Antigenic determinants of the HLA-B7 molecule; Bw6- and B7-specific determinants are spatially separate

Monoclonal antibodies that bind HLA-B7 were used to show that the B7-specific determinant is at a topologically different site from that of the broad polymorphic, Bw6 determinant. The relationship to other antigenic determinants defined by monoclonal antibodies was also assessed. These results were independently obtained in four ways: (1) by cellular blocking assays, in which there was no inhibition of 125I-B7 antibody binding in the presence of Bw6 antibody and no inhibition of 125I-Bw6 antibody binding in the presence of B7 antibody; (2) cellular binding assays under conditions of antibody saturation showed the binding of B7-specific and Bw6 antibodies were additive; (3) solid-phase radioimmune assays demonstrated enhancement between B7-specific and Bw6 antibodies; (4) analysis of antigen antibody complexes by size-exclusion high pressure liquid chromatography showed Bw6 and B7 antibodies could form tetramolecular complexes with papain-solubilized HLA-B7. Limitations were encountered in using cellular blocking assays to map antigenic determinants of HLA-B7. These assays can produce blocking in cases where two antibodies are not competing for an antigenic determinant. Mapping antigenic determinants with assays using purified HLA-B7 as the antigenic target, in addition to cell-based assays, provided a more accurate picture.     

Binding of [3H]Imipramine to Human Platelet Membranes with Compensation for Saturable Binding to Filters and Its Implication for Binding Studies with Brain Membranes

Abstract: Apparent specific binding of [³H]imipramine to human platelet membranes at high concentrations of imipramine showed deviation from that expected of a single binding site, a result consistent with a low-affinity binding site. The deviation was due to displaceable, saturable binding to the glass fibre filters used in the assays. Imipramine, chloripramine, desipramine, and fluoxetine inhibited binding to filters whereas 5-hydroxytryptamine and ethanol were ineffective. Experimental conditions were developed that eliminated filter binding, allowing assay of high and low-affinity binding to membranes. Failure to correct for filter binding may lead to overestimation of binding parameters, B_max and K_D for high-affinity binding to membranes, and may also be misinterpreted as indicating a low-affinity binding component in both platelet and brain membranes. Low-affinity binding ( K_D < 2 μ M ) of imipramine to human platelet membranes was demonstrated and its significance discussed.     

On the use of cells or membranes for receptor binding: Growth hormone secretagogues

Receptor binding techniques have been widely used in different biochemical applications, with isolated membranes being the most used receptor preparation in this type of assays. In this study, intact cells were compared with isolated membranes as receptor support for radioligand receptor binding assay. The growth hormone secretagogue receptor 1a (GHSR-1a) expressed in human embryonic kidney 293 (HEK293) cells was used as a model of G-protein-coupled receptors. Differences between using intact cells in suspension and using isolated membranes were evaluated for different aspects of the receptor binding assay: total binding variations while both receptor preparations remain on ice, modifications in incubation conditions, saturation, and competition using different agonists. Intact cells are more prone to variability. Although under optimized settings both preparations were equivalent, the K_d value for intact cells was three times higher than that using isolated membranes. However, no significant differences were observed in competition assays obtaining practically identical K_i values for all ligands tested. For the GHSR-1a, isolated membranes are the better choice if particular incubation conditions are required (less variability), whereas intact cells yield easy, fast, and physiological conditions for receptor binding assays.     

Production and characterization of antibody blocking epidermal growth factor: Receptor interactions

Membranes were prepared from the human epithelioid carcinoma cell line A-431 which has approx. 2 · 10⁶ epidermal growth factor receptors per cell. This membrane preparation which retained a high epidermal growth factor binding specific activity was used as an antigen to produce antisera in rabbits. Double-immunodiffusion experiments demonstrated that the immune serum contained precipitating antibodies to several components of detergent solubilized A-431 membranes.The immonoglobulin G fraction of this immune sera inhibited ¹²⁵I-labeled epidermal growth factor binding to receptors in: (1) intact human and mouse cells; (2) membrane preparations from A-431 cells and human placenta, and (3) solubilized A-431 membranes. Inhibition of ¹²⁵I-labeled epidermal growth factor binding was observed with divalent and monovalent fragments of immunoglobulin G prepared from the immunoglobulin G fraction. Also, the immunoglobulin G fraction blocked growth factor binding to membranes at low temperature (5°C).Anti-A-431 antibody blocked the induction of DNA synthesis in quiescent fibroblasts by epidermal growth factor in a manner similar to that of anti-epidermal growth factor antibody. Addition of either anti-A-431 or anti-epidermal growth factor antibodies to fibroblasts at times up to 5 h after the addition of epidermal growth factor completely reversed the hormone's mitogenic potential. At later times (after 12 h) addition of either antibody was without effect on the stimulation of DNA synthesis by epidermal growth factor. Anti-A-431 antibody did not block the induction of DNA synthesis in fibroblasts by fibroblast growth factor or serum.     

Clearance of Human IgG1-Sensitised Red Blood Cells In Vivo in Humans Relates to the In Vitro Properties of Antibodies from Alternative Cell Lines

We previously produced a recombinant version of the human anti-RhD antibody Fog-1 in the rat myeloma cell line, YB2/0. When human, autologous RhD-positive red blood cells (RBC) were sensitised with this IgG1 antibody and re-injected, they were cleared much more rapidly from the circulation than had been seen earlier with the original human-mouse heterohybridoma-produced Fog-1. Since the IgG have the same amino acid sequence, this disparity is likely to be due to alternative glycosylation that results from the rat and mouse cell lines. By comparing the in vitro properties of YB2/0-produced Fog-1 IgG1 and the same antibody produced in the mouse myeloma cell line NS0, we now have a unique opportunity to pinpoint the cause of the difference in ability to clear RBC in vivo. Using transfected cell lines that express single human FcγR, we showed that IgG1 made in YB2/0 and NS0 cell lines bound equally well to receptors of the FcγRI and FcγRII classes but that the YB2/0 antibody was superior in FcγRIII binding. When measuring complexed IgG binding, the difference was 45-fold for FcγRIIIa 158F, 20-fold for FcγRIIIa 158V and approximately 40-fold for FcγRIIIb. The dissimilarity was greater at 100-fold in monomeric IgG binding assays with FcγRIIIa. When used to sensitise RBC, the YB2/0 IgG1 generated 100-fold greater human NK cell antibody-dependent cell-mediated cytotoxicity and had a 10³-fold advantage over the NS0 antibody in activating NK cells, as detected by CD54 levels. In assays of monocyte activation and macrophage adherence/phagocytosis, where FcγRI plays major roles, RBC sensitised with the two antibodies produced much more similar results. Thus, the alternative glycosylation profiles of the Fog-1 antibodies affect only FcγRIII binding and FcγRIII-mediated functions. Relating this to the in vivo studies confirms the importance of FcγRIII in RBC clearance.     

Spectroscopic and thermodynamic evidence for antimicrobial peptide membrane selectivity

In our laboratory we developed a series of antimicrobial peptides that exhibit selectivity and potency for prokaryotic over eukaryotic cells (Hicks et al., 2007). Circular dichroism (CD), isothermal calorimetry (ITC) and calcein leakage assays were conducted to determine the mechanism of lipid binding of a representative peptide 1 (Ac-GF-Tic-Oic-GK-Tic-Oic-GF-Tic-Oic-GK-Tic-KKKK-CONH₂) to model membranes. POPC liposomes were used as a simple model for eukaryotic membranes and 4:1 POPC:POPG liposomes were used as a simple model for prokaryotic membranes. CD, ITC and calcein leakage data clearly indicate that compound 1 interacts via very different mechanisms with the two different liposome membranes. Compound 1 exhibits weaker binding and induces less calcein leakage in POPC liposomes than POPC:POPG (4:1 mole ratio) liposomes. The predominant binding mechanism to POPC appears to be limited to surface interactions while the mechanism of binding to 4:1 POPC:POPG most likely involves some type of pore formation.     

Generation by phage display and characterization of drug-target complex-specific antibodies for pharmacokinetic analysis of biotherapeutics

Anti-idiotypic antibodies play an important role in pre-clinical and clinical development of therapeutic antibodies, where they are used for pharmacokinetic studies and for the development of immunogenicity assays. By using an antibody phage display library in combination with guided in vitro selection against various marketed drugs, we generated antibodies that recognize the drug only when bound to its target. We have named such specificities Type 3, to distinguish them from the anti-idiotypic antibodies that specifically detect free antibody drug or total drug. We describe the generation and characterization of such reagents for the development of ligand binding assays for drug quantification. We also show how these Type 3 antibodies can be used to develop very specific and sensitive assays that avoid the bridging format.

Abbreviations: BAP: bacterial alkaline phosphatase; CDR: complementarity-determining regions in VH or VL; Fab: antigen-binding fragment of an antibody; HRP: horseradish peroxidase; HuCAL®: Human Combinatorial Antibody Libraries; IgG: immunoglobulin G; LBA: ligand binding assay; LOQ: limit of quantitation; NHS: normal human serum; PK: pharmacokinetics; VH: variable region of the heavy chain of an antibody; VL: variable region of the light chain of an antibody.     

Disulphide reduction alters the immunoreactivity and increases the affinity of insulin-like growth-factor-I receptors in human placenta.

We previously identified two forms of the insulin-like growth-factor-I (IGF-I) receptor in human placenta: a lower-affinity form reactive with an autoantiserum (B-2) to the insulin receptor and a higher-affinity non-immunoreactive form [Jonas & Harrison (1985) J. Biol. Chem. 260, 2288-2294]. Evidence is now presented that the lower-affinity immunoreactive forms are convertible into higher-affinity non-immunoreactive forms via reduction of receptor disulphide bonds. Treatment of placental membranes with increasing concentrations of dithiothreitol (DTT): (1) converted native Mr-290 000 heterotetrameric IGF-I receptors into Mr-180 000 dimers (determined by chemical cross-linking of 125I-IGF-I with disuccinimidyl suberate); (2) increased 125I-IGF-I binding, owing to an increase in receptor affinity; and (3) abolished the reactivity of Triton-solubilized IGF-I receptors with antiserum B-2 and transformed the curvilinear plot of IGF-I binding to a linear form. In isolated complexes between receptor and B-2 antibody, DTT increased 125I-IGF-I binding and released a single class of higher affinity IGF-I receptors of Mr 180,000. Thus DTT-treated IGF-I receptors have similar properties to the higher-affinity non-immunoreactive forms of the native receptor, except that reduced dimeric forms are not detected by cross-linking of 125I-IGF-I to native membranes. Cleavage of the inter-dimeric disulphide bonds is therefore not a prerequisite for higher-affinity binding or loss of immunoreactivity. These observations suggest that the thiol redox state of the IGF-I receptor in vivo is an important determinant of receptor conformation and therefore of the biological responses to IGF-I.     

Presence of D2-dofamine Receptors in Human Term Placenta

Abstract

We studied the binding of [³H]-spiperone on human term placental membranes. This binding reached plateau level after 30 min incubation at 37°C and was reversed (t_1/2 ～ 5 min) by addition of an excess of unlabeled spiperone. Scatchard analysis of saturation experiments with increasing doses of [³H]-spiperone (0–25 nM) showed one class of high affinity binding sites with a dissociation constant (Kd) of 14 ± 2 nM and a maximal binding capacity (Bmax) of 222 ± 9 fmoles/mg protein. The affinity of 5 competitors was determined in competitive binding assays. The D₂-dopamine antagonists were the most potent inhibitors: Ki for spiperone and haloperidol were 8 ± 2 and 56 ± 22 nM respectively. Dopamine inhibited [³H]-spiperone binding with a Ki of 570 ± 50 μM whereas Schering 23390 (D₁ antagonist) and propranolol (β-adrenergic antagonist) were without effect. The binding was also inhibited by 100 μM GTPγS (38 ± 8% inhibition), indicating that the dopamine receptor is coupled with a GTP binding protein. These results demonstrate for the first time the presence of D₂-dopamine receptors in human placenta.     

Higher cytolytic efficiency of an IgG2 alpha than of an IgG1 monoclonal antibody reacting with the same (or spatially close) determinant on a human high-molecular-weight melanoma-associated antigen

Serological and immunochemical assays showed that the monoclonal antibody (MoAb) 225.28S, an IgG_2α, and the MoAb 653.40S, an IgG₁, react with the same (or spatially close) antigenic determinant expressed on a set of molecules carrying a high-molecular-weight human melanoma-associated antigen. Neither monoclonal antibody mediates complement-dependent lysis of cultured melanoma cells, but both of them specifically mediate lysis of target cells in an antiglobulin cytotoxic assay and in an antibody-dependent cell-mediated cytotoxicity (ADCC) assay. In the latter two assays the IgG_2α displays a higher lytic activity than the IgG₁. The differential lytic activity of the IgG_2α and IgG₁ monoclonal antibodies was detected also when the sensitivity of the ADCC assay was increased either by boosting the cytolytic activity of the effector cells or by enhancing the susceptibility to lysis of target cells.     

Localization of the carboxyl terminus of Band 3 to the cytoplasmic side of the erythrocyte membrane using antibodies raised against a synthetic peptide

Polyclonal antibodies were raised in rabbits against a synthetic peptide which corresponds to the 12-amino acid carboxyl-terminal sequence of murine erythrocyte Band 3. Immunoblots of ghost membrane proteins showed that the antibody specifically recognized murine or rat Band 3 but not human or canine Band 3. The antibody also bound to murine ghost membranes applied directly to nitrocellulose but not to human ghost membranes. This shows that the carboxyl terminus of Band 3 is available for antibody binding in ghost membranes and that the carboxyl-terminal sequences of human and mouse Band 3 are not identical. The specificity of the antibody for the carboxyl terminus of Band 3 was confirmed by the loss of antibody binding after digestion of detergent-solubilized ghost membrane proteins with carboxypeptidase Y. In addition, carboxyl-terminal fragments of Band 3 generated by protease treatment of cells or ghost membranes were positive on immunoblots while amino-terminal fragments were negative. In contrast, protease-treated stripped ghost membranes did not contain a carboxyl-terminal fragment of Band 3 that was detectable on immunoblots. The carboxyl terminus of Band 3 was localized to the cytoplasmic side of the erythrocyte membrane since antibody binding as determined by immunofluorescence occurred in ghosts and permeabilized cells but not in intact cells. In addition, competition studies using enzyme-linked immunosorbent assays and immunoblots showed that cells and resealed ghosts competed poorly for antibody compared to ghost membranes, inside-out vesicles, or albumin-conjugated peptide.     

Specificity and distribution of antigenic determinants on the polyoma virus capsid and nature of the reaction of immunoglobulin G antibody with the capsid surface.

Antisera to the LP and SP34 strains of polyoma virus have been prepared and their reactions with purified virions studied by double diffusion in agar and direct assay of antibody binding. One or more common antigenic determinants appear on the capsids of both strains. The form of this determinant varies slightly on each of the strains tested. The SP34 strain also carries its own strain-specific antigenic determinant. Both strains of virus were able to bind 150 anti-LP IgG³ molecules per virion and 50 anti-SP IgG molecules per virion. The slow rate of dissociation of bound IgG antibody (k_dissociation = 2 × 10⁻⁶ s⁻¹), and the rapid rate of antibody binding (k_dissociation = 2 × 10⁷m⁻¹ s⁻¹), suggest that IgG antibody is bound to the capsid surface through two antigen-antibody bonds. 50 anti-SP IgG molecules per capsid, divalently bound, completely inhibit the binding of 150 anti-LP IgG molecules, and vice versa. Consideration of the symmetry and molecular dimensions of the IgG molecule and the polyoma virus capsid leads to a model of the divalent interaction of IgG antibody with the common antigenic determinant(s). In this model, one species of antibody binds divalently to opposed subunits of a hexamer morphological unit. The other species of antibody binds divalently to the subunits on either side of the point of tangency of any two morphological units.     

Localized conformational changes induced in a class I major histocompatibility antigen by the binding of monoclonal antibodies

We describe two monoclonal antibodies, R3/47 and YR1/1, directed against different epitopes of the expressed rat class I major transplantation antigen RT1Aa, that interact with each other so that the binding of one antibody, YR1/1, is greatly enhanced by the binding of the other. The positive interaction between R3/47 and YR1/1 also occurs when using RT1Aa molecules solubilized from cell membranes in detergent. It is therefore unlikely that the molecular environment of the membrane contributes to the interaction. The ability of R3/47 to modify the YR1/1 determinant on the RT1Aa molecule is mediated without any significant loss of potency by highly purified monomeric Fab fragments. This result suggests that the binding of R3/47 to the RT1Aa molecule alters the YR1/1 determinant by initiating a propagated conformational change in the antigen.     

Biochemical and genetic analysis of the Oka blood group antigen

The monoclonal antibody TRA-1-85 recognizes a cell surface antigen which is expressed by all human cell types tested, including red blood cells (RBCs), but not by mouse cells. All the human RBCs tested were TRA-1-85 positive except those with the rare phenotype Ok(a^–). Oka is a blood group antigen of very high frequency and only three unrelated Ok(a^–) people are known. The red cells of all three propositi were negative with the TRA-1-85 antibody. To confirm the relationship between the TRA-1-85 antibody and anti-Ok^a, the immune antibody found in the serum of Ok(a^–) individuals, Western blot analysis was used: the TRA-1-85 antibody and anti-Ok^a gave identical but complex patterns of re-activity in Western blot analysis of human cell lysates or membranes. This suggests that the anti-Oka and TRA-1-85 antibodies recognize the same cell-surface determinant and implies that Ok^a is not restricted in its expression to the surface of RBCs but is expressed on white blood cells (WBCs) of Ok(a⁺) individuals and all human cell lines tested to date. WBCs from one of the Ok(a^–) propositi were tested and found to be negative with the TRA-1-85 antibody. Finally, the species specificity of the TRA-1-85 antibody has been exploited by the use of somatic cell hybrids and DNA transfection techniques to examine the genetic control of the Ok^a antigen defined by the TRA-1-85 antibody. We report that the determinant is controlled by a single gene OK present on human chromosome 19.     

Binding of intrinsic factor to ileal brush border membrane in the rat

The rat ileal brush border membrane binds both free [¹²⁵I]-intrinsic factor (IF) and the IF-[⁵⁷Co]cobalamin (cbl) complex. This binding is observed with IF isolated from rat stomach, but not from IF isolated from hog, canine and human stomachs. The binding of rat-IF[⁵⁷Co]cbl can be blocked with free rat IF but not with hog IF. The IF-cbl complex binds at a higher affinity (Ka=0.15 × 10⁹ M^?1) compared to that of free IF (Ka=0.9 × 10⁹ M^?1). Rat IF-cbl also binds efficiently to human and canine ileal membranes. While antibody to the canine ileal receptor blocks the binding of rat, human or hog IF-[⁵⁷Co]cbl to human and canine ileal membranes, it does not affect the binding of rat IF-[⁵⁷Co]cbl to rat ileal membranes. These findings demostrate that the rat ileal receptor is different from canine and human ileal receptors.     

The inhibition of a membrane-bound enzyme as a model for anaesthetic action and drug toxicity.

Insulin-like growth factor (IGF)-binding sites copurifying with human placental insulin receptors during insulin-affinity chromatography consist of two immunologically distinct populations. One reacts with monoclonal antibody alpha IR-3, but not with antibodies to the insulin receptor, and represents Type I IGF receptors; the other reacts only with antibodies to the insulin receptor and is precipitated with a polyclonal receptor antibody (B-10) after labelling with 125I-multiplication-stimulating activity (MSA, rat IGF-II). The latter is a unique sub-population of atypical insulin receptors which differ from classical insulin receptors by their unusually high affinity for MSA (Ka = 2 x 10(9) M-1 compared with 5 x 10(7) M-1) and relative potencies for insulin, MSA and IGF-I (40:5:1 compared with 150:4:1). They represent 10-20% of the total insulin receptor population and account for 25-50% of the 125I-MSA binding activity in Triton-solubilized placental membranes. Although atypical and classical insulin receptors are distinct, their immunological properties are very similar, as are their binding properties in response to dithiothreitol, storage at -20 degrees C and neuraminidase digestion. We conclude that atypical insulin receptors with moderately high affinity for IGFs co-exist with classical insulin receptors and Type I IGF receptors in human placenta. They provide an explanation for the unusual IGF-II binding properties of human placental membranes and may have a specific role in placental growth and/or function.     

The free and the β2-microglobulin-associated heavy chains of HLA-A,B alloantigens share the antigenic determinant recognized by the monoclonal antibody Q1/28

Serological and immunochemical assays have shown that the monoclonal antibody Q1/28 recognizes an antigenic determinant which is expressed on the heavy chain of subsets of HLA-A, B antigens and is distinct from those defining the serological polymorphism of this system. Association of the HLA-A, B heavy chain with ₂-microglobulin is not required for expression of the antigenic determinant recognized by the monoclonal antibody Q1/28, since this antibody can immunoprecipitate a 45 000 m. w. component from radiolabeled lymphoid-cell glycoproteins immunodepleted with either an anti-human ₂-microglobulin xenoantiserum or the MoAb W6/32 to framework determinants of HLA-A, B, C antigens. Furthermore, the MoAb Q1/28 can immunoprecipitate a 45 000 m. w. component from an NP40lysate of radiolabeled Daudi cells, which lack the genetic information for ₂-microglobulin. The determinant recognized by the MoAb Q1/28 is relatively resistant to denaturing treatments and does not appear to be carbohydrate in nature. The MoAb Q1/28 is the first example of an antibody which recognizes an antigenic determinant expressed on both the ₂-microglobulin-associated and free HLA-A, B heavy chains.     

Establishment of a monoclonal antibody recognizing ultraviolet light-induced (6-4) photoproducts

We obtained a monoclonal antibody directed against UV-induced DNA damage. Analysis of the antigenic determinant in UV-irradiated DNA recognized by this antibody, 64M-1, revealed that it bound UV-irradiated oligo- or poly-nucleotides containing thymine-thymine or thymine-cytosine sequences. The antibody failed to bind DNA irradiated with 313 nm UV in the presence of acetophenone, which contained predominantly thymine dimers as DNA damage. The binding activity of this antibody to 254-nm UV-irradiated DNA decreased with 313-nm UV irradiation, and the decrease of this binding activity correlated with the decrease of fluorescence corresponding to (6-4) photoproducts. These results suggest that the antigenic determinant recognized by this monoclonal antibody is a (6-4) photoproduct. Using autoradiography with 3H-antibody, we could detect the formation of the (6-4) photoproduct in individual human cells irradiated with 254-nm UV doses as low as 20 J/m2.     

Separation and localization of two classes of auxin binding sites in corn coleoptile membranes

Summary Further evidence is presented for the discrete nature of the two classes of high affinity auxin binding sites in corn (Zea mays L.) coleoptile membranes, site 1 and site 2. Fractions can be obtained by differential centrifugation that exhibit binding kinetics characteristic of site 2, but not site 1. Membrane preparations containing both binding sites may be resolved on sucrose gradients into a light and a heavy band, whose binding kinetics and analogue binding specificities correspond to those deduced for site 1 and site 2 respectively in unfractionated membranes. Evidence from enzymic and chemical assays and from electron microscopy suggests that site 2, the auxin-specific binding site, is located in fractions enriched in plasma membrane, whereas site 1 is associated with Golgi membranes and/or endoplasmic reticulum.Abbreviations NAA 1-naphthylacetic acid - IAA 3-indolylacetic acid - TIBA 2,3,5-triiodobenzoic acid - SDH succinic dehydrogenase - IDPase inosine diphosphatase     

Characterization of Insulin-Like Growth Factor Binding to Human Granulosa Cells Obtained During in Vitro Fertilizationd

Abstract

Insulin and IGF-I affect in vitro ovarian stromal and follicular cell function in several species. We previously characterized insulin receptors on human granulosa cells obtained from in vitro fertilization procedures but were unable to demonstrate specific binding of IGF-I.

Following modification of the assay conditions, we now report specific, high affinity IGF-1 binding sites on human granulosa cells. Substitution of equimolar concentrations of sucrose for sodium chloride in the buffer solution increased binding of IGF but not insulin in equilibrium assays. Maximal specific IGF-I binding was 2.69 ± 0.30%/10⁵ cells (SEM, n=9) with half-maximal inhibition of binding at 2 ng/ml IGF-I. Unlabeled insulin recognized the type I IGF receptor with low affinity. An IGF-I receptor monoclonal antibody (αIR-3) inhibited ¹²⁵I-IGF-I but not ¹²⁵I-insulin binding. Affinity crosslinking followed by SDS/PAGE under reducing conditions revealed IGF-I binding at a molecular weight compatible with the αsubunit of the type I IGF receptor and with a pattern of inhibition by various ligands that paralleled the equilibrium binding assays.

IGF-I receptors are present on freshly isolated human ovarian granulosa cells obtained following pharmacologic stimulation with gonadotrophin according to the protocols of in vitro fertilization. The biologic function of these receptors currently is being investigated.