host genome and have been used successfully in transducing primary human monocytes and dendritic cells (50,51), although their efficacy in transducing human monocytic cell lines has been low (Ͻ1%) (52). By combining the ability of lentiviral vectors to stably transduce monocytic cell lines at a high efficiency and the potential for siRNA to mediate RNA interference, we have shown that stable gene silencing in human monocytic cell lines is achievable (Fig. 2). Transduced cells can be propagated under normal conditions without drug selection. The silenced phenotype was stable during 6 -8 weeks of continuous culture, and transduced cells could be used after long periods of storage in liquid nitrogen.
Despite recent progress in understanding the functions of PI3K family members in various cell types, study of distinct FIG. 4. Monocyte adherence induced by D 3 , but not LPS is dependent on p110␣. A, LPS-induced adherence. Cells were either treated or not with LY294002 (25 M) for 30 min at 37°C, before treatment with human serum-opsonized LPS (12.5-500 ng/ml). For each treatment group, PMA treatment (20 ng/ ml) was also done in parallel. After 24 h, non-adherent cells were rinsed away, and adherent cells were stained with crystal violet. Dye was then eluted with 100% methanol, and absorbance measured at 570 nm. To control for cell numbers plated per cell type, adherence is expressed as a percentage of PMA-induced adherence. PMA-induced adherence was similar among all cell types (p Ͼ 0.05, data not shown). All samples treated with LPS alone were not significantly different from each other (one-way ANOVA p ϭ 0.2827). The same applies to LY294002 plus LPS treated group (p ϭ 0.1962). All of the LY294002 plus LPS-treated samples were significantly different from LPS-treated samples (post-ANOVA Tukey test, p Ͻ 0.01 for all pairs), and the values observed were ϳ40% of the LPSonly group. B, D 3 -induced adherence. Cells were either treated or not with LY294002 (25 M) for 30 min at 37°C, before treatment with D 3 (100 nM). Cells were incubated with D 3 for 48 h before being assayed for adherence. All samples treated with LY294002 plus D 3 were not significantly different from each other (one-way ANOVA p ϭ 0.5945), whereas the D 3 treatment alone samples were significantly different from each other (oneway ANOVA p ϭ 0.0006). All of the LY294002 plus D 3 -treated samples were significantly different from cells treated with D 3 only (post-ANOVA Tukey test, p Ͻ 0.05 for all pairs), with the exception of HR-p110␣3 (p Ͼ 0.05). D 3 -induced adherence in p110␣-deficient cells is ϳ50% of control, equivalent to treatment with LY294002. All treatments were done in triplicate. Error bars indicate standard deviation, n ϭ 3.
functions of individual PI3K isoform in monocytes has been difficult due to their resistance to genetic manipulation. The PI3K inhibitors that are currently available, including LY294002 (53), wortmannin (54), and 3-methyladenine (19), are not useful in assigning function to specific isoforms, because at effective concentrations they inhibit virtually all classes of the PI3K family except for class II PI3K C2␣ (55). One way to obviate this problem has been to rescue PI3K inhibitor-induced phenotypes by delivering either class I PI3K or class III products such as PI 3,4,5-trisphosphate and PI 3-phosphate, respectively, by lipid carriers into cells (19). However, this approach cannot give information on the role of distinct class I PI3K isoforms. Another non-genetic approach to study the function of specific PI3K enzymes has been microinjection of inhibitory antibodies. This has proven to be useful for examining the roles of PI3K isoforms when combined with imaging studies of single cells, including the murine macrophage cell line J774 (56). However, not every type of cell can be subjected to this technique (57), and, due to the limited number of cells that can be studied, biochemical characterization of proteins is challenging. In addition, this method is transient in nature and cannot generate a stable phenotype for a prolonged period of time.
Genetic approaches to assigning function to individual class I PI3K p110 isoforms have been limited as well since gene knockouts of p110␣ or p110␤ in mice were found to be embryonically lethal (58,59). Consequently, it has not been possible to determine with precision the roles of these isoforms in immune cells. More recently, instead of gene knockouts, mutant p110␦ mice were created and have been a useful alternative (60). Thus far, however, no phenotypes in monocytic cells from either knockout or mutant p110␦ animals have been reported (60 -62). Due to the close and complex relationships between class I A regulatory and catalytic subunits (63)(64)(65), an ideal strategy in studying the functions of specific PI3Ks would be to reduce the expression of individual isoforms without disturbing the molecular balance of the regulatory and catalytic subunits. This has not always been achievable. For example, p110␣ embryonic knockout cells had increased p85␣ expression (58), and the massive accumulation of p85␣ monomers observed was suggested to exert a dominant negative effect on the remaining class I A p110 isoforms by binding non-productively to receptors (64,66). Conversely, in cells from p85␣ or p85␤ knockout mice, expression of p110 isoforms ␣, ␤, and ␦ was either reduced or normal depending on the cells studied (67)(68)(69)(70). In this context, lentiviral-mediated RNA silencing of p110 isoforms appears to be a superior approach, because we were able to specifically reduce p110␣ expression while not affecting levels of either other class I A p110 isoforms or p85␣ (Fig. 2, B and C). Taken together, the results shown indicate that lentiviral-delivered siRNA is an efficient method for gene silencing in monocytic cells. Furthermore, by virtue of the specificity offered by RNAi, studies of individual isoforms from protein families can be done with relative ease.
In contrast to lentiviral vector HR-p110␣3, transduction of the HR-p110␣1 vector did not reduce cellular levels of p110␣ (Fig. 2, B and C). Although HR-p110␣1 was originally designed as a candidate shRNA to mediate RNAi, the sequence did not bring about the desired result. Nevertheless, these cells served as useful controls for nonspecific effects of transduction and shRNA expression. There are several possible explanations for why this candidate siRNA might not have been effective. It has been suggested that target mRNA regions where hydrogen bonds form in secondary and tertiary structures can impede silencing (71,72). Another possible explanation for the lack of effect of the HR-p110␣1 vector may be that this construct targeted a region close to the AUG start codon. It has been suggested that these regions may be richer in sequences that bind regulatory proteins and this may limit the ability of the RISC complex to access the RNA target sequence (7,73). Nevertheless, this is not an absolute restriction, because it has been shown that targeting sequences close to the start codon may successfully induce RNA silencing (71).
Vitamin D 3 and LPS are both known to induce adherence in monocytic cells (29,32,33,37) and to activate PI3K (17,18,36). Through the use of in vitro kinase assays, we and others have previously shown that both LPS and D 3 activate PI3K in human monocytes and macrophages (17,18,36). Because in these studies the basic approach used was an anti-p85 antibody to immunoprecipitate the kinase, the results led to the conclusion that class I A PI3Ks are activated in response to LPS or D 3 , although the involvement of other PI3K family members could not be ruled out. Because p110␣ is a more robust PI3K in terms of biochemical kinetics than either p110␤ or p110␦ (60,74,75), we hypothesized that p110␣ might be the dominant class I A PI3K in mediating LPS and D 3 signaling. Based on this assumption, we predicted that by silencing PI3K p110␣ in THP-1 cells, a phenotype of diminished adherence induced by either LPS, D 3 , or both would be observed. As shown in Fig. 4  Cells were then washed once in binding buffer and stained with anti-CD11b RPE-conjugated antibody or isotype matched RPE-conjugated control antibody, according to manufacturer's instructions. After washing and resuspension in binding buffer containing 1.85% paraformaldehyde, 10,000 cells were analyzed for RPE fluorescence. For transduced cells, the analysis was restricted to GFP-positive cells. Data were collected after compensation of GFP and RPE fluorochromes on FL1 and FL2 channels. Mean fluorescence intensity (MFI) index is the ratio of (MFI of D 3 -treated samples stained with anti-CD11b antibody Ϫ MFI of isotype-matched control antibody)/(MFI of anti-CD11b antibodystained, untreated samples Ϫ MFI isotype-matched control antibody). Therefore, an MFI index of 1 indicates that there was no induction of CD11b by D 3 , or a complete inhibition by LY294002. All of the LY294002 plus D 3 -treated samples were significantly different from control cells treated with D 3 alone (post-ANOVA Tukey test, p Ͻ 0.05 for all pairs), but not from HR-p110␣3 plus D 3 alone (p Ͼ 0.05). Error bars indicate S.D., n ϭ 3. B), although adherence in response to both LPS and D 3 was sensitive to PI3K inhibitor LY294002, LPS-induced adherence was resistant to silencing of p110␣, whereas D 3 -induced adherence was not. This was examined over a range of LPS concentrations to control for the possibility that LPS might only utilize p110␣ at lower concentrations. Thus, the findings suggest that differential utilization of p110␣ reflects qualitative differences between LPS and D 3. These results lead to the somewhat surprising and interesting conclusion that these two agonists use signaling pathways in monocytes that activate distinct isoforms of class I A PI3Ks at least for some functional responses. It would appear most likely that LPS-induced adherence is mediated by p110␤ or p110␦, because LPS is known to activate class I A PI3K (17,36), and LPS-induced adherence was inhibited by LY294002 (Fig. 4A). At this point, we cannot rule out the possibility that LPS activates p110␣ for other signaling pathways not related to adherence.
Prior studies from this laboratory showed that in THP-1 cells treated with D 3 , the vitamin D receptor (VDR) associated with the p85 subunit of PI3K in a ligand-dependent manner (18). In addition, within 20 min of exposure to D 3 , a corresponding rise in PI3K activity was observed when PI3K assays were performed on either anti-p85 or anti-VDR immunoprecipitates, and PI3K activation was linked to changes in gene expression after 24 h (18). The findings in the latter and the present reports are consistent with a model in which class I A PI3K is activated through a steroid receptor. This model differs from the conventional paradigm in which class I A PI3K activation occurs downstream of transmembrane receptors such as growth factor receptors, immunoreceptors (reviewed in Ref. 76), and toll-like receptor 2 (77). However, recent progress in D 3 signaling research has resulted in the detection of a putative membrane-bound receptor (VDR mem ) that, based upon differences in binding properties, appears to be distinct from the nuclear VDR (VDR nuc ) (reviewed in Ref. 78). For example, selective binding of synthetic D 3 analogs to VDR mem and not VDR nuc has shown that the former mediates non-genomic rapid signaling effects of D 3 but not delayed classic genomic responses (79). Taken together, these findings suggest the interesting possibility that class I A PI3K may be activated by both VDR mem and VDR nuc , such that PI3K may regulate both rapid, non-genomic signaling as well as at least some delayed genomic effects of D 3 . Clearly in this model there would be ample opportunity for activation of PI3K through VDR mem to influence cellular responses to D 3 brought about through the classic VDR nuc . Further studies will be required to identify whether VDR mem complexes with and activates PI3K p85/p110␣.
The ␤ 2 integrin receptor CR3 (CD11b/CD18, ␣ M ␤ 2 ), is a marker of monocyte differentiation and can mediate adherence, phagocytosis, and leukocyte transmigration (reviewed in Ref. 80). CD11b is the ␣ subunit of CR3 and associates non-covalently with its partner CD18. Expression of CR3 is restricted to myeloid cells, and its level of expression depends on the state of differentiation with mature neutrophils and macrophages having the highest levels (81,82). THP-1 cells are known to express relatively low levels of CR3 in the basal state (83,84), and D 3 is known to augment expression further (18,29,30). We have previously reported that CR3 induction by D 3 in both THP-1 cells and human monocytes is sensitive to PI3K inhibitors (18). Using THP-1 cells made deficient in p110␣ by RNAi, we observed a significant reduction (59 to 63%) in D 3 -induced CD11b expression compared with several negative control cell populations (Fig. 5). Adherence of monocytes to plastic induced by PMA has been shown to be dependent on CR3 (85). Therefore, the defect in D 3 -induced adherence we observed in p110␣deficient THP-1 cells (Fig. 4B) may be partially due to the attenuation of CD11b expression. However, this seems unlikely to be the entire explanation, because unstimulated THP-1 cells do express basal levels of CD11b and are not adherent, hence a role for other adhesion proteins cannot be excluded in this model system.
In conclusion, we have demonstrated the ability of VSVpseudotyped lentiviral vectors to stably silence the PI3K p110␣ isoform in the monocytic cell lines THP-1 and U-937. Using transduced THP-1 cells deficient in PI3K p110␣, D 3 -induced, but not LPS-induced adherence was shown to be dependent on p110␣. D 3 -induced up-regulation of CD11b was also found to be dependent on p110␣. The ability of lentiviral vectors to transduce both dividing and non-dividing cells and stable expression of the transgene make this a versatile strategy for gene silencing based on RNAi. Moreover, the finding that lentiviral-transduced cells expressing shRNA can be further manipulated by transfection with reporter plasmids, for instance luciferase, significantly expands the utility of this approach. One important application of this technique may be in gene therapy research, such as silencing genes in myeloid leukemia cells or other difficult to transfect cells to better understand their biology and to identify potential therapeutic targets.