NANOG Metabolically Reprograms Tumor-Initiating Stem-like Cells through Tumorigenic Changes in Oxidative Phosphorylation and Fatty Acid Metabolism

Stem cell markers, including NANOG, have been implicated in various cancers; however, the functional contribution of NANOG to cancer pathogenesis has remained unclear. Here, we show that NANOG is induced by Toll-like receptor 4 (TLR4) signaling via phosphorylation of E2F1 and that downregulation of Nanog slows down hepatocellular carcinoma (HCC) progression induced by alcohol western diet and hepatitis C virus protein in mice. NANOG ChIP-seq analyses reveal that NANOG regulates the expression of genes involved in mitochondrial metabolic pathways required to maintain tumor-initiating stem-like cells (TICs). NANOG represses mitochondrial oxidative phosphorylation (OXPHOS) genes, as well as ROS generation, and activates fatty acid oxidation (FAO) to support TIC self-renewal and drug resistance. Restoration of OXPHOS activity and inhibition of FAO renders TICs susceptible to a standard care chemotherapy drug for HCC, sorafenib. This study provides insights into the mechanisms of NANOG-mediated generation of TICs, tumorigenesis, and chemoresistance through reprogramming of mitochondrial metabolism.

The trimethylguanosine cap structure of U1 snRNA is a component of a bipartite nuclear targeting signal

The ability of series of U1 snRNAs and U6 snRNAs to migrate into the nucleus of Xenopus oocytes after injection into the cytoplasm was analyzed. The U snRNAs were made either by injecting U snRNA genes into the nucleus of oocytes or, synthetically, by T7 RNA polymerase, incorporating a variety of cap structures. The results indicate that nuclear targeting of U1 snRNA requires both a trimethylguanosine cap structure and binding of at least one common U snRNP protein. Using synthetic U6 snRNAs, it is further demonstrated that the trimethylguanosine cap structure can act in nuclear targeting in the absence of the common U snRNP proteins. These results imply that U snRNP nuclear targeting signals are of a modular nature.

The 3'-untranslated region of hepatitis C virus RNA enhances translation from an internal ribosomal entry site

Translation of most eukaryotic mRNAs and many viral RNAs is enhanced by their poly(A) tails. Hepatitis C virus (HCV) contains a positive-stranded RNA genome which does not have a poly(A) tail but has a stretch of 98 nucleotides (X region) at the 3'-untranslated region (UTR), which assumes a highly conserved stem-loop structure. This X region binds a polypyrimidine tract-binding protein (PTB), which also binds to the internal ribosome entry site (IRES) in HCV 5'-UTR. These RNA-protein interactions may regulate its translation. We generated a set of HCV RNAs differing only in their 3'-UTRs and compared their translation efficiencies. HCV RNA containing the X region was translated three- to fivefold more than the corresponding RNAs without this region. Mutations that abolished PTB binding in the X region reduced, but did not completely abolish, enhancement in translation. The X region also enhanced translation from another unrelated IRES (from encephalomyocarditis virus RNA), but did not affect the 5'-end-dependent translation of globin mRNA in either monocistronic or bicistronic RNAs. It did not appear to affect RNA stability. The free X region added in trans, however, did not enhance translation, indicating that the translational enhancement by the X region occurs only in cis. These results demonstrate that the highly conserved 3' end of HCV RNA provides a novel mechanism for enhancement of HCV translation and may offer a target for antiviral agents.

Diversity in the signals required for nuclear accumulation of U snRNPs and variety in the pathways of nuclear transport

The requirements for nuclear targeting of a number of U snRNAs have been studied by analyzing the behavior of in vitro-generated transcripts after microinjection into the cytoplasm of Xenopus oocytes. Like the previously studied U1 snRNA, U2 snRNA is excluded from the nucleus when it does not have the 2,2,7mGpppN cap structure typical of the RNA polymerase II (pol II)-transcribed U snRNAs. Surprisingly, two other pol II-transcribed U snRNAs, U4 and U5, have a much less stringent requirement for the trimethyl cap structure. The gamma-monomethyl triphosphate cap structure of the RNA polymerase III-transcribed U6 snRNA, on the other hand, is shown not to play a role in nuclear targeting. Wheat germ agglutinin, which is known to prevent the import of many proteins into the nucleus, inhibits nuclear uptake of U6, but not of U1 or U5 snRNAs. Conversely, a 2,2,7mGpppG dinucleotide analogue of the trimethyl cap structure inhibits transport of the pol II U snRNAs, but does not detectably affect the transport of either U6 snRNA or a karyophilic protein. From these results it can be deduced that U6 enters the nucleus by a pathway similar or identical to that used by karyophilic proteins. The composite nuclear localization signals of the trimethyl cap-containing U snRNPs, however, do not function in the same way as previously defined nuclear targeting signals.

High affinity interaction between nucleocapsid protein and leader/intergenic sequence of mouse hepatitis virus RNA

The nucleocapsid (N) protein of mouse hepatitis virus (MHV) is the major virion structural protein. It associates with both viral genomic RNA and subgenomic mRNAs and has structural and non-structural roles in replication including viral RNA-dependent RNA transcription, genome replication, encapsidation and translation. These processes all involve RNA-protein interactions between the N protein and viral RNAs. To better understand the RNA-binding properties of this multifunctional protein, the N protein was expressed in Escherichia coli as a chimeric protein fused to glutathione-S-transferase (GST). Biochemical analyses of RNA-binding properties were performed on full-length and partial N protein segments to define the RNA-binding domain. The full-length N protein and the GST-N protein fusion product had similar binding activities with a dissociation constant (K(d)) of 14 nM when the MHV 5'-leader sequence was used as ligand. The smallest N protein fragment which retained RNA-binding activity was a 55 aa segment containing residues 177-231 which bound viral RNA with a K(d) of 32 nM. A consensus viral sequence recognized by the N protein was inferred from these studies; AAUCYAAAC was identified to be the potential minimum ligand for the N protein. Although the core UCYAA sequence is often tandemly repeated in viral genomes, ligands containing one or more repeats of UCYAA showed no difference in binding to the N protein. Together these data demonstrate a high-affinity, specific interaction between the N protein and a conserved RNA sequence present at the 5'-ends of MHV mRNA.

Preparation, characterization, and properties of monoclonal antibodies against the lac carrier protein from Escherichia coli

Monoclonal antibodies directed against the lac carrier protein purified from the membrane of Escherichia coli were prepared by somatic cell fusion of mouse myeloma cells with splenocytes from an immunized mouse. Several clones produce antibodies that react with the purified protein as demonstrated by solid-phase radioimmunoassay and by immunoblotting experiments; culture supernatants from the clones inhibit active transport of lactose in isolated membrane vesicles. Five stable clones were selected for expansion, formal cloning, and production of ascites fluid, and the antibodies secreted in vivo by each clone also were found to inhibit lactose transport. Antibody from hybridoma 4B1, an IgG2a immunoglobulin, inhibits active transport of lactose in proteoliposomes reconstituted with purified lac carrier and in right-side-out membrane vesicles. In contrast, the antibody has no effect on the generation of the proton electrochemical gradient by membrane vesicles nor does it alter the ability of vesicles containing the lac carrier to bind p-nitrophenyl-alpha-D-galactopyranoside. In order to achieve 50% inhibition of transport activity, a 2- to 3-fold molar excess of antibody to lac carrier is required, regardless of the amount of lac carrier in the membrane. Thus, the concentration of antibody required for a given degree of inhibition is proportional to the amount of lac carrier in the membrane. Finally, antibody-induced inhibition occurs within seconds, an observation suggesting that the epitope is accessible on the surface of the membrane.

Coronavirus translational regulation: leader affects mRNA efficiency

Cells infected with the murine coronavirus, mouse hepatitis virus (MHV), show decreased host protein synthesis concomitant with an increase in viral protein synthesis. We examined the in vitro translation property of the conserved MHV 5'-leader RNA sequence by constructing chimeric mRNAs in which the 72-nt 5'-leader of M protein mRNA (A59 strain) was positioned upstream of the human alpha-globin coding region in a T7 expression vector. Synthetic 5'-capped transcripts of these mRNA constructs were translated in cell-free extracts prepared from uninfected and MHV-infected murine DBT cells. Nonviral mRNAs translated readily in both uninfected and infected cell-free extracts. By contrast, replacement of the human alpha-globin 5'-untranslated region (UR) with the MHV 5'-leader increased translation ca. three- to fourfold in cell-free extracts from MHV-infected cells versus translation in extracts from uninfected cells. Chimeric globin mRNA containing the reverse complementary sequence of the viral leader RNA in the 5'-UR showed no such increase in translation, indicating sequence specificity for the effect. A 13-nt region (-UCUAAUCCAAACA-) immediately proximal to the start codon was found to be important for the increased translation of the MHV leader-containing mRNAs. These data indicate that the apparent down-regulation of host translation is not primarily due to an inhibition of host translation but also involves a significant stimulation of viral translation in cis by a structural feature of the MHV 5'-leader RNA sequence in conjunction with a virus-specified or virus-induced factor.

Beta-globin mRNAs capped with m7G, m2.7(2)G or m2.2.7(3)G differ in intrinsic translation efficiency

We examined the intramolecular effect of altered cap structures on translation efficiency of artificial beta-globin mRNAs. For these studies, synthetic dinucleotides of the form X(5')ppp(5')G [X = 7-methyl guanosine (m7G), 2,7-dimethyl guanosine (m2(2,7)G) or 2,2,7-trimethyl guanosine (m3(2,2,7)G)], were transcriptionally incorporated into mRNAs, containing rabbit beta-globin coding sequences, using T7 RNA polymerase and a beta-globin cDNA template. These synthetic mRNAs were assayed in reticulocyte lysate for activity relative to m7G-capped mRNA. m2(2,7)G-Capped mRNA was found to be 1.5-fold more active than m7G-capped mRNA. Messenger RNA capped with m3(2,2,7)G was less active with activity of 0.24 relative to its m7G-capped counterpart (activity = 1.0). These data suggest that m7G-capped mRNAs become more active as translation templates after addition of a single N2 methyl moiety, which is especially pertinent to gene expression in togaviridae. The latter are observed to synthesize m2(2,7)G and m3(2,2,7)G-capped mRNAs in addition to m7G-capped templates during the course of infection in animal cells.

Effect of hemin on site-specific phosphorylation of eukaryotic initiation factor 2

Initiation factor 2 (eIF-2) is phosphorylated in vitro by two different cyclic nucleotide-independent protein kinases. As previously shown, a protein kinase activity that comigrates with the major casein kinase activity from rabbit reticulocytes phosphorylates eIF-2beta. In addition, a second protein kinase that specifically phosphorylates eIF-2alpha has been identified. Both protein kinase activities demonstrate cyclic nucleotide-independent activity and are not inhibited by the inhibitor protein diagnostic for cyclic AMP-regulated protein kinase activities. Phosphorylation of eIF-2alpha is almost completely inhibited by 20--35 muM hemin, whereas phosphorylation of eIF-2beta is only partially inhibited. Hemin acts by decreasing the rate of incorporation of phosphate into eIF-2alpha. The protein kinase activity that modifies eIF-2alpha has been shown to have inhibitory activity in the cell-free protein-synthesizing system, whereas the protein kinase for eIF-2beta has no effect. The identity of the former enzyme with the hemin-controlled repressor and role of hemin in the control of initiation are discussed.

Preferred translation of human hepatitis B virus polymerase from core protein- but not from precore protein-specific transcript

In the human hepatitis B virus (HBV) genome, the 5' end of the polymerase coding sequence overlaps with the 3' end of the core protein coding sequence. Recent results obtained from genetic studies have suggested that translation of HBV polymerase initiates from the first ATG codon of the polymerase reading frame and is not a result of frameshift translation from the core protein reading frame, as in the case of retroviruses. By using in vitro-synthesized SP6 RNA transcripts, we now demonstrate that HBV core protein-specific mRNA can direct the synthesis of polymerase from the internal polymerase ATG codon in rabbit reticulocyte lysates and Xenopus oocytes. A related message with an additional 60 nucleotides at the 5' end (pre-core protein mRNA) was not as efficient as the core protein mRNA for translation of polymerase. Furthermore, translation of polymerase from the core protein mRNA was not inhibited by the cap analog m7GpppG. This result, together with the results described above, indicates that translation of HBV polymerase occurs in a novel, cap-independent manner.

Involvement of miR-30c and miR-301a in immediate induction of plasminogen activator inhibitor-1 by placental growth factor in human pulmonary endothelial cells

PAI-1 (plasminogen activator inhibitor-1) is a key physiological inhibitor of fibrinolysis. Previously, we have reported PlGF (placental growth factor)-mediated transcriptional up-regulation of PAI-1 (SERPINE1) mRNA expression via activation of HIF-1α (hypoxia-inducible factor-1α) and AP-1 (activator protein-1) in HPMVECs (human pulmonary microvascular endothelial cells), which resulted in elevated PAI-1 in humans with SCA (sickle cell anaemia). In the present study, we have identified the role of post-transcriptional mechanism(s) of PlGF-mediated accumulation of PAI-1 mRNA in HPMVECs by examining the role of microRNAs (miRNAs/miRs) in PlGF-induced PAI-1 mRNA stability. Our results show reduced expression of miR-30c and miR-301a, but not of miR-99a, in response to PlGF, which have evolutionarily conserved binding sites in the 3'-UTR (3'-untranslated region) of PAI-1 mRNA. Transfection of anti-miR-30c or anti-miR-301a oligonucleotides resulted in increased PAI-1 mRNA levels, which were increased further with PlGF stimulation. Conversely, overexpression of pre-miR-30c or pre-miR-301a resulted in an attenuation of PlGF-induced PAI-1 mRNA and protein levels. Luciferase reporter assays using wild-type and mutant 3'-UTR constructs confirmed that the PAI-1 3'-UTR is indeed a direct target of miR-30c and miR-301a. Finally, plasma levels of miR-30c and miR-301a were significantly down-regulated in patients with SCA compared with normal controls. These results provide a post-transcriptional regulatory mechanism of PlGF-induced PAI-1 elevation.

A spectroscopic study of the binding of N-7-substituted cap analogues to human protein synthesis initiation factor 4E

The binding of N-7-substituted cap analogues to eIF-4E from human erythrocytes is described. Data presented here indicate that there is a correlation between the tightness of binding of these cap analogues to eIF-4E and their potency as inhibitors of protein synthesis. This result indicates that the inhibitory activity of the cap analogues is strictly a function of the affinity of the analogue for eIF-4E under equilibrium conditions. The pH dependence of binding of the cap analogues to eIF-4E indicates that the enolate form of the cap is preferred, as originally postulated by Rhoads et al. [(1983) Biochemistry 22, 6084-6088]. Data indicate that there are differences in the mode of binding of alkyl-substituted and aryl-substituted cap analogues to eIF-4E arising from favorable interactions of the phenyl ring with the guanosine moiety. These differences may explain the enhanced recognition of the aryl-substituted cap analogues by eIF-4E.

Inhibition of eukaryotic translation by nucleoside 5'-monophosphate analogues of mRNA 5'-cap: changes in N7 substituent affect analogue activity

Nucleotide cap analogues of 7-methylguanosine 5'-monophosphate (m7GMP) were synthesized in which the 7-methyl moiety was replaced with 7-ethyl (e7), 7-propyl (p7), 7-isopropyl (ip7), 7-butyl (b7), 7-isobutyl (ib7), 7-cyclopentyl (cp7), 7-(carboxymethyl) (cm7), 7-benzyl (bn7), 7-(2-phenylethyl) [7-(2-PhEt)], and 7-(1-phenylethyl) [7-(1-PhEt)]. These derivatives were assayed as competitive inhibitors of capped mRNA translation in reticulocyte lysate. We observed that N7 alkyl and alicyclic substituents larger than ethyl significantly decreased the inhibitory activity of these cap analogues presumably by decreasing their affinity for cap binding proteins, which participate in the initiation of translation. This result defined a maximum size for this class of N7 substituents in the nucleotide binding domain of cap binding proteins. Like m7GMP, the N7-substituted GMP derivatives synthesized in this study were found to be predominantly in the anti conformation as determined by proton NMR analyses. However, bn7GMP and 7-(2-PhEt)GMP, which have aromatic N7 substituents, were more effective than m7GMP as competitive inhibitors of translation. The increased affinity of bn7GMP for cap binding proteins was further examined by synthesis of beta-globin mRNA containing 5'-bn7G, 5'-m7G, or 5'-e7G cap structures. These modified mRNAs were tested as translation templates. Messenger RNA capped with bn7G was observed to increase the translation activity of the template 1.8-fold relative to that of its m7G-capped mRNA counterpart. By contrast, e7G-capped mRNA was 25% less active than m7G-capped mRNA.2+V photo-cross-linking of m7G-capped mRNA to cap binding proteins

Inhibition of eukaryotic translation by analogues of messenger RNA 5'-cap: chemical and biological consequences of 5'-phosphate modifications of 7-methylguanosine 5'-monophosphate

New analogues of 7-methylguanosine 5'-monophosphate (m7GMP) were synthesized with modified 5'-phosphate moieties by replacement of -O with -H, -CH3, or -NH2. Additional analogues were synthesized with 8-methyl- or 8-aminoguanine base substitutions or ring-opened ribose (2',3'-diol). These compounds were analyzed by 1H and 31P NMR for solution conformation. In addition, they were also analyzed for biological activity as analogues of mRNA 5'-caps by competition as inhibitors of translation in reticulocyte lysate. Substitution of oxygen on the 5'-monophosphate moiety by -H and -CH3 diminished the activity of the cap analogue as a competitive inhibitor; however, replacement by -NH2 did not diminish the activity of the analogue as an inhibitor. It was inferred from this result that cap binding proteins require a hydrogen bond acceptor as opposed to having an exclusive requirement for a second anionic group on the alpha-phosphate moiety. Inhibition results obtained with C8-substituted m7GMP analogues indicated that the 8-amino derivative was a better inhibitor than the 8-methyl derivative of m7GMP. The former is primarily anti whereas the latter is primarily syn with respect to glycosidic bond conformation. This result further supports the model that the anti conformation is the preferred form of the cap structure for interaction with cap binding proteins. The 2',3'-diol derivative of m7GMP was inactive as an inhibitor of translation.

Mouse hepatitis virus nucleocapsid protein as a translational effector of viral mRNAs

The mouse hepatitis virus (MHV) nucleocapsid protein stimulated translation of a chimeric reporter mRNA containing an intact MHV 5'-untranslated region and the chloramphenicol acetyltransferase (CAT) coding region. The nucleocapsid protein binds specifically the tandemly repeated-UCYAA- of the MHV leader. This RNA sequence is the same as the intergenic motif found in the genome RNA. Preferential translation of viral mRNA in MHV infected cells is stimulated in part by this interaction and represents a specific, positive translational control mechanism employed by coronaviruses.

Differential expression of the murine eukaryotic translation initiation factor isogenes eIF4A(I) and eIF4A(II) is dependent upon cellular growth status

The murine translation initiation factor eIF4A is encoded by two genes: eIF4A(I), expressed in all mouse tissues, and eIF4A(II), a gene preferentially expressed in organs with low proliferative capacity. To investigate the hypothesis that regulation of the eIF4A isogenes is dependent upon cellular growth status, steady state expression of eIF4A(I) and eIF4A(II) mRNAs was quantitated in asynchronous cell populations and in cultures synchronized by nutrient starvation. Our data showed that changes in cell growth state were responsible for striking differences in eIF4A isogene-specific regulation. eIF4A(I) mRNA was 10-fold more abundant than eIF4A(II) in growing cells. In growth arrested cells eIF4A(I) mRNA levels remained unchanged, whereas eIF4A(II) mRNA levels increased approximately 3-fold. Following serum stimulation of growth arrested cells, eIF4A(I) mRNA levels increased 3- to 10-fold; conversely, eIF4A(II) mRNA levels decreased 2- to 3-fold. Thus, eIF4A(I) mRNA is synthesized and translated most efficiently in growing cells while eIF4A(II) mRNA synthesis and translation is associated preferentially with the growth-arrested (quiescent) state. This difference in expression patterns likely enables the cell to maintain required levels of this factor throughout its life cycle.

Tumor necrosis factor expression during mouse hepatitis virus-induced demyelinating encephalomyelitis

Neutralizing anti-tumor necrosis factor alpha (TNF-alpha) antibody treatment of mice infected with the neurotropic JHMV strain of mouse hepatitis virus showed no reduction of either virus-induced encephalomyelitis or central nervous system demyelination. TNF-alpha-positive cells were present in the central nervous system during infection; however, TNF-alpha could not be colocalized with JHMV-infected cells. In vitro, TNF-alpha mRNA rapidly accumulated following JHMV infection; however, no TNF-alpha was secreted because of inhibition of translation. Both live and UV-inactivated virus inhibited TNF-alpha secretion induced by lipopolysaccharide. These data show that TNF-alpha is not secreted from infected cells and indicate that if contributes to either JHMV-induced acute encephalomyelitis nor primary demyelination.

MicroRNA-18a improves human cerebral arteriovenous malformation endothelial cell function

BACKGROUND AND PURPOSE

Cerebral arteriovenous malformation (AVM) is a vascular disease that disrupts normal blood flow and leads to serious neurological impairment or death. Aberrant functions of AVM-derived brain endothelial cells (AVM-BECs) are a disease hallmark. Our aim was to use microRNA-18a (miR-18a) as a therapeutic agent to improve AVM-BEC function.

METHODS

Human AVM-BECs were tested for growth factor production and proliferation under different shear flow conditions and evaluated for tubule formation. Thrombospondin-1, inhibitor of DNA-binding protein 1, and vascular endothelial growth factor (VEGF) isotype mRNA levels were quantified by quantitative real-time polymerase chain reaction. Thrombospondin-1, VEGF-A, and VEGF-D protein expression was measured using enzyme-linked immunosorbent assay. Proliferation and tubule formation were evaluated using bromodeoxyuridine incorporation and growth factor-reduced Matrigel assays, respectively.

RESULTS

miR-18a increased thrombospondin-1 production but decreased inhibitor of DNA-binding protein 1, a transcriptional repressor of thrombospondin-1. miR-18a reduced VEGF-A and VEGF-D levels, both overexpressed in untreated AVM-BECs. This is the first study reporting VEGF-D overexpression in AVM. These effects were most prominent under arterial shear flow conditions. miR-18a also reduced AVM-BEC proliferation, improved tubule formation, and was effectively internalized by AVM-BECs in the absence of extraneous transfection reagents.

CONCLUSIONS

We report VEGF-D overexpression in AVM and the capacity of miR-18a to induce AVM-BECs to function more normally. This highlights the clinical potential of microRNA as a treatment for AVM and other vascular diseases.

Hepatitis C virus has a genetically determined lymphotropism through co-receptor B7.2

B-cell infection by hepatitis C virus (HCV) has been a controversial topic. To examine whether HCV has a genetically determined lymphotropism through a co-receptor specific for the infection by lymphotropic HCV, we established an infectious clone and chimeric virus of hepatotropic and lymphotropic HCV strains derived from an HCV-positive B-cell lymphoma. The viral envelope and 5'-UTR sequences of the lymphotropic HCV strain were responsible for the lymphotropism. Silencing of the virus sensor, RIGI, or overexpression of microRNA-122 promoted persistent viral replication in B cells. By cDNA library screening, we identified an immune cell-specific, co-stimulatory receptor B7.2 (CD86) as a co-receptor of lymphotropic HCV. Infection of B cells by HCV inhibited the recall reaction to antigen stimulation. Together, a co-receptor B7.2 enabled lymphotropic HCV to infect memory B cells, leading to inhibition of memory B-cell function and persistent HCV infection in HCV-infected hosts.

Differential activation of two protease-activated protein kinases from reticulocytes by a Ca2+-stimulated protease and identification of phosphorylated translational components

Two protein kinases have been partially purified from rabbit reticulocytes and shown to be activated by limited proteolysis with trypsin [S.M. Tahara and J.A. Traugh (1981) J. Biol. Chem. 256, 11558-11564; P.T. Tuazon, W.C. Merrick, and J.A. Traugh (1980) J. Biol. Chem. 255, 10954-10958]. Reticulocyte lysate was examined for protease activities which might be involved in activation of the protein kinases in vivo. Two neutral proteases, differentially activated by Fe2+ and Ca2+, were identified and partially purified. The Ca2+-stimulated protease specifically activated protease-activated kinase II; no effect was observed on protease-activated kinase I. The Fe2+-stimulated protease was not active on either protein kinase. The protease-activated kinases were examined using initiation factors (eIF) and 40-S ribosomal subunits as substrate. Protease-activated kinase I phosphorylated one subunit of eIF-3 (Mr 130000), eIF-4B and 40-S ribosomal protein S10. Protease-activated kinase II modified the beta subunit of eIF-2 (Mr 53000) and 40-S ribosomal protein S6. The substrate specificities are unique when compared with other cAMP-dependent and cAMP-independent protein kinases from reticulocytes.