Multiple genomic defects result in an alternative RNA splice creating a human gamma H chain disease protein

Heavy chain diseases (HCD) are human lymphoproliferative disorders in which a clonal B cell population produces Ig molecules made of truncated H chains without associated L chain. We characterized the rearranged H chain gene and its mRNA from the leukemic cells of a patient (RIV) with gamma-HCD. The abnormal RIV serum Ig consisted of shortened, dimeric gamma 1-chains which had an amino terminus within the hinge region. RIV lymphoblasts possessed a foreshortened (1200 bp) gamma 1-mRNA which had sequences for only the leader, hinge, second, and third constant region domains (CH2 + CH3), but lacked variable (VH) and CH1 information. Sequence of the productive gamma 1 allele revealed it had undergone VH-JH and H chain class switch recombinations. However, normal RNA splice sites had been eliminated by a DNA insertion/deletion (VH acceptor site), mutations (JH donor site), or a large deletion (CH1 region). Inserted sequences were of non-Ig and apparently non-genomic origin. These DNA alterations resulted in aberrant mRNA processing in which the leader region was spliced directly to the hinge region, accounting for the HCD protein.     

Recombinant human IgG1-murine IgE chimeric Ig. Construction, expression, and binding to human Fc gamma receptors

We have constructed a set of chimeric Ig by exchanging corresponding H chain C domains between human (hu) IgG1 and murine (m) IgE. We used this set of Ig to dissect the interaction of individual Ig domains with human Fc gamma receptors. Only one of the chimeras, epsilon/C gamma 2,3 (an mIgE with C epsilon 3 and C epsilon 4 replaced by C gamma 2 and C gamma 3 from huIgG1), binds tightly to the human Fc gamma RI on U937 cells. We found that epsilon/C gamma 2,3 has only twofold lower affinity for Fc gamma RI as compared to huIgG1. The gamma/C epsilon 4 (huIgG1 with C epsilon 4 replacing C gamma 3) binds weakly to Fc gamma RI. The other chimeric Ig, epsilon/C gamma 3, epsilon/C gamma 2, and gamma/C epsilon 3, as well as mIgE do not bind detectably to Fc gamma RI. From these data we conclude that the C gamma 2 domain is crucial for binding and contains the majority of the binding site for Fc gamma RI on IgG1. The C gamma 3 domain makes a smaller contribution to the binding, and the C gamma 1 domain and the hinge region have very little effect on the Fc gamma RI-IgG1 interaction. The chimeric epsilon/C gamma 2,3 and huIgG1 both mediate the formation of rosettes between K562 cells and antigen-sensitized E with similar concentration dependences. These results suggest similar ability to bind to Fc gamma RII. The other chimeric Ig do not cause rosettes in this assay system. Hence, both C gamma 2 and C gamma 3 seem to be required for binding to Fc gamma RII, but the C gamma 1-hinge region has no detectable effect.     

IFN-gamma receptor-Ig fusion proteins. Half-life, immunogenicity, and in vivo activity.

Two mouse IFN-gamma receptor (MoIFN gamma R)-Ig fusion proteins, which were constructed for the purpose of creating new efficient mouse IFN-gamma (MoIFN-gamma) inhibitor molecules, were studied in vivo to determine their plasma half-life and immunogenicity, and to show their biologic activity. The hybrid proteins show 40-h blood persistency. They do not provoke an antibody response when injected into mice, and they are biologically active in vivo, as demonstrated by the prevention of streptozotocin-induced diabetes. The two fusion proteins are efficient MoIFN-gamma antagonists and can be used in mouse models of human diseases to investigate the role of MoIFN gamma in these pathologic states.     

Fc gamma RI and Fc gamma RII on monocytes and granulocytes are cytotoxic trigger molecules for tumor cells

As part of an effort to define the cytotoxic trigger molecules on human myeloid cells, the ability of the different Fc receptors for IgG (Fc gamma R) to mediate killing of tumor cell lines by monocytes and granulocytes was examined. This was accomplished by studying cytolysis of hybridoma cell (HC) targets bearing surface antibody directed toward the different Fc gamma R. The HC line, HC IV.3A, which bears Ig directed to the low affinity Fc gamma R (Fc gamma RII) on monocytes and neutrophils was lysed by human monocytes. The extent of lysis of HC IV.3A was approximately equal to that of anti-Fc gamma RI (the high affinity Fc gamma R on human monocytes) bearing HC lines (HC 32.2A and HC 62A) and was not augmented by treatment of the monocytes with interferon-gamma (IFN-gamma). In contrast, neutrophils lysed HC IV.3A and HC 32.2A only after activation with IFN-gamma. Since Fc gamma RI is not detectable on untreated neutrophils and is induced by IFN-gamma on these cells, lysis of HC 32.2A by IFN-gamma-activated neutrophils correlated with receptor induction. On the other hand, Fc gamma RII was present at equal levels on untreated and IFN-gamma-treated neutrophils, but only IFN-gamma-treated neutrophils mediated cytotoxicity via Fc gamma RII. In this case, enhanced killing appeared to be due to events other than an increase in Fc gamma RII number. Neither untreated nor IFN-gamma-treated neutrophils mediated the lysis of the anti-Fc gamma RIII bearing HC 3G8A. Thus, binding to the tumor target via this Fc receptor does not lead to lysis and may initiate signals distinct from those triggered through Fc gamma RI or Fc gamma RII. Surprisingly, HC bearing high amounts of mouse IgG1 antibody of irrelevant specificity were also lysed by monocytes. This lysis was blocked by soluble IV.3 antibody and thus appeared to be due to binding of the Fc portion of the surface Ig to Fc gamma RII on monocytes. Furthermore, monocytes from donors with a form of Fc gamma RII incapable of binding aggregated mouse IgG1 did not lyse these HC, but displayed normal lysis of HC IV.3, demonstrating that this structurally different Fc gamma RII remained a functional trigger molecule. Overall, these studies have demonstrated the specificity of Fc receptors in triggering monocyte- and granulocyte-mediated antibody-dependent tumor cell killing and have begun to dissect functional similarities and differences among the three defined Fc gamma R on human myeloid cells.     

Cloning and complete nucleotide sequence of mouse immunoglobulin gamma 1 chain gene.

The 6.6 kb DNA fragment coding for the immunoglobulin γ1 chain was cloned from newborn mouse DNA using λgtWES·λB as the EK2 vector. The complete nucleotide sequence (1823 bases) of the γ1 chain gene was determined. The cloned gene contained the entire constant region gene sequence as well as the poly(A) addition site, but not the variable region gene. The results indicate that the variable and constant region genes of immunoglobulin heavy chain are separated in newborn mouse DNA. The constant region genes of other gamma chains (that is, γ2a, γ2b and γ3) are not present in the cloned DNA fragment. The sequence demonstrates that the γ1 chain gene is interrupted by three intervening sequences at the junction of the domains and the hinge region, as previously shown in the γ2b and α chain genes and in the γ1 chain gene cloned from myeloma. The results suggest that the intervening sequence was introduced into the heavy chain gene before divergence of the heavy chain classes, and also support the hypothesis that the splicing mechanism has facilitated the evolution of eucaryotic genes by linking duplicated domains or prototype peptides not directly adjacent to one another. Comparison of the nucleotide sequence of the γ1 chain gene around the boundaries of the coding and intervening sequences with those of other mouse genes revealed extensive divergence, although short prevalent sequences of AG-GTCAG at the 5′ border of the intervening sequence and TCTGCAG-GC at the 3′ border were deduced. A limited homology of nucleotide sequences was found among domains and between the hinge region and the 5′ portion of the CH2 domain. Comparison of 3′ untranslated sequences from the γ1 and γ2b chain genes and the mouse major β-globin gene shows significant homology and a palindrome sequence surrounding the poly(A) addition site.     

Human Fc gamma RI and Fc gamma RII interact with distinct but overlapping sites on human IgG.

Cellular receptors for IgG (Fc gamma R) mediate important protective functions. By using site-specific mutants of a chimeric antibody (mouse V H domain and L chain; human IgG3 C H domains), we have demonstrated that human Fc gamma RI interacts with a site in the lower hinge of human IgG (residues 234 to 237) and that this interaction dictates Fc gamma RI-mediated superoxide generation. Mutations at position 235 resulted in the most profound reductions in Fc gamma RI recognition. We have also mapped an interaction site for Fc gamma RII to the same region; however, mutations at position 234 and 237 resulted in the greatest reductions in Fc gamma RII recognition. The two receptors appear to recognize overlapping but nonidentical sites on the lower hinge of IgG. Deviations from the optimal motif 234-Leu-Leu-Gly-Gly-237 may then explain the human IgG subclass specificity profile for human Fc gamma RI and Fc gamma RII.     

Chimeric Fc receptors identify functional domains of the murine high affinity receptor for IgG.

Chimeric Fc gamma R have been generated between the mouse high affinity receptor for IgG (Fc gamma RI) and the low affinity receptor for IgG (Fc gamma RII) by exchanging the first two domains of the three-domain extracellular structure of Fc gamma RI with the homologous two-domain extracellular structure of Fc gamma RII. Studies of the affinity and specificity of binding of mouse Ig classes to these receptors defined functional regions of Fc gamma RI and showed some surprising results. After removal of the third extracellular domain of Fc gamma RI, the remaining two domains (domains 1 and 2) retained the capacity to bind Ig in the form of immune complexes, however, they bound monomeric IgG2a with a reduced affinity. Surprisingly, these two domains in the absence of the third domain bound not only IgG2a but also IgG1 and IgG2b, i.e., the third domain of Fc gamma RI suppresses the intrinsic capacity of the first two domains to act as a low affinity Fc gamma RII-like molecule. Linking the third extracellular domain of Fc gamma RI to the two extracellular domains of Fc gamma RII resulted in a receptor that retained the specificity and affinity of Fc gamma RII. Thus, the removal of domain 3 from Fc gamma RI resulted in the conversion of Fc gamma RI to an "Fc gamma RII-like" receptor. These findings indicate that domains 1 and 2 of Fc gamma RI form an Ig-binding motif, and although domain 3 is not essential for Fc binding by Fc gamma RI, it plays a crucial role in determining the specific high affinity interaction of Fc gamma RI with IgG2a.     

Molecular cloning of rabbit gamma heavy chain mRNA.

A cDNA library of rabbit spleen mRNA was screened for immunoglobulin heavy chain sequences. In this paper we report the nucleotide sequence of two cDNA clones containing part of the constant region of the rabbit gamma heavy chain mRNA. The sequence encodes part of the CH2 domain (amino acids 268 to 340), the entire CH3 domain (amino acids 341 to 447) and the 3' untranslated region. This nucleotide sequence has been compared to the corresponding sequences of mouse gamma 1, gamma 2a and gamma 2b genes. The homologies between rabbit gamma chain gene sequence and each of the mouse gamma chain gene sequences are of the same magnitude order. This comparison shows that the CH2 domains are more homologous to each other than CH3 domains or 3' untranslated sequences. The presence of species specific nucleotide positions suggests that mouse gamma chain genes could have evolved from a common ancestor shortly after the mouse-rabbit species separation. Genomic blot analysis of rabbit liver DNA with the rabbit C gamma probes shows a limited number of related sequences, with little restriction site polymorphism between individual rabbits.     

Evidence for a critical role for the cytoplasmic region of the interleukin 2 (IL-2) receptor gamma chain in IL-2, IL-4, and IL-7 signalling.

The high-affinity interleukin 2 receptor (IL-2R) consists of at least three distinct subunits: the IL-2R alpha chain (IL-2R alpha), beta chain (IL-2R beta), and gamma chain (IL-2R gamma). It has been shown that the cytoplasmic region of IL-2R beta, but not of IL-2R alpha, is essential for IL-2 signalling to the cell interior. In the present study, we examined the functional role of the IL-2R gamma cytoplasmic region in the IL-3-dependent mouse hematopoietic cell line BAF-B03, which expresses the endogenous IL-2R alpha and IL-2R gamma, or its subline F7, which additionally expresses human IL-2R beta cDNA. We show that overexpression of a mutant IL-2R gamma, lacking all but 7 amino acids of its cytoplasmic region, results in the selective inhibition of IL-2-induced c-fos gene activation and cellular proliferation in F7 cells. When two chimeric receptor molecules in which the cytoplasmic regions of IL-2R beta and IL-2R gamma had been swapped with each other (IL-2R beta/gamma and IL-2R gamma/beta) were coexpressed in BAF-B03, the cells responded to IL-2. These results indicate the critical importance of the IL-2-induced functional cooperation of the two cytoplasmic regions. Finally, we provide evidence that the IL-2R gamma cytoplasmic region is also critical for the IL-4 and IL-7-induced growth signal transduction in BAF-B03.     

Cloning and sequencing of immunoglobulin variable-region genes using degenerate oligodeoxyribonucleotides and polymerase chain reaction

A procedure is described for using the polymerase chain reaction (PCR) to amplify and clone the cDNA from mouse immunoglobulin (Ig) variable (V) regions. This method uses a set of universal 5'-oligodeoxyribonucleotide primers that are degenerate and allow for the amplification of Ig V-region sequences from gamma and mu heavy chains and from kappa light chains. Selective first-strand cDNA synthesis is performed using Ig constant region primers and then a PCR is achieved by using the appropriate universal 5'-primer. The universal Ig heavy-chain primer was used to amplify the V-region cDNA from gamma and mu isotypes and the universal light-chain primer was used to amplify three separate kappa light V-region sequences. This procedure was used to obtain Ig V-region gene sequences from hybridomas secreting IgG1/kappa, IgG2b/kappa and IgM/kappa isotypes.     

Studies on the Fc gamma receptor of the murine macrophage-like cell line P388D1. II. Binding of human IgG subclass proteins and their proteolytic fragments.

Binding studies of human IgG proteins to murine P388D1 cells indicated that they bind to an apparently homogeneous Fc receptor population. The association constant was 0.89 x 10(6)M-1 at 22 degrees C and was comparable to the binding affinities of homologous murine IgG2a and IgG2b. The number of receptor sites was found to be approximately 6 x 10(5)/cell. Fc gamma 1 and Fc gamma 3 fragments bound with an affinity comparable to that of the parent proteins. The P388D1 receptors could discriminate between the human IgG subclasses; the relative cytophilic activity was IgG3 greater than IgG1 greater than IgG4 and IgG2 was devoid of binding activity. Fragments corresponding to the C gamma 2 and C gamma 3 domains of human IgG1 were both unable to bind to the P388D1 receptors either alone or in equimolar combination. This suggests that the cytophilic site may be formed cooperatively by interaction between the two domains. The integrity of the hinge region appeared to be essential for full expression of cytophilic activity since reduction of the hinge-region disulfides in both human IgG1 and its Fc fragment markedly decreased their binding affinity. In addition, a mutant IgG1 molecule lacking the hinge region was significantly less cytophilic than its normal counterpart.     

Recombinant 55-kDa tumor necrosis factor (TNF) receptor. Stoichiometry of binding to TNF alpha and TNF beta and inhibition of TNF activity.

The extracellular domain of the 55-kDa TNF receptor (rsTNFR beta) has been expressed as a secreted protein in baculovirus-infected insect cells and Chinese hamster ovary (CHO)/dhfr- cells. A chimeric fusion protein (rsTNFR beta-h gamma 3) constructed by inserting the extracellular part of the receptor in front of the hinge region of the human IgG C gamma 3 chain has been expressed in mouse myeloma cells. The recombinant receptor proteins were purified from transfected cell culture supernatants by TNF alpha- or protein G affinity chromatography and gel filtration. In a solid phase binding assay rsTNFR beta was found to bind TNF alpha with high affinity comparable with the membrane-bound full-length receptor. The affinity for TNF beta was slightly impaired. However, the bivalent rsTNFR beta-h gamma 3 fusion protein bound both ligands with a significantly higher affinity than monovalent rsTNFR beta reflecting most likely an increased avidity of the bivalent construct. A molecular mass of about 140 kDa for both rsTNFR beta.TNF alpha and rsTNFR beta.TNF beta complexes was determined in analytical ultracentrifugation studies strongly suggesting a stoichiometry of three rsTNFR beta molecules bound to one TNF alpha or TNF beta trimer. Sedimentation velocity and quasielastic light scattering measurements indicated an extended structure for rsTNFR beta and its TNF alpha and TNF beta complexes. Multiple receptor binding sites on TNF alpha trimers could also be demonstrated by a TNF alpha-induced agglutination of Latex beads coated with the rsTNFR beta-h gamma 3 fusion protein. Both rsTNFR beta and rsTNFR beta-h gamma 3 were found to inhibit binding of TNF alpha and TNF beta to native 55- and 75-kDa TNF receptors and to prevent TNF alpha and TNF beta bioactivity in a cellular cytotoxicity assay. Concentrations of rsTNFR beta-h gamma 3 equimolar to TNF alpha were sufficient to neutralize TNF activity almost completely, whereas a 10-100-fold excess of rsTNFR beta was needed for similar inhibitory effects. In view of their potent TNF antagonizing activity, recombinant soluble TNF receptor fragments might be useful as therapeutic agents in TNF-mediated disorders.     

Circulating cellfree Fc gamma 2b/gamma 1 receptor in normal mouse serum: its detection and specificity

By using intact or Fab fragments of rat monoclonal antibodies against murine Fc gamma 2b/gamma 1 receptor, in a solid phase radioimmunoassay, we demonstrated the occurrence of circulating cellfree Fc gamma 2b/gamma 1 receptors (Cf-Fc gamma 2b/gamma 1R) in normal mouse serum. These Cf-Fc gamma 2b/gamma 1R were removed from serum by affinity chromatography by using Sepharose columns coupled with IgG but not by Sepharose coupled with F(ab')2 fragments. Furthermore, the material retained by and eluted from the Sepharose IgG column reacted with the monoclonal antibody; these results support a direct relationship between the antigenic and the functional Cf-Fc gamma 2b/gamma 1R that were detected in serum. This Cf-Fc gamma 2b/gamma 1R was found in all the 39 normal mouse sera that were tested. The results seemed to indicate that aging may be associated with increased levels of Cf-Fc gamma 2b/gamma 1R and that levels of circulating Fc gamma R may be under genetic regulation. By forming complexes with circulating IgG within the blood stream, such Cf-Fc gamma 2b/gamma 1R may modulate some of the functions in which the Fc portion of Ig is involved.     

Structure and mapping of the gene encoding mouse high affinity Fc gamma RI and chromosomal location of the human Fc gamma RI gene.

We describe the isolation and characterization of the gene encoding the mouse high affinity Fc receptor Fc gamma RI. Using a mouse cDNA Fc gamma RI probe four unique overlapping genomic clones were isolated and were found to encode the entire 9 kb of the mouse Fc gamma RI gene. Sequence analysis of the gene showed that six exons account for the entire Fc gamma RI cDNA sequences including the 5'- and 3'-untranslated sequences. The first and second exons encode the signal peptide; exons 3, 4, and 5 encode the extracellular Ig binding domains; and exon 6 encodes the transmembrane domain, the cytoplasmic region, and the entire 3'-untranslated sequence. This exon pattern is similar to Fc gamma RIII and Fc epsilon RI but differs from the related Fc gamma RII gene which contains 10 exons and encodes the b1 and b2 Fc gamma RII. Southern blot analysis had shown that the mouse Fc gamma RI gene is a single copy gene with no RFLP in inbred strains of mice, but analysis of an intersubspecies backcross of mice showed that unlike other mouse FcR genes which are on mouse chromosome 1 the locus encoding Fc gamma RI, termed Fcg1, is located on chromosome 3. Interestingly, the Fcg1 locus is located near the end of a region with known linkage homology to human chromosome 1. Analysis of human x rodent somatic cell hybrid cell lines indicates that the human FCG1 locus encoding the human Fc gamma RI maps to chromosome I and therefore possibly linked to other FcR genes on this chromosome. These results suggest that the linkage relationships among these genes in the human genome are not preserved in the mouse.     

DNA rearrangements affecting both variable and constant regions of Ig H chain genes in MPC11 mouse myeloma variants

We have examined the mechanisms that account for short Ig H chain production in two variants of the mouse myeloma cell line MPC11 (IgG2b, kappa) by mRNA sequencing and restriction enzyme mapping. One variant, F5.5, has a thymidine residue inserted into the (CH3) domain, of the Ig H chain, resulting in premature termination and translation of a gamma 2b H chain of 50,000 m.w. A second variant, E5.7A12, contains gamma 2a-derived sequences that extend from near the 3' end of the CH2 domain to the intervening sequence between the CH2 and CH3 domains, consistent with a microrecombination event (defined as either a double cross-over or gene conversion event). In this variant, the 5' end of the CH3 domain has been deleted, but the remainder of the gamma 2b(CH3) domain is present, resulting in the translation of a gamma 2b-gamma 2a-gamma 2b H chain of 52,000 m.w. Additional rearrangements affecting sequences in or adjacent to the variable region accompany H chain constant region alterations in these cell lines and subclones of these cell lines. In F5.5, novel sequences have recombined within one of two duplicated copies of the VH gene. In a sister clone of E5.7A12 that has ceased H chain production (E5.7A14), new sequences have recombined within 300 bp 5' of the enhancer element. Both F5.5 and E5.7A12, like their immediate unstable precursor cells, fail to assemble H-H dimers, halting the Ig assembly process at the heavy-light stage, and do not secrete H chains. We speculate that defects in H chain assembly and secretion, as exemplified by the parents of these variants (i.e., intermediates of these secondary variants), reactivate the Ig gene rearrangement machinery and result in the formation of these putatively equally unstable secondary variants.     

The nucleotide sequence of a human immunoglobulin C gamma1 gene

We report the nucleotide sequence of a gene encoding the constant region of a human immunoglobulin gamma 1 heavy chain (C gamma 1). A comparison of this sequence with those of the C gamma 2 and C gamma 4 genes reveals that these three human C gamma genes share considerable homology in both coding and noncoding regions. The nucleotide sequence differences indicate that these genes diverged from one another approximately 608 million years ago. An examination of hinge exons shows that these coding regions have evolved more rapidly than any other areas of the C gamma genes in terms of both base substitution and deletion/insertion events. Coding sequence diversity also is observed in areas of CH domains which border the hinge.     

Efficient neutralization and disruption of rhinovirus by chimeric ICAM-1/immunoglobulin molecules.

The intercellular adhesion molecule 1 (ICAM-1) is used as a cellular receptor by 90% of human rhinoviruses (HRVs). Chimeric immunoadhesin molecules containing extracellular domains of ICAM-1 and constant regions of immunoglobulins (Igs) were designed in order to determine the effect of increased valency, Ig isotype, and number of ICAM-1 domains on neutralization and disruption of rhinovirus structure. These immunoadhesins include ICAM-1 amino-terminal domains 1 and 2 fused to the hinge and constant domains of the heavy chains of IgA1, IgM, and IgG1 (IC1-2D/IgA, -/IgM, and -/IgG). In addition, all five extracellular domains were fused to IgA1 (IC1-5D/IgA). Immunoadhesins were compared with soluble forms of ICAM-1 containing five and two domains (sICAM-1 and ICI-2D, respectively) in assays of HRV binding, infectivity, and conformation. In prevention of HRV plaque formation, IC1-5D/IgA was 200 times and IC1-2D/IgM and IC1-2D/IgA were 25 and 10 times more effective, respectively, than ICAM-1. The same chimeras were highly effective in inhibiting binding of rhinovirus to cells and disrupting the conformation of the virus capsid, as demonstrated by generation of approximately 65S particles. The results show that the number of ICAM-1 domains and a flexible Ig hinge are important factors contributing to the efficacy of neutralization. The higher efficiency of chimeras that bound bivalently in disrupting HRV was attributed to higher binding avidity. The IC1-5D/IgA immunoadhesin was effective at nanomolar concentrations, making it feasible therapy for rhinovirus infection.     

Structural analysis of the murine IgG3 constant region gene.

We have determined the complete sequence of the gamma 3 heavy chain constant (C gamma 3) region gene of the BALB/c mouse including the 5''-flanking region up to the switch site and the 3''-flanking region past the end of the membrane exons. The C gamma 3 coding region, typical of other IgGs, is divided into six exons corresponding to the protein domains (C gamma (3)1, hinge, C gamma (3)2, and C gamma (3)3) and to the membrane carboxyl terminus (M1 and M2). The predicted amino acid sequence of the gamma 3 chain has three potential N-linked carbohydrate addition sites (including one in the membrane spacer segment), as compared with a single occurrence in the other mouse IgGs. Between the switch recombination region and the body of the C gamma 3 gene, there is a remarkable homology with a sequence between C mu and C delta which provides a rationale for an alternative, T cell-independent class-switch mechanism. We have used a computer to analyze the secondary structure of the gamma 3 mRNA precursor for the membrane form. We predict that this RNA precursor (approximately 12 000 bp) folds into four leaf-like domains which correspond to the variable region, the large IVS, the body of the constant region, and the membrane exons. This organization may have a role to play in the function of the mRNA precursor.