Effects of mutations in potential phosphorylation sites on transcytosis of FcRn

The neonatal Fc receptor (FcRn) was identified because of its role in the transmission of passive immunity from mother to young; it is expressed in the yolk sacs of fetal rats and mice (Roberts et al., 1990; Ahouse et al., 1993) and in the intestines of neonates (Rodewald and Kraehenbuhl, 1984; Simister and Rees, 1985; Simister and Mostov, 1989b), and is necessary for the transmission of maternal IgG to fetal and neonatal mice (Israel et al., 1995). FcRn is likely to have an analogous function in the human placenta (Story et al., 1994), where it is expressed in the syncytiotrophoblast (Kristoffersen and Matre, 1996; Leach et al., 1996; Simister et al., 1996).

In the yolk sac endoderm, intestinal epithelium and probably in the syncytiotrophoblast, FcRn transports maternal IgG in an apical to basolateral direction. It has been suggested that transcytosis of FcRn in the opposite direction may be used to recycle the receptor (Rodewald, 1980), or secrete IgG in the adult intestine (Israel et al., 1997). Transcytosis in both the apical to basolateral and basolateral to apical directions has been reported in recent studies of endogenous human FcRn in BeWo (Ellinger et al., 1999) and T84 cells (Dickinson et al., 1999), of a mouse FcγRII-B2/rat FcRn chimera (Stefaner et al., 1999) and human FcRn (Praetor et al., 1999) expressed in Madin-Darby canine kidney (MDCK) cells, and of rat FcRn in rat inner medullary collecting duct (IMCD) cells (McCarthy et al., 2000). These observations raised the question of whether the mechanisms of transcytosis in the two directions are the same. Transport of IgG in either direction is inhibited by compounds that prevent endocytic compartments acquiring the acidic pH needed for FcRn to bind its ligand (Dickinson et al., 1999; McCarthy et al., 2000). In contrast, some other compounds affect apical to basolateral and basolateral to apical transcytosis differently. The calmodulin antagonist W-7, for example, inhibits basolateral to apical but not apical to basolateral transcytosis of the FcγRII-B2/FcRn chimera (Stefaner et al., 1999). Similarly, phosphatidylinositol 3-kinase inhibitors block basolateral to apical transport by FcRn more completely than apical to basolateral transport (McCarthy et al., 2000). Nocodazole inhibits only basolateral to apical transcytosis of the FcγRII-B2/FcRn chimera in MDCK cells (Stefaner et al., 1999), but inhibits transcytosis of FcRn in both directions in IMCD cells (McCarthy et al., 2000). In this study we have investigated the effects on apical to basolateral and basolateral to apical transcytosis of mutations that affect the phosphorylation of FcRn.

The α chain of rat FcRn contains no tyrosine residues within its cytoplasmic domain, but has one threonine and five serine residues (Simister and Mostov, 1989b). Phosphorylation on serine or threonine regulates the intracellular sorting of several transmembrane proteins, including the polymeric immunoglobulin receptor (pIgR) (Casanova et al., 1990; Okamoto et al., 1994), the major histocompatibility complex class II-associated invariant chain (Anderson and Roche, 1998; Anderson et al., 1999), peptidylglycine α-amidating monooxygenase (PAM-1) (Yun et al., 1995; Steveson et al., 1999) and CD4 (Pitcher et al., 1999). To determine whether rat FcRn was phosphorylated, we expressed it in Rat1 fibroblasts, and metabolically labeled the α chain with ³²P-orthophosphate. We used similar labeling experiments on truncation and substitution mutants to identify two phosphorylation sites in the cytoplasmic domain of FcRn. We analyzed the effects of mutations on endocytosis, transcytosis and on distribution of FcRn between the apical and basolateral surfaces of IMCD cells. Elimination of the major phosphorylation site inhibited apical to basolateral transcytosis, whereas basolateral to apical transport was unaffected by mutations in the cytoplasmic domain. These observations suggest that FcRn is sorted by different mechanisms during apical to basolateral and basolateral to apical transport.

Rat inner medullary collecting duct (IMCD) cells, fibroblast-like Rat1 cells, and their transfected derivatives were cultured as described previously (Simister and Mostov, 1989a; McCarthy et al., 2000). Except where noted, IMCD-derived lines were plated on 24 mm Transwells (polycarbonate, 3 μm pores; Corning Costar, Acton, MA, USA) and grown for 2-3 days before experiments to allow a polarized monolayer with resistance greater than 300 Wcm² to form (McCarthy et al., 2000). Dulbecco’s Modified Eagle’s Medium (DMEM) buffered with 25 mM Hepes instead of bicarbonate was used for endocytosis and transcytosis assays so that the pH could be maintained outside a CO₂ incubator.

Stop codons were substituted for those encoding amino acids 305, 315, 325 and 335 of the rat FcRn α chain (Simister and Mostov, 1989b) by PCR with Pfu DNA polymerase (Stratagene, La Jolla, CA, USA), to make DNA encoding the truncation mutants 304t, 314t, 324t and 334t (Fig. 1). Alanine codons were substituted for those encoding Ser-313, Ser-315 and Ser-319, individually or in combination, using overlap exchange PCR (Higuchi et al., 1988; Horton et al., 1989), to make DNA encoding FcRn-S313A, -S315A, -S319A and -S313A/ S319A. Likewise, codons for aspartic acid were substituted for serine codons to give DNA encoding FcRn-S313D and -S319D. All sequences were verified.

DNAs that coded for mutant and wild-type rat FcRn α chains were subcloned into pRc/RSV (Invitrogen, San Diego, CA, USA) or SRα-MSVtkneo(ΔHind) (a kind gift of Dr Owen Witte). Constructs in pRc/RSV were transfected into IMCD or Rat1 cells using a calcium phosphate method (Gorman et al., 1990). Constructs in SRα-MSVtkneo(ΔHind) were transfected into 293T retroviral producer cells, and the supernatants were used to transduce Rat1 cells (Yang et al., 1999). Cells resistant to G418 were selected and individual colonies were expanded to establish cell lines.

Rat1-derived cells were grown to confluence in 6-well plates. For ³²P-orthophosphate labeling, cells were starved for 1 hour in phosphate-free DMEM with 10% dialyzed fetal bovine serum (FBS, Hyclone Laboratories, Logan, UT, USA), then labeled with 0.5 mCi/ml of [³²P]orthophosphate (ICN, Costa Mesa, CA, USA) for 4 hours. Alternatively, cells were starved for 1 hour in Met⁻, Cys⁻DMEM, 10% dialyzed FBS, and labeled for 4 hours with 0.1 mCi/ml of ³⁵S-Met+Cys (NEN-Dupont, Boston, MA, USA). The cells were washed with phosphate-buffered saline (PBS), pH 7.4, then lysed with SDS and immunoprecipitated with rabbit anti-rat FcRn α-chain (Simister and Mostov, 1989b) essentially as described (Simister and Mostov, 1989a), but with the phosphatase inhibitors 1 mM Na₃VO₄, 50 mM NaF, 5 mM EDTA and 5 mM EGTA (Anderson and Roche, 1998) in all solutions. The precipitated proteins were analyzed on 8% polyacrylamide denaturing, reducing gels. The gels were soaked in 1 M sodium salicylate, dried and exposed at −70°C to Kodak XAR 5 film (Rochester, NY, USA), with intensifying screens. The 32P- and 35S-labeled FcRn α chain bands were quantified using a Gel Doc 1000 workstation and Molecular Analyst 2.1.1 software (Bio-Rad).

Rat1 cells expressing wild-type FcRn were labeled with [³²P]orthophosphate or ³⁵S-Met+Cys, then lysed and immunoprecipitated with rabbit anti-rat FcRn α-chain, as above. The immune complexes were precipitated with protein A-trisacryl (Pierce), and digested overnight at 37°C on the beads with 3 mU of endoglycosidase H or 600 mU of peptide N-glycosidase F (Boehringer Mannheim). After digestion, proteins eluted in sample buffer (Laemmli, 1970) were analyzed by SDS-PAGE as above, on 8% polyacrylamide gels.

The apical or basolateral surfaces of IMCD-derived cells were labeled twice for 30 minutes with sulfosuccinimidyl 6-(biotinamido) hexanoate (sulfo NHS-LC-biotin; Pierce, Rockford, IL, USA), on ice, essentially as described (Okamoto et al., 1992). The biotinylation was quenched with medium, and the filters were cut from the Transwells (three for each surface). The cells were lysed in SDS. Lysates from each set of three Transwells were pooled, normalized for total protein and precipitated with rabbit anti-rat FcRn α-chain and protein A-trisacryl, then with streptavidin-agarose as described (McCarthy et al., 2000). Proteins were eluted from streptavidin-agarose into reducing sample buffer and resolved on 8% polyacrylamide denaturing gels. Gels were electroblotted onto polyvinylidene difluoride membranes (Novex, San Diego, CA, USA). Blots were probed with horseradish peroxidase-conjugated streptavidin (2 μg/ml in phosphate-buffered saline, 10% nonfat powdered milk, 0.5% Tween 20, pH 7.4; Pierce) and Renaissance Chemiluminescence Reagent (NEN-Dupont, Boston, MA, USA). The FcRn α chain bands (upper and lower bands) were quantified as above.

The Fc fragment of human IgG (Jackson Immunoresearch, West Grove, PA, USA) was labeled with Na125I (NEN-Dupont) using Iodogen (Pierce). IMCD-derived cells were grown on 12-well plates until nearly confluent but still fibroblast-like in appearance. The cells were cooled on ice for 1 hour, washed twice with ice-cold DMEM, pH 6.0, and allowed to bind ¹²⁵I-Fc (100 ng/ml) in DMEM, 1 mM KI, 1.5% fish gelatin (Sigma, St Louis, MO, USA), 25 mM Hepes (DMEM-KIGH), pH 6.0, for 6 hours (IgG binds FcRn at a mildly acidic pH, but not at pH 8.0; Jones and Waldman, 1972). After the cells had been washed four times with ice-cold DMEM, pH 6.0, they were incubated at 37°C with prewarmed DMEM, pH 6.0, for 0, 2, 5, 15 and 30 minutes. Then the cells were cooled on ice and the medium was removed and counted (cpmmed). The cells were washed with ice-cold DMEM, pH 6.0, then incubated on ice for 45 minutes with chymotrypsin and proteinase K, 50 μg/ml each in phosphate-buffered saline pH 8.0, to digest ¹²⁵I-Fc from the cell surface (cpm_surf). Finally, the cells were dissolved in 0.1 M NaOH, and counted (cpm_int). For each time point of each cell line, the percentage of ¹²⁵I-Fc internalized was calculated as cpm_int×100/(cpm_med+cpm_surf+cpm_int).

IMCD-derived lines were grown on Transwells. The loading surface (apical or basolateral) was washed twice with DMEM, pH 6.0, and the non-loading surface with DMEM, pH 8.0. The cells were cooled on ice at 4°C for 1 hour. The medium at the loading surface was then replaced with ¹²⁵I-Fc (100 ng/ml) in ice-cold DMEM-KIGH, pH 6.0. In some experiments unlabeled human γ-globulin (1.3 mg/ml) was included as a competitor. The plates were incubated at 14°C for 2.5-3 hours to allow Fc to be taken up while transcytosis was inhibited (Song et al., 1994; Luton and Mostov, 1999), then returned to ice. ¹²⁵I-Fc in the non-loading compartment was measured in a CliniGamma 1272 gamma counter (LKB Wallac, Piscataway, NJ, USA), and wells in which more than 1% of the applied Fc had crossed the monolayer were rejected. The loading surface was then washed four times with ice-cold DMEM, pH 6.0. The non-loading surface medium was replaced with DMEM, pH 8.0, prewarmed to 37°C. Cells were incubated at 37°C for 10, 20, 40 and 60 minutes, after which times the medium from the non-loading surface was removed and replaced with fresh DMEM, pH 8.0. At the last time point, the loading surface medium was removed. The Transwell membrane was cut out and cells were lysed with 0.1 M NaOH. All media and the cell lysate were precipitated in ice-cold 10% trichloroacetic acid (TCA) and the amount of TCA-insoluble ¹²⁵I-Fc was measured (cpm₁₀, cpm₂₀, cpm₄₀, cpm₆₀, cpm_loading, cpm_cell). For each well at each time point (t), the cumulative percentage of ¹²⁵I-Fc transported was calculated as (cpm₁₀+. +cpm_t)×100/(cpm₁₀+cpm₂₀+cpm₄₀+cpm₆₀+cpm_loading+

cpm_cell). FcRn-S313A clones from two independent transfections were assayed.

A polyclonal antibody against rat FcRn α chain (Simister and Mostov, 1989a) precipitated two ³⁵S-labeled proteins from Rat1 cells expressing FcRn, with apparent molecular masses of 51 and 58 kDa (Fig. 2). The lower band was shifted to an apparent molecular mass of 42 kDa by endoglycosidase H treatment, but the upper band was resistant. Both bands migrated with the same decreased apparent molecular masses after digestion with peptide N-glycosidase F (the multiple bands probably reflect incomplete deglycosylation of some molecules). The higher molecular mass form of FcRn was labeled with ³²P (Fig. 2).

Phosphate incorporation into wild-type FcRn and truncation and substitution mutants was compared, using ³⁵S-methionine and -cysteine labeling to normalize for differences in expression level between cell lines (none of the mutations removed methionine or cysteine residues). FcRn-304t, which included only the 4-amino-acid putative stop-transfer sequence of the cytoplasmic domain, and FcRn-314t, which included 10 more membrane-proximal amino acids, were not labeled with ³²P-orthophosphate (Fig. 3). FcRn-324t and FcRn-334t were both phosphorylated. We substituted alanine residues to disrupt the three potential phosphorylation sites near amino acid 315 or between positions 315 and 324. FcRn-S313A and -S319A incorporated, respectively, 27% (±17, n=4) and 70% (±30, n=4) of the ³²P incorporated into wild-type FcRn, whereas FcRn-S315A was phosphorylated to a similar extent to wild type (Fig. 3).

For the Fc-loading stage of the cohort assay we used low temperatures (Song et al., 1994; Luton and Mostov, 1999) to inhibit microtubule-dependent transcytosis of FcRn (McCarthy et al., 2000). In preliminary studies, apical to basolateral transport of Fc across IMCD cells expressing wild-type FcRn was greatly reduced at 14°C compared with 37°C (the amount of Fc transported in 3.5 hours at 14°C was approximately 25% of that transported at 37°C and no different from untransfected IMCD cells at 14°C; data not shown.) When cells loaded at 14°C were warmed to 37°C under conditions allowing release of Fc from either surface (pH 8.0 both sides), only approximately 25% of the Fc was released at the non-loading surface (data not shown.) To increase transcytosis of Fc in studies of FcRn mutants, conditions that reduced release at the loading surface were used (pH 6.0 in the loading compartment; pH 8.0 in the non-loading compartment).

Fig. 4A-D shows apical to basolateral transport measured under these conditions. Each panel is representative of 3-5 experiments. An average of 42% (±3, n=5 experiments) of the Fc loaded from the apical surface of cells expressing wild-type FcRn was released at the basolateral surface in 1 hour (Fig. 4A-D). FcRn-304t was impaired in transcytosis, transporting only 20% (±4, n=5) of loaded Fc (Fig. 4A,C,D). FcRn-S313D, FcRn-S315A, FcRn-S319A and FcRn-S319D were similar to wild-type FcRn in transcytosis: FcRn-S313D transported 54% (±10, n=4) of loaded Fc, FcRn-S315A 52% (±13, n=4), FcRn-S319A 36% (±6, n=4) and FcRn-S319D 48% (±11, n=4 Fig. 4A-D). In contrast, both FcRn-S313A and FcRn-S313A/S319A were impaired in transcytosis to approximately the same extent as the tailless FcRn-304t: cells expressing these mutants transported only 18% (±8, n=3) and 16% (±5, n=4) of loaded Fc, respectively (Fig. 4A-D). A second FcRn-S313A clone gave similar results (not shown). FcRn-304t and FcRn-S313A returned more Fc to the apical medium than wild-type FcRn, 62% (±5, n=6) and 59% (±11, n=5), respectively, compared with 39% (±9, n=6, not shown). About 10% of the loaded Fc remained associated with all the cells studied (data not shown). The amounts of TCA-soluble ¹²⁵I released at the apical and basolateral surfaces of cell expressing wild-type and mutant receptors were similar, indicating that the intracellular breakdown of ¹²⁵I-Fc was not altered by the mutations (not shown).

In the basolateral to apical direction wild-type FcRn transported 31% (±6, n=3) of the Fc loaded (Fig. 4E,F). FcRn-304t transported similar amounts to wild type, 28% (±4, n=4) (Fig. 4E). There was little or no difference between any of the substitution mutants and wild-type FcRn in basolateral to apical transcytosis (Fig. 4E,F). Basolateral to apical transport by both the wild-type receptor and FcRn-304t was inhibited by excess unlabeled human γ-globulin (Fig. 5A,B).

To determine whether the defect seen in transcytosis by FcRn-S313A was due to impaired endocytosis, we compared the abilities of IMCD cells expressing FcRn-S313A, FcRn-S313D, FcRn-304t and wild-type FcRn to internalize ¹²⁵I-Fc. FcRn-S313A and FcRn-S313D both took up approximately 30% more Fc than wild-type FcRn (Fig. 6), whereas FcRn-304t was impaired in endocytosis (60% less Fc was taken up in 30 minutes than by wild-type FcRn). The FcRn-S319A mutant showed wild-type endocytosis (data not shown).

We selectively biotinylated the apical and basolateral surfaces of IMCD cells to determine whether the S313A and S313D mutations affected the steady state distribution of FcRn. Most of the wild-type FcRn at the plasma membrane (77%±2, n=4) was on the basolateral surface, as previously noted (McCarthy et al., 2000), whereas only 29% (±2, n=4) of FcRn-304t was basolateral. As shown in Fig. 7, 58% (±13, n=4) of the surface FcRn-S313A, 62% (±7, n=3) of FcRn-S313D and 43% (±25, n=3) of FcRn-S319A were basolateral.

Rat FcRn α chain could be labeled with ³²P in Rat1 cells, but an essentially tailless mutant could not. This indicates that the receptor is phosphorylated in the cytoplasmic domain, as has been suggested (Stefaner et al., 1999). Further truncation mutants were analyzed to determine which of six potential sites (Ser-305, Ser-313, Ser-315, Ser-319, Thr-343 and Ser-344) was phosphorylated (Fig. 1). FcRn truncated after Leu-314 failed to incorporate ³²P, whereas FcRn truncated after Pro-334 was labeled to an extent similar to the wild-type receptor. The most straightforward inference was that the major phosphorylation site or sites lay between amino acids 315 and 334, a region that contains Ser-315 and Ser-319. Nonetheless, we continued to investigate Ser-313 also, because it might have been affected by the truncation after position 314. Therefore, we substituted alanine residues for Ser-313, Ser-315 and Ser-319 individually. Phosphorylation was reduced in FcRn-S313A and -S319A by approximately 70% and 30%, respectively; FcRn-S315A was phosphorylated normally. We concluded that FcRn was phosphorylated on both Ser-313 and Ser-319.

The molecular masses of the high, low and deglycosylated forms of FcRn were higher by approximately 4 kDa than those reported for mature, immature and unglycosylated FcRn expressed in Rat1 cells (Simister and Mostov, 1989a). This is probably because prestained molecular mass markers were used in the present study, which are less accurate than the ¹⁴C markers used previously. Peptide N-glycosidase F, which removes all N-linked glycans (Chu, 1986), shifted the high and low molecular mass forms to the same size, indicating that the two forms differ only in their glycan moieties. The sensitivity of the low molecular mass form of FcRn to endoglycosidase H suggested that it was the high mannose form usually found in the endoplasmic reticulum. We inferred that the endoglycosidase H-resistant upper band contained complex-type oligosaccharide chains modified in the Golgi. Although the high molecular mass form was greatly enriched at the cell surface compared with the low molecular mass form (Fig. 7), the low molecular mass form was clearly present. Only the high molecular mass form of FcRn was phosphorylated, perhaps reflecting different trafficking pathways for the high and low molecular mass forms.

Neither Ser-313 nor Ser-319 lies within the consensus recognition sequence for a known protein kinase (Kreegipuu et al., 1999). However, the variety of sequences that a kinase can phosphorylate may not be represented adequately by a consensus sequence (Kreegipuu et al., 1998). Because Ser-319 is next to a pair of aspartate residues in positions 317 and 318 it is probably phosphorylated by an acidophilic protein kinase; Ser-313 lies 4 amino acids to the other side of this diacidic patch. Ser-313 is conserved in all four species from which FcRn has so far been cloned (Simister and Mostov, 1989b; Ahouse et al., 1993; Story et al., 1994; Kacskovics et al., 2000). Ser-319 is not conserved in bovine FcRn (Kacskovics et al., 2000), but is present in mouse FcRn (Ahouse et al., 1993) and conservatively replaced by threonine in human FcRn (Story et al., 1994). These patterns of conservation are consistent with Ser-313 and Ser-319 having important functions. It should be noted, however, that Ser-305 and Thr-343 are also present in rat, mouse and human FcRn α-chains, although they do not appear to be phosphorylated.

In IMCD cells, we were unable to detect phosphorylation of FcRn (data not shown). It is possible that the receptor was not phosphorylated in these cells, but we doubt this because the effects of Ser-313 mutations on Fc transport (below) are most readily explained as effects on a phosphorylation site. Rather, we suspect that the fraction of FcRn phosphorylated was too low for us to detect.

Transcytosis of Fc by IMCD cells expressing mutant or wild-type FcRn was compared using a cohort assay to allow for different levels of receptor expression. Transcytosis by tailless FcRn was greatly impaired in the apical to basolateral direction, but apically directed transcytosis was similar to that mediated by the wild-type receptor. Thus, apical to basolateral transcytosis of FcRn depended upon cytoplasmic determinants, but basolateral to apical transport did not. Basolateral to apical transport measured by the cohort assay was specific because it was inhibited by competing unlabeled IgG, and because it did not occur when medium at pH 8.0 was used in the basal compartment to prevent FcRn binding ligand. It was surprising that FcRn-dependent transport occurred in the absence of the cytoplasmic domain, in part because the cytoplasmic domain is involved in endocytosis: a tailless FcγRII-B2/FcRn chimera is internalized at approximately 30% of the wild-type rate (Stefaner et al., 1999), and tailless FcRn underwent endocytosis at approximately 40% of the level of the wild-type receptor in IMCD cells. The different requirements of Fc uptake and transport could imply that basolateral to apical transcytosis did not require endocytosis of FcRn (perhaps because FcRn bound Fc in endosomes and got there without going to the cell surface), or at least that endocytosis was not the rate-limiting step. We prefer the latter explanation because the requirement for the basal medium to be at pH 6 is consistent with FcRn binding Fc at the cell surface. Whichever explanation is correct, transfer of FcRn from a basal compartment to the apical cell surface must occur either by a pathway selective for determinants in the transmembrane or extracellular region of the receptor, or by a default pathway requiring no sorting signals. In Caco-2 cells tailless aminopeptidase N also undergoes basolateral to apical transcytosis (Vogel et al., 1995).

Consistent with basolateral to apical transcytosis being independent of the cytoplasmic domain, substitution of alanine or aspartate for Ser-313, Ser-315 and Ser-319 had little effect on Fc transport. Only FcRn-S313D consistently transported less than wild-type FcRn, and the difference was generally slight.

Apical to basolateral Fc transport was not markedly or consistently affected by the replacement of Ser-319 with alanine or aspartate. Nor did these mutations significantly affect endocytosis or the distribution of FcRn between the apical and basolateral regions of the plasma membrane. If Ser-319 was indeed phosphorylated in IMCD cells, this did not appear to regulate transcytosis. In contrast, mutation of Ser-313 of rat FcRn to alanine dramatically reduced basolaterally directed transcytosis of Fc, to the level seen with the tailless mutant. Although we were unable to detect FcRn phosphorylation in IMCD cells, our observation of normal (or slightly enhanced) transport when we substituted aspartate for Ser-313 is consistent with a requirement for phosphorylation for transcytosis.

We investigated the cause of the defect in transcytosis further. First we measured endocytosis by FcRn-S313A, because phosphorylation enhances the internalization of some membrane proteins, including the pIgR (Okamoto et al., 1994) and CD4 (Pitcher et al., 1999). The S313A mutant was not impaired in endocytosis. The steady state distribution of FcRn-S313A between the apical and basolateral surfaces of IMCD cells was, like wild-type FcRn (McCarthy et al., 2000), predominantly basolateral. In cells expressing FcRn-S313A and wild-type FcRn the fractions of the Fc cohort that were degraded during the assay were also similar. One difference was in the percentage of Fc released back into the apical medium, which was higher for FcRn-S313A than wild-type FcRn. Because endocytosis of FcRn was not impaired by the mutation, increased release suggests increased recycling of FcRn to the apical plasma membrane. Together, these observations show that Ser-313 is not required for endocytosis by FcRn, its distribution between the apical and basolateral cell surfaces, or its ability to protect Fc from degradation. Rather, they suggest that Ser-313 is necessary to allow FcRn to be diverted from an apical recycling pathway into a basolaterally directed transcytotic pathway. The fact that FcRn-S313D did not transport significantly better than wild-type suggests that phosphorylation of Ser-313 would not force FcRn to undergo transcytosis.

Recent observations in mouse hepatocytes (Borvak et al., 1998) and mammary epithelial cells (Cianga et al., 1999) suggest that in these cells FcRn predominantly recycles IgG back to surface from which it enters. FcRn may also recycle IgG taken up from the blood by capillary endothelial cells (Ghetie and Ward, 1997). Our observation that Fc loaded into IMCD cells without an external pH gradient was mostly returned to the input compartment suggests that in IMCD cells apical recycling predominates over transcytosis. We speculate that in cells that exclusively recycle FcRn, Ser-313 is not phosphorylated.

We thank Owen Witte for the vector SRα-MSVtkneo(ΔHind). This work was supported by NIH grants HD27691 and HD01146.