Interaction the CDR2-Containing Region for Igm in the Polymeric Ig

Fine Specificity of Ligand-Binding Domain 1 in the Polymeric Ig Receptor: Importance of the CDR2-Containing Region for IgM Interaction This information is current as of September 24, 2021. Målfrid Røe, Inger N. Norderhaug, Per Brandtzaeg and Finn-Eirik Johansen J Immunol 1999; 162:6046-6052; ; http://www.jimmunol.org/content/162/10/6046 Downloaded from

References This article cites 50 articles, 22 of which you can access for free at: http://www.jimmunol.org/content/162/10/6046.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 24, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Fine Speciﬁcity of Ligand-Binding Domain 1 in the Polymeric Ig Receptor: Importance of the CDR2-Containing Region for IgM Interaction1

Målfrid Røe, Inger N. Norderhaug, Per Brandtzaeg, and Finn-Eirik Johansen2

The human polymeric Ig receptor (pIgR), also called transmembrane secretory component, is expressed basolaterally on exocrine epithelia, and mediates specific external transport of dimeric IgA and pentameric IgM. The extracellular part of pIgR consists of five Ig-like domains (D1-D5), and a highly conserved D1 region appears to mediate the initial noncovalent ligand interaction. While the human pIgR binds both dimeric IgA and pentameric IgM with high affinity, the rabbit counterpart has virtually no binding capacity for pentameric IgM. This remarkable disparity constitutes evidence that the binding site of the two ligands differs with regard to essential receptor contact elements. Therefore, we expressed human/rabbit chimeric pIgRs in Madin-Darby canine Downloaded from kidney cells and found that human pIgR D1 is crucial for the interaction with pentameric IgM when placed in the context of a full-length receptor regardless of its backbone species. D1 contains three complementarity-determining region-like loops (CDR1– 3), and to further map human D1 regions involved in pentameric IgM binding, we transfected Madin-Darby canine kidney cells with human/rabbit chimeric receptors in which the regions containing the CDR-like loops had been interchanged. Our results showed that the region containing the CDR2-like loop is the most essential for pentameric IgM binding. The region containing the

CDR1-like loop also contributed substantially to this interaction, whereas only little contribution was provided by the region http://www.jimmunol.org/ containing the CDR3-like loop, although it appeared to be necessary for maximal pentameric IgM binding. The Journal of Immunology, 1999, 162: 6046–6052.

ecretory Ig (SIgA3 and SIgM) Abs play a major role in for pIgR in both pIgA and pentameric IgM, it is unknown how the adaptive defense at mucosal surfaces, the largest body area polymers themselves are involved in this binding site. The fact that S exposed to the external environment (1–4). Dimers and pentameric IgM has been shown to interact with free SC with an larger polymers of IgA (collectively called pIgA) and pentameric affinity that is 8–30 times that determined for pIgA (16, 17) sug- IgM synthesized by subepithelial plasma cells become specifically gests that the pIgR binding site of the two polymers is structurally bound by the human pIg receptor (pIgR), also known as the trans- somewhat different. This idea is supported by remarkable species by guest on September 24, 2021 membrane secretory component (SC). This receptor is expressed differences shown by pIgR with regard to pentameric IgM inter- on the basolateral surface of secretory epithelial cells (5, 6). The action (see below). pIgs are next internalized and transported across these cells to their The pIgR is a glycoprotein of 100–120 kDa (depending on the apical domain, where the extracellular ligand-binding portion of species) with five Ig-like extracellular domains (D1-D5) that are the receptor is proteolytically cleaved and released to the lumen, structurally most similar to the IgV regions (18). Binding of pIgA either complexed to the ligand as bound SC or unoccupied as free to pIgR appears to be a sequential process in which an initial SC (4, 7). noncovalent ligand interaction with D1 progresses to other do- It remains an enigma that two so structurally different polymers mains (Ref. 19 and Norderhaug et al.4) and is followed (in most as pIgA and pentameric IgM can bind specifically to the same species) by disulfide binding between one of the IgA heavy chains receptor, although their shared J chain has been shown to be es- and D5 (reviewed in Ref. 20). Several lines of evidence suggest sential for this interaction (8–15). However, by itself this polypep- that D1 carries the primary site of interaction with pIgA. First, both tide shows only marginal affinity for free SC (10). Therefore, al- proteolytic and recombinant fragments of the receptor that contain though the J chain in a crucial way contributes to the binding site D1 have been shown to retain the capacity to bind pIgA, although this initial interaction with certain synthetic receptor peptides appears to be “promiscuous” with regard to Ig class (19, 21–23) (see Laboratory for Immunohistochemistry and Immunopathology (LIIPAT), Institute of Pathology, University of Oslo, The National Hospital, Rikshospitalet, Oslo, Norway below). Second, mAbs that recognize an epitope within D1 have Received for publication November 16, 1998. Accepted for publication March been shown to compete with pIgA for binding to the receptor (21). 3, 1999. Furthermore, we have shown recently that binding of pentameric The costs of publication of this article were defrayed in part by the payment of page IgM to the human pIgR depends preferentially on a strong inter- charges. This article must therefore be hereby marked advertisement in accordance action with D1, while binding of pIgA in addition depends on with 18 U.S.C. Section 1734 solely to indicate this fact. determinants within D2 and/or D3 to support the initial noncova- 1 This study was supported by the University of Oslo, the Research Council of Nor- lent interaction with D1.4 way, the Norwegian Cancer Society, and Anders Jahre’s Foundation for Promotion of Science. M.R. has been a Research Fellow of the University of Oslo. Located in D1 are three loops corresponding to the complemen- 2 Address correspondence and reprint requests to Dr. Finn-Eirik Johansen, LIIPAT, tarity-determining regions (CDR1-CDR3) of IgV regions (24), the Institute of Pathology, Rikshospitalet, N-0027 Oslo, Norway. E-mail address: [email protected] 3 Abbreviations used in this paper: SIg, secretory Ig; CDR, complementarity-deter- 4 I. N. Norderhaug, F.-E. Johansen, P. Krajci, and P. Brandtzaeg. Domain deletions in mining region; MDCK, Madin-Darby canine kidney; pIgA, polymeric IgA; pIgR, the human polymeric Ig receptor disclose differences between its dimeric IgA and polymeric IgR, SC, secretory component. pentameric IgM interaction. Submitted for publication.

Copyright © 1999 by The American Association of Immunologists 0022-1767/99/$02.00 The Journal of Immunology 6047 sequence that determines their Ag-binding specificity. The CDR1- like loop in pIgR D1 is highly conserved among different species, 82–100% when conservative amino acid changes are not taken into account (25). The CDR2- and CDR3-like loops in D1 show less interspecies homology, but retain some invariant residues that may play important roles in ligand binding. Thus, a study by Coyne et al. (25), based on a mutational approach with modeling of the rabbit pIgR D1 sequence on known Ig variable structures, suggested that all three loops participate in the specific noncovalent binding of human pIgA. The order of importance was not exam- ined, but based on the studies mentioned above, the CDR1-like loop is probably the most important one for the initial interaction. In humans, pIgR binding of pIgA also depends on structural elements outside D1 (Ref. 19 and Norderhaug et al.4), and the importance of interactions between pIgA and elements in CDR2 and CDR3 may therefore be less significant than for rabbit D1. Human pIgR does not show stable binding of monomeric IgA or other monomeric Ig isotypes, but interacts with both pIgA and pentameric IgM with high affinity, as mentioned above (16, 26). In Downloaded from mice and rabbits, on the other hand, the pIgR binds primarily pIgA (27). To characterize the interaction of human pentameric IgM with pIgR, we exploited this species difference and generated several chimeric receptors to characterize the pentameric IgM binding site of the receptor. We found that D1 of human origin could

transfer its pentameric IgM-binding afﬁnity to the rabbit pIgR, thus http://www.jimmunol.org/ substantiating its crucial role for both pIgA and pentameric IgM interaction. Furthermore, we demonstrated that the D1 regions containing CDR1- and CDR2-like loops of human origin could transfer substantial pentameric IgM-binding capacity to the rabbit pIgR. However, all three human pIgR CDR-like regions were required for maximal pentameric IgM-binding capacity.

Materials and Methods Immunoglobulins FIGURE 1. A, Comparison of the deduced amino acid sequences of the by guest on September 24, 2021 experimentally interchanged regions A, B, and C of human (H) and rabbit Polyclonal human pIgA (28), mainly IgA1, crude monomeric IgA, and (R) pIgR domain 1 (upper and lower lines, respectively). The CDR-like monoclonal pentameric IgM, were isolated and characterized, as previ- loops are indicated by boxes. When used in the text, capital letters for ously described (29). region A, B, or C refer to the amino acid sequence of human origin, and Plasmid constructions corresponding lower case letters rabbit origin; and the backbone species is denoted by -h and -r for human and rabbit, respectively. Crossovers were All constructs were cloned in the eukaryotic expression vector chosen from stretches of highly conserved amino acid sequence to mini- pCDNA3neo (Invitrogen, San Diego, CA). The subcloning of the full- mize chances of steric incompatibility and incorrect folding of the chimeric length cDNA encoding the human pIgR (30) has been described previously proteins. B, Point mutations generated to create restriction enzyme cloning (29). Subcloning of the murine and rabbit pIgR was performed by PCR with pIgR cDNA of mouse (gift from C. Kaetzel, University of Kentucky) sites for swapping of regions A, B, or C. Recognition sequences are un- (31) and rabbit (gift from K. Mostov, University of California at San Fran- derlined, and restriction enzymes used are written below. The nature of the cisco) (32) origin as template, respectively. The human/rabbit chimeras mutations is shown in parentheses. For human region C to rabbit D2, SacI were constructed by PCR with primers that introduced silent mutations in was used, and for rabbit region C to human D2, ApaI was used. the overlaps to create restriction enzyme cloning sites. Chimeric expression constructs were made with standard molecular biology techniques and en- each construct that showed uniform, strong staining were randomly se- coded the amino acid sequences outlined in Fig. 1 (details of construction lected for further analysis. will be provided upon request). Binding of radioiodinated pIgA and pentameric IgM Cell culture Preparations of pIgs were 125I labeled with Chloramine-T-catalyzed iodi- MDCK (strain II) cells were grown in DMEM (Bio Whittaker, Walkers- nation and purified by gel-filtration chromatography on a PD-10 column ville, MD) supplemented with 5% FCS, 50 ␮g/ml of gentamicin, and 1 mM (Sephadex G-25 M) (Pharmacia Biotech, Uppsala, Sweden). Stably trans- L-glutamine (Life Technologies, Paisley, U.K.). fected MDCK cells grown to confluence in microtiter plates (number 3590; Costar, Cambridge, MA) were incubated with 125I-labeled pIgA or pen- Transfection and clonal selection of pIgR constructs tameric IgM, and with various concentrations of unlabeled corresponding pIg in DMEM/5% FCS/10 mM HEPES at 4°C for 2 h. The medium was MDCK cells were stably transfected by the DNA-calcium phosphate pro- then harvested and the cells were washed four times with ice-cold PBS and cedure (33) or by electroporation, in which 107 cells were exposed to 10 ␮g lysed in 2 M NaOH. The radioactivity was determined in an automatic DNA in 0.4 ml PBS at 250 ␮F, 675 V/cm. Clones expressing the neomycin gamma counter (1470 Wizard; Wallac, Turku, Finland). Nonspecific bind- resistance marker were selected in the presence of 0.5 mg/ml G418 (Ge- ing of 125I-labeled pIg was determined by incubation with nontransfected neticin; Sigma, St. Louis, MO). Stable cell lines were established by iso- MDCK cells. lating resistant colonies with cloning cylinders. Clones expressing the pIgRs were identified by immunofluorescence staining, in which cells were Epithelial cell translocation of pIgA and pentameric IgM first incubated with pIgA (19 ␮g/ml) for1hatroom temperature, and then with a rabbit anti-human IgA FITC conjugate (F 0204; Dako, Glostrup, Transfected and nontransfected MDCK cells were cultured on 3-␮m Trans- Denmark) diluted 1/50 (1 h, room temperature). At least two clones for well-COL filters (Costar) for 6 days, and transepithelial resistance for each 6048 INTERACTIONS BETWEEN pIgs AND pIgR

filter was measured to be Ն180 ⍀ before the start of transcytosis experiments. Polarized cells were incubated with 50 nM pIgA or pentameric IgM together with 50 nM IgG in the basal medium at 37°C for 20 h. The apical medium was harvested and analyzed by ELISA to determine the concentrations of IgA, IgM, and IgG, as previously described (29). Results Rabbit pIgR binds pentameric IgM poorly compared with human pIgR Previous studies on binding of pIgA and pentameric IgM to pIgR from different species have demonstrated that mouse and rabbit pIgR binds pentameric IgM (from any species) poorly but pIgA from all species quite well, whereas human pIgR binds both pentameric IgM and pIgA with high affinity (see above). We used this disparity to localize regions of the human receptor essential for the binding of pentameric IgM. Preliminary experiments were first performed to show that we were able to reproduce the previously observed species difference. MDCK cells were stably transfected with cDNA encoding the human, murine, or rabbit pIgR. Binding of pIgA and pentameric IgM to the transfected receptors was de- Downloaded from termined with 125I-labeled ligands. The rabbit pIgR bound pentameric IgM poorly compared with pIgA, while the human pIgR bound both pentameric IgM and pIgA with high affinity, pentameric IgM in fact showing the better binding (Fig. 2A). The relative binding of pentameric IgM and pIgA to murine pIgR demonstrated binding characteristics in this species intermediate to the binding http://www.jimmunol.org/ characteristics for human and rabbit receptors (Fig. 2B). Based on these results, we chose to use human-rabbit chimeras to identify the pentameric IgM binding site of the human pIgR, and to use ligand concentrations of 5, 10, and 20 nM to determine relative binding of pentameric IgM and pIgA to these chimeras. D1 of human pIgR is essential for binding of pentameric IgM The initial noncovalent interaction of human pIgR with pIgA occurs in D1 (21–24), whereas the receptor elements required for by guest on September 24, 2021 binding of pentameric IgM are less well characterized. To study whether human pIgR D1 is essential for binding of pentameric IgM as well, we constructed chimeric receptors, namely a human receptor with D1 from rabbit and vice versa. Binding of 125I-labeled pIgA or pentameric IgM to MDCK cells stably transfected with these constructs showed that the human chimeric receptor containing rabbit D1 (rD1-h) exhibited reduced binding affinity for pentameric IgM compared with pIgA, whereas the rabbit chimeric receptor containing human D1 (hD1-r) bound pentameric IgM FIGURE 2. A, Saturation-binding experiments with MDCK cells expressing the human, rabbit, and D1 swapped chimeric pIgRs depicted in B. with relatively high affinity (Fig. 2, A and B). Thus, the human B, Schematic diagram of the human, mouse, and rabbit constructs, and the pIgR D1 is the primary determining element for the noncovalent relative binding of 125I-labeled pentameric IgM (pIgM) and pIgA. The first initial receptor interaction with pentameric IgM, in the same fash- extracellular domain (D1) and the region-spanning domain two to five ion as with pIgA. (D2–5) are indicated. White and grey boxes denote human and rabbit sequences, respectively. TM, transmembrane region; C, cytoplasmic tail. Translocation of pIgA and pentameric IgM by pIgR-transfected Transfected MDCK cells were incubated with 125I-labeled pIgM or pIgA MDCK cells for2honiceandwashed four times, and the cell-bound radioactivity was To study the functional capacity of the different wild-type and determined in a gamma counter. Values for nonspecific binding, deter- chimeric receptors, transfected MDCK cells were grown on filters mined by binding to untransfected MDCK cells, were subtracted. Binding to confluent monolayers, and the translocation of pIgA and pen- values represent the average of the ratio of fmol pIgM-to-fmol pIgA bound tameric IgM from the basal surface to the apical medium was at three different unsaturated ligand concentrations (5, 10, and 20 nM) expressed as mean (ϩ1 SD) of at least two separate experiments. Binding characterized. Cells were incubated at 37°C with 50 nM pIgA or values that differ from the wild-type human pIgR (p Ͻ 0.05; two-tailed pentameric IgM in the basal medium, together with 50 nM IgG as Student’s t test) are indicated by a star; and binding values that differ from an internal control. After 20 h, the apical medium was harvested, the wild-type rabbit pIgR (p Ͻ 0.05; two-tailed Student’s t test) are indi- and the concentrations of IgA, IgM, and IgG in each fraction were cated by a closed circle. analyzed by ELISA. The capacity for pentameric IgM transport by the human pIgR was found to be quite similar to that for pIgA (Fig. 3). By contrast, the rabbit pIgR showed a relatively high capacity former ligand), whereas rD1-h, like the wild-type rabbit pIgR, for pIgA transport, while it translocated only trace amounts of translocated mainly pIgA, but also some pentameric IgM. These pentameric IgM. Like the wild-type human pIgR, hD1-r translo- results harmonized with the binding results and showed that the cated both pIgA and pentameric IgM (although favoring the chimeric receptors were functional. The level of IgG sampled from The Journal of Immunology 6049

FIGURE 3. Transport of pIgA and pentameric IgM (pIgM) by MDCK cells transfected with human pIgR, rabbit pIgR, and D1 interchange constructs (see Fig. 2). The transport was measured over a time period of 20 h after basolateral addition of 50 nM pIgA or pentameric IgM with 50 nM

IgG as an internal control. The apical medium was harvested and analyzed Downloaded from by ELISA. The apical levels of IgG when coincubated with pIgA or pentameric IgM are shown as separate columns. Apical levels of IgA (columns 1–4) and IgM (columns 6–9) were signiﬁcantly higher than background Ͻ (columns 5 and 10, respectively) for all pIgR transfectants (p 0.05; FIGURE 4. Schematic diagram of various chimeric domain 1 (D1) con- two-tailed Student’s t test). IgG transfer (columns 11–20) did not differ structs with a human pIgR backbone (D2–5), and their relative binding of signiﬁcantly for any of the cell lines. Data are from one of three similar 125I-labeled pentameric IgM (pIgM) and pIgA compared with the relative

ϩ http://www.jimmunol.org/ experiments, expressed as mean ( 1 SD) of three ﬁlters. binding of the two ligands to wild-type human or rabbit pIgR (at the bottom). White and grey boxes (on the left) denote the employed human- or rabbit-derived sequences, respectively. D1 is divided into three regions the apical medium was virtually the same in cells expressing the labeled A, B, and C, containing CDR-1, CDR-2, and CDR-3, respectively. different pIgR constructs and in nontransfected MDCK cells (Fig. Binding values (on the right) represent the average of the ratio of fmol 3). This IgG most likely reﬂected passive paracellular diffusion, a pIgM-to-fmol pIgA bound at three different unsaturated ligand concentra- possibility supported by the fact that even smaller amounts of the tions (5, 10, and 20 nM) expressed as mean (ϩ1 SD) of at least two larger pIgA and pentameric IgM molecules were translocated separate experiments. Binding values that differ from the wild-type human across nontransfected cells (Fig. 3). pIgR (p Ͻ 0.05; two-tailed Student’s t test) are indicated by a star, and

binding values that differ from the wild-type rabbit pIgR (p Ͻ 0.05; two- by guest on September 24, 2021 Several human D1 elements participate in pIgR binding of tailed Student’s t test) are indicated by a closed circle. For details of chi- pentameric IgM meras and experimental setup, see Figs. 1 and 2, respectively. To define more precisely the sites for pentameric IgM interaction within D1 of the human pIgR, we constructed a series of chimeras designed to assess the relative role of the various regions contain- ABc-h) resulted in only slightly reduced pentameric IgM binding. ing the different CDR-like loops (25). D1 was divided into three However, AbC-h showed a reduction in pentameric IgM binding regions (A, B, and C) that contained the CDR-like loop 1, 2, or 3, of about 83% compared with the human pIgR, suggesting that the respectively (Fig. 1). These human regions (denoted by the capital human CDR2-containing region was most important for pentam- letters A, B, and C) were replaced alone, or in combination, with eric IgM binding. The human chimeras that had two CDR-like the same region(s) from rabbit D1 (denoted by the corresponding loops replaced with the comparable rabbit regions (Fig. 4; abC-h, lower case letters), resulting in six different mutants (Fig. 4). The aBc-h, and Abc-h) all bound pentameric IgM poorly; aBc-h (which use of these chimeras allowed us to define the subdomain inter- retained the most pentameric IgM-binding activity) showed a re- changes between human and rabbit D1 that decreased or abolished duction of about 75%, and Abc-h about 90%, compared with the the binding of pentameric IgM to the human pIgR. For unknown human pIgR, whereas abC-h did not bind pentameric IgM at levels reasons, chimera aBC-h was extremely difficult to express stably in higher than the wild-type rabbit receptor. Taken together, these the MDCK cells. However, the clones we were able to isolate results demonstrated that elements within the CDR1- and CDR2- showed a surface expression of the mutant receptor at similar lev- containing regions of the human pIgR are particularly important, els to the clones expressing the human wild-type pIgR based on and that their combined interchange for equivalent rabbit regions similar values in a cell-based ELISA with a mAb against human (abC-h) was sufficient to abolish pentameric IgM binding to the pIgR-D3 (data not shown). The aBC-h clones, however, did not human pIgR. Exchanging the CDR3-containing region of the hu- show significant pIgA or pentameric IgM binding at low ligand man receptor alone (ABc-h) or in combination with region A or B concentration; therefore, the relative binding of the two ligands for (aBc-h and Abc-h) resulted in only a small reduction in pentameric this construct was calculated at ligand concentrations of 20, 40, IgM binding compared with the appropriate parental pIgR con- and 80 nM (Fig. 4). structs (human wild type, aBC-h, and AbC-h, respectively). Binding of 125I-labeled pIgs revealed that all chimeras bound pIgA. The chimeras containing only one region of rabbit origin A human CDR2-like loop confers pentameric IgM-binding (aBC-h, AbC-h, and ABc-h) all showed some pentameric IgM capacity to rabbit pIgR D1 binding, but this was reduced compared with that of the wild-type A comparable series of D1 chimeras was constructed on a rabbit human receptor. Replacing the human CDR1-containing or CDR3- backbone (D1 regions of human origin denoted by capital letters, containing region with the equivalent rabbit region (Fig. 4; aBC-h, and rabbit origin by lower case letters, as above) to define the 6050 INTERACTIONS BETWEEN pIgs AND pIgR

differences (27, 35–38) have suggested that the initial noncovalent ligand-receptor interaction to some extent involves different structural elements. In this study, we describe for the first time domain regions of the human pIgR that are uniquely essential and sufficient to mediate the interaction with pentameric IgM. We constructed chimeric receptors interchanging parts of human pIgR (which binds both pIgA and pentameric IgM) and rabbit pIgR (which in essence binds only pIgA), and found that D1 of human pIgR was sufficient in a full-length context to mediate pentameric IgM binding. Furthermore, we determined that the CDR2-containing and, to a smaller extent, the CDR1-containing region of human D1 were crucial to allow pentameric IgM binding to the pIgR, whereas the region containing a CDR3-like loop at best provided a minor contribution to this end. The characterization of the interaction between pentameric IgM and pIgR is of both basic and clinical interest. Phylogenetically, SIgM appears to be the first secretory Ab class that evolved (39), so the receptor configuration specific for pentameric IgM may be

more ancient than that for pIgA. Selective IgA deficiency is the Downloaded from most common primary immunodeficiency in humans, with a prev- alence of one case in 500–700 subjects in our part of the world FIGURE 5. Schematic diagram of various chimeric domain 1 (D1) con- (40). IgA deficiency predisposes particularly for upper respiratory structs with a rabbit pIgR backbone (D2–5), and their relative binding of tract infections and is sometimes associated with an immunoregu- 125I-labeled pentameric IgM (pIgM) or pIgA compared with the relative latory disorder (e.g., allergy, autoimmunity, or celiac disease) (41).

binding of the two ligands to wild-type human and rabbit pIgR (at the However, most subjects (about two-thirds) remain healthy, which http://www.jimmunol.org/ bottom). White and grey boxes (on the left) denote human and rabbit- may be partially explained by compensatory SIgM Abs. Thus, derived sequences, respectively. Binding values (on the right) represent the when IgA-producing immunocytes are reduced or completely average of the ratio of fmol pIgM-to-fmol pIgA bound at three different lacking in the gut, IgG- and especially IgM-producing cells are unsaturated ligand concentrations (5, 10, and 20 nM) expressed as mean substantially increased (42, 43). Therefore, IgA-deficient subjects (ϩ1 SD) of at least two separate experiments. Binding values that differ from the wild-type human pIgR (p Ͻ 0.05; two-tailed Student’s t test) are generally have increased intestinal and salivary IgM levels (44, indicated by a star, and binding values that differ from the wild-type rabbit 45), but such mucosal IgM compensation does not always take pIgR (p Ͻ 0.05; two-tailed Student’s t test) are indicated by a closed circle. place in the upper respiratory tract of patients with infectious prob- For details of chimeras and experimental setup, see Figs. 1 and 2, lems there (46). respectively. In agreement with previous studies (17, 27), we confirmed in by guest on September 24, 2021 preliminary experiments that the rabbit pIgR virtually did not bind pentameric IgM, whereas the murine pIgR showed intermediate minimal interchange between human and rabbit pIgR that could affinity for this polymer (Fig. 2). Similar results have been ob- confer efficient pentameric IgM-binding capacity to the rabbit tained for rat pIgR (27). In contrast to humans (5), these species pIgR (Fig. 5). All expressed chimeras showed binding properties express the pIgR on their hepatocytes, and therefore perform ef- similar to their human backbone analogues. Of those containing ficient removal of pIgA from the circulation (5). Clearly, binding only one human D1 region, aBc-r demonstrated substantial pen- of pentameric IgM to the pIgR on hepatocytes would compromise tameric IgM binding, while Abc-r and abC-r behaved more like the the protective role of pentameric IgM in the systemic circulation wild-type rabbit pIgR (Fig. 5). Of the chimeras containing two by removing such circulating Abs. In contrast, the high avidity CDR-like loops of human origin, ABc-r bound pentameric IgM binding of pentameric IgM by the human pIgR would not be dis- similar to human wild-type pIgR, whereas AbC-r showed only advantageous, because in humans this receptor is absent from moderate pentameric IgM binding (Fig. 5). For unknown reasons, hepatocytes, although they express other binding sites for IgA we were unable to express the aBC-r construct stably in our (3, 47). MDCK cells, comparable with the difficulties expressing aBC-h The extracellular ligand-binding part of pIgR consists of five (see above). Ig-like domains, and it has been suggested that the noncovalent Together our results showed that the human pIgR D1 region interaction between pIgA and pIgR is mediated by elements in D1 containing the CDR2-like loop is sufficient to endow the rabbit (see later). By constructing different chimeric receptors, we found pIgR with pentameric IgM-binding capacity. Elements in the hu- that as for pIgA binding, the D1 of the human pIgR is responsible man CDR1-containing region further enhanced substantially the for the initial noncovalent pentameric IgM interaction; transfer of ability of the rabbit pIgR to bind pentameric IgM. The fact that this domain to the nonbinding rabbit pIgR conferred a significant AbC-r also bound pentameric IgM suggests that the CDR1- and increase in pentameric IgM-binding properties, while the recipro- CDR3-like regions contain elements that contribute to pentameric cal transfer nearly abolished pentameric IgM binding. Thus, D1 of IgM binding, and that their combination suffices to endow the rab- the human pIgR is necessary and sufficient to bind both pIgA and bit pIgR with some pentameric IgM-binding capacity. This sup- pentameric IgM, at least when this binding is supported by inter- ports the notion that all three regions of human pIgR D1 contribute actions between the ligand and D2-D5 of either human or rabbit to its pentameric IgM-binding activity. pIgR. In vivo, pIgR is constitutively expressed by secretory epithelial Discussion cells at all exocrine tissue sites (5). We used MDCK cells trans- Epithelial transport mediated by pIgR is specific for pIgA and fected with different pIgR cDNAs to determine whether their abil- pentameric IgM, but both affinity studies (16, 26, 34) and species ity to transport pIgA and pentameric IgM was directly related to The Journal of Immunology 6051 their polymer-binding capacities. We found that the interspecies IgM-binding efficiency compared with the chimeras that contained exchange did not affect the processing and sorting of the different only one of the first human regions (Abc-r, Abc-h, or aBc-h). receptors in MDCK cells, and they were able to translocate the Taken together, these data suggested that the CDR1-like loop con- bound ligand corresponding to its binding affinity regardless of the tains a structural element essential for pIg binding, but that the species backbone. It has been shown previously that although the specificity to discriminate between pIgA and pentameric IgM rabbit pIgR, in contrast to the human receptor, does not stabilize mainly resides in the CDR2-like loop. the binding of human pIgA by a disulfide-exchange reaction, its Importantly, we found no IgG binding to our chimeric receptors. transport efficiency was nevertheless as great or greater than that of On the other hand, all combinations of human/rabbit chimeric D1 the human pIgR (48). This observation is consistent with the view constructs retained relatively high pIgA-binding capacity. Thus, that disulfide bonding to the receptor does not enhance ligand rabbit pIgR elements in the CDR2-like region, for example, sub- translocation, but may rather play an important physiologic role in stituted efficiently for the same human elements without jeopar- stabilizing SIgA Abs and protect them against degradation in ex- dizing the pIgA binding, whereas the binding of pentameric IgM ternal body fluids (5, 49–53). Furthermore, signals for intracellular was diminished. Therefore, although pIgA and pentameric IgM trafficking have been mapped to the highly conserved cytoplasmic interacting sites in the pIgR are overlapping, the exact amino acids tail of the pIgR (54). responsible for the initial ligand contact appear to differ. There is Bakos et al. (19) also studied the binding of human pIgs to a high degree of amino acid sequence identity in D1 between the different pIgR-derived peptides, and found that only pIgR frag- two species studied, but certain differences must account for the ments containing D1 bound to pIgA and pentameric IgM. Inter- observed differences with regard to pentameric IgM binding. Finer estingly, a synthetic peptide from D1 (SC (15–37)) bound not only mutational analysis and more detailed structural information of Downloaded from pIgA and pentameric IgM, but also to monomeric IgA and IgG pIgR D1 will be necessary to accurately determine the amino acids equally well. Thus, these findings suggest the presence of a com- involved in binding of the two ligands. Recently, an amino acid mon promiscuous Ig binding site in D1. This domain contains motif in C␣3 was identified as essential for dimeric IgA binding to three loops corresponding to the CDRs of Ig variable domains the human pIgR (55). However, considerable structural work is (CDR1-CDR3), the sequence that determines their Ag-binding still required to characterize the intact SC or pIgR binding site in

specificity and affinity. The synthetic peptide (SC (15–37)) shown the polymers. The only crucial common element identified up until http://www.jimmunol.org/ to exhibit promiscuous Ig-binding capacity contains the amino ac- now is the J chain (5). Several studies have suggested that this ids comprising the CDR1-like loop (19). A mAb that recognizes polypeptide is directly involved in the binding site (11, 14, 15), as this pIgR peptide specifically blocked the binding of pIgA to free well as in correct assembly of IgA and IgM polymers (56, 57). Our SC (19, 21), suggesting that this area is an essential initial binding finding that the binding of pIgA and pentameric IgM to pIgR D1 site. Furthermore, the human SC (15–37) peptide has also been involves unique elements, in addition to overlapping structures, shown to compete with human, bovine, rabbit, and rat SC for bind- does suggest that both the J chain and the respective heavy chains ing of human pIgA (24), indicating that this conserved structure contribute to the pIgR binding site. plays an Ig-binding role in all species. The CDR2- and CDR3-like loops in D1 are less homologous Acknowledgments by guest on September 24, 2021 among species than the CDR1-like loop, but retain some invariant We thank the technical staff at Laboratory for Immunohistochemistry and residues that might be important in ligand binding. As a matter of Immunopathology for invaluable assistance. fact, a study by Coyne et al. (25) suggested that all three loops of the rabbit pIgR D1 participate in the noncovalent pIgA binding. References Replacing regions corresponding to the CDR2- and CDR3-like 1. Brandtzaeg, P., K. Baklien, K. Bjerke, T. O. Rognum, H. Scott, and K. Valnes. loops with their counterparts from D2 resulted in complete abro- 1987. Nature and properties of the human gastrointestinal immune system. In gation of binding activity (25). Our results suggested that several Immunology of the Gastrointestinal Tract. K. Miller and S. Nicklin, eds. CRC Press, Boca Raton, p. 1. human pIgR D1 regions likewise contribute to pentameric IgM 2. Childers, N. K., M. G. Bruce, and J. R. McGhee. 1989. Molecular mechanisms binding. We divided D1 into three regions, each containing one of immunoglobulin A defense. Annu. Rev. Microbiol. 43:503. CDR-like loop, and exchanged them between human and rabbit 3. Mestecky, J., I. Moro, and B. J. Underdown. 1999. Mucosal immunoglobulins. In Mucosal Immunology. J. Mestecky, J. Bienenstock, J. McGhee, M. Lamm, pIgR to study each region in the context of a full-length receptor. W. Strober, and P. Ogra, eds. Academic Press, San Diego, p. 133. Although all three regions were necessary for maximal pentameric 4. Goldblum, R. M., L. A. Hanson, and P. Brandtzaeg. 1996. The mucosal defense system. In Immunologic Disorders in Infants & Children, 4th Ed. E. R. Stiehm, IgM binding, we found that the region containing the CDR2-like ed. W. B. Saunders Co., Philadelphia, p. 159. loop was most important in the pentameric IgM-binding process. 5. Brandtzaeg, P. 1985. Role of J chain and secretory component in receptor-me- Replacing this region of the human receptor with the rabbit coun- diated glandular and hepatic transport of immunoglobulins in man. Scand. J. Im- munol. 22:111. terpart (AbC-h) significantly reduced the pentameric IgM-binding 6. Mostov, K. E., J. P. Kraehenbuhl, and G. Blobel. 1980. Receptor-mediated tran- capacity. In support of this observation, we also found that a sim- scellular transport of immunoglobulin: synthesis of secretory component as mul- ilar human replacement in a rabbit receptor (aBc-r) sufficed to tiple and larger transmembrane forms. Proc. Natl. Acad. Sci. USA 77:7257. 7. Musil, L. S., and J. U. Baenziger. 1987. Cleavage of membrane secretory com- confer substantial pentameric IgM-binding capacity. Combining ponent to soluble secretory component occurs on the cell surface of rat hepatocyte the two human D1 regions that contained the CDR1- and CDR2- monolayers. J. Cell Biol. 104:1725. 8. Eskeland, T., and P. Brandtzaeg. 1974. Does J chain mediate the combination of like loops on the backbone from either species (ABc-h or ABc-r) 19S IgM and dimeric IgA with the secretory component rather than being nec- reconstituted pentameric IgM binding to a level approaching that essary for their polymerization? Immunochemistry 11:161. of the human wild-type receptor. The third region, including the 9. Brandtzaeg, P. 1975. Immunochemical studies on free and bound J chain of human IgA and IgM. Scand. J. Immunol. 4:439. CDR3-like loop, did not appear to be necessary for maximal pen- 10. Brandtzaeg, P. 1975. Blocking effect of J chain and J-chain antibody on the tameric IgM interaction, because the chimeras with this region of binding of secretory component to human IgA and IgM. Scand. J. Immunol. rabbit origin (ABc-h or ABc-r) showed binding levels similar to 4:837. 11. Brandtzaeg, P., and H. Prydz. 1984. Direct evidence for an integrated function of that of the human wild-type pIgR. However, a positive effect of the J chain and secretory component in epithelial transport of immunoglobulins. Na- CDR3-containing region was demonstrated by the chimeras, in ture 311:71. 12. Ma, J. K., A. Hiatt, M. Hein, N. D. Vine, F. Wang, P. Stabila, C. van Dolleweerd, which it was combined with one of the two first human regions K. Mostov, and T. Lehner. 1995. Generation and assembly of secretory antibod- (AbC-r, AbC-h, or aBC-h); this exchange increased the pentameric ies in plants. Science 268:716. 6052 INTERACTIONS BETWEEN pIgs AND pIgR

13. Hendrickson, B. A., L. Rindisbacher, B. Corthesy, D. Kendall, D. A. Waltz, 35. Mach, J. P. 1970. In vitro combination of human and bovine free secretory com- M. R. Neutra, and J. G. Seidman. 1996. Lack of association of secretory component with IgA of various species. Nature 228:1278. ponent with IgA in J chain-deficient mice. J. Immunol. 157:750. 36. Thompson, R. A. 1970. Secretory piece linked to IgM in individuals deficient in 14. Vaerman, J. P., A. E. Langendries, D. A. Giffory, C. S. Kaetzel, C. M. Fiani, IgA. Nature 226:946. I. Moro, P. Brandtzaeg, and K. Kobayashi. 1998. Antibodies against the human 37. Radl, J., F. Klein, P. van den Berg, A. M. de Bruyn, and W. Hijmans. 1971. J chain inhibit polymeric Ig receptor-mediated biliary and epithelial transport of Binding of secretory piece to polymeric IgA and IgM paraproteins in vitro. Im- human polymeric IgA. Eur. J. Immunol. 28:171. munology 20:843. 15. Vaerman, J. P., A. Langendries, D. Giffroy, P. Brandtzaeg, and K. Kobayashi. 38. Delacroix, D. L., and J. P. Vaerman. 1982. Secretory component (SC): prefer- 1998. Lack of pIgR/SC mediated epithelial transport of a polymeric IgA devoid ential binding to heavy (greater than 11S) IgA polymers and IgM in serum, in of J chain: in vitro and in vivo studies. Immunology 95:90. contrast to predominance of 11S and free SC forms in secretions. Clin. Exp. 16. Brandtzaeg, P. 1977. Human secretory component. VI. Immunoglobulin-binding Immunol. 49:717. properties. Immunochemistry 14:179. 39. Vaerman, J. P. 1994. Phylogenic aspects of mucosal immunoglobulins. In Hand- 17. Socken, D. J., and B. J. Underdown. 1978. Comparison of human, bovine and book of Mucosal Immunology. P. Ogra, M. Lamm, J. McGhee, J. Mestecky, rabbit secretory component-immunoglobulin interactions. Immunochemistry 15: W. Strober, and J. Bienenstock, eds. Academic Press, San Diego, p. 99. 499. 40. Amman, A. J. 1991. Selective IgA deficiency. In Basic and Clinical Immunology, 18. Brandtzaeg, P., F.-E. Johansen, P. Krajci, and I. B. Natvig. 1998. Secretory com- 7th Ed. D. P. Stites and A. L. Terr, eds. Lange, East Norwalk, p. 329. ponent (the polymeric Ig receptor). In Encyclopedia of Immunology, 2nd Ed. 41. De Laat, P. C., C. M. Weemaes, R. Gonera, P. J. Van Munster, J. A. Bakkeren, I. M. Ritt and P.-J. Delves, eds. Acadamic Press, London, p. 2152. and G. B. Stoelinga. 1991. Clinical manifestations in selective IgA deficiency in 19. Bakos, M. A., A. Kurosky, and R. M. Goldblum. 1991. Characterization of a childhood: a follow-up report. Acta Paediatr. Scand. 80:798. critical binding site for human polymeric Ig on secretory component. J. Immunol. 147:3419. 42. Nilssen, D. E., P. Brandtzaeg, S. S. Frøland, and O. Fausa. 1991. Subclass com- 20. Mostov, K., and C. S. Kaetzel. 1999. Immunoglobulin Transport and the Poly- position and J-chain expression of the “compensatory” gastrointestinal IgG cell meric Immunoglobulin Receptor. J. Mestecky, J. Bienenstock, J. McGhee, population in IgA deficiency. Clin. Exp. Immunol. 81:237. M. Lamm, W. Strober, and P. Ogra, eds. Academic Press, San Diego, p. 181. 43. Savilahti, E. 1973. IgA deficiency in children: immunoglobulin-containing cells 21. Bakos, M. A., A. Kurosky, C. S. Woodard, R. M. Denney, and R. M. Goldblum. in the intestinal mucosa, immunoglobulins in secretions and serum IgA levels. 1991. Probing the topography of free and polymeric Ig-bound human secretory Clin. Exp. Immunol. 13:395. Downloaded from component with monoclonal antibodies. J. Immunol. 146:162. 44. Brandtzaeg, P., I. Fjellanger, and S. T. Gjeruldsen. 1968. Immunoglobulin M: 22. Frutiger, S., G. J. Hughes, W. C. Hanly, M. Kingzette, and J. C. Jaton. 1986. The local synthesis and selective secretion in patients with immunoglobulin A defi- amino-terminal domain of rabbit secretory component is responsible for nonco- ciency. Science 160:789. valent binding to immunoglobulin A dimers. J. Biol. Chem. 261:16673. 45. Haneberg, B., and D. Aarskog. 1975. Human fecal immunoglobulins in healthy 23. Bakos, M. A., S. G. Widen, and R. M. Goldblum. 1994. Expression and purifi- infants and children, and in some with diseases affecting the intestinal tract or the cation of biologically active domain I of the human polymeric immunoglobulin immune system. Clin. Exp. Immunol. 22:210. receptor. Mol. Immunol. 31:165. 46. Brandtzaeg, P., G. Karlsson, G. Hansson, B. Petruson, J. Bjorkander, and

24. Bakos, M. A., A. Kurosky, E. W. Czerwinski, and R. M. Goldblum. 1993. A L. A. Hanson. 1987. The clinical condition of IgA-deficient patients is related to http://www.jimmunol.org/ conserved binding site on the receptor for polymeric Ig is homologous to CDR1 the proportion of IgD- and IgM-producing cells in their nasal mucosa. Clin. Exp. of Ig V␬ domains. J. Immunol. 151:1346. Immunol. 67:626. 25. Coyne, R. S., M. Siebrecht, M. C. Peitsch, and J. E. Casanova. 1994. Mutational 47. Brandtzaeg, P. 1982. Review and discussion of IgA transport across mucosal analysis of polymeric immunoglobulin receptor/ligand interactions: evidence for membranes. Recent Adv. Mucosal Immun. 267:285. the involvement of multiple complementarity determining region (CDR)-like 48. Tamer, C. M., M. E. Lamm, J. K. Robinson, J. F. Piskurich, and C. S. Kaetzel. loops in receptor domain I. J. Biol. Chem. 269:31620. 1995. Comparative studies of transcytosis and assembly of secretory IgA in Ma- 26. Weicker, J., and B. J. Underdown. 1975. A study of the association of human din-Darby canine kidney cells expressing human polymeric Ig receptor. J. Im- secretory component with IgA and IgM proteins. J. Immunol. 114:1337. munol. 155:707. 27. Underdown, B. J., I. Switzer, and G. D. Jackson. 1992. Rat secretory component 49. Lindh, E. 1975. Increased resistance of immunoglobulin A dimers to proteolytic binds poorly to rodent IgM. J. Immunol. 149:487. degradation after binding of secretory component. J. Immunol. 114:284. 28. Brandtzaeg, P., and K. Baklien. 1977. Characterization of the IgA immunocyte 50. Chintalacharuvu, K. R., A. S. Tavill, L. N. Louis, J. P. Vaerman, M. E. Lamm, population and its product in a patient with excessive intestinal formation of IgA. and C. S. Kaetzel. 1994. Disulfide bond formation between dimeric immuno- by guest on September 24, 2021 Clin. Exp. Immunol. 30:77. globulin A and the polymeric immunoglobulin receptor during hepatic transcy- 29. Natvig, I. B., F. E. Johansen, T. W. Nordeng, G. Haraldsen, and P. Brandtzaeg. tosis. Hepatology 19:162. 1997. Mechanism for enhanced external transfer of dimeric IgA over pentameric 51. Brown, W. R., R. W. Newcomb, and K. Ishizaka. 1970. Proteolytic degradation IgM: studies of diffusion, binding to the human polymeric Ig receptor, and epi- of exocrine and serum immunoglobulins. J. Clin. Invest. 49:1374. thelial transcytosis. J. Immunol. 159:4330. 52. Renegard, K. B., G. D. F. Jackson, and J. Mestecky. 1998. In vitro comparison 30. Krajci, P., K. H. Grzeschik, A. H. Geurts van Kessel, B. Olaisen, and of the biologic activities of monoclonal monomeric IgA, polymeric IgA and P. Brandtzaeg. 1991. The human transmembrane secretory component (poly-Ig secretory IgA. J. Immunol. 160:1219. receptor): molecular cloning, restriction fragment length polymorphism and chro- mosomal sublocalization. Hum. Genet. 87:642. 53. Crottet, P., and B. Corthesy. 1998. Secretory component delays the conversion of Ј 31. Piskurich, J. F., M. H. Blanchard, K. R. Youngman, J. A. France, and secretory IgA into antigen-binding component F(ab )2: a possible implication for C. S. Kaetzel. 1995. Molecular cloning of the mouse polymeric Ig receptor: mucosal defense. J. Immunol. 161:5445. functional regions of the molecule are conserved among five mammalian species. 54. Mostov, K. E. 1994. Transepithelial transport of immunoglobulins. Annu. Rev. J. Immunol. 154:1735. Immunol. 12:63. 32. Mostov, K. E., M. Friedlander, and G. Blobel. 1984. The receptor for transepi- 55. Hexham, J. M., K. D. White, L. N. Carayannopoulos, W. Mandecki, R. Brisette, thelial transport of IgA and IgM contains multiple immunoglobulin-like domains. Y.-S. Yang, and J. D. Capra. 1999. A human immunoglobulin (Ig)A C␣3 domain Nature 308:37. motif directs polymeric Ig receptor-mediated secretion. J. Exp. Med. 189:747. 33. Wigler, M., A. Pellicer, S. Silverstein, R. Axel, G. Urlaub, and L. Chasin. 1979. 56. Krugmann, S., R. J. Pleass, J. D. Atkin, and J. M. Woof. 1997. Mutagenesis of DNA-mediated transfer of the adenine phosphoribosyltransferase locus into J chain residues critical for IgA dimer assembly. Biochem. Soc. Trans. 25:323S. mammalian cells. Proc. Natl. Acad. Sci. USA 76:1373. 57. Randall, T. D., J. W. Brewer, and R. B. Corley. 1992. Direct evidence that J chain 34. Bouvet, J. P., J. Pillot, and S. Iscaki. 1990. Secretory component-binding prop- regulates the polymeric structure of IgM in antibody-secreting B cells. J. Biol. erties of normal serum IgM. Scand. J. Immunol. 31:437. Chem. 267:18002.